Beruflich Dokumente
Kultur Dokumente
WATSON
G.N.WATSON
PRESS
A TREATISE ON
The completeness of the theoretical account, combined with the wide scope of the
of practical examples and the extensive numerical tables, have resulted in a book which is indispensable to pure mathematicians, to applied mathematicians, and to physicists alike.
collection
'In Professor Watson's treatise, which is a of erudition and its often too rare accompaniment, clear exposition, we have a rigorous mathematical treatment of all types of Bessel functions, their properties,
monument
representations, asymptotic expansions, integrals containing them, allied functions, series, zeros, tabulation, together
integral
.indis-
who have
occasion to
THEORY OF
BESSEL FUNCTIONS
W.
B. F.
A TREATISE ON THE
THEORY OF
BESSEL FUNCTIONS
BY
G. N.
WATSON
SECOND EDITION
CAMBRIDGE
AT THE UNIVERSITY PRESS
1966
First paperback
edition
1966
~--fcU^6\\*s
Hazell Watson
PREFACE
THIS
book has been designed with two objects in view. The first is the development of applications of the fundamental processes of the theory of
functions of complex variables.
For this purpose Bessel functions are admirably adapted; while they offer at the same time a rather wider scope for the application of parts of the theory of functions of a real variable than
is
provided by
The second object is the compilation of a collection of results which would be of value to the increasing number of Mathematicians and Physicists who
encounter Bessel functions in the course of their researches. The existence of such a collection seems to be demanded by the greater abstruseness of properties
of Bessel functions (especially of functions of large order) which have been
While my endeavour has been to give an account of the theory of Bessel functions which a Pure Mathematician would regard as fairly complete, I have
consequently also endeavoured to include
special, which,
all
although without theoretical interest, are likely to be required in practical applications; and such results are given, so. far as possible, in a form appropriate for these purposes. The breadth of these aims, combined
with the necessity for keeping the size of the book within bounds, has made it necessary to be as concise as is compatible with intelligibility.
is, for the most part, a development of the theory of funcexpounded in the Course of Modern Analysis by Professor Whittaker and myself, it has been convenient to regard that treatise as a standard work of reference for general theorems, rather than to refer the reader to original
tions as
sources.
draw attention here to the function which I have regarded namelt the function which was defined by Weber and used subsequently by Schlafli, by Graf and Gubler and by Nielsen. For historical and sentimental reasons it would have been pleasing to have felt justified in using Hankel's function of the second kind; but three
It is desirable to
The
first is
my
of mathematicians who use Weber's function has greater weight than the authority of the mathematicians who use any other one function of the second kind. The second is the parallelism which the use of Weber's function exhibits
between the two kinds of Bessel functions and the two kinds (cosine and
sine)
VI
of trigonometrical functions.
interpolation
PREFACE
The third is the existence of the device by which made is possible in Tables I and III at the end of Chapter XX, which seems to make the use of Weber's function inevitable in numerical work.
It has been my policy to give, in connexion with each section, references any memoirs or treatises in which the results of the section have been previously enunciated; but it is not to be inferred that proofs given in this book are necessarily those given in any of the sources cited. The bibliography at the end of the book has been made as complete as possible, though doubtless omissions will be found in it. While I do not profess to have inserted every memoir in which Bessel functions are mentioned, I have not consciously omitted any memoir containing an original contribution, however slight to the theory to
of the functions; with regard to the related topic of Riccati's equation, I have
been eclectic to the extent of inserting only those memoirs which seemed to be relevant to the general scheme.
In the case of an analytical treatise such as this, it is probably useless to hope that no mistakes, clerical or other, have remained undetected; but the
of such mistakes has been considerably diminished by the criticisms and the vigilance of my colleagues Mr C. T. Preece and Mr T. A. Lumsden, whose labours to remove errors and obscurities have been of the greatest value. To these gentlemen and to the staff of the University Press, who have given every assistance, with unfailing patience, in a work of great typographical
number
complexity, I offer
my
grateful thanks.
G. N.
August
21, 1922.
W.
XII
XIX;
waned
since 1922,
and I
prepared to undertake such a task to the detriment In the preparation of this new edition I have therefore limited myself to the correction of minor errors and misprints and to the emendation of a few assertions (such as those about the unproven character of Bourget's hypothesis) which, though they may have been true in 1922, would have been
definitely false
had they been made in 1941. thanks are due to many friends for their kindness in informing me of errors which they had noticed; in particular, I cannot miss this opportunity of expressing my gratitude to Professor J. R. Wilton for the vigilance which
My
list
of corrigenda. G. N.
W.
March
31, 1941.
CONTENTS
CHAP.
I.
PAGE
1826
II.
H
38
III.
IV.
DIFFERENTIAL EQUATIONS
85
132
V.
VI. VII.
160 194
VIII.
IX.
....
.
225
271
X.
XI.
XII.
308
358 373
XIII.
383 450
.
XIV.
MULTIPLE INTEGRALS
XV.
XVI.
477
NEUMANN
VARIABLES
SERIES
XVII.
KAPTEYN SERIES
OF FOURIER-BESSEL AND DINI
XVIII. SERIES
576 618
.
XIX.
SCHLOMILCH SERIES
XX.
654
665 753
INDEX OF SYMBOLS
LIST OF
AUTHORS QUOTED
....
789
791
GENERAL INDEX
796
point,
large,
:
and to be curious
in
is
to be granted to
him that
ii.
will
make an abridgement.
2
Maccabees
30, 31.
CHAPTER
I
1826
The theory of Bessel functions is intimately connected with the theory of a certain type of differential equation of the first order, known as Riccati's equation. In fact a Bessel function is usually defined as a particular solution
of a linear differential equation of the second order (known as Bessel's equation)
which
is
The
this
by John Bernoulli in 1694. In paper Bernoulli gives, as an example, an equation of this type and states that he has not solved it"f\
In various letters! to Leibniz, written between 1697 and 1704, James Bernoulli refers to the equation, which he gives in the form
dy = yydx + xxdx,
and
states,
more than
1697): "
Vellem porro ex Te
Ego
in mille
lusit."
Thus he writes (Jan. 27, dy yydx + xxdx. meam operam improbum sed Problema performas transmutavi,
once, his inability to solve
scire
it.
num
et hanc tentaveris
Five years later he succeeded in reducing the equation to a linear equation of the second order and wrote to Leibniz (Nov. 15, 1702) " Qua
petuo
quam
simpliciter differentialis
sed separavi
illas
dy yydx + a?dx
differentio-differentialem||
had been made, it was a simple step to solve the last and so to obtain the solution of the equation of the first order as the quotient of two power-series.
this discovery
series,
When
equation in
* Acta Eruditorum publicata Lipsiae, 1694, pp. 435 437. i f "E8to proposita aequatio differentialis haec x dx + y*dx = a?dy quae an per separationem indeterminatarum construi possit nondum tentavi " (p. 436). t See Leibnizens gesamellte Werki, Dritte Folge (Mathematik), in. (Halle, 1855), pp. 5087. Ibid. p. 65. Bernoulli's procedure was, effectively, to take a new variable u defined by the
formula
1 du u dx
_ ~*
by
y.
in the equation
il
to replace
this equation
and a
in 4-3.
[CHAP.
form of the solution was communicated to Leibniz by (Oct. 3, 1703) in the following terms*:
X
V
3
,
X 3.4.7
7
X 3.4.7.8.11
11
XVa
3.
X19
.
4.7.8.
11
12.
15^3.
X*
3.4
X*
3.4.7.8
X 12 3.4.7.8.11.12
x 3.4.7.8.11.12.15.16
lti
quae
series
ratio progressionis
qua
_a?
y
x1
3.3. 7
2x11
373. 3. 7. 11
13-r15
3.3.3.3.5. 7777II
+
far
Of
and consequently James Bernoulli seems to have received hardly the full credit which his discovery entitled him. Thus, twenty-two years later, the paperf in which Count Riccati first referred to an equation of the type which now bears his name, was followed by a note:,: by Daniel Bernoulli in which it was
to
ax dx + uudx = bdu
11
was a hitherto unsolved problem. The note ended with an announcement in an anagram of the solution " Solutio problematis ab 111. Riccato proposito characteribus occultis involuta 24a, 6b, 6c, Sd, 33e, hf, 2g, 4>h, 33i, 61, 21m,
:
26n, 16o, Sp, oq, I7r, 16s, 25t, 32m, 5#, By, +, -,
, =,
4, 2, 1."
to
The
4m/(27/i 1),
where
is
any
integer, for
namely any one of which the equation is solution will be given in 4*1, 4*11.
to the
work of Riccati by Daniel Bernoulli, combined equation was of a slightly more general type than
m.
(Halle, 1855), p. 75.
in
vm.
(1724), pp.
6673.
The form
xmdq = du + uu dx:q,
where
other
= xn
% Ibid. pp. 73
75. Daniel Bernoulli mentioned that solutions had been obtained by three members of his family John, Nicholas and the younger Nicholas. The reader should observe that the substitution
u=
gives rise to
|j
bdz
z
dx
an equation which is easily soluble in series. Exercitationes quaedam matheviaticae (Venice, 1724), pp. 77 pp. 465473.
80;
1*2]
John Bernoulli's equation* has resulted in the name of Riccati being associated
not only with the equation which he discussed without solving, but also with
still
where P, Q,
It is
if
R are given
functions of x.
P=0,
supposed that neither P nor R is identically zero. If U=0, the equation is linear; the equation is reducible to the linear form by taking 1/y as a new variable.
last
The
it is
and
this equation is
sometimes reducible
by an elementary transformation
Mention should be made here of two memoirs by Euler. In the first it is proved that, when a particular integral y x of Riccati's generalised equation is known, the equation is reducible to a linear equation of the first order by replacing y by y x + 1/u, and so the general solution can be effected by two quadratures. It is also shewn (ibid. p. 59) that, if two particular solutions are known, the equation can be integrated completely by a single quadrature; and this result is also to be found in the second of the two papers. A brief discussion of these theorems will be given in ChaDter iv.
||
1*2.
In 1738 Daniel Bernoulli published a memoir! containing enunciations of a number of theorems on the oscillations of heavy chains. The eighth ** of
these
is
as follows
"
Defigura catenae uniformiter oscillantis. Sit catena AG flexilis suspensa de puncto A, eaque oscillationes
si turn
AMF;
fueritque
ejus
longitudo catenae
valorisff ut
fit
/
= 1:
FM x, sumatur n
'
n
*
Jl + 4ww
P_
4 9n
.
l^_
4 9 16n4
.
4.9.16. 25ns
See James Bernoulli, Opera Omnia, n. (Geneva, 1744), pp. 1054 1057 it is stated that the is the determination of a solution in finite terms, and a solution which resembles the solution by Daniel Bernoulli is given. t The term Hiccati's equation was used by D'Alembert, Hist, de I' Acad. R. de Set. de Berlin,
;
'
X Iiutitutiones Calculi Integralis, n. (Petersburg, 1769), 831, pp. 8889. the reduction, see James Bernoulli's letter to Leibniz already quoted.
||
In connexion with
Novi Coram. Acad. Petrop. vni. (17601761), [published 1763], p. 32. 163164. If " Theoremata de osoillationibus corporum filo flexili connexorum et catenae verticaliter suspensae," Comm. Acad. Set. Imp. Petrop. vi. (1732 3), [published 1738], pp. 108 122.
Ibid. ix. (17621763), [published 1764], pp.
** Loc.
cit: p.
116.
is n.
[CHAP.
Ponatur porro distantia extremi puncti F ah linea'verticali = 1, dico fore ab eadem linea verticali aequalem distantiam puncti ubicunque assumpti
x n
xx
i
<*
**
**
4>nn
4.9w3
4.9.16n 4
4.9.16.25w8
f ctc
He
= proxime 0691
I....
Habet autem
The last series is now described as a Bessel function * of order zero and argument 2 V(#/n); and the last quotation states that this function has an infinite number of zeros.
viii,
Bernoulli published f proofs of his theorems soon afterwards; in theorem he obtained the equation of motion by considering the forces acting on
the portion
FM
of length
later
x.
The equation
by
Euler \
many years
following
is
of the chain.
The
Let p be the line density of the chain (supposed uniform) and let T be the tension at height x above the lowest point of the chain in its undisturbed position. The motion being transversal, we obtain the equation bT=gpbx by resolving vertically for an element of chain of length bx. The integral of the equation is Tgpx.
The
is
the (hori-
'*-( r 2)If
we
substitute for
T and
dt*
If
y dx\ dxj
'
is
the length of the simple equivalent pendulum for any one normal vibration,
we
write
where
A and
f are constants
and then
n (x/f)
d
is
A
If
-\ ( dx\ dx)
x/f= ii, we obtain the solution
1
+- =
f
series,
in the
form of Bernoulli's
v?
namely
u ul v= l H
*
1.4
+ u* 1.4.9 1.4.9.16
On
the Continent, the functions are usually called cylinder functions, or, occasionally, func-
Journal fur Math. lxix. (1868), p. 128; see also Math. Ann.
m.
(1871), pp.
609610.
t Comm. Acad. Petrop. vn. (17345), [published 1740], pp. 162179. X Acta Acad. Petrop. v. pars 1 (Mathematics), (1781), [published 1784], pp. 157177. Euler took the weight of length e of the chain to be E, and he denned g to be the measure of the distance (not twice the distance) fallen by a particle from rest under gravity in a second. Euler's
notation has been followed in the text apart from the significance of g and the introduction of
p
and
5 (for d).
1*3]
1826
D are constants.
If a is the
where
C and
Since
is finite
zero.
"
j. +
1
./
TT4/2 ~
4. 9/3
""'
By an extremely ingenious analysis, which will be given fully in Chapter xv, Euler proceeded to shew that the three smallest roots of the equation in a/f are 1-445795, 7*6658 and 18-63. [More accurate values are 1*4457965, 7*6178156 and 18*7217517.]
In
'the
(u-r-\+v0, but
law of formation of successive coefficients is rather incomplete. The law of formation had, however, been stated in his Institutiones Calculi Integrality, n. (Petersburg, 1769), 977,
pp. 233-235.
1*3.
The
1764.
membrane were
Idz
1
investigated by Euler}: in
He
ld?z_d?z
e*
dt2
d*z
3
'
where z
is
coordinates are
e is
To obtain a normal
where
a,
solution he wrote
z = u sin (at
A,
fi,
are constants
is
and u
*w,l^w,
The
of Euler's memoir;
p\
at the origin
is
is finite
given on
p.
256
?-P
\l
{
2(w+l)e
is
||
+ 2.4(w+l)(n
2/3
+ 3)e4 ~)
+ 1.
as Bessel's equation for functions
0, 1, 2,
is
. .
now known
and
may
have
..
Save
for
coefficient of order /S
now
called a Bessel
The
* Acta Acad. Petrqp. v. pars 1 (Mathematics), (1781), [published 1784], pp. 178190. t See also 935, 936 (p. 187 et seq.) for the solution of an associated equation which will be
discussed in 3*52.
X Novi Comm. Acad. Petrop. x. (1764), [published 1766], pp. 243260. The reason why Euler made this change of notation is not obvious.
||
If
/3
were not an integer, the displacement would not be a one-valued function of position,
{p</>
+ B).
6
circular
[CHAP.
membrane of radius a with a fixed boundary* are to be determined from the consideration that u vanishes when r = a.
This investigation by Euler contains the earliest appearance in Analysis of
Only a few year* after Euler had arrived at the general Bessel coefficient on vibrating membranes, the functions reappeared, in an astronomical problem. It was shewn by Lagrangef in 1770 that, in the elliptic motion of a planet about the sun at the focus attracting according to the law of the inverse square, the relations between the radius vector r, the mean anomaly and the eccentric anomaly E, which assume the forms
in his researches
e sin
E,
Z A E=M+ n=l
in
j,
sinnM,
n=\
which a and
.
and
_9 ~ "m=o
2+"*"ro (n
!
w)'!
n
'
^ ~~
n
(_)m H + ( ^ l+2 t
n-t-m-i ro
n+2m e
mt
2'
w!(w + m)!
in
for
1, 2, 3.
The
Terns,
f
p.
6 that
dA n
n at
a
memoir by
is
Poisson
A remarkable investigation of the approximate value of A n when n is large and < e < 1 is due to Carlini J: though the analysis is nob rigorous (and it would be difficult to make it rigorous) it is of sufficient interest for a brief
account of
it
to
be given here.
in. (1866), pp.
* Cf. Bourget,
113138.]
(Milan, 1817).
col.
X Ricerche sulla convergenza delta serie che serva alia soluzione del problema di Keplero This work was translated into German by Jacobi, Astr. Nach. xxx. (1850),
197254 [Werke, yii. (1891), pp. 189245]. See also two papers by Scheibner dated 1856, reprinted in Math. Ann. xvn. (1880), pp. 531544, 545560.
14]
It is easy to
1826
equation
An
c
2
is
a solution of the
differential
A n =2nn 2
/udt e /n
and then
2
(^+2 ) + tt-
u-
(l-2 ) = 0.
large.
;
Hence when n
If u
(n a
)
is
large either
u or u2 or du/dt must be
(n a ) respectively and and du/dt to l>e (n 2a ) and on considering the highest powers of n in the various terms of the last differential equation, we find that a= 1. It is consequently assumed that u admits of an expansion in descending powers of n in the form
we should expect
where
?<
u lt u 2
...
are independent of n.
first
On
/-ero
we
find that
...
= (l- f 2 )/*2
fl
+ 2oi)+Mo=0,
,
where v n'=du
/dt
so that
= +-
..
2,
and therefore
judt-n^ogj^f2)vA(1-e
and, since the value of
2
)
+ l}-ilog(l- f 2 ) + ...,
is
A n shews that Jude ~ n log ^e when be taken and no constant of integration is to be added.
From
Stirling's
must
formula
it
now
n exp
B
v/(^
'
and
by
Carlini.
much
which
further
by Meissel
This method of approximation has been carried Cauchy* has also discussed approximate
8"42), for
formulae for
An
in the case of
Carlini's
approximation
is
The
much more
corre-
sponding approximation
Bn
The
he uses
taking
investigation given
is
by Laplace
is
of considerable
importance when
the value of
all
Bn
or,
e is a pure imaginary. But Laplace goes on to argue that an approximation established in the case of purely imaginary variables may be used sans crainte in the case of real variables. To anyone who is acquainted
supposing that
'
'
with the
modem
xxxvm.
(1854), pp.
t.
990993.
v.
[first
published 1827].
[CHAP.
on the principle
crease
up
to a certain point
of the
sum
of the series
may
For other and more recent applications of this principle, see Stokes, Proc. Camb. Phil. 362366 [Math, and Phys. Papers, v. (1905), pp. 221225], and Hardy, Proc. London Math. Soc. (2) n. (1905), pp. 332339 Messenger, xxxiv. (1905)., pp. 97101. A statement of the principle was given by Borel, Acta Mathematica, xx. (1897), pp. 393
Soc. vi. (1889), pp.
;
394.
The
The
iL<" = 2
in
2
,=o
(n
'
which n
is
large
/u
and
t.
greatest term
is
mfi, where
is
+ /x)
(n + 2/i
- 2) ^ (n + 2M ) n2 e2
and so
fi
is
approximately equal to
).
it is
where
).
"
Hence
since* q
is
.BB< 1
. .
.}
~2ttMX/{7r/(l-?)},
nearly equal to
Stirling's
1.
Now, by
theorem,
6"- 1 exp {ws/(l
7T% 2
+ g2 )|
'
{l+ x/(l+ 2 )}
)
and
so
BJ l)r
2 v'(l
nn3
+ c2 )) ^*
J
exp
{1
+ 2 )} + ^(1 + ,')}
{
WU
The
drew from
jD,l
~_ V
/ 2V(1
**) \*
6"expW(l-e )}
{l
ttw*
+ V(l-e )}
2
'
This approximate formula happens to be valid when e < 1 (though the reason for this restriction is not apparent, apart from the fact that it is obviously necessary), but it is difficult to prove it without using the methods of contour
The
cf.
formula 1 + 2 2<f'
~ N/W(l-g)} maT
Bromwieh, Theory of
It is also
*3
Modern Analysis,
21*51.
(0 \r)
= (-*)-* M0 -TI
);
1*5]
integration
1826
and
La TMorie analytique
It is
de la Chaleur; in this work Bessel functions of order zero occur in the discussion of the symmetrical motion of heat in a solid circular cylinder.
shewn by Fourier ( 118 120) that the temperature x from the axis of the cylinder, satisfies the equation
v,
at time
t,
at distance
e~
ni
gx*
f
tf<*
x*
\ ')
'
22
2 2 .42
2 2 .42 .6 2
where g
at the
= rnCD/K and
+ K(dvldx) =
is
theorem that and no complex roots. His proof is slightly incomplete because he assumes that certain theorems which have been proved for polynomials are true of integral functions; the defect is not difficult to remedy, and a memoir by HurwitzJ
Fourier proceeded to give a proof
307
309) by
Rolle's
m hasf an infinity of
real roots
formula ( 313) for dervate; generalisations of this formula Another formula given by Fourier, namely
22 2 2 42
.
mentioned that Fourier discovered the continued fraction the quotient of a Bessel function of order zero and its
will
be discussed in
5-6, 9*65.
2 2 42 6 2
.
1 =
f*
|
had been proved some years earlier by Parseval; are now known as Bessel's and Poisson's integrals
*
22, 23). memoir deposited in the archives of the French Institute on Sept. 28, 1811, and crowned on Jan. 6, 1812. This memoir is to be found in the Mim. de VAcad. des Sci., iv. (1819), [published 1824], pp. 185555; v. (1820),
(
The
[published 1826], pp. 153246. t This is a generalisation of Bernoulli's statement quoted in 1*2. t Math. Ann. xxxin. (1889), pp. 246266.
648. This paper also contains the formal M4m. des savant itrangers, i. (1805), pp. 639 statement of the theorem on Fourier constants which is sometimes called Parseval's theorem another paper by this little known writer, Mim. des savans Strangers, i. (1805), pp. 379 398, con-
10
[CHAP.
The expansion of an arbitrary function into a series of Bessel functions of order zero was also examined by Fourier ( 314 320); he gave the formula for the general coefficient in the expansion as a definite integral.
was examined much more recently by Hankel, 471494; Schlafli, Math. Ann. x. (1876), pp. 137142; Diui, Serie di Fourier, i. (Pisa, 1880), pp. 246269 Hobson, Proc. London Math. Soc. (2) vn. (1909), pp. 359388; and Young, Proc. London Math. Soc. (2) xvni. (1920), pp. 163200.
validity of Fourier's expansion
The
Math. Ann.
vm.
(1875), pp.
1/6.
The unsymmetrical motions of heat in a solid sphere and also in a solid cylinder were investigated by Poisson* in a lengthy memoir published in 1823.
In the problem of the sphere f, he obtained the equation
is
a constant, n
is
a positive
R is that
sin2n+1 coda
Jo
?i
= 0,
1,
2 in detail.
is
order n
+ ^.
(ibid. p.
340
et seq.)
is
V
.'o
co)
sin2n G>cifo>,
where
w= 0,
1, 2, ...
and \
The
integral is
now known
as Poisson's integral (
In the case n
0,
and
is
its
an important approximate formula for the last integral (ibid., pp. 350 352) when the variable
a>
Let J
(k)
=it
'
(k)=
ir
cos
Then J
(k) is
*&ffl + 1+ i),_a (
249403. 300 etseq. The equation was also studied by Plana, Mem. della R. Accad. delle Sei. di Torino, xxv. (1821), pp. 532 534, and has since been studied by numerous writers, some of whom are mentioned in 4*3. See also Poisson, La TMorie Mathimatique de la Chaleur (Paris,
+ Ibid.
p.
t See also Rohrs, Proc. London Math. Soc. not used by Poisson.
v. (1874), pp.
136
187.
The notation J
(k)
was
1*6]
1826
11
When k is
may
we may expect
that Jo{k)sJk
and
B are constants.
2
(k)
- sin k
for
'
()
=-
"J
)}
ofo.
Write
ir
- to
a
.
and then
cos k
(k)
- sin k
/'
(k)
cos 2
Ao cos
(2k sin 2 i)
o?<u
^j
/"
I
V(2 * }
-2-)
"^'
and similarly
sin ../(*)
+ cos . J
'
()
= ^^
fl
-^Y"
nrnx^dx.
But
fo *-^oo y
lim
VW
(l-^V.^
\
2kJ sin
f J sin
^V.^K/^),
by a
well
known formula*.
[Note. It is not easy to prove rigorously that the passage to the limit is permissible the simplest procedure is to appeal to Bromwich's integral form of Tannery's theorem,
Series, 174.]
cos k
(k)
-sink.
'
(k)
1 = j^
1 4- t ),
sin
J
;
(k)
+ cos k J
.
'
(k)
= -jip- (1
rj
k ),
where
**-*-()
and
ri
k -*-0
as k-+-x>
and therefore
J ^ =
J(*k)
K1 +
ffc
C S
^
*+^+
sin
^'
i
It
^o'
(*)
^'3-
where
A =.5=1. The
of this expansion was not established, until nearly forty years later,
when
it
was
investi-
196.
The
(k) *J(irk)
eft
+ rn
is
-*[
.rz.l.B -jA
2.
2B"- (I. 2 + 1)
A'
2.
3B'"-(2. 3 + j) A"
& p
+
&
Ji
*
+
+-
1 J
J'
15, 1914),
+ 8m *[_
* Cf.
p. 71, for
+-
Watson, Complex Integration and Cauchy's Theorem (Camb. Math. Tracts, no. a proof of these results by using contour integrals.
12
and
so,
[CHAP.
we
find that
A'A---H,
A" -" A
A *, 2 g2
A'"*= A -
9 25 2.3.83
'
R
'
'
is
OM(*ooe.)A,-^J
j
o
C08 *
+ 8l"2T8^~2.3.88ife + -; 8m
9.25
,1
*J-
But, since the series on the right are not convergent, the researches of Lipschitz and subsequent writers are a necessary preliminary to the investigation of the significance of
be mentioned that an explicit formula for the general term in the expansion B. Hamilton, Trans. B. Irish Acad. xix. (1843), p. 313; his result
"["
-^ 1
[0]
([
(20 - $n,r
[0]
- }n),
and he described the expansion as semi-convergent; the expressions to be interpreted as 1/n and (-|)( f) ... (-+)
!
is
y = Ax%
Jo
e- hxcoSu> da>
+ Bx$
m) da>,
Jo
where
and
B are
constants.
It follows at
dxr
is
x dx
= A\
.'o
e~ hxcoau da)
+B
e hxcoa '\og
(a?
sin2 w)
cZo>.
Jo
likely that
This result was quoted by Stokesf as a known theorem in 1850, and it is he derived his knowledge of it from the integral given in Poisson's memoir; but the fact that the integral is substantially due to Poisson has
t Camb. Phil. Trans, ix. (1856), p. [38], [Math, and Phys. Papers, hi. (1901), J See Encyclopidie des Set. Math. n. 28 ( 53), p. 213.
p. 42].
1'7]
1826
13
17.
bear his
The memoir* in which Bessel examined in detail the functions which now name was written in 1824, but in an earlier memoir f he had shewn
is
+ 2
Bn cosnM,
ne sin u)
where
Bn = Bn should
sin
u sin (nu
du
1'4.
If = k zvrJo
I
2"
cos (hu
He
not adapted for defining the most worth study when h is not an integer (see 10-1) the function which is of most interest for non-integral values of h is not Ikh but the function defined by Lommel which will be studied in Chapter in.
Bessel's integral
;
that
After the time of Bessel investigations on the functions became so numerous it seems convenient at this stage to abandon the chronological account
logical order.
account of researches from the time of Fourier to 1858 has been compiled by Wagner, Bern MittheUungen, 1894, pp. 204266 a briefer account of the early history was given by Maggi, Atti delta It. Accad. dei Lincei, (Tra/isunti), (3) iv. (1880), pp. 259263.
historical
;
An
* Berliner Abh. 1824 [published 1826], pp. 1 52. The date of this memoir, " Untersuchung des Theils der planetarischen Storungen, welcher aus der Bewegung der Sonne entsteht," is Jan. 29, 1824.
t Berliner Abh. 181617 [published 1819], pp. 4'J 55. J This integral occurs in the expansion of the eccentric anomaly
nA n
= 21\ t
p.
383.
CHAPTEE
21.
II
of the Bessel
is
coefficients.
The
of a set of functions
known
;
as Bessel coefficients.
the method which will be adopted in this work is to in a certain expansion. This procedure is due coefficients the as them define properties of the functions from his defimany derived who Schlomilch*, to
nition,
and proved incidentally that the functions thus defined are equal to the definite integrals by which they had previously been defined by Bessel f. It should, however, be mentioned that the converse theorem that Bessel's integrals are equal to the coefficients in the expansion, was discovered by Hansen:}:
fourteen years before the publication of Schlomilch's memoir. results had been published in 1836 by Jacobi ( 222).
Some
similar
The generating
is
It will be shewn that this function can be developed into a Laurent series, n qua function of t; the coefficient of t in the expansion is called the Bessel is denoted by the symbol Jn (z), coefficient of argument z and order n, and it
so that
*('"?)(1)
I
=-oo
t"Jn
(z).
expanded into an and for all values of t, ielt can be expanded into an absolutely converwith the exception of zero, e~ of t. When these series are multiplied powers descending gent series of convergent series, and so it may be absolutely together, their product is an is to say, we have an expansion of the arranged according to powers of t that and t, t = excepted. form (1), which is valid for all values of z
To
ascending powers of
* ZeiUchriftfiir
namely that of
Math, und Phys. n. (1857), pp. 137 165. For a somewhat similar expansion, Mem. Soc. Ital. (Modena), xvm. (1820), p. 503. It must be
pointed out that Schlomilch, following Hansen, denoted by Jftn what we now write as Jn {2z) but the definition given in the text is now universally adopted. Traces of Hansen's notation
are to be found elsewhere, e.g. Schlafli, Math. Ann. 22. t Berliner Abh. 1824 [published 1826], p.
theil, [Schriften
m.
(1871), p. 148.
+ Ermittelung der Absoluten Storungen in Ellipsen von beliebiger Excentricitdt und Neigung, der Sternwarte Seeburg : Gotha, 1843], p. 106. See also the French translaab$olues (Paris, 1845), p. 100, and Leipziger tion, Memoire sur la determination des perturbations
i.
2*1, 2 '11]
15
If in (1)
\jt
for
t,
we get
n -x
= 5 i-tyj_ n { Z
n
x
),
on replacing n by - w. Since the Laurent expansion of a function a comparison of this formula with (1) shews that
(2)
is
unique*,
J-n{z)
is
= (-YJn (*),
where n
any integer
Jn (z)
as an integral.
(2) it
is
From
(3)
may be
{t
=J
{z)
+ J
w-1
+{_
Jn {2)
The Analyst, of elementary applications of these functions to problems of Mathematical Physics has been compiled by Harris, American Journal of
i.
(1874), pp.
The
function of order unity has been encountered by Turriere, Nouv. Ann. de Math. (4) 433441, in connexion with the steepest curves on the surface z=y (5.r2 -y).
2*11.
The ascending
series
for
Jn (z).
the form of an ascending series of powers
series for
by considering the
exp
(h zt)
I flT v
r =o
- **>'"'L
ml
rl
l=t>
term of the first series on the associated with the general term of the second series gives rise to a term involving tn is the term for which r = n + ; and, since n > 0, there is always one term for which r has this value. On associating these
right which,
When n is a
when
terms
coefficient of t n in the
product
is
m=o(n-rm)l
ml
We
(1)
Jn(z)= S
* Z)
series in t, in which some of the were not zero. If we then multiplied the expansion by t and integrated it round a circle with centre at the origin, we should obtain a contradiction". This result was noticed by Cauchy, Comptes Rendus, xm. (1841), p. 911.
'
16
[CHAP.
II
where n
The
first
Jn () =
z*
2
.
2^1 { 1 ~ 2
Jo (?)
(n
+
1
2-
* + 1) (n +
2)
'"]'
In particular
(3)
22
^ 22 42
.
22
42 6 2
.
"
To obtain the Bessel coefficients of negative order, we select the terms involving irn in the product of the series representing exp tyzt) and exp(- \z\t\ where n is still a positive integer. The term of the second series which, when
n associated with the general term of the first series gives rise to a term in t~ is the term for which ra = n + r ; and so we have
J -n{Z) ~
rZo
r\"
(n
+ r)l
'
2'1 (2),
namely
J_^) = (-)
It is to
^(*)-
be observed that, in the series (1), the ratio of the (w + l)th term 2 oo for all to the rath term is - s /{ra (n + ra)}, and this tends to zero as ra that follows it values of z and n. By D'Alembert's ratio test for convergence, it so and and n, the series representing Jn (z) is convergent for all values of z
,
is
0,
1,
2,
3,
. . .
Jn
(z) is
;
and so
it
a transcendental function
to say it is not expressible as a finite combination of exponential, logarithmic and algebraic functions operated ou by signs of
transcendent, that
indefinite integration.
From
ance
(cf.
(1)
some importis
Whether
we have
ao
i
i2W
*
and
so,
n!
mZom\(n
+ l)m
'
when n ^ 0, we have
!
(4)
J.
^
l.
exp (J
|>).
854
This result was given in substance by Cauchy, Comptes Rendus, a similar but weaker inequality, namely
;
xm.
lJ
.l< -^P(l*l ,
2*12]
17
By
()
Jn (z)
except the
first, it is
found that
j.()-^(i+),
||
where
It should
(*
' ,'>- 1
l
any, bounded
*-plane.
be observed that the series on the right in 2*1 (1) converges uniformly in domain of the variables z and t which does not contain the origin in the
if 8,
For
A and
<
A.I,
the terms in the expansion of exp (\zt) exp ($z/t) do not exceed in absolute value the corresponding terms of the product exp ($RA) exp QR/8), and the uniformity of the convergence follows from the test of Weierstrass. Similar considerations apply to the series obtained
by term-by -term
(z),
or
or both z and
t.
The equations*
(1)
Jn-i (*)
(2)
known
as recurrence formulae.
To prove the
namely
,*'-= 5 tj (z), n
with respect to
t
;
we get
$z(l
+ l/t*)e
2
iz(t ~
m=
2 nt^Jniz),
nt^Jniz).
t
so that
^(1 + 1/e ) 5
If the expression on the left
is
Jn (z)= 2
arranged in powers of
and
coefficients of n_1
tf
are equated in the two Laurent series, which are identically equal,
it is
evident
that
(z))
=nJn (z),
is
the
first
of the formulae f.
Throughout the work primes are used to denote the derivate of a function with respect to
argument.
f Differentiations are permissible because ( 2*11) the resulting series are uniformly convergent. of coefficients is permissible because Laurent expansions are unique.
The equating
18
[CHAP.
;
II
m=
= -oo
Jn '(z),
t
so that
\(t
- Ijt) 2
=-oo
Jn (z) = 5
w=-oc
Jn
'
(z).
By equating
The
(3)
coefficients of tn
(2) immediately.
results of adding
(4)
zJn (z) + nJn (z) = zJn__ (z), zJn'(z)-nJn (z) = -zJn+1 (z).
to
^{^ (*)}=*"</-!(*),
^{z-Jn (z)} = -z-"Jn+1 (z).
is trivial
(6)
/,'(#)
= -/(*).
(1) and (4) from which the others may be derived were discovered by Abh. 1824, [1826], pp. 31, 35. The method of proof given here is due to Schlomilch, Zeitschrift fur Math, und Phys. ir. (1857), p. 138. Schlomilch proved (1) in this manner, but he obtained (2) by direct differentiation of the series for Jn (2).
The formulae
Bessel, Berliner
A
(g)
{ibid. p.
is
^T-)=
\m
where
Cm
is
a binomial
coefficient.
(6),
By
(d
/
n
{*
zTz)
Jn(z)}=2 n m Jn-m(z),
(10)
(is)
is
where n
any positive
integer.
The formula
(10)
is
due
As an example
*Ji
= 4J"
(z)
= 4 1 (-y-*nJm (z),
since
zJ^+i (z) -*
as iV-*- 00
by
211
(4).
2-13, 2'2]
19
*./,(*)
=4 I
n-l
{-)n
nJ2n (z),
functions ( 3'o7).
2*13.
The
Jn {?)>
When
the formulae 2'12 (5) and (6) are written in the forms
(z) is
seen to be
d
dz
that
is
to say
and
so
we have
(1)
W)
+z
dJAz)
The
analysis
is
simplified
+ n) Jn (z) = zJn _
(^
{z),
(^ - n
1)
and so
n + 1)
[z-i
(*
that
is
z-> (^
The same
(S
Jn +
(z) is
formulae
+ n + 1) Jn +
(z)
(z).
2*2.
We
(1)
now prove
that
Jn (z) =
JLit
"'
I
cos {n$
- z sin B) d6.
'o
Jn (z),
and he
Mem.
Soc. ItaL.(Modena),
xvm.
(1820), p. 504.
20
[CHAP.
II
modify
(1)
in-
2tt
for
Jn (z)=IT
cos (n0
- z sin 0) dO.
first
JO
27r,
the
equation
may
be transformed
/*2ir+a
I
(3)
J n (z) = 2~
is
cos(n0-zsm0)d0,
~x
where a
any angle.
(1),
To prove
(1)
by t~n
and
We
thus get
2771 J
=-
all
Z7rt
so
m = n;
and
may
be a circle of unit radius and write t = e~ i0 so that be taken to decrease from 2tt + a to a. It is thus found that
to
t
rtor+a
(5)
Jn(*)^-/
*<**-*) <$
77" J a
by
Jn (z) = -L
and equation
(2),
e*(*-z n )
+ e - Une - z sin 9) Q
J
may be
deduced,
is
now
obvious.
Jn (z) =
If
f
I
cos
If... n0
I
sin
IT
Jo
TT J
be replaced by 7r in these two integrals, the former changes sign when n is odd, the latter when n is even, the other being unaffected in each case and therefore
s Jn (&) J i ; =i
n n & sm ( z sm 0) d0
(odd).
.
(6)
T JO
sin
w0 sin (s sin 0) d0
I
integration
is
contour. It
"
2*21]
21
Jn {z)K)
2
'**
I
cos
n$ cos (z sin
6)
dd
("even).
= ir
,^
I
cos w
d0
Jo
If
be replaced by tt /n (*)
77
and
(7), it is
found that
(w odd),
8)
--
(-)*<
cos
n?7
sin (s cos
17) <fr;
(9)
/ii
(z)
=-(-)**
(n even).
The
that
last
two
It
results are
[Note.
(1805), pp.
639648,
2s
+ 22742 ~
a*
2 2.42.62 +
1
,,
[*
cos (
and
will
so, in
the special case in which n = 0, (2) will be seen in 2-3 that two integral representations of
identical
8m *) <**
It
Jn (z),
namely
Bessel's integral
for
when w=0,
so a special
name
thL
case
is
The reader
2-12 (4)
(1)
and
from Bessel's
integral.
2 '21.
formulae involving definite integrals which are closely connected with PaTseval's integral formula are worth notice. The first, namely
(1)
is
Two
due to Bessel f.
/>
J
dfy
+ iz sin 6f +
dd = 0,
f"
(y cos 6
+ iz sin 6f dd =
2r ( w + *)r(i)
(y2
difficulty.
The other
(2)
is
definite integral,
J
(1)
(2 <Jz)
=IT
/"" e<i+)cos
cos {(l
- z) sin
1
0} dd,
J
1
a special case of
obtained by substituting
-z and
+z
for 2
and y
respectively.
* Journal fUr Math. xv. (1836), pp. 1213. integrals actually given by Jacobi had limits
[G<?s.
Math. Werke,
100102]
1,
the
2/*-.
and w with
See also Anger, Neueste Schriften Comptes Rendus, xxxvm. (1854), pp. t Berliner Abh., 1824 [published Ges. in Danzig, v. (1855), p. 10, and
and Cauchy,
910
913.
See also Anger, Neueste Schriften der Naturf.
p. 151.
(2) xli. (1876), p.
1826], p. 37.
22
Catalan's integral
THEORY
may
OJF
BESSEL FUNCTIONS
[CHAP.
II
no that
./
(2i\'*)
-i_-J-.
=
J_.
2
(0
/"
-,
+
t-^-Wdt
ex
p^ + ! )^ = l
N
;
[* exp {*+*-}
<W,
to be a unit circle
2"22.
ie
in
The simplest method of obtaining them the fundamental expansion 21 (3). We thus get
M=l
to write
= J 0) + 2 2 Jm 0)cos 2nd
n=\
2i
1)
0.
On adding and
we
find
(1)
cos (z sin 0)
(2)
sin(*sin0)=
Write \ir
(3)
r)
for 6,
and we get
t?)
cos
cos
= Jo (z) + 2 2
=i
(4
sin (s cos
t/)
1)
rj.
The results (3) and (4) were given by Jacobi, while the others were obtained later by Anger t. Jacobi's procedure was to expand cos (s cos 17) and sin (z cos q) into a series of
cosines of multiples of
integrals
t}, and use Fourier's rule to obtain the which are seen to be associated with Bessel's integrals.
coefficients in the
form of
In view of the fact that the first terms in (1) and (3) are not formed according to the same law as the other terms, it is convenient to introduce Neumann's factor* en> which is defined to be equal to 2 when n is not zero,
and
to be equal to 1
*
when n
is zero.
The employment
[Ges.
Math.
t Neueste Schriften der Xaturf. Gen. in Danzig, v. (1855), p. 2. % Neumann, Theorie der BesteVschen Functionen (Leipzig, 1867), p.
7.
2-22]
will
23
(5)
cos (^ sin 0)
=o
ao
(6)
sin (z sin 0)
= X
=o
find
e2w+1
If
we pub
in (5),
we
0)
If
1=2
m Jm (z).
of. Bessel coefficients when m is any positive any polynomial is thus expansible. This is a special case of an expansion theorem, due to Neumann, which will be investigated in Chapter xvi.
is
that zm
we differentiate (5) and (6) any number of times before putting 6 = 0, we obtain expressions for various polynomials as series of Bessel coefficients. shall, however, use a slightly different method subsequently ( 27) to prove
We
integer.
For the present, we will merely notice before 6 is put equal to 0, there results
(8)
that, if (6)
be differentiated once
*= 2
%-ir
e 2n+1
(2n
+ l)J2n+1 (z),
(5)
after
2
two differentiations of
i
and
(6),
then
zsinz
zcosz
(z) -...},
(10)
(z)
52
!i
(z)-
...}.
associated with
en
Jn (?)
is
**"/
(z).
2-12 (2),
we have
If
we
we get
Z
(t
(11)
)di:
Soc. xvi. (1910),
American Math.
pp. 225230.
* Studien ilber die Bessel'schen
It will
Functionen (Leipzig, 1868), p. 41. be seen in Chapter xvi. that this is a form of " Lommel's function of two variables."
24
2
3.
[CHAP.
II
Shortly before the appearance of Bessel's memoir on planetary perturbations, Poisson had published an important work on the Conduction of Heat*, in the
course of which he investigated integrals of the types f
Jo
2"-1" 1
Odd,
Jo
where n
is
He
and gave the investigation, which has already been reproduced in 1*6, to determine an approximation to the latter integral when z is large and positive, in the special case n = 0.
solutions of certain differential equations^
We
shall
now prove
that
it
describe the expressions on the right as Poissons integrals for n case n = 0, Poisson's integral reduces to Parseval's integral ( 2'2).
It is easy to prove that the expressions
Jn (z);
I
in powers of z
fir
oo
7tJ
cos (* cos
0)8^0(20 = - t 7T m =o
V (2W)!
\m sm
fir
cos2'0sin=0<W
Jo
=
j!
~(>0r
'
2.4.6... (2n
'
+ 2m)
~ -1.3.5...(2-1) 2
and the result
is
+ m)r
obvious.
* Journal de VEcole R. Polyteckuique, xn. (cahier 19), (1823), pp. 249403. previously been t Ibid. p. 293, et seq. ; p. 340, et seq. Integrals equivalent to them had examined by Euler, hut. Calc. Int. u. (Petersburg, 1769), Ch. x. 1036, but Poisson's forms are more elegant, and his study of them is more systematic. See also 3'3.
R=rn+1
then
udu,
Jo
equation
dr1
Nielsen,
r-
Handbuch der
51, calls
them
is
but the above nomenclature seems preferable. The series to be integrated is obviously uniformly convergent ; the procedure adopted
due
2-3, 2*31]
25
=
it
Jo
when we
mean
by
by replacing
0.
We thus get A
(3) slight modification of this formula,
namely
(1
/(*)
r(n
4|^V,
(cf.
_ PY - iM>
(4)
0d0 = 2
'
\
0d0
Jo
Jo
=2
f*
cosCssin^cos*1
Jo
^,
was obtained
and each of these expressions gives rise to a modified form of Poisson's integral.
Poisson's integrals
-o
(zm)\
integrated.
(
J (z)-(-Y I
Jm (s)
2'31.
The
Bessel J,
Jn (z)
is
somewhat elaborate;
substantially as follows
It is seen
on differentiation that
sin2"
-1
-^ cos
I
cos (z cos 0)
9
8
,-
sin2*1
**
" "
1
=
|
(2w - 1) sin8"-8
made
- 2n sin " +
p. 340,
* Poisson actually
sinSH+1 6
;
the statement
but, as
he points out on
(p. 293) concerning the integral which contains odd powers may be replaced by even powers throughout
his aualysis.
f Studien Uber die BesseVscken Functional (Leipzig, 1868), p. 30. + Berliner Abh. 1824 [published 1826], pp. 3637. Jacobi, Journal filr Math. xv. (1836), p. 13, [Get. Math. Werke, vi. (1891), p. 102], when giving his proof ( 2*32) of Poisson's integral formula, objected to the artificial character of Bessel's demonstration. w. B. jr. 2
26
[CHAP.
-1)1
6d0 - %n
^2
0d0
+ 2n + i J
If
r*
0d0 = 0.
now we
write
rin + ^ra) /^
the last formula shews that
z<f>
s * <>
(n
- 1) - 2w0 (w) +
it is
*<-6
+ 1) = 0,
so that
(w)
and
Jn (z)
satisfy the
=J <f>(0)
<f>
(1)
^J
l r
0d0 = -
-J
'
0d0
=and
so,
sm (z cos
0) cos
Odd
(z)
(z),
when w
= 0,
1, 2, 3, ....
2*32.
The problem
integral
method
si^ = ^
We
shall
it
where
fi
cos
0.
assume this formula for the moment, and, since no seems to have been previously published, we shall
2323.
fi,
first
is
Odd
= zn
cos (zfi) (1
.
/a
)"-* dfi
= (-) n
*
J_
COS foa
i
- \ mr)
-^ d. di
}
vi. (1801),
Journal fiir Math. xv. (1830), pp. 1213. [Ges. Math. Werke,
i.
(1836), pp.
195196.
2*32, 2*321]
If
27
we now use
becomes
{Zfi
1.3.5...(2n-l)f If
cos
\n-jr)
.dsinnd dsinnd
dfi
=1
3 5
.
. .
(2n
- 1)
Jo
cos (z cos
= 1.3.5...(2n-l)wJ n (5),
by
to
and
(9), since
cos^cos 6
is
\nir)
is
equal
( )***
2*321.
sin(^cos 9) according as n
as a special case of a formula due to Lacroix*; but the proof which Lacroix gave of his formula is open to objection in that it involves the use of infinite series to obtain
a result of an elementary character. A proof, based on the theory of linear differential equations, was discovered by Liouville, Journal de Math. vi. (1841), pp. 69 73; this proof will be given in 2*322. Two years after Liouville, an interesting symbolic proof was published by Boole, Camb. Math. Journal, in. (1843), pp. 216224. An elementary proof by induction was given by Grunert, Archio der Math, und Phys. iv. (1844), pp. 104 109. This proof consists in shewing that, if
<p- i(W) -
/"(* dQa --2/i9B -M(H-l) 5 J
then
2 n+1 = (l- M )-
B rfM ,
and that ( -
n~
)
'
1.3.5.;. (2
Other proofs of this character have been given by Todhunter, Differential Calculus (London 1871), Ch. xxviil, and Crawford t, Proc. Edinburgh Math. Soc. xx. (1902), 15, but all these proofs involve complicated algebra. pp. 11
depending on the use of contour integration is due to Schlafli, Ann. di Mat. (2) The contour integrals are of the type used in establishing 202. Lagrange's expansion; and in 2*323 we shall give the modification of Schlafli's proof, in which the use of contour integrals is replaced by a use of Lagrange's expansion.
v.
A proof
To prove
i.l!5^-l) g^ 1 -^'
"
1+ '->- ,
\
=0
a"
m ~t"2J
=
and
*
m=0
=sin(2wx0),
this is the transformation required
J.
183. See also a note written by edition), pp. 182 Catalan in 1868, Mim. de la Soc. R. des Sci. de Liege, (2) in. (1885), pp. 312316. t Crawford attributes the formula to Rodrigttes, possibly in consequence of an incorrect statement by Frenet, Recueil d'Exercices (Paris, 1866), p. 93, that it is given in Rodrigues' dissertation,
Diff.
i.
TraiU du Calc.
(Paris, 1810,
2nd
Corresp. sur VEcole R. Poly technique, in. (18141816), pp. X I owe this proof to Mr C. T. Preece.
361385.
28
2*322.
[CHAP.
II
and
then obviously
(l-V)Z)y+(2n-l) My=0.
Differentiate this equation
(1
- fi2 )
Dtt+l
i/
but
(l"^-^-^^^-"^.
(^ +
D*~ 1 y= A
;
- (iD^+^D^-^^O
so that
Z)- y = 0.
Hence
where
sin nd +
BcoB n6,
A and B are constants since D y is obviously an odd function of 6, B is zero. To determine A compare the coefficients of 6 in the expansions of D*- 1 y and A sinnd in ascending powers of 6. The term involving 6 in D*~ l y is easily seen to be
n"1
n~ l
(-^)
so that
-3
<9'
%J = (-)- 1 1.3.5...(2n-l),
result,
d-sin2-i0
.1.3.5...(an-l)
v
'
dp*~ l
2*323.
We first recall
is
so that
(,)
| = *
Now
it
take
/(2) 3 -|(l-2*),
v/(l-*
8
),
The
singularities of 2
,J{\
pansion of
22 )
de
when A-*-0. (2) reduces to ^(1 ja function of h are at A=>e *' ; and so, when 6 is real, the exin powers of h is convergent when both h and 2 are less than unity.
-/*2
to sin 6
| \
Now
and so
Jfl
J)
(l^-*)*-(l^*)
is
/_\-i
Hence
pansion of
it
follows that
d -1 sin 2B_1 d
,
d/t"
. . -1
J(l
- 22 )
z
(62/8/*)
in
powers of
h.
But
it is
evident that
,n
^*
(l-^y-i-q-^-*)-* coefficient of A*
j?
* Cf.
Modem
Analyst*, 7-32.
2-322-2*33]
29
233.
An
cos
nxdx =
2
Jo t-0
which am is the coefficient of tn+2m in the expansion of Jn (t)/J (t) in ascending powers of t, has been studied by Jacobi*. To establish it, integrate the expression on the left n times by parts it transforms ( 2*32) into
;
1.3.5...(2n-l) Sl
and,
f(n) (C S
*} 8inm * dx
>
when
1
sin2n #
f
is
XM ,
J.
(n,
2n
2n(n-l)
We
now
integrate
4a?,
...
by
parts,
and by
we
When
/(cos x)
is
a polynomial in cos
is
a?,
the process
certainly valid.
To determine
[
/(cos x)
according as n
is
and
(9) that
Jn (t) = 2
so that am has the value stated.
(_)-,+
{(_), J%
It has been stated that the expansion is valid when /(cos x) is a polynomial in cos a;; it can, however, be established when /(cos x) is merely restricted to be an integral function of cos x, say
* 6w cos n a?
is less
seems
to be of
no practical importance.
*
2526
xxvm.
(1849), col.
[Qes. Math. Werke, vi. (1891), pp. 117118]. 94 [Ges. Math. Werke, vn. (1891), p. 174].
See
30
2*4.
[CHAP.
II
The Bessel
Jn (y + z)
expressed in terms of Bessel coefficients of y and which was first given by Neumann* and Lommelf, is
00
may be
This formula,
(1)
J(y
+ z) = 2 Jm (y)Jn-r,i(z).
m-oo
is
Jn {y
+ *) = 5^-.
f<
'~
po+)
M_1 * {y+*
oo
{t
dt
~
I i
00
X
oo
m -n- Jm (y)e*z 1
<t
1/t
>dt
no+)
= _i-.
analysis
Jm (y)\
p--i eft-Milt
2
wi -oo
Jm (y)Jn-m{z),
line of the
on changing the order of summation and integration in the third and this is the result to be established.
;
Numerous
2'5.
coefficients.
Special cases of
early as 1843.
The
Neumann's addition formula were given by HansenJ as first system of formulae is obtainable by squaring the
2*1 (1), so that
fundamental expansion
.-!/,
r=
oo
Km-
By expressing
the product on the right as a Laurent series in t, and equating n the coefficient of tn in the result to the coefficient of t in the Laurent expansion of the expression on the
left,
we
find that
J(2s)
In particular, taking
(1)
?i
= Jr {z)Jn-*(z).
have
= 0, we
r=l
Jo (2s)
= Jo* (:) + 2S
-
r=0
(-)' e r
Jr
(z).
* Theorie der BesseVschen Functionen (Leipzig, 1867), p. 40. f Studienilber die BesseVschen Functionen (Leipzig, 1868), pp. 26
27
Math.
Ann. in. (1871), pp. 135137. + Ermittelung der absoluten Stiirungen (Gotha, 1843), p. 107 et seq. Hansen did not give (4), and be gave only the special case of (2) in which n = l. The more general formulae are dne to Loramel, Studien Uber die BesseVschen Functionen (Leipzig, 1868), p. 33.
For
brevity,
JH2 (z)
is
written in place of
{</
(z)) 2 .
2'4-2'6]
31
From
(2)
we
find that
Jn (2z) -S/f (#) Jn-r (*) + 2 (-Y Jt (*) Jn+r (z), r=0 r=l
Bessel coefficients of negative order are removed
when the
by using
2*1 (2).
Similarly, since
j
ni
{z)\
1/t)}
= exp [z (t =1
>
exp [\z (-
1/t)}
it
follows that
(3)
J >(z) + 2% Jr*(z) = l,
r-0
(4)
Equation (4) is derived by considering the coefficient of tm in the Laurent expansion the result of considering the coefficient of tm+1 is nugatory.
;
A
(5)
(3),
namely
that,
when x
is real,
\J,{*)\*1,
|^,(*)j<W2,
where r
2"6.
= 1,
2, 3, ...,
Neumann's
integral for
Jn
(z).
It is evident
Jn (z) = land so
Jn2 (z)
1 = |-5
f*
j
f*
I
*>+*> e -(nfl+sin +,
dOdj).
new
variables defined
8-<f>
= 2X
0+<f>
2yJr,
d(0,<f>)
2.
9 (x Vr )
It follows that
J 2 (*) =
*"""*
2^*11
is
y
d*;cty,
where the
field
of integration
unaffected
yfr
if
both
% and
yfr
are increased by
tt,
or
if
increased by
gration
may
simultaneously decreased by ir, the evidently be taken to be the rectangle for which
ir is
while
field
of inte-
0<X^ 7r _ TT ^
l/r
7T.
32
[CHAP,
- 2^-
f f*
* * cty dy
=If
result
(1
1 C"
if
Jo
<%
acute or obtuse,
we
replace
% by |tt T
0,
according as
t
is
we
obtain the
Jn*
may
W = -f
If J
This formula
(2)
Jn
(z)
JO
which is the result actually given by Neumann*. It was derived by him by some elaborate transformations from the addition-theorem which will be given in H'2. The proof which has just been given is suggested by the proof of that addition- theorem which was published by Graf and Gublerf.
We obtain a different form of the integral with respect to x instead of with respect to yjr.
Jr? (s)
so that
i
if
^f
r*
1 if
Jo (2* sin
yjr)
(3)
^""^"Srl
a
/o(2*sin^)cos2n^(fy
f*
(2z sin
i/r)
cos 2nyjr
dijr,
2'61.
Neumann's
series
a for J
(z).
taking the formula 26(1), expanding the Bessel coefficient on the right in powers of z and then integrating term-by-term, Neumann shewed
By
that
lf
(-)* zm+wt
sin 2n+^B,
JQ
_ ~
TO =
-) (2n
-f
2m) (^zYn+im
! '
m\{tn+m)\{(n + m)l\ %
Theorie der BesseVtchen Funetionen (Leipzig, 1867), p. 70. t Einleitung in die Theorie der BesteVtchen Funktionen, n. (Bern, 1900), pp. 81 85. of Neumann's treatise. J The formula is an immediate consequence of equation 16 on p. 69 was given, was first pub Math. Ann. in. (1871), p. 603. The memoir, in which 4his result lished in the Leipziger Berichte, mi. (1869), pp. 221256.
2*61, 2'7]
33
W
where
Jn{Z)
(n\f I
l(2n
l)
1.2.(2n
l)(2n
+ 2)
J'
1\
l = 2w + 2' 2m-
(2)
3) 4)'
+ l)(2tt + 3)(2n + 5)
,
6)
This expansion
is
a special case of a more general expansion (due to any two Bessel functions as a series of powers with
( 5*41).
2*7.
Schlomilch's expansion of z m in
series
of Bessel
coefficients.
We shall now obtain the result which was foreshadowed in 2*22 conis any cerning the expansibility of zm in a series of Bessel coefficients, where = has already been given in 2*22 (7). positive integer. The result for
In the results 2*22(1) and (2) substitute for cos 2nd and sin(2n+ 1)0 These expansions are*
cos2n#= 2
,=o
n.(n + s -1)!
(-)'
wo\t (n-s)!(2s)!
,
(2 sin ey,
The
(
cos (x sin 6)
- J, (x) + = 1 Jm+
2 ^1
J (x)
j
sin (x sin 9)
(X)
(-)
gj$| (* - )""(
series in sin 6
If
we rearrange the
series
that
(
it is
permissible to do
so),
(assuming
-/i\
" (2n
ro TT\i
Hobson,
r^-^r
l).(n +
s)!
J+i(*)
* Cf.
34
If
[CHAP.
II
we
l=Jo(s) + 2
n=l
5 J2n (4
(
I
'
"-.
Jn^7)\
Jm+1
(z >-
*' 2>
>
The first of these is the result already obtained bined into the single formula
(1)
the others
may be com-
(i .
I C + *.)( + r = w=0
. -Dl
Jm>
(m _ 1(
2|
3...)
m = 1, 2, 3, were given by Schlomilch* shewed how to obtain the general formula which was given explicitly some years later by Neumann f and LommelJ.
The
particular cases of (1) for which
He
also
permissible
The rearrangement of the double series now needs justification the rearrangement if we can establish the absolute convergence of the double series. If we make use of the inequalities
;
is
IA +1
(*)|<||^exr(iM*),
{z
^'<1,
-
<>.),
sin 6) T
we
,
see that
I2sin4j*+i (2w
+ l) .(n+)!,
Nl
|2sin^p +
|i*|
fc+
,.
,
l2N
= sinh
and so the series of moduli a similar manner.
is
(|
2 sin
1)
exp (
2
|
),
convergent.
The
may
be treated in
The somewhat elaborate analysis which has just been given is avoided Lommel's proof by induction, but this proof suffers from the fact that it
supposed that the form of the expansion
tion.
If,
in
is
is
verifica-
following
i tor- n=0
* ZeitschriftfUr Math,
< + * t
>-iT
n. (1857), pp.
1 )'
jM ,(A
und Phys.
140141.
f Theorie der BesseVschen Functionen (Leipzig, 1867), p. 38. t Studien liber die BesseF?chen Functionen (Leipzig, 1868), pp. 35
in
36.
Lommel's investigation
2*71]
35
m = 1],
we have
S (m + 2n) (m + n 1)
n=0
i
x.
n
\
^i
* J~+ (*)
2
=0
(wi -Mi)!
,,
n\
_
Since (m
"
=0
(m+l + 2n).Q + n)
w! as
-* oo
It is obvious
in-
m = 2,
3, 4,
An
Let
t
is
as follows
2
,
let
so that
when
= 1),
We then
have
wi!
/"
expi-ia^-l/O}---^*
^/
when we
n=0
calculate the
lw
-p<-^'-'/',
*
r
\
+2n)
-'
i!
-ir'
'--"
the inter-
zh
sum
;
change of the order of summation and integration is permitted because the series converges uniformly on the contour and the required result is obtained.
Note.
When m
is zero,
-y-
has to be replaced by
2*71.
2n r JH
(z).
The formulae
(1)
1 (?nypJin {z)= 2
/^"S
i^?* -*
9
1
,
(2)
n=0
in
which p
is
and 1*
is
a numeri-
cal coefficient, are evidently very closely connected with the results of 27.
* Messenger, xxxii. (1903), p. 8; a proof
The formulae
had been
pre-
on the same lines for the case Nieuw Archief voor Wiskunde, xx. (1893), p. 120.
=l
36
[CHAP.
(1857), p. 141,
II
II.
and he
{)
is
w>
<* i-r-c*(-**r
2m
m ~A=o
.m!
where m Ck is a binomial coefficient and the last term of the summation is that for which k hm - 1 or \{rti 1). To prove the first formula, take the equation 2*22 (.1), differentiate
to
0,
zero.
/ <
- ).<*.). *.
xn
/o
x*.
/x R**" w _[__
(* sin *)"! ;
j,. o =
V Z [^ Mi;
d *p
(-yfaiiPT]
(2m)
,
j,,,,
The terms
m>p,
when expanded
in ascending powers of
6,
it is sufficient
to evaluate
[_d6*>JLo
(2^Ti
J-o"
JL
(a*)l".iJoLM*l
P
fAz^ 2m 2m
!
(20""*
iJ-o
m=0 v * wl7
Jfc-0
= 2(-)l S
=o
jP-Zff,
and the end of the summation with respect to evident, and equation (2) is proved in a
which were stated by
are equal.
similar
(1) is
now
The reader
Schlomilch
4)
...
2*72.
From
( 2'7)
of
of even order,
Bessel coefficients,
by using Neumann's
Thus,
if
am (2sin0)=
,
v 2
(2m
v
, /a J Wi+m (2zsm0),
.
-.
0,
I?.? &>*- I
ir
(^ + 2)-(2m + n-l),
Jo
=o
so that
(I) (1)
(when
m > 0)
!
/-*<*>
2-72]
37
<^-S.L-ri^<*
when
and
this is true
m = 0,
for it
25
As
special cases,
we have
**=
(3)
~
tJ.
.
4=2
.
4n* (4
- 22) J* (*),
- 22 )
(4n* - 4*)
./* (*),
*6= Il-f-'-J I
4 5 b =3
4n* (4n2
If
we
it is
tA\
uh- 1
.!(- 1)1
Z (2m + 2n-l).(2m+-2)!
ST
CHAPTER
3*1.
III
BESSEL FUNCTIONS
The generalisation of
Bessel's differential equation.
II,
The
are functions
of which z is
required to be an integer.
We
shall
now
in the course
shewed that the function, so defined, is a solution of the linear differential equation which is to be discussed in this section. Lommel's definition of the Bessel function Jv {z) of argument z and order v wasf
(*)
= rfr +ffirft)
0d0
'
and the integral on the right is convergent for general complex values .of v for which R(v) exceeds |. Lommel apparently contemplated only real values of v, the extension to complex values being effected by HankelJ; functions of order less than - \ were defined by Lommel by means of an extension of the recurrence formulae of 212.
on comparing 3*3 with 1*6 that Plana and Poisson had investigated Bessel functions whose order is half of an odd integer nearly half a century before the publication of Lommel's treatise.
will observe,
The reader
We shall now replace the integer n which occurs in Bessel's differential equation by an unrestricted (real or complex) number v, and then define a Bessel function of order v to be a certain solution of this equation it is of
;
Jn (z) when
v assumes
n.
We
(1)
z>*l
will
+z
^+
(z>- v>)
y = 0,
v.
which
Functionen (Leipzig, 1868), p. 1. + Integrals resembling this (with v not necessarily an integer) were studied by Duhamel, Cours d Analyse, n. (Paris, 1840), pp. 118121.
%
Math. Ann.
i.
(1869), p. 469.
v,
ft.
is customary on the Continent, though it has not yet come into general use in this country. It has the obvious advantage of shewing at a glance whether a result is true for unrestricted functions or for functions of integral order only.
This distinction
3-1]
BESSEL FUNCTIONS
39
Let us now construct a solution of (1) which is valid near the oiigin; the form assumed for such a solution is a series of ascending powers of z, say
y= 2
m-0
+m cm z* ,
be determined, with the pro-
coefficients c m are to
differential operator
will
be called V,
w+s+*-*
is
It
V, 2
cm z* +m
= 2
c w {(a
+ m)
-i/2 }s" +m
+ 2 cm z a+m+a
namely
The expression on the right reduces to the first term of the first series, c (a8 Vs) za if we choose the coefficients cm so that the coefficients of
,
2 2
=0 =0 =0
(3)
c,{(a
+ ra) -v }+cM_2 =
2
If,
we have
2
m-0
cTO ^+'tt
=c
(aa -i/2 )2
a
.
From
values of
be a solution
of (1) only
z.
=+
for exceptional
consider the mth. equation in the system (3) written in the form
cm (o
Now
when
w > 1.
It can
be
- v + m) (a + v + m) +
Cm^
negative integer, that
is
Cm-*
= 0,
and
so it determines cm in terms of
greater than 1
unless
a-i>ora + i/isa
integer (when
a= v)
(o
or unless 2v
unless
2v
We
3*5),
*
moment
1,
and then
When
+ m) p*
2
when
2, 3, ....
now
the constants a and c m have been determined by the following analysis, the series
is easily seen to be convergent and differentiable, so that the formal procedure actually produces a solution of the differential equation.
>
40
[CHAP.
and that
Ill
= 0,
c 2Ml is ex-
pressible in terms of c
c"*
by the equation
(-r*
(a-v + 2)(a-v + 4,)...(a-v + 2m)(a+v +
of equations (3)
is
2)(a
The system
from (4) that
(5)
is
now
satisfied;
c zv
1+ 2
-i
(-)" (**)""
(v
+ 1) {v + 2)
If
. . .
(1).
we take
+ m) a= v, we
(v
obtain a second
formal solution
(6)
c
'z-"\i+ s
,-i
'
(-)m (^y
m\(- v + 1) (- v + 2)
...
(-v + m)]'
In the latter, c has been written in place of c because the procedure of obtaining (6) can evidently be carried out without reference to the existence of (5), so that the constants c and c ' are independent.
,
Any
values independent of z
may be
and
'
but, in view of the desirability of obtaining solutions reducible to v -* n, we define them by the formulae *
(')
i \ 2T(v+l)' and (6) may now be
i
Jn (z) when
co
= oirv..
2-T{-p + l)'
(-) w (i-g)~" +m
The
series (5)
written
|
,=o
*
(
m\r(v + m +
l)'
m!r(-i/ + m + l)'
In the circumstances considered, namely when 2v is not an integer, these series of powers converge for all values of z, (z = excepted) and so term-by-term
differentiations are permissible.
The operations involved in the analysis f by which they were obtained are consequently legitimate, and so we have obtained two solutions of equation (1). The
first
of the
is
first kindX',
is
denoted by
(z).
the symbol
(z).
Since v
the present, 2v
is
evidently J_
(z) is defined
by the equation
3-6.
'
'
3*11]
BBSSBL FUNCTIONS
solution of Bessel's equation has been given
41
An interesting symbolic
form
by Cotter*
in the
[l+zv D- 1 z- 2
where Dsd/dz while
"- 1
),
and
B are constants.
may be
[D(zD~2v)+z]zv y=0,
[zD-2p+D~ 1 z]z"y= -2vB, zD (z~ vy) + z~ 2v D~ V+V= - 1vBz- % \ x tv z- vy+D- z-^- x D- x z v+l y=A +Bz~
which gives Cotter's
3'11.
result.
is
two cases of Bessel's generalised equation were temporarily omitted from consideration, namely (i) when v is half of an odd integer, (ii) when v is an integerf It will now be shewn that case (i) may be included in the general
In
3*1,
.
v.
When
where r
If
(1) ' v
is
v is half of
an odd
integer, let
S = (r + $r,
a positive integer or zero.
we
take a = r
that
ci
(m>l) '
v
,,
and so
(2)
Ctm
is
( y q
2.4...(2m).(2r + 3)(2r + 5)...(2r +
27/i
+ l")
by r
which
If
the value of
c^ given by
c
3'1
when a and
v are replaced
+ \.
we
take
1
-2~r"+4r(r + f)'
we
m!r(r +
m + f)'
Jr+ ^(z),
so that the definition of
which
is
naturally denoted
by the symbol
31
If,
however,
we take a = r
\cm
(S) (3)
foi.l(-2r)
before, c lt c3 ,
cv -i are all zero,
=0,
(m>1)
is
As
and
is
...
+ Czr
(-)w
0,
Car+r>
by an arbitrary value of
when
ra
> r,
C2m+i
=
(2r
-r c
2r
+i
.
1)
...
(2m- 2r)
* Proc. R. Irish.
Acad. xxvn. (A), (1909), pp. 157161. f The cases combine to form the case in which 2v is an integer.
42
If
[CHAP.
Ill
Jv (z)
be defined by
2-"-*
3*1 (8)
when v r
\,
the solution
now
con-
structed is*
Co
It follows that
Jv (z)
is
necessary
is
when
v=(r + );
that the
negative root of the indicial equation gives rise to a series containing two
arbitrary constants, c
and
c^+i,
i.o.
equation.
3*12.
A fundamental system
It is well known that, if yx and ya are two solutions of a linear differential equation of the second order, and if y( and y2 denote their derivates with respect to the independent variable, then the solutions are linearly inde'
pendent
if
y* y*
yl
does not vanish identically; and
then, either one of the
if the Wronskian does vanish identically, two solutions vanishes identically, or else the ratio of
is
a constant.
identically,
form
cx
yx
+ d y2
where
cx
and
ca are
8K{yi,y.},
when
it is
We now
If
proceed to evaluate
we
V v J_ v {z) = 0,
which
V v J,(z) = Q
results,
we obtain an equation
may be
[zm{j z j
* In
Sci. di Torino, xxvi. (1821), pp.
v {z),
j-^yi-q,
delle
connexion with series representing this solution, see Plana, Mem. della R. Aeead.
519538.
( 23), p.
Sci. Math. n. lfr Proofs of the theorems quoted in the text are given by Forsyth, Treatise on
74.
3*12]
BESSEL FUNCTIONS
43
a)
where
m[j
G is
C,
v {z),
J-Az))=-,
z
a determinate constant.
To evaluate we have
we observe
that,
when
v is not
an
integer,
and
\z\ is
small,
"
{z)
l>TT)
{l
m>5
W>
1 )}
and hence
-.
V
'
If
we compare
is
right which
(2)
this result with (1), it is evident that the expression on the 0(z) must vanish, and so*
m
is
= _ *!E^r.
TTZ
Since sini^r
/_ {z)
Jv (z),
(2),
31
(1).
When
an
integer, n,
we have seen
that,
21
and when v
is
made equal
to
in 3*1 (8),
we
find that
nK)
Since the
first
mZ
m\r(-n + m + Ty
the series
is
n terms of the
easily reduced to
( )n Jn(z)y so that the two definitions of J-n(z) are equivalent, and the functions Jn (z), J_n (z) do not form a fundamental system of solutions of Bessel's equation for functions of order n. The determination of a fundamental
system jn this case will be investigated in
363.
Jv (z) is defined, for all values of v, by the and J (z), so defined, is always a solution of the equation Vt/ = 0. When v is not an integer, a fundamental system of solutions of this equation is formed by the functions Jy (z) and /_ (z).
To sum
up, the function
expansion of
31
(8);
a basic number []
is
by
Gamma
function
Tp
(u
+ l) = [u].rp (v).
rp
The
by replacing the numbers which occur in the by basic numbers. It has been shewn that very many theorems
p. 104.
This result
z
-*
<x>
making
is due to Lommel, Math. Ann. iv. (1871), and using the approximate formulae which
He
by
will be investigated in
Chapter vn.
44
[CHAP.
Ill
concerning Bessel functions have their analogues in the theory of basic Bessel functions, but the discussion of these analogues is outside the scope of this work. Jackson's main
results are to
6572
be found in a series of papers, Proc. Edinburgh Math. Soc. xxi. (1903), pp. xxn. (1904), pp. 8085; Proc. Royal Soc. Edinburgh, xxv. (1904), pp. 273276; Trans. Royal Soc. Edinburgh, xli. (1905), pp. 128, 105118, 399-408; Proc. London Math. Soc. (2) I. (1904), pp. 361366; (2) II. (1905), pp. 192220; (2) ill. (1905), pp. 123.
;
The more obvious generalisation of the Bessel function, obtained by increasing the number of sets of factors in the denominators of the terms of the series, will be dealt with in 4 4. In connexion with this generalisation see Cailler, Mem. de la Soc. de Phys. de
#
Geneve, xxxiv. (1905), p. 354; another generalisation, in the shape of Bessel functions of two variables, has been dealt with by Whittaker, Math. Ann. lvil (1903), p. 351, and Peres,
170.
3'13.
The
closed
which defines
domain
when
\v\
N and
\z\
is
-I*
m(v + m)
whenever
m (m A )
r
8
being excepted)
an analytic function of z for all values ofz(z = possibly an analytic function of v for all values of v. An important consequence of this theorem is that term-by-term differentiations and integrations (with respect to z or v) of the series for Jv (z) are
Hence f
Jv (z)
is
and
it is
permissible.
An
(i)
W=tvt)( 1+ >
|
where
and
+l
is
{jJi]}v+2 numbers v + 1
|
6\
<exp
1'
1,
|,
1+3|,
....
This result may be proved in exactly the same way as 2'11 pared with the inequalities which will be given in 3*3.
Finally, the function
z",
(5)
it
should be com-
which
is
a factor of
Jv (z),
* Bromwich, Theory of Infinite Series, 82. + Modern Analysis, % 5-3. % Math. Ann. lii. (1899), p. 230; Nyt Tidsskrift,
(1902), p. 494.
ix.
Math. Ann.
lv.
3-13, 3-2]
tion.
BESSEL FUNCTIONS
define
it to
45
is
We
given
its principal
value so that
ir < arg z
When
it is
ir.
values of arg
z,
mention
will
3*2.
for
(1)
^(fJ + ^W-v/rW z
J _ (z)-Jv+1 (z)=2J,'(z),
v
l
(2) (3)
(4)
-vJ
{z)
= - zJv+1 (z).
2*12, for
These are of precisely the same fonn as the results of being the substitution of the unrestricted number v
the integer
first
that
it V - A v
(-)j*h
**
7\.
_ \m ^2 v
+ai
Zo2"-
1+im
.m\r{v + m)
on the
left,
we
In like manner,
dz
i \
T rJ z-k
'WJ-rfj mt 2".i!r(v +
=1
oc
\\-A 5
(-) m z*m
l)
/_\r^2wi
2 K+2m-i
W _ 1)! T
\m+i n+i
(v
(l/
+W+
1)
/
.
+m+1 w = 2'
ml r
w + 2)
whence
(4)
is
may be
tracting (3)
*
and
when
StUdien Uber die BesseVschen Functionen (Leipzig, 1868), pp. 2, 6, 7. Formula (3) was given v is half of an odd integer by Plana, Mem. delta R. Accad. delie Sci. di Torino, xxvi. (1821),
p. 533.
46
We
(5)
can
generalised formulae
>
(AJ [zvj
W
{z)]
= *-~ j- <*>'
(6)
(^)
obtained
all
by repeated
differentiations,
when
Lommel
integral
The formula
of Bessel functions.
By
and
(7)
expressing
/_!
0) and
J_
x
(z) in
(z)
by
(3)
(4),
we can
J. (z) J^ v
/vtv + JL T-/\r/\2 = - J
(z)
(z)
v .,
(z)
312.
aud
(2) is that, if
An
(8)
Q y () = J"2 (2),
then
<?v-,w-^ + i(^y;wi.
was discovered by Lommel, who derived various consequences of it, Studien Functionen (Leipzig, 1868), pp. 48 et seg. See also Neumann, Math.
this formula
Ann.
111.
(1871), p. 600.
and imaginary parts of the function J +M (x), where v, /* and x been discussed in some detail by Lommel J, and his results were have real, are subsequently extended by B6cher.
The
real
VtH.{x)
and
v>ll
(x)
by the
Lommel
(1)
dx*
~ {K
Vi
*(a)
iS(x)}
+ 2(fr)
(2)
- 0,
w+ltM (x)
.
(3)
* See, e.g.
S, +I>M (x)
Lommel, Milnchener Abh. xv. (18841886), pp. 644647. t Math. Ann. iv. (1871), p. 105. Some associated formulae are given
X Math. Ann. in. (1871), pp. 481486. Annals of Math. vi. (1892), pp. 137160. The reason for inserting the factor on the right
I!
in 3*63.
is
will
be
established in 3-3.
3-21, 3-3]
BESSEL FUNCTIONS
47
with numerous other formulae of like character. These results seem to be of no great importance, and consequently we merely refer the reader to the memoirs in which they were published.
by Boole* many
years ago.
type.
We
(1)
It
shall
now shew
that,
, then
when 2v is a positive integer (zero inon the right is a solution of Bessel's equation and expression was adopted by Lommelj as the definition of Jv (z) for positive
;
values of v
+ \.
that the function, so denned,
is
and that
(1).
it satisfies
defined
Jv (z)
...
by suc-
cessive applications of 3 - 2
To deduce
trans-
(-y*Q*Y
I> + m + l)
provided that
series
l= i
(2wz)!
converges uniformly with respect to t throughout the interval (0, ), and so it 1 may be integrated term-by-term on adding to the result the term for which
;
* Phil.
Tram, of
the
p. 239.
Tripos, 1894.
t Journal de VEcole R. Poly technique, xn. (cahier 19), (1823), pp. 300 et teq., 340 et $eq. Strictly speaking, Poiason shewed that, when 2v is an odd integer, the expression on the right
multiplied by Jz is a solution of the equation derived from Bessel's equation by the appropriate change of dependent variable.
t Studien
liber die
et seq.
48
[CHAP.
Iir
m = 0,
<"~*(1
t^dt,
which
is
convergent,
we
find
that,
when
whence the result stated follows by making the substitution t = sin2 6 and using the fact that the integrand is unaffected by writing ir 6 in place of 6-
The simplest procedure seems terms omitted and integrate by parts, thus
(2m)!
the analysis necessary to establish the last equation is a little to be to take the series with the first two
m _2
Jo
m=2>'
+i
(2m)!
./
Jov+iU
/
(2mj!
(1
'>
J*
j*
+**ui
(2m)i
(1
The interchange
of the order of
summation and
On
m0, m = 1
J- <" "
I*
G - ' dt
<*>
cos <*>
*
*
<3>
* <*> -
nMm fj
2
- ,r * cos (zt)
<5)
(z)
r(-
+i)r( C'
<6 >
J " =
i>^f)Ta)J.
1
>0,,sin8
"^
namely
(*>
= ^TlVffe
^s
^'
sometimes usefuiy^t
valid,
only
when R{v)>
3*31]
BESSEL FUNCTIONS
(4)
49
by Nielsen*
An expansion involving Bernoullian polynomials has been obtained from with the help of the expansion
f-*-(i-.-)i
'= in which
<f> n ()
rfl
'1
*" +,tf+1)
(n
+ 1)!
[Note.
Accad. delle
Integrals of the type (3) were studied before Poissou by Plana, Mem. delta R. Sci. di Torino, xxvi. (1821), pp. 519 538, and subsequently by Kummer,
Journal
Math, xn. (1834), pp. 14-4 147; Lobatto, Journal fiir Math. xvn. (1837), pp. 363371; and Duhamel, Cours d' Analyse, II. (Paris, 1840), pp. 118 121.
filr
(2),
- v2 )* cos vx
dv,
was investigated by Lommel, Archiv der Math, und Phys. xxxvu. (1861), The converse problem of obtaining the differential equation satisfied by
pp.
349360.
zv
f*
{v-af- (v-P)"- 1 dv
was also discussed by Lommel, Archiv der Math, und Phys. XL. (1863), pp. 101126. In connexion with this integral see also Euler, Inst. Cole. Int. II. (Petersburg, 1769), 1036,
and
3*31.
From
(i)
33 (6)
it
\,
then
!-/.<*)!
rj+ffrft-j
/' exp
'
'
sin "
ede
By
3-*2
(1)
and
(4),
we deduce
in a similar
manner that
(2)
|JpW
<TT
}exp|/W|
J]
<-|<-i>,
(>-*).
(3)
i^wK^lJi + j-^y^i/wi
using the expressionf
z for
By
J^ (z)
it
may be shewn
that
(1) is valid
when
= \.
These inequalities should be compared with the less stringent inequalities obtained in 313. When v is complex, inequalities of a more complicated character can be obtained in the same manner, but they are of no great importance.
*
Math. Ann.
Modern
*
50
3*32.
[CHAP. HI
The
(1 >
(cos *)
rfft
in
which
Gn v (t)
2at + o
-"
)
in
ascending powers of
R (v) > \
and n
is
the formula is valid when a, is due to Gegenbauer* any of the integers 0, 1, 2, .... When n = 0, it obviously
Jn +
(z)
= (_i)
(ff/
we
Pn (cos 0) sin dd
To prove Gegenbauer's
'
'*
and integrate
times
by parts
the result
is
^
Now
,,()
(-2i)nr(,
+ ,,4-^ra)J_ 1
!
e
1
<**
pCn "(0,
it is
known thatj
(- 2) n T ( v + n +
$)
dn (l- t*)*+-l
whence we have
T (2i/)
(1
-)-*
(3)
J
t/
(2) (
e
1
(1
I;
c w<.
is
evident.
is
this
=h
aid of the
The reader
recurrence formula
* Wiener Sitzunasberichte, lxvii. See also Bauer, (2), (1873), p. 203; txx. (2), (1875), p. 15. Milnchener Sitzungsberichte, v. (1875), p. 262, and 0. A. Smith, Giornale di Mat. (2) xn. (1905), The function Cn " (t) has been extensively studied by Gegenbauer in a series of pp. 365 373.
memoirs
in the
Wiener Sitzungsberichte
some
of the
more important
results obtained by
him
are
given in Modern Analysis, 15'8. + Proc. London Math. Soc. xxxv. (1903), pp. 198-206.
See 6-17, 10
5.
t Cf. Modern Analysis, 15*8. Proc. London Math. Soc. iv. (1873), pp. 100, 263.
"
3*32, 3*33]
BESSEL FUNCTIONS
namely*
*"'
51
***
i*"* denotes
;
(7(^)) - r(^i^+V^ (P )
is
in
which
due to Filon,
Phil.
Mag.
(6) vi.
(1903), p. 198
the reader.
3*33.
It has
(1 )
R (v) > 0,
cos ^r)]
3-
Jv (r ) =
3
*
I
I
""1
^ sin
0d^d0,
r = Z* + z* - 2 Zz cos < and Z, z, <f> are unrestricted (complex) variables. This result was originally obtained by Gegenbauer by applying elaborate integral transformations to certain addition formulae which will be discussed in Chapter xi. It is possible, however, to obtain the formula in a quite natural
where
in the
geometry of the
when
= 0,
is
Jv (Z) =
we proceed
unit sphere
this point
TT 1
%-r-r
and
(* eiz sin8"
as longitude
(V) Jo
f sin *"
8
.'
yjr dyjr,
to regard
;
yfr
we denote
(I,
by
the direction-cosines of the vector from the centre to m, n) and the element of surface at the point by dto.
then transform Poisson's integral by making a cyclical interchange of the coordinate axes in the following manner
:
We
J* ()
e ^FJ-\ I"!*
iv *
*9
sin2 '
sin8-- 1 fd0dyfr
ttJ
{v))} m> o
[(
(W
= -^A
* It is supposed tbat
Ofy d0-
t Wiener Sitzungsberichte, ucxiv. (2), (1877), pp. 128 129. t This method is effective in proving numerous formulae of which analytical proofs were given by Gegenbauer ; and it seems not unlikely that he discovered these formulae by the method
in question
;
cf.
12*12, 12-14.
The
xiii.
The symbol
which
jj i is positive.
m ^ means
52
[CHAP.
Ill
an integral periodic function of yjr. and so the limits of yfr may be taken to be a and a + Itt, where a is an arbitrary (complex) number. This follows from Cauchy's theorem.
integration with respect to
Now
We
thus get
(w)
(ktffY
-,
/'i"'/" a
+2w
e iv sin
e cos
/.
* cos2 " -1
sin Bdslr d$
TTl
(v)Jo J*
W
'
= -te K
irl
Wio
o
I *l
.'o
cos2 ""1
sin ddylrdd.
We
so that
now
define a
r cos
= Z z cos
exp
[i
<f>,
r sin
z sin
cos
yfr
<f>,
(a)
(Aw)"
-
/*" f 2 "
1
I
L,
(Z z cos
<f>)
sin
iz sin
2
<f>
sin
t/t
sin 0]
con
'- 1
6 sin Bdyfrde.
The only
difference
between
v
this formula
Jv {m) =
is in
l, 7Tl (v) Jo
exp [iw
.'
sin
sin 0ety> d0
When
(f>
m,
.
%{Zz cos
The
last expression is
cos
I,
yfr
sin
0.
iZ cos
iz (cos
<j>
cos
+ sin
<f>
sin
cos
\jr),
(IwY
y,,
ff ex P [*^ cos
** ( C(>s
<}>
cos
+ sin
<f>
sin
2
cos
yfr)]
2"
sin
"" 1
^ sin
0d\frd0,
is
established.
The
manner of
I'
Acad, des
1789, p. 372,
3*4.
We
when v is half of an odd integer, the function Jv (z) is expressible by means of algebraic and trigonometrical functions of z.
It will
is
appear later
;
not so expressible
such a value, then ,TV (z) but of course this converse theorem is of a much more
(
474)
that,
when
v has not
is
now about
to be proved.
3*4]
[Note. Solutions in
finite
BESSEL FUNCTIONS
terms of differential equations associated with
;
53
Jn+
(z)
were ob-
it
in.
(1762
;
)t
(z) is
while
Terns.
2*
Jn+i
(.s)
was solved
in finite
1823 [1820], pp. 245257 and Mecanique Celeste, v. (Paris, 1825), pp. 8284 della R. Accad. delle Sci. di Torino, xxvi. (1821), pp. 533534; by Paoli, di
Fis. (Modena), xx. (1828), pp. 183188; and also by Stokes in 1850, Trans. Camh. Phil. Soc. ix. (1856), p. 187 [Math, and Phys. Papers, n. (1883), p. 356]. The investigation
which will now be given is based on the work of Lommel, Studien Fimctionen (Leipzig, 1868), pp. 51 56.]
liber die
BesseVsehen
It
is
then,
by
^(^ST.^d-^
n
!
dt
\Zv '_!
nly/ir
\_
r+x r=0 z
;
dtr
J_i'
t*)
when we
since (1
n is a polynomial of
To simplify the last expression we observe that if dr (l -t2 n /dtr be cal) culated from Leibniz' theorem by writing (1 -t2 n = (1 - t)n (l + t) n the only ) term which does not vanish at the upper limit arises from differentiating n
,
times the factor (1 - t)n and therefore from differentiating the other factor r n times; so that wo need consider only the terms for which r^n.
,
Hence
and similarly
[tOfff]
=(-), 0.
-5^
[*&=!*j
_(_>-., Q, .
J^
/"_
v
It follows that
Um
* KZ)
Jir
rn
^+1 .(r-n)!(2n-r)!
2
V+l Oan-r V L
-,
-
~\
,.-s' +1
.(r-w)!(2n-r)!
'
-J [>
*~"<n + r)l
(- t-r-n-x (n
+ r)
;-|
This result
(2)
may be
WL
cos^-i^r' _o
2
(-M* ++!)!
(2?is
1
'
+ l)!(n-2r-l)!(2^'+'
this
formula
143.
J
54
[CHAP.
Ill
In particular we have
(3)
Jj (*)
= ()
sin
m,
(z)
(-)
-^ - cos z)
power
is
(1),
-n
-(n
+ r)l
*
(~i)r n .(n+r)\
,~ rl(n-r)\(2zy
rl(n-r)l(2zy
we can express
\zdz)
z
by
iz
,
and
;
so
we must have
,_"o
(y-(~ + r)
r (n
!
- r)
(2*)
y.W<*
WW
V****.
*
'
for, if not,
where fa
(z)
1/z;
impossible*.
Hence
it
r -o
r!(n-r)!(2^"
,=
r!
(w-r)! (2-) r
= (-r (2tt>
Consequently
(4)
*+>
{*-*
(
J
i)
(^).
j_j (*>i
= (-)(27r^./_n_
J
</_,
| 2
rS(n
r)!(2
* Cf.
p. 51.
From
m-_r(j)
J,
(-r <**)
(
,to4-# ...(m-i)'
it is
obvious that
(e)
cos
z.
3*41]
BBSSEL FUNCTIONS
55
and hence
(5)
j^ w -(i)*r
00 .(,
+ j,?
*
77
yr^L!^_
v
'
r-o
(2r+l)!(>i-2r-l)!(2)2
+1
_
In particular, we have
(6)
We
The explicit expression of a number of these functions can be written down from numerical results contained in a letter from Hermite to Gordan, Journal fur Math. lxvi.
(1873), pp.
303311.
3*41.
is
Functions of the types /(n+j)(s) occur with such frequency in various branches of Mathematical Physics that various writers have found it desirable to denote them by a special functional symbol. Unfortunately no common notation has been agreed upon and none of the many existing notations can
be said to predominate over the others. Consequently, apart from the summary which will now be given, the notations in question will not be used in this work.
In his researches on vibrating spheres surrounded by a gas, Stokes, Phil. Trans, of the clviii. (1868), p. 451 {Math, and Phys. Papers, iv. (1904), p. 306], made use of
Royal Soc.
the series
"
n( + l ) 2.imr
(-!)( + !)
( + 2 ) " 2. 4. fair) 2
K "'
which
is
d* dr2
d
dr
n(n + l)
r2
Sm e ~ *</ (r) + Sn
;
'
'*/.
" *'),
fa
is'
where
<S'
and
Sa
'
so that
operator
dr1
+ 2 d 4-,a
r
M (" + 1 )
r2
'
dr
and by the
partial operator
'' l 2 cr*
2 d
v)
dff)
55 c6
\ {
sin 6 52}
+ m2
In this notation Stokes was followed by Rayleigh, Proc. London Math. Soc. iv. (,1873), 103, 253283, and again Proc. Royal Soc. i.xxii. (1903), pp. 4041 [Scientific v. (1912), pp. 112114], apart from the comparatively trivial change that Rayleigh would have written/,, fair) where Stokes wrote/, (/).
pp. 93 Papers,
'
56
[CHAP.
it
Ill
necessary to take
Sn = '
n+
)
Sn
and then
t>
ir
It follows
from
3-4 that
rn+1
i (ir\ Q?-' =
n [e- ir i +
/
( %
r \ n e~
ir
,
-4-
\ rcrj
and that
ir In order to have a simple notation for the combinations of the types e* /(tr) which write convenient to are required for solutions finite at the origin, Lamb found it
** ~ l ~ 2(2n + 3)
in his earlier papers, Proc.
+2
4.
(jSh
+ 3)
(2/i+5)~
"*'
(1884), pp.
pp.
pp.
London Math. Soc. xm. (1882), pp. 5166; 189 212; xv. 139149 xvi. (1885), pp. 2743 Phil. Trans, of the Royal Soc. clxxiv. (1883), 519549; and he was followed by Rayleigh, Proc. Royal Soc. lxxvii. A, (1906), 486499 [Scientific Papers, v. (1912), pp. 300312], and by Love* Proc. London
; ;
Math. Soc. xxx. (1899), pp. 308321. With this notation it is evident that
.=p^,=(-)-...(
!
!.
+ .).( i)'^.
Soc.
Subsequently, however,
ingly in his treatise on
pp. 11
Lamb
found
it
Hydrodynamics and
London Math.
xxxn.
(1901),
(d
l)
"" 4(2 + 3) (S + 5) ~ J
e~"
-j-\
=*n(*) -l'V'i(*)
so that
* n () =
>t+ ^ pr+i
"
, '
= vnw<A
()
to Y~
while Rayleigh, Phil. Trans, of the Royal Soc. cciii. A, (1904), pp. 87110 [Scientific Papers, Love, v. (t912) pp. 149161] found it convenient to replace the symbol /(*) by x*(*)Phil. Trans, of the Royal Soc. ccxv. A, (1915), p. 112 omitted the factor (-) and wrote
dye-**
/rf\"sin2
while yet another notation has been used by Sommerfeld, Ann. der Physik
und
Chemie, (4)
xxvm.
(1909), pp.
665736, and two of his pupils, namely March, Ann. der Physik und xxxvu. (1912), p. 29 and Rybczyriski, Ann. der Physik und Chemie, (4) xli.
;
this notation is
^(.HW^-^
C(*)(MH'+i (*)
and
it is
(-ai)"T'
on
electric
+ (-)* tf--*(*>l
waves which was the
E n (z)
as
- ) . 1
...
(2n - 1)
^
J
but, as
82
^ n (z)
by Heine, Handbuch der Kugelfiinctionen, i. (Berlin, 1878), to be twice the expression on the right in his treatise, but
not in his memoir, Journal filr Math. lxix. (1869), pp. 128
141.
3*5]
BESSEL FUNCTIONS
57
Sommerfeld's notation is a slightly modified form of the notation used by L. Lorenz, who used vn and vn +( ) n iwn in place of >frH and fn see his memoir on reflexion and refraction
;
of light, K.
(1898),
I.
3*5.
It has
been seen
whenever v
is
Jv {z)
and
./_(z).
When
is
J_ n (z) = () n Jn (z)is
linearly independent of
will give
Jn (z);
JH (z)
The
the
now be
full details
by BOcherf.
method of constructing HankeFs solution was discovered by Forsyth method of Frobenius, Journal fiir Math, lxxvi. (1874), pp. 214235, for dealing with any linear differential equation. Forsyth's solution was contained in his lectures on differential equations delivered in Cambridge in 1894, and it
alternative
his procedure is based on the general
An
has since been published in his Theory of Differential Equations, iv. (Cambridge, 1902), 102, and in his Treatise on Differential Equations (London, 1903 and 1914), pp. 101
Chapter
vi.
note
1.
and
if
n be any integer
(positive,
j {z)-(-yj_ v {z)
v
is
and
this function
vanishes
when
n.
Consequently, so long as v
n,
v
the function
v
J {z)-(-TJ_
v
is
(z )
n
;
= n.
We
and we
shall
now evaluate
lim
^-(-y^-fr)
shall
shew that
it is
* Math. Ann. i. (1869), pp 469472. t Annals of Math. vi. (1892), pp. 85 90.
65-71
X The essence of Hankel's investigation is the construction of an expression which satisfies the equation when v is not an integer, which assumes an undetermined form when v is equal to
the integer
limit
when
v-*~n.
w.
B. F.
58
order n and that
[CHAP.
Ill
independent of
Jn (z)
so that it
may be taken
J, (*)
- (- V J- v (*) _ Jv () - Jn (M) v n v n
dv
l
;
)n
(
du
.
as v
- n,
Hence
^ w
n
n.
exists; it is called
To
we
distinguish
it it
denote
may be
(1)
YH (s) =
*.(*)now
to be
j
lim
v-*v
{z)-(-yj_ v ( Z )
v
and
also
(2)
( K
Vt dJ"-,(fY
>
dv
It has
shewn that Y
(z) is
J v (z)
the
a matter of indifference .
Hence the
equations
V v Jv {z) =
with respect to v
-
may be
written
d*
dJv (z)
dv
d bJv (z)
dz
dv
^ dJ v (z)
dv
_n
dz%
When we
combine the
we
find that
dJv {z)
dv
*
K
}
dJ_ v (z)
dv
*2v{Jv
(z)-(-rJ-Az)}>
it is
The reader
3-1.
+ See
It is
as total differential
coefficients while differentiations with respect to v are written as partial differential coefficient-.
Of course,
in
many parts of the theory, variations in v are not contemplated. Yn (z), which was actually used by Hankel, is used in this work
ljir
to denote a
( 3*54).
3-51]
so that
BESSEL FUNCTIONS
59
Now make
functions of
v,
v --n.
and so we have
where v
is
to
We
so that
It is to be noticed that
-W
'
=
whence
(4)
M ^
im
J-A*)-(-YJAz) fi + n
due
to
Lommel *,
Y_ n (*)=(-)Yn (s).
Again,
while, because
Jv (z) is a
dv
monogenic function of v at v
0,
we have
dJ. v (zy
J-0
^(-^Jv-O
dv
J"""'
and hence
it
follows that
A result equivalent to
3*51.
this
The expansion of
Y
is
(z) in
an ascending
series.
Y (z),
it is is
con-
simpler
more compact.
We
SUidien
1
iiber die
what
is
BesseVschen Functionen (Leipzig, 1868), p. 87. Lommel actually proved sometimes called Neumann's function of the second kind. See 3-58 (8).
n. (Paris, 1840), pp. 122
t Court
d Analyse,
124.
60
[CHAP. HI
(*)
= 2["I
^w-^'Tt
is
x ]
lo g
fa) ~ J- log r (v + m +
1) 11
=2 2
where
yfr
denotes, as
Gamma-
function*.
Since
0< ^ (i + 1) < m
in.
when w=l,
is
2, 3, ...
may be
^(m+1)
is
replaced by
The convergence
The
(1)
Y.(*)-2Sl
'J**'
.
{log(^)-^(m + l)},
(s)
log (*)
- X
(3)
Y. () -2
fry
yy * -2 JS^ yy
^
(m
+
1)J
I+ g +
1 + raj !
The reader
iY.(*)
is
+ (log2- 7 )J.(*)
The expansion
of
this function is
Neumann,
This function was adopted as the canonical function of the second kind of order zero by Theorie der BesseVschen Functional (Leipzig, 1867), pp. 42 44; see 3 57. But the series was obtained as a solution of Bessel's equation, long before, by Euler \.
own notatiou
is
V~
X
n*
\JW X + 1.8.27'^
.
1.8.27.64^
tT 3
('
+et -
+A +a
*
1 ( \
- 3- r + JL- X2 _
nn
1
.
_L_
1.4. 9
C
4m'
g + 1.4.9.
.*
Ion8
- etc
Ix
nn
X+ 1.4n* X
Ch. xn.
9 f
+1
16
*'
integer,
Modern
Atialy-iis,
It is to be
remembered
that,
when
m is a positive
then
*U)=-7,
t Inst. Calc. Int.
11.
V'('
1 )=
+ 2+ - +-->
233235.
See also Acta Acad. Petrop.
v.
....
3*52]
BESSEL FUNCTIONS
He
61
22=5.6-4.2,
w
<r
2(1
\1
+* 1+...+!) = ^, mj
m
(r
m+ i = (2m+l)<r-m
ii
m.1
3*52.
The expansion of
Yn (z)
in
an ascending
series
and
the definition
of
<*) We
shall
of the
fl
(^),
where n
any positive
integer.
[Cf.
equation (4) of
It is clear that
?^.<*>dv
m =()dv
m+
w=0 m!r(i/
ra!(w
+ w+l)
l
6V2
n
/J
'
/;
OT =o
+ ra)!
6V2
rv
is
when
i/-*-w,
where
is
to say
i
a)
The
of
|^>] o"
L
J^=n
_li ^ + n+wj
[3/_(.s)/3i/],,_ n is a little more tedious because of the pole (- v + m + 1 ) at v = w in the terms for which m = 0, 1. 2, ., n 1. We break the series for /_ (z) into two parts, thus
evaluation of
yjr
" V
'
TO
l m!
r(-i/ +
m + l)
*! r(-i/
+m+
l)'
V(-v + m+l)
Now, when
[~
t
by
r (v in) sin (v m) it *
1
'
^
"
ra
<
n,
= [(J*)~ r+m r(i-m) {tt-i ^r m) sin (> ra)7r + cos (v m) tr tt' - (*)-+ r (n - m) cos (n - m) tt.
(i
Math. Ann.
i.
(1869), p. 471.
62
[CHAP. HI
pM*)1
v
+
,;.
that
is
to say
( **>
()
L
dp
_U(1)
m!
+(
x
{
>
.t m!(n + m)!
(m+1)},
lo g(i*)-"f
when we
replace
On combining
we have Hankel's
a
formula,
namely
w
=2
{
m=o
*!
m=0
w
w!(n + m)!
1}
7 + log g *)}/,(*)
-(^r
1
(n
~r,"
(h^
,
|l
(12
1
,111
m
1
w + m)
} {
term (m =s 0) of the
last
is
11,
12
d
Lommel*)
to write
(4)
%(*) =
-^
Z)
-Mz)logz,
so that
<5
>
when
^ (s)
is
on the right.
We
(6)
thus have
7l
(z).
The complete
contained a logarithmic factor) by Euler, Inst. Calc. Int. n. (Petersburg, 1769), 935, 936 solutions of this equation are
;
**
J (2aM),
x
#*
Yi (2aM).
;
(ibid.
solutions of
**
J2 (4a* **),
(4a* .v*).
3-53, 3-54]
3-53.
BESSEL FUNCTIONS
definition
63
The
o/Y v
(z).
Hitherto the function of the second kind has been defined only when its order is an integer. The definition which was adopted by Hankel* for unrestricted values of v (integral values of
2v excepted)
is
(1)
(*)
= 2ne
JAz)C0S
- J
-' (z)
.
sin2i/7r
This definition fails both when v is an integer and when v is half of an odd integer, because of the vanishing of sin 2vtt. The failure is complete in
the latter case
;
is
the definition of
3'5.
To prove
lim
(z)
ire"**
lim
-n J
/_ (z)~\ n v
\
J-AzY
~n
Jv (z)
J
(2)
It is
Yn
(*).
now
of that equation,
when
(i)
:
Y (z), defined either by (1) or by the limiting form a solution of Bessel's equation for functions of order v both v has any value for which 2v is not an integer, and when (ii) v is an
evident that
is
integer
the latter result follows from equation (2) combined with 3*5 (3). function Y (z), defined in this way, is called a Bessel function of the second kind (of Hankel's type) of order v ; and the definition fails only when
The
is
+1
an integer.
The reader should be
careful to observe that, in spite of the change of form, the qua function of v, is continuous at v = n, except when z is zero; and, in fact, Jv {z) and Y(z) approach their limits J (z) and Y (z), as v-*~n, uniformly with n respect to z, except in the neighbourhood of z = 0, where n is any integer, positive or negative.
Note.
function
Y (z),
3*54.
The
Hankel
( 3*53) was modified slightly by Weberf and Schlafli} in order to avoid the inconveniences produced by the failure of the definition when the
is
Math. Ann.
i.
(1869), p. 472.
f Journal fur Math, lxxvi. (1873), p. 9 ; Math. Ann. vi. (1873), p. 148. These papers are dated Sept. 1872 and Oct. 1872 respectively. In a paper written a few months before these, Journal filr Math. lxxv. (1873), pp. 75105, dated May 1872, Weber had used Neumann's
function of the second kind (see 3-57, 3-58). + Ann. di Mat. (2) vn. (1875), p. 17 ; this paper
is
64
[CHAP.
Ill
Weber
second kind
when
v is
an
integer).
<*\
Subsequent writers, however, have usually omitted this factor \tr, e.g. Graf and Gubler in their treatise f, and also Nielsen, so that these writers work with Weber's function.
The symbol
and so
to
it is
is,
Yv {z) to denote
after the
manner
of Nielsen J,
and to adopt
of Hankel's function of
certain formulae.
We
(1)
thus have
Yv (z) =
"
(jOcgg
'" --*--(*>
= cos^;
Yn (0
.
(2)
Yn (z) =
lim
,-*. n
J M^^JLrl-A^ = I
Sin VTT
7T
Schlafli's fuuction has been used by B6cher, Annals of Math. vi. (1892), 8590, and by McMahon, Annals of Math. vm. (1894), pp. 5761; ix. (1895), pp. 23 30. Schafheitlin and Heaviside use Weber's function with the sign changed, so that the function which we (with Nielsen) denote by Yv (z) is written as Yv (z) by Schafheitlin and (when v = n) as GM (z) by Heaviside||.
[Note.
pp.
Gray and Mathews sometiinesIT use Weber's function, and they denote symbol Yn
.
it
by the
(see 6*1)
which
i.
is
the
expression (with the factor \ir inserted) was actually given by Schlafii.
+ Einleitung in die Theorie der Bexsel'xchen Funktioiien,
number
it
index, thus
Y v (z), Handbuch
There
used by
and we
pp. 31
reserve
Neumann
( 3*58).
See, e.g.
(181)5),
Proc. lloyul Sue. liv. (1893), p. 138, and Electromagnetic Theory, n. (Loudon, 1899), p. 255;
made from his Electrical Papers, n. (London, 1892), TreatUe on Bessel Functions (London, 1895), pp. G5 66.
p. 445.
3-55]
BESSEL FUNCTIONS
in his later work, used
65
Lommel,
function
Neumann's function
86,
he used the
7rFM
where
(*)
Yn (z)
is
for the function much more complicated; see Julius, Archives Nterlandaises, xxviii. (1895), pp. 221 225, in this connexion.]
3'55.
The definition given by Heine* of the function of the second kind possesses some advantages from the aspect of the theory of Legendre functions it enables certain generalisations of Mehler's formula ( 5*7 1), namely
;
lim
to be expressed in a
Pn (cos $/n) = J
The
(0),
'
compact form.
to
function,
the symbol
equal to
Kn
(z), is
-\Yn {z);
by
Schlafli.
The use of Heine's function seems to have died out on the Continent many years ago the function was occasionally used by Gray and Mathews in their treatise t, and they term it On (z). In this form the function has been extensively tabulated first by AldisJ: and
Airey, and subsequently in British Association Reports, 1913, 1914 and 1916.
This revival of the use of Heine's function seems distinctly unfortunate, both on account of the existing multiplicity of functions of the second kind and also on account of the fact (which will become more apparent in Chapters vi and vn) that the relations between the
functions
cosine
JH (z)
sine
and
;
and
of
Jn (z)
and
On (z)
as canonical functions
It
is
com-
%ir sins as
canonical functions.
must
also be pointed
On (z) has
two writers, namely Heaviside, Proc. Royal Soc. and Dougall, Proc. Edinburgh Math. Soc. xvm.
Note.
been used in senses other than that just explained by at least liv. (1893), p. 138 (as was stated in 3*54),
(1900), p. 36.
error in sign on p. 245 of Heine's treatise has been pointed out by Morton, Nature, lxiii. (1901), p. 29 ; the error is equivalent to a change in the sign of y in formula
3*51 (3) supra. It was also stated by Morton that this error had apparently been copied by various other writers, including (as had been previously noticed by GraylT) J. J. Thomson, Recent Researches in Electricity and Magnetism (Oxford, 1893), p. 263. A further error
i. (Berlin, 1878), pp. 185 248. Treatise on Bessel Functions (London, 1895), pp. 91, 147, 242. t Proc. Royal Soc. lxvi. (1900), pp. 32 43. Phil. Mag. (6) xxn. (1911), -pp. 658663.
An
||
From
and also
for using
the historical point of view there is something to be said for using Hankel's function, Neumann's function ; but Heine's function, being more modern than either,
its
favour.
1 Nature,
66
[CHAP.
Ill
noticed by Morton in Thomson's work seems to be due to a most confusing notation employed by Heine for on p. 245 of his treatise Heine uses the symbol to denote the function called - \n J'q in this work, while on p. 248 the same symbol denotes - \n (Y -iJ ).
3*56.
(z)
and"Y v {z).
The recurrence formulae which are satisfied by Yv (z) are of the same form by Jv (z) they are consequently as follows
;
(1)
(2)
(3) (4)
(z),
zYv+1 (z),
replaced throughout by the
may be
3*2 (3)
and
(4) in the
forms
(*)}
{? Jj z
vrr,
(*)}
= - *' J-v+
>
(*)
we multiply
whence
Equation (4)
is
the formulae
{*- J
By
(z)}
{*- J. v
(z)}
= r- /__> (zy
(1).
we
The formulae are, so far, proved on the hypothesis that v is not an integer but since Yv (z) and its derivatives are continuous functions of v, the result of proceeding to the limit when v tends to an integral value w, is simply to
replace v
by
n.
ire rri
is
Again, the effect of multiplying the four equations by equal to 7re " 1,,ri sec {y l)7r, is to replace the functions
(
sec
vir,
which
Y by the
functions
Y throughout.
In the case of functions of integral order, these formulae were given by Lommel,
Studien
iiber die Bessel'schen
p. 87.
The reader
will find it
instructive to establish
them
from the
series of 3-52.
Neumann's
3*56, 3*57]
3'57.
BESSEL FUNCTIONS
67
Neumanns function of the second kind. The function which Neumann* adopted as the
it is
integrals of Poisson's type than the functions of the second kind which have
but this
is its
only merit.
We
first
<0)
(z).
The second solution of Bessel's equation for functions of order known to contain logarithms, Neumann assumed as a solution the
zero being
expression
J
where
(z) log z
+ w,
,
is
a function of z to be determined.
have
But, by
212 (11),
and
so,
2z
(z)
2z J,
(s)
= 8 2 {-T~'nJ2n (*)
have
since
2 and V
is
easily justified.
Hence a
is
n-l
2(-y->Jm (z)/n,
(0)
(z),
F< \z)
=J
(z) log z
(-)""1
^-^
/-*Oas+0, (the
it is
the origin),
solutions,
evident that
and hence
(z) is
series for w being an analytic function of z near J (z) and F (z) form a fundamental system of expressible as a linear combination of J (z) and
,0)
Yw (z); a comparison of the behaviours of the three functions near the origin shews that the relation connecting them is
(2)
*
F<> (*)
= J Y, (z) +
(log 2
- 7 ) Jo (*).
44.
Neumann calls
this function
of the second
is
kind
But, because.
is
un-
and it has not survived. t Neumann's function is distinguished from the Weber- Schlafii function by the position of the suffix which indicates the order.
desirable
68
3*571.
It
[CHAP.
Ill
{0)
(z).
h
is
eixcoa
<
log (x sin 2
<o)
dm
;
a solution of Bessel's equation for functions of order zero and argument x and subsequently Stokes obtained an expression of the integral in the form of an ascending series (see 3*572).
The
associated integral
f*
(4iz
cos2 0)
,0 >
dd
;
was identified by Neumann f with the function F (z) and the analysis by which he obtained this result is of sufficient interest to be given here, with some slight modifications in matters of detail.
From
2-2 (9)
we have
2
/*
(-)M /*(*)
nir J
and
so, if
changed,
of
n=i
=7rJ
cos (z cos 0) 7
f*
I
z
W=1
do
d0
from this result combined with Parseval's integral F (z), we at once obtain the formula
,0
(1
F<> (*)
= !/"
<*0,
is
obvious.
The change of the order of summation and integration has now to be examined, because 1n~ l cos 2n0 is non-uniformly convergent near 6=0. To overcome this difficulty we observe that, since 2 ( - ) n J.in {z)jn is convergent, it follows from Abel's theorem that
%
S(-)^.W/= lim I (-)Ji<)/n= lim * j /*% os (s C os0)?Wcos2 ^ca?. =l o^-i-o n=l M -*M5Tn-U0
* Journal de. VEcole 2?. Poly technique, hi. (oahier 19), (1823), p. 476. The solution of an associated partial differential equation had been given earlier (ibid. See also Duhamel, p. 227). Cowr d? Analyse, n. (Paris, 1840), pp. 122124, and Spitzer, Zeitsehri/t filr Math, und Phys. n. (1857), pp. 165170. t Theorie der Bessel'sr.hen Functionen (Leipzig, 1867), pp. 45 49. See also Niemoller, Zeitschriftfllr Math, und Phys. xxv. (1880), pp. 6571.
3-571, 3*572]
BESSEL FUNCTIONS
69
Now, since a is less than 1, 2 (on cos 2n6)/n does converge uniformly throughout the range of integration (by comparison with 2a"), and so the interchange is permissible that is to say
;
=1
2 - 2
/"*""
/
/"i*
I
Jo
cos(cos#) 2
lh"
/
n-1
n dO
,
1 = --
n Jo
cos
cos d) log (1
- 2a cos 26 + a2
flW.
Hence we have
n-l
(")* (*'
=_
lim
o-*.l-0
i "
*""
|
cos
(a
cos d) log (1
- 2a cos 20
a2 )
<J9.
y o
a-*-l-0 J
It is evident that
- 2a cos 26 +
log (1
j
- 4a sin2 6 ( 1 - a) 2 ^ 0,
a2 )
(z
and so
Hence,
'It
if
- 2a cos 26
A
(
cos 6)
when
0^6^ ,
\
we have
cos
I/,o
**
Jo
{log (1
6)}
,
d6
.
-A
i
/**"
\
~ ( |-22
I
a"cos20
+log (l/a>- 2
log (2 sin
6)\d6
J
io
*i
"
a< 1.
Hence, when
I
a< 1,
log (l/a)-*-0,
cos
(z
cos 0) {log (1
as a--l
0; and
2
n=j
Consequently
(-)w,7*(*) n
= _
I ( ** cos ii a-*i-o ,r y o
(*
CoS
tf)
dd
-1
/"*"
cos
(z
and the interchange is finally justified. The reader will find it interesting to deduce combined with 3*5 (5).
3*572.
this result
The
by Stokes \
in 1850
was -rf
+-
-r-
- m2 y =0, where
m is a constant. This is Bessel's equation for functions of order zero and argument imz. Stokes stated (presumably with reference to Poisson) that it was known that the general
solution was
/itr
o
d&.
The value
of this limit
t If 2
p. 42.]
is real,
A l\
if
not,
A < exp
(z)
j
'
70
It is easy to see that,
is
[CHAP.
Ill
by Stokes
it is
equal to
(imz)
+ 2D 2
m 2" z2n
,_
.,
f in
I
=o v* n )
J o
may
be
deduced that
-j*%o82^logsin^^=^^|^log2 + i
Neumann's
ff
(I
+ ^ + ... + i)}.
3*58.
definition of
(n)
(z).
The Bessel
by
Neumann*
(1)
in terms of
<0)
(z)
dYi
z
- nY^
(z)
= - zYW (*),
is
which is a recurrence formula of the same type as 2*12(4). It from this equation that
(2)
evident
FW(jr)-(^)(^)*FW().
satisfies
Now Y m (z)
the equation
and, if
theorem,
<3>
we get
'
*"
&T
is
''
{l)
(a)"
{Z)]
and so
{zlzf
^^
{Z)]
~ nYW
{zlz)
^^
{Z)
This equation
(4)
at once reducible to
V,Fw(5)-0,
and
so
(w)
(s) is
Again, (3)
may be
l
(*)}
- (2w + 2) z-*-
F<n+1
>
(*)
+ *~ n F(w
>
(2:)
= 0,
when
The function
is
undefined
order
not an integer.
is simplified
t The analysis
3*58, 3*581]
so that
BESSEL FUNCTIONS
71
dF(n+I >(*)
Tz
n+ 1 + -^t
F<+-> (z)
- F
(*)
= 0,
d7
(Z)
2
)
(z).
When we combine
formulae
(6)
F'"- 1 (z)
F< +1 (z)
>
=
=
F<"> (),
(7)
F<"--> (*)
dF< "'^ as
Consequently
(w
(*) satisfies
and
Yn
Jn (z),
F (s)
(s).
It follows from
357
(2) that
<
(8)
solution of the equation V H (y) = in the form of a definite integral, which reduces to the integral of 3*571 when =0, has been constructed by Spitzer, Zeitschrift fiir Math, und Phys. ill. (1858), pp. 244-246; cf. 3-583.
3*581.
Neumann's expansion of
(n)
(z).
The
it is
(1)
m-1
i
|_ -^
On ro-i
,
r /\
v. where
sn
= T + B + 5+
1
111
Z
,=i
m (n + m)
Ln (z)
/_\
...+-,
= 0.
define the functions
To
we
first
and
Un (z)
by
the equations
(2)
(3)
so that
- U (z). We shall prove that Ln (z) and Un (z) L^ (z) = - Zn (z) + (n/z) Ln (z), (4)
F<> (*)
=Z
W.J
(z)
n-l
9n-m-i
*>
.(.) log
'
(z),
and then
(1) will
358
(2).
84
Otti,
and Haentzschel,
(1886), pp. 25
35
33.
72
[CHAP.
Ill
L n (z)
zn
\
)
dz dz\
1 _
d (JJs)) S dz\ zn
Mz) _
z n+1 n - r"- 1
2 -->.nl _d
m.
.!
...
Jn (z)
n z} 2
Jm (z)
vl
z1
io~^r-|
1+ ^^ri| ^HriJ 2
and the
first
part of (4)
n
is
proved.
part,
we have
)
d [Un(z)]_ dz\ zn
)
TO =i
= - * -=- + 7n *
(z)
=i
~ -^ [mJr*^ (n +
( )m
- (n + m) J1l+M (z)}
*7n + l(*)
(4) is proved.
It follows from
358 (2)
that
)
U^ {z) _
j'
and since the expression on the right vanishes when n = 0, it induction that it vanishes for all integral values of n. Hence
evident by
therefore established.
Un (z)>
3*581 (3) as
The function
coefficients, coefficients,
n (z),
Schlafli* as a
To
3*57 (1);
g-<'>-. s v.i(;v)- it is
and by straightforward differentiation, the expression on the right satisfies the same recurrence formula as that of 3*581 (4) for Un (z) equation (1) is then evident by induction.
establish this result, observe that
true
when n =
by
3*51 (3)
and
that,
Note.
by evaluating the
coefficient
(3).
3*582-3*6]
BBSSBL FUNCTIONS
will
73
The reader
(2)
(3) (4)
now
3fn (*)
F> (m)
i*Yn{)
The
iivbegral
3 '583.
of 3-571 for F() (z) was generalised by Lommel, Sttidien ilber die BesseVschen Functionen (Leipzig, 1868), p. 86, with a notation rather different from Neumann's; to obtain Lommel's result in Neumann's notation, we first
Jv (*),
we have
^^-Jy(z)logz=
and
so,
^^^ f^
coa(z Sme)coH^B{log(ico^0)-ylr(u+i)}dB,
from 3*582
(3),
rW (>) "
r(n+l?r(i)
8
J**(*sin 0)co*** 6
{log
() \/r (1)
(
jw () =
it is
r(w + ^r(i) j
de
in
which
coefficients
Ln (z)
is
3*6.
vn) on integral and asymptotic expansions of Jv (z) and Yv (z), two combinaBessel functions, namely J, (z) iY, (z), are of frequent occurrence.
also present themselves in the theory of "Bessel functions
3*7).
The combinations
seemed desirable to Nielsen* to regard the pair of as standard solutions of Bessel's equation, and he describes them as functions of the third kind; and, in honour of Hankel, Nielsen denotes them by the symbol H. The two functions of the third kind are defined by the equations f
functions
Jv (z)iYv (z)
(1)
From
'
combined with
354
(1),
we have
t9\
(!),.,
v
is
J- v {z)-e-Jv (z)
J. v (z)-e^Mz)
by
their limits.
When
an
Since Jp {z) and F (z) satisfy the same recurrence formulae ( 3*2, 3'56), the functions enter linearly, and since the functions of the third kind which in
* Ofversigt over det K.
bitch
1902,. p. 125.
Hand-
Hf(z)
Hv
2
(z).
74
[CHAP.
v
Ill
Jv (z) and Y
:
(z), it
follows that
these same recurrence formulae are satisfied by functions of the third kind.
*.>)+*&<.)-**.
*?,+*-**(.),
(6)
.f^-^w.-^w
-&-- -<
Vv ff(z)=0,
(2) >
.f^'-^w-.dr*,.,,
-Jr- --<<>.
Vv ff(*)=0,
(7)
(8)
(1907), pp.
Note. Bayleigh on several occasions, e.#. PAt7. J/ogr. 350359 [Scientific Papers, iv. (1904), p. 290; v.
(1912), pp.
symbol
which Nielsen
calls
\ triH
(2)
(s).
3*61.
The reader
on the
(1)
/(*)
H\z) + Hf(z)
g
W^
nnw
2t
'
J-' (jr)l
(3)
2
,/<*) cob
'
(jr)
r- J_ r <*)
<*)
sm
" #f (*)
(5)
{x)
(z)
"
= J-M-c-iJriz)
t
= Y.
_
(z)-e-"- i
sin vir
(z)
'
sin vtt
(6)
r.,(*)-e-'r,,(*)
sini/7r
^sini/7r
From
(7)
(5)
and
H (z) = e- J5T
if
(,) . --
iff
(*).
3-61-3-63]
3*62.
BESSEL FUNCTIONS
75
ze mvi
.
z and
is
unaltered if z
replaced by
z,
we must expect
by
J v ( z)
Jv (z).
To avoid the slight difficulty produced by supposing that the phases of both of the complex variables z and z have their principal values*, we
shall construct Bessel functions of
argument
ze mni
where
is
any
integer,
arg z has
and
it is
supposed that
+ arg z.
it is
Jv (z)/zv
is
obviously con-
when
the phase of z
as that
is
unrestricted,
same convention
by which
z" is defined;
Jy (zemni - emviri J
)
(z),
v (z).
(2)
J_ v (ze
miri
e- m
J-
The
now be
defined for
all
values of the argument by means of the equations 3"54 (1), 36 (1); and then the construction of the following set of formulae is an easy matter:
(3)
(zemvi )
,Hiri
= e~ m
(z)
(4)
(5)
r_ (ze
= e~ m
J
'
(z),
= sin(l-w)i/7r ff(1)
sini/7r
"
'
_ e_
yit
sinmi/Tr
Rm
"
sinv7r
7r
(6) v '
Hf (zem) - a* H
v
(2)
"
(z) v
'
+ 2e J^T
8
"
sini/7r
Jy (z) vV '
(l)
''
"
Of these results, (3) was given by Hankel, Math. Ann. vin. (1875), p. 454, in the special when m=\ and v is an integer. Formulae equivalent to (5) and (6) were obtained by Weber, Math. Ann. xxxvu. (1890), pp. 411, 412, when m=l see 6-11. And a memoir by Graf, Zeitschrift fur Math, und Phys. xxxvm. (1893), pp. 115120, contains the general
;
formulae.
3*63.
equation.
It has
3 12) that
(z) and J_ (z) form a fundamental system of and only when, v is not an integer. We shall
when
v is
an integer.
= Arg z =f t,
according as I (z)
0.
76
[CHAP,
III
[Jv (z),
(z)}
(z)}
_2
TTz'
This result
is
siderations of continuity
(1)
shew that
-2/(,
and
not.
Hence
(z)
(z)
deduce that
aa {j, (*),T
(#)}-
z cos vrr
and, in particular*,
(3)
m{Jn (z),Yn
found that
(z)}
= 2Jz.
When we
v (z), it is
(4)
5129
{H (z),
hT
v.
(z)}
=-2im{J
(z),
Yv
(jr)}
= - 4i/(w),
so that the functions of the third kind also form a fundamental system of
solutions for all values of
Various formulae connected with (1) and (3) have been given by Basset, Proc. London Math. Soc. xxi. (1889), p. 55 they are readily obtainable by expressing successive differ' ' ential coefficients of Jv (z) and v (z) in terms of Jv (z), Jv (z) f and v (z), v (?) by re;
which the
earlier ones
are frequently required in physical problems) are expressed in the notation used in this work by the following formulae
(5)
Jv
JJ
(z)
IV
JV
(z)
Yv
17
(z)
Jv" Jv"
(z)
--
J.
(6)
()
()-
(,)
()-
A (l -^
(^-l),
x
(7)
Jv
j;
(z)
(8)
(z)
w - JV
()
(z)
jv>"
()
A (^ _ y
(9)
JJ'iz)
JV"
Yv"{z) Jv'"
(.)-- (i
-^- + --^r-)
J,
(10)
Jv
j;
(z)
YvW{z)-Yv
jy>(*)- r;
(z)JvW(z) =
./,<">(*)=
J^
(i-^3),
+i)
(ii)
(,)
(,)
A (^i -
- sin vw
and J
Yv may be replaced by J_ v if the expressions on the right Yv may be replaced by W^ ET throughout if the
2i.
and Hankel, Math. Ann. vnr.
(1875). p. 457.
iv. (1871), p.
106,
3*V]
BESSEL FUNCTIONS
associated formula, due to
77
iv. (1871), p. 106,
An
12 )
and Hankel,
Math. Ann.
(
vm.
(1875), p. 458, is
This
3'7.
is
32
(7).
The
which differs from Bessel's equation only in the coefficient of y, is of frequent occurrence in problems of Mathematical Physics; in such problems, it is usually desirable to present the solution in a real form, and the fundamental systems
(iz)
and J_ (iz) or
Jv (iz)
and
(iz)
It is
is a real function of z which is a solution customary to denote it by the symbol / (z) so that
oo
Iw(s)wg
is
y+2m <**)
J: 9 m\r( v +
m + iy
regarded as a complex variable, it is usually convenient to define its phase, not with reference to the principal value of argiz, as the consideration of the function Jv (iz) would suggest, but with reference to the principal value of arg z, so that
/ (*)
When z
= e-W Jv (ze^%
/ (*)
= e J (ze~^),
The introduction
argument "
is
of the symbol
due
to Basset f
and
it is
Iv (z) to denote "the function of imaginary now in common use. It should be menbeen used by
tioned that four years before the publication of Basset's work, Nicolas} had suggested the use of the symbol v (z), but this notation has not
other writers.
The relative positions of Pure and Applied Mathematics on the Continent as compared with this country are remarkably illustrated by the fact that, in Nielsen's standard treatise , neither the function /(*), nor the second solution v {z\ which will be defined immediately, is even mentioned, in spite of their importance in
physical applications.
it is
is
and
VTT
easy to r prove
Ccf.
{/.(*X
/-(*)}
=-
28in
irz
Lommel gave the corresponding formula for Neumann's function of the second kind. f Proc. Camb. Phil. Soc. vi. (1989), p. 11. [This paper was first published in 1886.] Basset in this paper, defined the function of integral order to be i+ Jn (iz), but he subsequently changed it, in his Hydrodynamics, u. (Cambridge, 1888), p. 17, to that given in the text. The more recent definition is now universally used.
+ Ann. Sci. de Vticole norm. sup.
supplement,
p. 17.
(Leipzig, 1904).
78
It follows that,
[CHAP.
Ill
In the case of functions of integral order, a second solution has to be con3*54. structed by the methods of 3'5
The
function
K
is
second solution,
<>
this
work as the
*.- ['-^](cf.
An
equivalent definition
3*5) is
It
may be
The
verified,
is
by the methods of
n.
3*5,
that
Kn (z)
is
a solution of (1)
when
v,
the order v
equal to
function
(z)
by
K
it
(Z)
= A-7T
may be
verified that
#(*)=
deduce from (6) that
iT,(z)
lim
(z\
= i^'ir v
(iz)
= $>me-i"n H_ v ()-
fact that it is a v (z) lies in the physical importance of the function as z<x> through zero to exponentially tends which equation (1) solution of
The
positive values.
in 723.
The definition of n (z) is due to Basset, Proc. Camb. Phil. Soc. vi. (1889), p. 11, and the infinite integrals by his definition is equivalent to that given by equations (4) and (5) ; 61 5. Basset subsein 614, discussed be will $ which he actually defined the function
(Cambridge, 1888), quently modified his definition of the function in his Hydrodynamics, n. rdl-y(z) 37, (*)1 1 , Ai equivalent to c pp. 1819, and his final definition is
.
^Ti^
'
cV~J v -n
satisfies
(z),
in their work,
1/2", so
equivalent to
'
2|_
ov
Cv
Jv= n
by the
London Math.
3*71]
(cf.
BESSEL FUNCTIONS
79
Modern Analysis, 1771) but this function suffers from the serious disadvantage that vanishes whenever 2 is an odd integer. Consequently in this work, Macdonald's function will be used although it has the disadvantage of not satisfying the same recurrence formulae as Iv (z).
it
An inspection of formula (8) shews that it would have been advantageous if a factor %n had been omitted from the definition of v (*) but in view of the existence of extensive tables of Macdonald's function it is now inadvisable to make the change, and the presence of the
factor is not so undesirable as the presence of the corresponding factor in Schlafli's function ( 354) because linear combinations of /() and v (z) are not of common occurrence.
3'71.
(z).
We
shall
now
those constructed in
give various formulae for Iv (z) and v (z) analogous to 3'23*6 for the ordinary Bessel functions. The proofs
(1)
/_, (,)
K.
v
(z)
/_, (z)
K^
(z)
= zlv_
(z),
zKJ (z)
^L)
{z"Iv (.)}
*Kv (z)} =
{-r*K
>
(z),
\zdzj
z"
\zdz)
z"
]~
~^~'
(7)
(8)
/,'(,)/,(*),
K '(z)
K_
v
K,{z),
I_ n
(z)
= In (z),
{z)
=K
(z).
The
(9)
when R(v +
\)
>
Iv (z) =
cosh (z cos 0)
'
sm^BdO
{zt) dt
- i>?|rr (i)
(1
" t2y
cosh
<**)'
,(1
^os s i n2 ,
dQ
r^r|^)/rcosh
"
{z cos d) sin2v0d0
-t7^fc)/.
<r4oo8hw&
'
80
[CHAP.
Ill
n0
(10)
r (A = I n+h (,)
-j^ [* X
rl(n
_ r)|(2 ,)r
>
+^
(11)
r!( W
-r)!(2^J'
7_ (w+i)
(s)
^ r!(n ^
r)!(2(8)rJ
(12)
/f + , (.) =
r- 2
o;T
^-^
+ 2
(13)
*,-)'.-.
iTo (*)
(14)
- - log (*) 7
.
<*)
^
-
* <,+ 1),
\ n (15)
<")n+1
/1 2 \n+am
'
2
*
f
,
,
v.
{log (**)
W (m + - &
1)
(n +
m +1
)],
(16)
e*
008 '
+ 7}
dO,
= emv7ri Iv (z),
(*"*) =
{!,(*),
tti
^^
=
1/z.
/, (,),
1/*,
J (*)
JT^ (,) +
(z)
The
by Stokes
(cf.
3"572).
The integrals involved in (9) and the series in (14) were discussed by Riemann in his memoir "Zur Theorie der Nobili'schen Farbenringe," Ann. der Physik und Chemie, (2) xcv.
(1855), pp. 130
v=0; he
power
series for 7
The recurrence formulae have been given by Basset, Proc. Camb. Phil. Soc. vi. (1889), 19; by Macdonald, Proc. London Math. Soc. xxix. (1899), pp. 110115; and by pp. 2
11.
(1920), pp.
819.
Functions of this type whose order is half an odd integer, as in equations (10) and (12), were used by Hertz in his Berlin Dissertation, 1880 [Oes. Werke, 1. (1895), pp. 7791]
3-8]
3'8.
BESSEL FUNCTIONS
Thomson's functions ber
(z)
81
their generalisations.
and
bei (z)
and
which occurs in certain electrical problems consists of Bessel functions whose arguments have their phases equal to \tr or f ir.
A class of functions
The
first
examined by W. Thomson *
they
may be
(1)
defined
by the equation f
ber (x)
+ i bei (x) = J
(xi Ji)
=I
(x
^/i),
where x is real, and ber and bei denote real functions. For complex arguments we adopt the definitions expressed by the formulae
(2)
ber (z)
bei (z)
-J
(zi
Vi
*)
= h (* V
*')
Hence we have
(3)
first,
second and
by
Russell:}:
and Whitehead.
The functions of the second kind of order zero were defined by Russell by a pair of equations resembling (2), the function IQ being replaced by the function thus
(5)
ker
(z)
kei (z)
=K
(zy/
i).
Functions of unrestricted order v were defined by Whitehead with reference to Bessel functions of the first and third kinds, thus
(6)
(7)
ber, (z)
her, (z)
i bei, (z) =
i hei (z)
Jv (ze** *),
1
= H (ze**).
(z),
ker (z) =
\tr hei
kei (z)
The
(9)
ker (*)
- log (z)
ber (z) +
fr bei
(z)
* Presidential Address to the Institute of Electrical Engineers, 1889. [Math, and Phys. Papers, in. (1890), p. 492.] f In the case of functions of zero order, it is customary to omit the suffix which indicates the order.
Mag. (6) xvii. (1909), pp. 524552. Quarterly Journal, xui. (1911), pp. 316 342. Integrals equal to ker (z) and kei (z) occur in a memoir by Hertz,
t Phil.
II
(3)
xxn. (1884),
p.
450
\_Ges.
Werhe,
i.
(1895), p. 289].
82
(10)
[CHAP.
Ill
=-
log (\z)
bei (*)
It has also been observed by Russell that the ber2 ()+bei 2 () have simple coefficients, thus
first
but this result had previously been obtained, with a different notation, by Nielsen m 5-41) the coefficient of ($z)* in the expansion on the right is l/[(m !) 2 (2i) !].
;
.
(cf.
of the general
is
functions have been obtained by Russell; for such formulae the reader
memoir and
also to a paper
by Savidge*.
Formulae analogous to the results of 3*61, 3"62 have been discussed by Whitehead it is sufficient to quote the following here
;
(12)
(13)
ber_ (z)
bei_ {z)
cos vtr
ber (z)
bei (z)
sin vrr
[hei (z)
(14)
(15)
+ cos vir
hei v
(z).
The reader will be able to construct the recurrence formulae which have been worked out at length by Whitehead.
The
by B. A. Smith f.
3*9.
The
definition
of cylinder functions.
Various writers, especially SonineJ and Nielsen, have studied the general
theory of analytic functions of two variables 9$ {z) which satisfy the pair of
recurrence formulae
(i)
rM(l)+H(') =
*_.<#) -tfH.(*) =
7rW,
W(*).
(2)
which z and v are unrestricted complex variables. These recurrence formulae are satisfied by each of the three kinds of Bessel functions.
in
Functions which satisfy only one of the two formulae are also discussed by Sonine in his elaborate memoir a brief account of his researches will be given in Chapter x.
;
* Phil.
Mag.
(6)
xk.
(1910), pp.
4958.
425.
Handbuch
1,
42
et seq.
3*9]
BESSEL FUNCTIONS
call
83
It will
any function ^(z), which satisfies both of now be shewn that cylinder functions
When we
(3)
(4)
(2),
we
find that
z^:{z) + v%\{z)
and
so, if
be written
for z (d/dz),
(5) (6)
It follows that
I*) 3? (Z)
that
is
to say
Hence
where a and
6 are
^ (*) = a v Jv (z) +
independent of
z,
bv
Yv (z),
v.
When
we
substitute in (3)
we
x
find that
ay Jy _
and
so,
(z)
+ fcX-i {z) s
<*_,
we must have
and,
Hence ay and
(2) are satisfied.
bv
solution of (1)
(*)
and
(2) is
%
-er^v)
i () Jv (z)
cr2 ()
F, (z),
where
It
and
r(i/)
may be
(9)
difference equation, which is more general than (1), has been examined by Barnes, Messenger, xxxiv. (1905), pp. 52 71 ; in certain circumstances the solution is expressible by Bessel functions, though it usually involves hypergeometric functions.
is used by Nielsen to denote Jv (z), Yv (z), ffvW () more general functions discussed in this section. This procedure is in accordance with the principle laid down by Mittag-Leffler that it is, in general, undesirable to associate functions with the names of particular mathematicians. The name cylinder function is derived from the fact that normal solutions of Laplace's
Note.
The name
and
HW
V
cylinder function
(z)
as well as the
e**>Jm {kpf r
(cf.
sin
m<t>
4-8
18*5).
84 Some
[CHAP.
call
Ill
who
called
Jn (z)
Jn {z)
a Fourier function.
Although Bessel coefficients of any order were used long before the time of Bessel 1-3, 1*4), it seems desirable to associate Bessel's name with them, not only because it has become generally customary to do so, but also because of the great advance made by Bessel on the work of his predecessors in the invention of a simple and compact notation
(cf.
Bessel's
name was
que usus
celeberrima."
A more recent controversy on the name to be applied to the functions is to be found in a series of letters in Nature, lx. (1899), pp. 101, 149, 174; lxxxi. (1909), p. 68.
* E.g. Nicolas, Ann. Sci. de f Journal fiir cylinder function.
supplement,
to be responsible for the
term
CHAPTER IV
DIFFERENTIAL EQUATIONS
4*1.
The
(!)
-^ = azn + by 2
when
0;
4,-4: &
J-2
12 --
is.-
while a and 6 have any constant values f, then the equation is soluble by algebraic, exponential and logarithmic functions. The values of n just given are comprised in the formula
means of
/9\ \*)
4m 2m
where
Bernoulli's
equation,
of solution is as follows If n be called the index of the proved that the general equation J of index n is transformable
:
N, where
(3)
N=- n + 1'
also proved that the general equation of index
v,
and
it is
is
transformable
where
n
=-
is
4.
The
Hence, by
(4),
- 4 is integrable.
-1
(3)
integrable.
and
is
and
(2).
it
(4) alternately, we arrive at the easy to see that these cases are
Exercitationes
pp. 473
475.
quaedam muthematicae (Venice, 1724), pp. 7780 Acta Eruditorum, 1725, The notation used by Bernoulli has been slightly modified and in this analysis
; ;
is
zero.
If either
That
is,
and
86
4*11.
[CHAP. IV
Now
Take
we
proceed to give the analysis by which the requisite transformations are effected.
4*1 (1) as the
[Note.
n.
= Z, y = + !-"*1
Yis
The substitutions
n+ 1
not included
among
;
the values of
its
The
factor
is
presence
was not inserted by Bernoulli the effect of that the transformed equation is more simple than if it were omitted.]
in the denominator
dY
-T7t Y*dZ
=+ T
Y*zn
'
that
is
where
N = - n/(n + 1)
and
N.
Again in
41
(1)
make
the substitutions
where
=n4
is
and
v.
The
and so the
equation
But
4*12.
- 2.
When
41
(2),
the processes described in $4'1, 411 are continually applied to oo in which the index tends, when not soluble by is - 2. The equation with index - 2 is consequently
a finite
number
To
- 2, namely
write y
= v/z,
dz
= a + v + bv2
unci this is
Hence,
equation
is still
of elementary functions.
4-114'13]
DIFFERENTIAL EQUATIONS
414) y = _ _ J?
87
If
we
write
(cf.
is
dz^
z*
-U
>
which
is
it is
immediately soluble.
Euler does not seem to mention the limiting case of Riccati's equation explicitly, although he gave both the solution of the homogeneous linear equation and the transformation which connects any equation of Riccati's type with a linear equation.
It will
appear subsequently
that
is
examined
and
that the only cases in which terms are the cases which have now been to say, those in which the index has one of the values
(
4*7 4'75)
0;
-f, -|;
-f, -;
...,
_2
which a or b
This converse theorem, due to Liouville, is, of course, much more recondite than Bernoulli's theorem that the equation is soluble in the specified cases.
4'13.
practical
soluble cases
method of constructing a solution of Riccati's equation in the was devised by Euler* and this method (with some slight changes
in notation), will
now be
explained.
41
(1),
by taking new
2q
variables
and
(!)
= -v/b,
is
ab
= -c\
cT
+r,2
~ &zn~^ =
is
'
and the soluble cases are those in which 1/q Define a new variable w by the equation
(3)
an odd
integer.
= c^-i + I<^, w dz
dw
is _
-^ + 2cz*-> -^ + (q - 1 ) cz*--w = 0.
is
00
w = z-m-v 2 A r z~Qr
provided that'
Ar_+l
_ (2qr + q +
1) (2qr
+qpp.
1)
Ar
*
8qc(r
vm. (17601761)
l)
[1763],
363; and
ix.
(17621763)
[1764], pp.
154169.
88
[CHAP. IV
and so the series terminates with the term A m z~ qm if q has either of the values l/(2m + 1); and this procedure gives the solution* examined by Bernoulli.
The general solution of Riccati's equation, which is not obvious by this method, was given explicitly by Hargreave, Quarterly Journal, vn. (1866), pp. 256258, but Hargreave's form of the solution was unnecessarily complicated; two years later Cayley, Phil. Mag. (4)
xxxvi. (1868), pp. 348351 [Collected Papers, vn. (1894), pp. 912], gave the general solution in a form which closely resembles Euler's particular solution, the chief difference between the two solutions being the reversal of the order of the terms of the series involved.
Cayley used a slightly simpler form of the equation than (2), because he took constant multiples of both variables in Riccati's equation in such a way as to reduce it to
(5)
W-^~ f c
= 0.
4'14.
We
given in
this
and we shall now explain Cayley 'sf method of solving equation, which is to be regarded as a canonical form of Riccati's
413
(2);
equation.
When we make
(1)
the substitution %
r)
= d (\ogv)/dz,
*?_<-,*-.,,_<);
if U^ and Uz are a fundamental system of solutions of this equation, the general solution of the canonical form of Riccati's equation is
and,
where
d and C
z.
differentiations with
respect to
To express
and
in a finite form,
v
we
write
= w exp
(cz q
/q),
by
is
4'13
sfl
(4).
in
is
?-l
q(q-l)
czq+
^ Vx ^ ^q(q-l)2q(2q-l)
-^*_
(g-l) (3g-l) *
,
/,>
iq
_
and we take
*
to
be exp
(czq/q) multiplied
the index n of the Riccati equation is - 2, equation (4) is homogeneous, Mag. (4) xxxvi. (1868), pp. 348351 [Collected Papers, vn. (-1894), pp. 912]. the memoirs by Euler which were cited in 4*13. cf. 1-1. X This is, of course, the substitution used in 1702 by James Bernoulli
When
t Phil.
Cf. also
4*14]
DIFFERENTIAL EQUATIONS
equation (1)
is
89
so
Now
unaffected
we take
series
?(0-l)2$(2gr-l) - l)(5i- 1)_ Z + + _il_- 1JJ3 ? q(q- l)2q(2q- l)Sq(Sq-lf -J' terminate when q is the reciprocal of an odd positive
"I
?(9-l)
Ci
J72 is
U U
lt
and so
write
If q were the reciprocal of an odd negative integer, equation (1) in the form
we should
*-"W.)^(./.)-a
whence
it
follows that
d
where
lt
^ and y
2
V F = z exp( +
The
series
-?-;.
czfl
will
be examined in
detail in 4'4
much
greater
4'42.
following solutions of Riccati's
equation,
The reader should have no difficulty in constructing the when it is soluble in finite terms.
Equation
Values of
C72
(i)
(dr,/dz)
+ t,*=l
exp (z)
(ii)
(iii)
(dr,/dz)+Ti*=z-*i 3
{dr)ldz)
(l+3z l s )exv(3zW)
/
+ rP=Z-W
(1
+ W* + ^z2 6 ) exp ( + 5z
'
1/5
Equation
Values of
Vlt V%
(i)
(ii)
(dtj/dz)
+ i =z- i
rf
exp
1 /*)
1 /5
(dT,;dz)+q*=z-M
(dT)/dz)+t)Z=z-
z(l
z (1
(iii)
+ 5-
It is to
be noticed that the series Di, U^ (or Vu V2 as the case may be) are supposed term before the first term which has a zero factor in the numerator 4-42 and Glaisher, Phil. Trans, of the Royal Soc. clxxii. (1881), p. 773.
,
w.
B. F.
90
[CHAP. IV
who have studied equation (1) are Kummer, Journal fur Math. xn. 144147, Lobatto, Journal fur Math. xvn. (1837), pp. 363371, Glaisher (in the memoir to which reference has just been made), and Suchar, Bull, de la Soc. Math, de France, xxxn. (1904), pp. 103116; for other references see 43.
(1834), pp.
diately soluble;
when q0, the equation (1) is homogeneous and immeand that the second order equation solved by James Bernoulli ( 1*1) is obtainable by taking q2 in (1), and so it is not included among the soluble cases.
will observe that
The reader
4*15.
of Riccati's equation.
The form
(i)
by
Schlafli *
was
Zt
This
is easily
ift
~ 1ra~lu
*-
4*13(2) by taking
t~a/a
as a
new
independent variable.
To
at
Iff
F(a,t)=X
1=9
w!r(o + m + l)'
is
t).
=c
F(a,
t)
+ c*t-aF(- a,
The
solution of (1)
is
then
Ct^Fia + 1,
t)
+ c2 F{- a - 1,
and
t)
^F^^ + cJ-HFi-a,*)
Bessel's equation is thus
Riccati's equation
but a somewhat tedious investigation is necessary ( 4*43) between Cayley's solution and Schlafli's solution.
<f>
:
Note.
The
function
,
z,
1+- + H2'
.
1 a3 a2 e(z+l)^2.3'z(z + l)(z+2y'"'
which
is evidently expressible in terms of Schlafli's function, was used by Legendre, Elements de Oeometrie (Paris, 1802), note 4, in the course of his proof that is irrational.
Later the function was studied (with a different notation) by fragment in his Math. Papers (London, 1882), pp. 346 349.
Clifford; see
a posthumous
Ann. di Mat.
(2)
i.
(1868), p. 232.
The reader
James
Bernoulli's solution in
series ( 1*1) is to
be associated with Schlafli's solution rather than with Cayley's solution, t This notation should be compared with the notation of 4-4.
4*15, 4*16]
It is obvious that
DIFFERENTIAL EQUATIONS
Jv (*) = (\zy F (v, - \z*),
91
and it has recently been suggested* that, because the Schlafli-Clifford notation simplifies the analysis in the discussion of certain problems on the stability of vertical wires under gravity, the standard notation for Bessel functions should be abandoned in favour of a
notation resembling the notation used by Schlafli-Clifford a procedure which seems comparable to a proposal to replace the ordinary tables of trigonometrical functions by tables
:
of the functions
4*16.
A solution of Riccati's equation, which involves definite integrals, was given by Murphy,
Trans. Camb. Phil. Soc. in. (1830), pp; 440
443.
is
and,
if
a be written
for l/(r+2)
and
(when
ABa2 =l)
is
y =4*
where
If 1/A
h~^[(f> (A)
exp
(*i//A)
+ < (1/A)
exp
(A*V)] dh,
<6(A)=e*A- ^
e~ h ha
- 1 dh= I
Jo
, n =oa(a
for
be written for A in the second part of the integral, then the last expression given reduces to wit multiplied by the residue at the origin of h~* < (A) exp (* 1/0/A), and the
( 4*15) is evident.
was published by Challis, Quarterly Journal, vn. (1866), pp. 51 which shewed how to connect two equations of the type of 4*13 (2), namely
investigation
An
53,
an odd positive integer, and in the other it is an odd negative This investigation is to be associated with the discovery of the two types of
The equation
which
is easily
t- H
f-
bz" u 2
- czm 0,
i
new
variables,
it
transformed into an equation of Riccati's type by taking - + and z*u as was investigated by Rawson, Messenger, vn. (1878), pp. 69 72. He trans-
formed
dv -jt
a+a
y+
bz m
-a
+a 2 y -cz" =Q,
by taking bu=czaly; two such equations are called cognate Riccati equations. A somewhat similar equation was reduced to Riccati's type by Brassinne, Journal de Math. xvi. (1851), pp. 255256.
The connexions between the various types of equations which different writers have adopted as canonical forms of Riccati's equation have been set out in a paper by Greenhill, Quarterly Journal, xvi. (1879), pp. 294 298.
* Greenhill, Engineering, cvh. (1919), p. 334 see also Engineering, cix. (1920), p. 851.
Phil.
Mag.
(6)
xxxvm.
(1919), pp.
501528
92
The reader should
[CHAP. IV
by Siacci, Napoli Rendiconti, (3) vn. a monograph on Riccati's equation, which apparently contains the majority of the results of this chapter, has been produced by Feldblum, Warschau
(1901), pp. 139
143.
And
5, 7,
and 1899,
no. 4.
4*2.
equation.
is
An
(1)
^ = P + Qy + Rf,
;
where P, Q, R are any given functions of z. This equation was investigated by Euler*. It is supposed that neither P nor R is identically zero for, if
either
It
P or R is zero,
the equation
is easily
integrable by quadratures.
Sci.
Hath. n.
that a
nxxdx nyydx-k-xxdyxydx
was studied by Manfredius, De constructions aequationum differentialum primi gradus (Bologna, 1707), p. 167. "Sed tamen haec eadem aequatio non apparet quomodo construibilis sit, neque enlni videmus quoinodd illam integremus, nee quomodo indeterminatas ab
invicem separemus."
The equation
order,
(1) is easily
(2)
--*--"
\r\
dR) du
(4)
(where
(5)
p p lt p2
,
z),
we
,
write
u
is
= e^' dz
(6)
^^.fly-y., y
dz
is
which
of the
same type
as (1).
'
consequently
The equations
1896, pp.
*
1
33.
[Jahrbuch
Nov. Comvi. Acad, retrop. vin. (1760 1761) [1763], p. 32 ; see also a short paper by W. W. Johnson, Ami. of Math. in. (1887), pp. 112 115. t This is the generalisation of James Bernoulli's substitution ( 1-1). See also Euler, Inst.
Calc. Int. n. (Petersburg, 1769), 831, 852, pp. 88, 104.
4*2, 4*21]
DIFFERENTIAL EQUATIONS
Eider's theorems concerning the generalised Riccati equation.
93
4*21.
been shewn by Euler* that, if a particular solution of the is known, the general solution can be obtained by two quadratures if two particular solutions are known the general solution is obtainable by a single quadrature f. And it follows from theorems discovered by Weyr and Picard that, if three particular solutions are known, the general solution can be effected without a quadrature.
It has
To prove the
y be a particular solution of
g-P+fc + JV.
and write y = y
+ 1/v. The
dv
equation in v
is
^ + (Q+2Ry )v + R =
first
0,
is
exp {f(Q
.dz
= 0,
and, since
v = l/(y y
theorem
is
manifest.
To prove
y-yi
The
result of substituting (y x
y<>-yi
w y )/(w 1)
for
in the equation is
,
*?
d}h
(w-lfdz^w-ldz
and,
w-1 dz~
is
when we
and
P + Qy
+ Ry*,
P + Qy + Ry*
reduced to
wTz
so that
=Ry- Ry^
{f(Ry
w = c exp
y
is
Ry ) dz],
x
where
we
let
see that
Hence, from the equation defining w, expressed as a function involving a single quadrature.
third result, let y
let c'
To prove the
.
yjrJf.o
=l
C
'
y*-y<>
y. 2
y-yi
and
this is the integral in a
-y
'
form
free
from quadratures.
* Nov.
Comm. Acad.
t Ibid. p. 59, and ix. (17621763) [1764], pp. 163164. Math. xl. (1850), p. 361.
94
[CHAP. IV
form
Hence
giving
it is
evident that,
lf 2
,
if
3
,
the values
C C C G
it is
equal to
(0 -(7# )(Ci
1
-(Q"
it
In spite of the obvious character of this theorem, been noticed until some forty years ago*.
these correspond to the methods of reducing a linear equation to its normal form by changes of the dependent and independent variables respectively.
Various properties of the solution of Riccati's equation in which P, Q, R are rational functions have been obtained by C. J. D. Hill, Journal fiir Math. xxv. (1843), pp. 23 37 Autoune, Comptes Rendus, xcvi. (1883), pp. 1354 1356; cxxvm. (1899), pp. 410 412; and Jamet, Comptes Rendus de V Assoc. Francaise (Ajaccio), (1901), pp. 207 228; Ann. de la
21.
hagen,
The behaviour of the solution near singularities of P, Q, Nieuw Archie/ voor Wiskunde, (2) vi. (1905), pp. 209
The equation
of the second order
R has
248.
whose primitive
is
of the type
C1C1
+ C2& + C3C3'
where c x c2 , c3 are constants of integration (which is an obvious generalisation of the primitivo of the Riccati equation), has been studied by Vessiot, Ann. de la Fac. des Sci. de
,
Toulouse, ix. (1895), no. 6 and by Wallenburg, Journal fiir Math. cxxt. and Comptes Rendus, cxxxvii. (1903), pp. 1033 1035.
(1900), pp.
210
217-
* Weyr, Abh. biihm. Ges. Wiss. (G) vm. (18751876), Math. Mem. i. p. 30 ; Picard, Ann. Sci. de VEcole norm. sup. (2) vi. (1877), pp. 342 343. Picard's thesis, in which the result -is contained, is devoted to the theory of surfaces and twisted curves a theory in which Riccati's
(4)
4*3]
4*3.
DIFFERENTIAL EQUATIONS
Various transformatiwis of Bessel's equation.
95
are now about to investigate are derived from by elementary transformations of the dependent and inde-
pendent
variables.
first
The
(i)
type which
we
p -*,.BiE
t dz*
where
c is
an unrestricted constant.
The equation
is
of frequent occurrence
is
usually an integer.
The equation has been encountered in the Theory of Conduction of Heat and the Theory of Sound by Poisson, Journal de VlZcole Polytechnique,- xn. (cahier 19), (1823), pp. 249403; Stokes, Phil. Tram, of the Royal Soc. 1868, pp. 447464 [Phil. Mag. (4) xxxvi. (1868), pp. 401421, Math, and Phys. Papers, iv. (1904), pp. 299324]; Rayleigh, Proc. London Math. Soc. iv. (1873), pp. 93103, 253283 [Scientific Papers,
i.
(1899)',
The
special equation in
which
p=2
Ellis,
may be
d
d? (uz-i)
its
general solution
is
(2>
= z^p+i
(ciz).
unrestricted,
Consequently the equation is equivalent to Bessel's equation when p is and no advantage is to be gained by studying equations of the form (1) rather than Bessel's equation. But, when is an
termsf"
(cf.
4),
and
it
is
then
frequently desirable to regard (1) as a canonical form. various types of solutions of (1) will be examined in
The
relations
between
detail in
441-44:}.
is derived from (1) by a transformation of the dependent variable which makes the indicial equation have a zero root The roots of the indicial equation of (1) are + 1 and p -p, and so we write u = vz-p we are thus led to the equation
of equation
dz*
z dz
CV
'
is
W
* See Plana,
= zP + i<@pH (ciz).
Mem.
Mem. della R. Accad. delle Sci. di Torino, xxvi. (1821), pp. 519538, andPaoli di Mat. e di Fis. della Soc. Italiana delle Sci. xx. (1828), pp. 183188.
(1)
t This was known to Plana, who studied equations has just been made.
and
(5)
in the paper to
which reference
96
Equation
der Physik
396];
in the
(3),
[CHAP. IV
which has been studied in detail by Bach, Ann. Sci. de Vjtcole norm. sup. 47 68, occurs in certain physical investigations; see L. Lorenz, Ann.
und Chemie, (2) xx. (1883), pp. 1 and Lamb, Hydrodynamics (Cambridge,
form of continued fractions
(2)
21 [Oeuvres
Scientifiques,
1906), 287
291.
de VAcad. R. de Belgique,
(1876), pp.
(cf. 5*6, 9 '65) have been examined by Catalan, Bulletin xxxi. (1871), pp. 68 73. See also Le Paige, ibid. (2) xli.
10111016, 935939.
(3),
by a change of independent
variable,
an equation
Jr-c^-^o,
solution
is
and
its
(6)
= (Wqyi<$im) (cip/q).
absorbed into the symbol
^?,
When
a constant factor
is
the solution
may be
taken to be
P^iKwiciplq).
Equation
(5),
Mem.
(4)
Studien
xxxvi. (1868), pp. 348351 [Collected ilber die BesseVschen Functionen (Leipzig, 1868), pp. 112
been studied by Plana, 519538; Cayley, Phil. Mag. Papers, vn. (1894), pp. 912]; and Lommel,
in 4 14, has
118.
The system
now been
cussed systematically by Glaisher*, whose important memoir contains an interesting account of the researches of earlier writers.
The equations have been studied from a different aspect by Haentzschel f who regarded them as degenerate forms of Lame's equations in which both of
the invariants g 3 and g3 are zero.
pp.
The following papers by Glaisher should also be consulted Phil. Mag. 433438 Messenger, vm. (1879), pp. 2023 Proc. London Math.
:
pp.
197202.
It
may
be noted that the forms of equation (1) used by various writers are as follows:
cPy
k(k + l)
(Poisson),
S-*-*^*
Equation
(5)
(Wisher).
has been encountered by Greenhill J in his researches on the stability of a under the action of gravity. When the cross-section constant, the special equation in which <?= is obtained, and the solution of it leads to
bridge
Tram, of the Royal Soc. clxxii. (1881), pp. 759 828 and Dublin Math. Journal, ix. (1854), pp. 272 290. t Zeitschriftfiir Math, und Phy*. xxxi. (1886), pp. 25 33. % Proc. Camb. Phil. Soc. iv. (1883), pp. 6573.
* Phil.
by Curtis, Cam-
4-31]
4*31.
DIFFERENTIAL EQUATIONS
Lommel's transformations of Bessel's equation.
occasions; his earlier researches* were of a
97
Lommel on two
somewhat
special
much more
general.
&y i?- s + *-
2i/-l dy
,
"(2)
y- *"#,(*).
direct transformations to construct the equation
at
Lommel proceeded by
general solution
whose
result,
is gfi"~
i^v
(y^), where
a,
#,
are constants.
His
which
(3)
is
it will
be
(4)
= z*-*<@v (yz*).
When /3=0,
and when y=0,
unless
degenerates into
(dv is zero.
The
be
solution of (3)
was given
explicitly
by Lommel
in
numerous
special cases.
It will
sufficient to
(7)
(8)
S+ ^+
(1
-"
S + 5"*=0;
tt=0;
*- + V,W')-**.{iV*).
u=
(1
_ v) g_i
(9)
^+/3y*-=0;
2
*Wmw (y*).
**^ (**).
(10)
l^ + 0;
<^ zu =0-
-**#S (Jld),
u=**#4 (3*),
(11)
*^j ().
An
64, 10-2.
98
120
Math. Ann.
in.
(1871), pp.
475487.
98
[CHAP. IV
Lommel's later researches appeared at about the same time as a memoir by Pearson*, and several results are common to the two papers. Lommel's procedure was to simplify the equation f
d*{ylx(z)\
2v-ld{y/ X (z)}
+(s)
d{ir{z)Y
of which the solution
( 12 )
d+iz)
+ y (*)~
X
'
is ( 4*3)
On
(13)
+ (2 ,-i)^ + 2^x>)_aGf) Tf ^
<
i.l
y=
0.
Now
(2)
by the equation
It will
be adequate to take
(H)
If
^W-^WfeWI'^W^
it is
1
.
we eliminate x ( 2 )>
v ;
ck
<(*)
dz
|_4 (</>(^j
2<(*)
y
is
=o
As a
(17
IS
>
special case, if
we take
<f>
(z)
= 1,
it is
a?
[a
(18)
is
(20)
*
y {*<*)}* W*<*)).
f
The
Inactions
(*)
and
\p (z)
are arbitrary.
4-32]
DIFFERENTIAL EQUATIONS
following are special cases of (17):
99
The
(21)
(22)
^ + (^~^)y = H + ^^ =
will find
0;
y -#,(),
0;
y-*^^>
worked
The independent researches of Pearson proceeded on very similar lines except that he started from Bessel's equation instead of from the modified
form of it. The reader out in his paper.
many
and
(8),
namely
du
VZ
du
Ct
has been investigated by Kepinski, Math. Ann. lxi. (1906), pp. 397 405, and MyllerLebedeff, Math. Ann. lxvi. (1909), pp. 325330. The reader may verify that Kepinski's formula
is
w.
The
int.
when
i/=
de
I'
Acad, des
Sci.
de
198
205.
4'32.
equation,
Twenty years before Lommel published his researches on transformations of Bessel's Malmsten* investigated conditions for the integrability in finite terms of the
equation
which
(15).
is
3+;-(^>
obviously a generalisation of Bessel's equation
;
and
it is
a special case of
^ 4*31
To reduce the
equation,
variables defined
by the formulae
is
We choose p
so that p
this
may
(1)
considered by
= - ir fan.
2pq-q + l+qr=0,
to
(m + 2)q = 2,
cPuV
d?
*
4A
L(
+ 2) 2+
2 2 g {4*+(l-r) }-l ~]
4*
_P
Camb. and Dublin Math. Journal, v. (1850), pp. 180 182. The case in which =0 had been previously considered by Malmsten, Journal fUr Math, xxxix. (1850), pp. 108 115.
100
[CHAP. IV
By
terms
if
to{4 + (l-r)}-i-*( +
l),
where n
(2)
is
an integer
so that
m + 2= + ^+(l-r)>}
+
is
The equation
A =0 and
m=
-2.
4'4.
series
of hypergeometric
type.
is
A compact notation, invented by Pochhammer* and modified by Barnesf, convenient for expressing the series which are to be investigated. We shall
now and subsequently
(a)
write
= o(a+ l)(a+
will
2)
...
(a
+-
1),
(a)
1.
be used
,
is,
in general,
...,
ap
Pl
p2
...,
pq
z)=
rt=0
n
J
j-^
KPi)n \Pi)n
zn
\Pq)n
In particular,
I
CO
< g >"
n=0W!(p)n
.71
=0!(p)n'
*F,(,/9;
,;*)=*
-^7^^
three series are called generalised hyper-
The
first
geometric functions.
It
may be
(a
z) is
a solution of the
differential equation
and,
when p
is
is
^"".^(a-p +
It is evident that
l; 2 -p;
z).
i , 0^*3
174178, 197218.
(1891), pp. 227, 586, 587.
Cf. 4-15.
is
(1890), p.
84
xxxvm.
The
the insertion
p and
4'4, 4*41]
DIFFERENTIAL EQUATIONS
Various solutions in
series.
101
4'41.
We
shall
now examine
d2 u
dz*
-j
r + \) c*u = p(p - u,
z*
will for
supposed
for
supposed that p
It is already
is
known
$p+
(ciz),
and
F, (p
\c*z-)
z~p
-oF^-p;
\c*z*).
in the forms
dz*
dz
z*
When ^
is
(^
-p -
1) (^
+ p)
= 0.
When we solve these in series we are led to the following four expressions for u
zP+^.tF^p + l; 2p + 2; -2cz); zp+i e-^^F^p + 1; 2p + 2; 2cz)
Now, by
sT*'e**
l
(- p
-2p; -2cz);
2cz).
left
are
z^ zp+3
.
....
tr*,
is
z1 ~p,
z-'P, ....
not an integer,
x
we must have
x
*. ,F, (p
2p + 2
- 2cz) = e~ n F
(p
2p + 2
2cz)
(2)
2cz)
to
Kummerf. When
is
(1) has
been proved
for general
obvious on replacing
is
p by p 1
integer.
in (1).
We now
* It follows
an
a special investigation
is
also necessary
when p
is
half of an odd
integer.
102
[CHAP. IV
and there
is
..., the solutions which contain z~p by series involving logarithms (3*51, 3*52), only one solution which involves only powers of z. By the
,
still
holds.
When p
0, 1, 2, ...
of z involved in the solutions shews that (1) that there are no relations of the form
z-p J\(\-p\ \<?z*)
x
holds; but
it is
not obvious
^z-Pe-^F.i-p; -2p;
where
lcit
2cz)
+ k2 zr+*
<t
+ ; \&z*) F {p + %; \c*z%
1
(p
zero.
We
and
(2)
(1)
Note.
vii.
In addition to Rummer's researches, the reader should consult the investigaby Cayley, Phil. Mag. (4) xxxvi. (1868), pp. 348351 [Collected Papers,
(1894), pp. 912] and Glaisher, Phil. Mag. (4) xliii. (1872), pp. Trans, of the Royal Soc. clxxii. (1881), pp. 759828.
433438;
Phil.
4'42.
The equation
*Vi(p+l; 2p+2;
which forms part of equation
general formula due to
(i)
-2cz)
= e-ez F
1
(p
+ l; 2p + 2;
2cz),
(1) of
441,
is
Kummer*
i^(; p;
0-Wp-;
/>;
-),,
which holds for all values of o and p subject to certain conventions (which will be stated presently) which have to be made when a and p are negative integers.
We first suppose that p is not a negative integer and then the coefficient of n in the expansion of the product of the series for e* and (p a p ) is
&
mto(n
- m)\
ml(p)m
"
=
if
()
n\(p)n
'
we
first
in the numerator;
use Vandermonde's theorem f and then reverse the order of the factors and the last expression is the coefficient of fn in X X (a; p; *).
The
*
values J.
Journal fur Math. xv. (1836), pp. 138
141;
t See, e.g. Chrystal, Algebra, n. (1900), p. 9. $ Another proof depending on the theory of contour integration has been given by Barnes,
Tram. Camb.
254257.
4'42]
DIFFERENTIAL EQUATIONS
p
is
103
When
and a < p
! \ |
j.
The
(=
p
N) and
The
;
let
is
valid.
series
^(N;
terminating
sists of
series,
N+ 1
is an infinite series which conterms followed by terms in which the earlier factors p + N, + 2, ... in the sequences in the numerators can be cancelled
while
(p
+N
p,
1,
+ 2,
...
in the denominators.
X
When
iFt (p
is
+ iV p
= M,
and where
. . .
to the limit
when p
-*.
if,
may then be
written*
tJFxO-JV;
"1
-)1,
term in
*', i.e.
the last term in which the numerator does not contain a zero factor, while
1 means that the series is to proceed normally as far as the term and then it is to continue with terms in f Jf+1 f->/+2 ... the vanishing factors in numerator and denominator being cancelled as though their ratio were one of equality.
the symbol
in % M ~N
,
With
(3)
Ai-N; -M;
+ ( - )M ~ N
?)1=i^,(-iV;
-M;
f)
1
1
;
N
MKM~Ay.
-
** +I
^ (^ " ^ +
,
3/
?).
When we
(4)
replace iV
by
M N and by
) 1
we have
;
^(N-M; -M;
+ ( ->"
If
= iJFi <tf -
Jf
-M; -?)1
1
;
3/
2
;
t).
As an ordinary
case of (1)
^(M-N+l; M+2;
and from
(5)
this result
(A +l;
r
if
+ 2; -
),
combined with
;
(2), (3)
and
(4)
we deduce that
^ (-
iV;
- Jf 01 - rf
^ (JV-
if;
if;
) 1.
an integral
462],
This could have been derived directly from (1) by giving p value, and then making p tend to its limit.
* Cf. Cayley, Messenger (old series), v. (1871),
(instead of o)
pp.77 82
[Collected Papers,
vm.
(1879), pp.
2023.
104
[CHAP. TV
(6)
0*J
(p
+1
2p
2cz)
& (p +
f;
\&z%
is also
which
and which
due to
Kummer*. If we suppose
that 2p
is
(-2r
1
(p
ix
m= o(n-m)\ml(2p
n
(-)*
2) m
(2p
l)m (-
+
n
2) n m=0
Now
2W
Cm (^ +
l
- 2p - l)n-,
,
expansion of (1
2t)~p~
(1
t) n+2p+1 and
1
so it
1
is
equal to
/>+)
2~
of the
f(o+)
.
(1
dt
Y~\
0-
~ 0~p_1 w~ n~
dM,
where w
= t/(l t) and the contours enclose the origin but no other singularities integrands. By expanding the integrand in ascending powers of u, we
is
zero if n
is
odd, but
it is
equal to v,,
(p
4- l)i
.,
when n
is
Hence
it
follows that
%
=o2
2n
m .n!(p + f) n
(cz)
'
and
be proved.
When we make p
lim
,F,
N, we
find
by
(p +
2p + 2
- 2cz) = F (l-N;
l
l
- 22V; - 2cz) 1
It follows that
&Q-N;
we change the we find that
If
(7)
icV) = ^.
(l-iV; 2-22V;
- 2c*) 1
2 tfx (f
- N icV) = e.
;
(1
- N;
- 2tf; - 2cz)
1
~1
+ e-^.
F (l-N; 2-2N;
1
2cz)~\,
* Journal fUr Math. xv. (1836), pp. 138 141. In connexion with the proof given here, see Barnes, Tram. Camb. Phil. Soe. xx. (1908), p. 272.
4*43]
DIFFERENTIAL EQUATIONS
105
the other terms on the right cancelling by a use of equation This is, of (1) J_y+ j (icz) in finite terms with a different notation.
inte-
4*43.
The equation
(1 >
'g+g + C + ^-o.
which is a generalisation of Bessel's equation for functions of order zero, occurs in the theory of the reflexion of sound by a paraboloid. It has been investigated by Sharpe* who has shewn that the integral which reduces to
unity at the origin
(2)
is
y=C\
J
cos {z cos
+A
where
(3)
1
Jo
"cos
(4
This
is
( 2-3).
To
in-
= tanh
<f>,
and
*>)
it
becomes
(4)
=G
r CSW + * tanh<
Jo
coshrf> <f>
1,
dd> ^
it
values of
for
which \I(A)\<
andf
2
C=The
cosh
hvA.
ir
integral has been investigated in great detail by Sharpe and he has given elaborate rules for calculating successive coefficients in the expansion of y in powers of z.
A simple
is
y- to ,^(jTi^;
The reader should have no
* Messenger, x. (1881), pp.
1;
2iz).
174185
6679
pp. 101136.
t See,
e.g.
65.
106
4*5.
[CHAP. IV
The
is
soluble
been effected by Lommel* its possibility depends on the fact that cylinder functions exist for which the quotient <$ (z)/^ v (z) is independent of z.
by means
Each of the functions Jn (z) and Yn (z), of integral order, possesses (z), H property [ 2*31, 3*5]; and the functions of the third kind p
this
{2)
(z)
whether v
is
an integer or
not.
Now when
C1 )
the form
}( "- m,
^,-m(7 V*)>
if
of order
m = 2v.
and
This
v
(i)
if v is
an
integer, n,
m = 2n,
or
(ii)
if
= n+\
and
if
m = 2n + 1.
,
Hence
^n denotes either Jn or Yn
we obtain Lommel's
we have
From
z^ n
this equation
Jn (y \Jz)
Yn (y \lz) are
solutions off
,o\
'
so that
(r
= ic exp (riri/n).
we obtain 2w
= 0,
1, 2,
. .
n-
1)
By
giving 7
r
all
possible values
solutions of (2),
and these
denotes
{1)
n+i
we have
and hence
\P)
z* n+
lH
{l)
dzm+1
zn+i
'
= cm+1 e- (n+ *
{riri/(n
),ri
,
so that
(r
y
and the solutions
*
= -ic exp
+ )},
= 0,
1, 2,
. .
2rc)
so obtained
Studien
tiber die
pp. 624635.
The more
general equation
167 189.
4-5]
DIFFERENTIAL EQUATIONS
107
pp.
In view of (1), which holds when m is an integer, Lommel, Math. Ann. n. (1870), p. 635, has suggested an interpretation of a "fractional differential coefficient." Thus he would
JgJ length by Heaviside in various papers.
exp(yV)
to
mean 90o
(yijz).
The
some
(3) of
form
(* + a) (^
+ a - 2j3i>) u = - 'ptftPu,
(<yz^).
the solution of the equation being u = zpv- a(@ v by induction that, with this value of u,
n-\
II
For
it is
easy to verify
(^
r=0
= (-) n/92nc2ns 2n
<,
w,
and so solutions of
-i 4)
(^
+ a - 2r) (^ + a -
2/3v
- 2r/8) u = (-)n/3^cmz^ u
= 0,
n-
where
(r
1,
. . , .
1)
By
we
This equation resembles an equation which has been encountered by Nicholson* in the
investigation of the shapes of
(5)
2 Au.d u\
that
is
to say
$(3-l)(S + 4/*-2)
j*,
($
+ 4p-3) u=*-i*u.
(4)
If
we
form of
sets
of values for a, ,
JL
1
2.
2
1
-1
* Proc.
~\
* 3
10
Soc. lxxxix.
587
519. See also Dendy and Nicholson, Proc. Royal the special cases of (5) in which /*=0 or 1 had been solved
Chemie,
501512.]
108
[CHAP. IV
These four cases give the following equations and their solutions
(6)
^=;
**{#! +^ 4 ()},
22tt;
(7)
~ S} =
4
{?
=^
J
{^2 (2^)+^2(2tV^},
(8)
{*V S} *
-*'*{^(l )+^t-(t il .
-^{^(~*) +*(-*)}.
(9)
t S} =2
,2
;
tt;
of Bessel functions
These seem to be the only equations of Nicholson's type which are soluble with the aid in the case fi 2, the equation (5) is homogeneous. Nicholson's general
is
equation
2 + 2/x
1+2/i
'
z*-** \
'
V4-2/*'
4-2 M
4-2M
(4-2M )V'
4*6.
in symbolic forms,
when p
( 4*3)
is
These forms
are intimately connected with the recurrence formulae for Bessel functions.
It has
been seen
is
39 (6) we have
(ciz) is expressible as
+ Pe-^ls/z,
may
/3
are constants,
it
2>
u=zP+1
[jdz)
ae + &ey .
6*
c*
r
6
A
(3)
is
w
a'
where
This
* Phil.
equation
p. 92.
Royal Soc. clxzii. (1881), p. 813. It was remarked by Glaisher that Earnshaw, Partial Differential Equations (London, 1871), See also Glaisher, Quarterly Journal, xi. (1871), p. 269, formula (9), and p. 270.
(3) is
substantially given by
4 * 6]
DIFFERENTIAL EQUATIONS
109
Note. A result equivalent to (2) was set by Gaskin as a problem* in the Senate House Examination, 1839; and a proof was published by Leslie Ellis, Camb. Math. Journal, n. (1841), pp. 193195, and also by Donkin, PhU. Trans, of the Royal Soc. cxlvii. (1857), pp. 4357. In the question as set by Gaskin, the sign of c 2 was changed, so that the solution involved circular functions instead of exponential functions.
Next we
(4)
W4Y.JL
\zdz)
zp+1
it is
In operating on a function with the operator on the right, is multiplied by I/* *- 2 before the
5
supposed
application of the
convenient to write
and then
(5)
/('*) (e"Z)
.
e*
./(*
+ a) Z,
in
which a
The proof
is
a constant and
Z is
any function of z.
S, as is
Equations (1914),
of this formula presents no special difficulties when/(<>) is a polynomial in the case in the present investigation. See, e.g. Forsyth, Treatise on inferential jrwwm*
33.
2j,
+ 4)<* -
2p
+ 6). ..*,
(5).
successive functions er (beginning with those on the left) past the operators one at a time, by repeated applications of
reverse the order} of the operators in the last result, and by a reversal of the previous procedure we get
We
now
2) * -<*-*>]
.
= J-[( 23 lYJL~\
z^
\_\
dz) z*-*]
'
Th e P bIem W<18 * he 8eCODd Part f uestion 8 Tu eday 1 afternoon, Jan. n Cambridge University Calendar, 1839, p. 319.
8,
JJ:;iZ S
t
It
(1876)>
pp 240
'
~243
'
349
"350
Rmjal Soc
by
s -,
-, obey
'
110
[CHAP. IV
1,
and
be proved.
If
we
replace
p by p +
we
find that
When we
transform (2) and (3) with the aid of (4) and (6), (] ) is expressible in the following forms
we
see that
~zp+A
dz)
*-
>
<
>
^
,
'
The
. 2pdv - -E-j- cH = 0,
[(3) of 4-3],
<9 >
which correspond to
(10)
-^Ub) v = ^ (-^J
+1
(aV*
+ 0'e-a
),
/nx
12 >
more
1^,
dya f + fier"
<
^s)
?!
(7) is due to Boole, Phil. Trans, of the Equations (London, 1872), ch. xvn. on Differential Royal IX. (1854), p. 281. pp. 423425; see also Curtis, Cambridge and Dublin Math. Journal, II. Journal, (1841), pp. 169, Ellis, Camb. Math. first Leslie given by The solution (9) was 193, and Lebesgue, Journal de Math. xi. (1846), p. 338; developments in series were
A different and
direct
;
method of obtaining
Treatise
obtained from
it
by Bach, Ann.
Sci.
de
I'jtcole
norm, sup.
~-c
z%>~i
v=0
7), p. 29.
pp. 364371,
(13)
/3
v==c
IV
^\dc'c)
ae"
+ 3e ~
'
*)-
r)
This
is
~-
if) - g-
functions of
following operators
-^'[(M^m-i}
it
cz;
is
then manifest.
4*7]
4*7.
DIFFERENTIAL EQUATIONS
Liouville's classification
111
in
Before we give a proof of Liouville's general theorem (which was mentioned 412) concerning the impossibility of solving Riccati's equation "in finite
"
except in the classical cases discovered by Daniel Bernoulli (and the we shall give an account of Liouville's* theory of a class of functions known as elementary transcendental functions; and we shall introduce a convenient notation for handling such functions.
limiting form of index -2),
terms
l(z)
log z,
,
l2
(z)
= I (I (z))
l3
(z)
= I (l2 (*)),
= e(ea (z)),
{*/(*)},
. .
e(z)
= ez
ea (z)
= e(e(z)),
e3 (z)
*if(*)-*f(*)-ff(*)dM,
9,/(#)-9{9/(#)},
*/<*)- 9
....
A
lr
function of z
if it is expressible
<f>(z),
e r y}r(z),
is then said to be an elementary transcendental function.% as an algebraic function of z and of functions of the types <irx( z )> where the auxiliary functions <f>(z), yjr(z), %(z) are
expressible in terms of z and of a second set of auxiliary functions, and so on; provided that there exists a finite number n, such that the nth set of auxiliary
functions are
all
algebraic functions of
z.
The order
(I)
is
then defined
inductively as follows:
Any
algebraic function of z
is
of order zero.
r,
(II) If
fr (z)
*M*\
(into
efr {z),
,fr (z),
first
fr (z),
/(*),.../,(,)
is
which at
least
one of the
is
three enters)
said to be of order r
1.
(III)
Any
function
possible order.
order
r,
not of order r
+ 2.
In connexion with this and the following sections, the reader should study Hardy, Orders of Infinity (Camb. Math. Tracts, no. 12, 1910). The functions discussed by Hardy were of slightly more restricted character than those now under consideration,
i
since, for
not required, and also, for his purposes, postulate the reality of the functions which he investigates.
r is
symbol
it is
convenient to
It may be noted that Liouville did not study properties of the symbol s in merely remarked that it had many properties akin to those of the symbol I.
detail,
but
T Journal de t It
is
Math. n. (1837), pp. 56105 in. (1838), pp. 523-547 supposed that the integrals are all indefinite.
;
iv.
(1839), pp.
423456
explicite."
powers of
:,
112
4*71.
[CHAP. IV
The
in
which
% (z)
is
a transcendant of orderf
:
n,
has been
made by
Liouville,
who
If equation
(I) has
solution which is
a transcendant of order
m+
1,
where
m > n,
(2)
= <M*) */*(*),
<M
where
M (z) is
of order
ft,
and
the order of
/a,
and
fi is
such that
n^p^m.
;
/m+i ( z )
m + 1, let it be fm+i(z) then an algebraic function of one or more functions of the types lfm (z) sfm ( z )> efm (z) as well as (possibly) of functions whose order does not exceed m. Let us concentrate our attention on a particular function of one of the three types, and let it be called 6, ^ or according to its type.
If the equation (1) has a solution of order
is
(I)
m + 1,
We shall first shew how to prove that, if (1) has a solution of order then a solution can be constructed which does not involve functions of
S-.
function of
in
order
m+1
which occurs
Then
/qx {6)
it is
*F
/ n F X K) " Tz*~
W+
dz>
1
df (z)
dz
d*
fm
(z)
d6dz
+ (_!_
it
\d_
\fm {z)
dP^ldz\fm (zj
W-'-xW
The expression on the right in (3) is an algebraic function of 6 which vanishes identically when 6 is replaced by lfm (z). Hence it must vanish
identically for all values of 6
;
equating
it
to
m+ 1
Journal de Math.
is
435442.
t This phrase
if
u were of order
less
than
w,
which
is
contrary to hypothesis.
4'71]
DIFFERENTIAL EQUATIONS
is is
113
replaced
by
In particular, the expression on the right of (3) vanishes when + c, where c is an arbitrary constant and when this change
;
made the
expression on the
left
d-F(z,0 +
which
is
c)
'-
therefore zero.
That
is
to say
When we
c,
we
'"
find that
dF(z,
oc
+ c)
'
o-F(z,
dc>
+ c)
'
are solutions of (1 ) for all values of c independent of z. If performing the differentiations, these expressions become
we put
after
dF(z,
)
'
d3F(z,
d0
Fe>
d&
(I).
) "'
'"'
will
be called
Now
either
and
Ft
they do not.
If they do not,
we must have*
F = AF,
where
is
On 0. F=<PeA0
,
integration
we
find that
where 4> involves transcendants (of order not exceeding + 1) which are algebraically independent of 0. But this is impossible because e Ae is not an algebraic function of and therefore and e form a fundamental system ; of solutions o/'(l).
Hence
where
in
Fee
is
expressible in terms of
F and F
and
B are
constants.
its solution is
F = O^* +
where
<I>,
or
/= e* {4>
+ 4>a 0},
4>,
and
while a and
/9
are
a?-Aa;-B = 0.
The only value
a
=$=
of
is
obtained when
Similarly, if
it
fm+1 (z)
* Since
we can prove
that
must be a
linear function of ^.
F mutt involve
0,
Ft cannot be
identically zero.
114
[CHAP. IV
and ^,
involves
we may
write
>
fm+i(z)^%0
where the functions
of order
m + 1 at most, and
which be
...
m+1
involved in
them
...,
...
in
fm+l (z)
let it
is
...%,%,
and
Ox {z)0t {z)
P (z).* 1 (z)
* Q (t).irPiQ (z).
it
follows that
d_d_
d0!
d_
d_
1
d_
d*S 2
_a_
d02
'"
d0p'd&
"*
S^
/**.(*)
is
a solution of (1)
i.e.
But ^p
the types
Q (z) is either
function of order
m+1
case,
a function of order not exceeding m, or else it is a which involves functions of the type and not of
repeat the process of reduction to functions of lower
and ^.
In the former
order,
we
and in the
latter case
we
see that
is
an
We
a transcendant
else it
has a
an algebraic function of functions of the type ef^iz) and is of order /* and <M (z) is of an order which does not exceed ft.
next prove that, whenever (1) has a solution which is a n, then it has a solution which involves
it
We
shall
as a factor.
We concentrate
e/j, (z),
of
the form
and then the postulated solution may be written in the form where is an algebraic function of and any function (other than of order /* + 1 which occurs in G is algebraically independent of .
;
G (z,
),
itself)
Then
it is
rf*G
&G
rPG
d*G
+ e[/"
{/,'(*))]
|g-G.x<*>.
The expression on the right is an algebraic function of which vanishes when is replaced by e/^iz), and so it vanishes identically, by the arguments used in (I). In particular it vanishes when is replaced by c , where c is independent of z. But its value is then
'-G(*,ce).x(),
4*71]
so that
/a\ (6)
DIFFERENTIAL EQUATIONS
d*G(z,c%) ;
115
~. , , -G(z,c&). X (z) =
.
0.
When we
c,
we
find that
dG(z,c&)
dc
'
&G (z,
9c
c0)
''
'
z.
If
we put
= 1, these
dG(z,S)
^ d*G(z,e)
*
d@
Hence, by the reasoning used in
a@ 2
we have
'
"
(I),
G9 = AG or else
and
B are
constants.
G Q% A
or
and
1
in the latter
+ O,**
&{&
+ &t \og&} =
&{&
& f?(?)},'
t
;
where
4>,
Q>lt
<t>8
while
1)
- Ax - B = 0.
is
either contains
a power of
or else
is
the
sum
latter case*,
G(z, c%)-c*G(z, 0)
is
is
a power of 0.
r
By
we
2 , ...
are
all
the
which occur in the postulated solution, we can derive from that solution a sequence of solutions of which the sth contains ... 0,; and the rth 0,, 0, ... 0, only by factors which are powers of n 2 member of the sequence consequently consists of a product of powers of 0i, r multiplied by a transcendant which is of order ft at most; this solution is of the form
transcendants of order
1
,
(*)exp|2 7log*V>
which
is
of the form
^M (z)
ef^ (z).
;
if it is,
then
4> 0* is 2
116
4*72.
|CHAP. IV
We
(!)
[in
the equation
;i-X<*)
is
which x ( z )
is
an elementary tran-
then
it
where
/*
^ n.
be called
,.
Liouville's theorem,
which we
is
shall
now
prove,
is that,
for
equal
to n.
Let
d
and then
If
t
log
dz
is
C'
of order
at
most
let
the order of
is
N=n,
by
t,
proved.
If
N>n,
satisfied
namely
is
greater than
n.
Now
tys-i i z),
an algebraic function of at least one transcendant of the types s/jr-i (z), ef N _ x (z) and (possibly) of transcendants whose order does
t is
not exceed
If
t
N-
1.
We
call
the
first
three transcendants
0,
^,
@ respectively.
contains more than one transcendant of the type 0, we concentrate our attention on a particular function of this type, and we write
t
= F(z,0).
if
By arguments
N>
n,
then
F(z,0 + c)
is
The corresponding
expJJF(*,
solution of (1)
is
+ c)dz,
z.
and
this
is
a solution for
all
c,
values of c independent of
Hence, by
we
^c
is
[ex V
jF(z,0 + c)dz]j^
and we have
u2 = u JF
l
dz,
so that
diu
U
da,
*Tz-
*dz~
= U F *
4-72,4-73]
DIFFERENTIAL EQUATIONS
solutions of (1)
is
117
a constant*; and so
a constant.
J(C/F,).
;
an algebraic function of and similarly it is an algebraic function of all the functions of the types and ^ which occur in t. Next consider any function of the type which occurs in t we write
m, is
;
Hence
&), and, by arguments resembling those used in 471 and those used earlier in this section, we find that the function u., defined as
3
ris
r
= G (z,
a solution of (1)
and we have
so that
__
i(j
_=
,,.eG 9
This Wronskian
is
a constant,
C and
lt
so
,-V{W(?.)l.
Consequently u
the types
! is
an algebraic function, not only of all the transcendants of and ^, but also of those of type which occur in t and therefore
x
is
of order
iV.
This
is
is
of order
fi
l,
where f<L^N,ifN> n.
The
of
d (log Uj)/dz 4
#
And
this is the
73.
We
shall
now shew
We
has no integral (other than a null-function) which first reduce the equation to its normal form
is
an algebraic function of
z.
by writing
y=uz~^, p
= v-%-
t Journal de Math. iv. (1839), pp. 429435 ; vi. (1841), pp. 47. Liouville's first investigation was concerned with the general case in which x (z) is any polynomial the application (with various modifications) to Bessel's equation was given in his later paper, Journal de Math. vi.
;
(1841), pp.
113,
36.
118
[CHAP. IV
This
of the form
dhi
-"%<*>
,
where
(2)
X (.).(l)-i.
an algebraic integral then (1) also Let the equation which expresses this integral, u,
;
as an algebraic function of z be
(3)
64(u,z)
is
= 0,
;
where 64
and
it is
supposed that 64
ik
irreducible*.
Since u
(4)
is
a solution of (1)
we have
0.
The equations
of (3) satisfy
(4).
and
(4)
have a common
root,
and hence
all
the roots
and (4) (qua functions of u) would than 64 itself, and this would be a factor other reducible, which is contrary to be would Hence 64
uM
Then,
if s is
Let
all
,, ,
tt/
...
+ W *+
2
+ Ujg*
is
a rational function of z ; and there for which this sum is not zerof.
is
s be taken,
and
let
=i
Also
let
= s (s - 1)
Since
. . .
(8
- r + 1) j^ .-* (-^)'
are
all
where r =
prove that
<*\
1, 2, ... 5.
** 2
... m,,
solutions} of (1),
it is
easy to
dWo
-w
1
(6)
l ,
(r
1, 2, ...
- 1)
That
in z or in both u and -J^ has no factors which are polynomials in u or would be zero.
Because
(4) is satisfied
by
all
the roots of
(3),
qua equation in
u.
4 *73]
DIFFERENTIAL EQUATIONS
119
Since
so that
is
a rational function of
z, it is
W=
and
X A n z+2
BW
'\H>
positive
where
An
Bn>q
are constants, k
- aq)
which occurs in
be lj(z
aq ) p
Kz -
9 ) in
power of
Hence there is a higher power on the left of (7) than on the right. This contradiction shews that there are no terms of the type 2? n n>9 (z - a q )~ in a and so
= X A n z\
n=
-it
We may now
have a
last
assume that
if it
term
A K ^ 0,
must
it is
lt
W W W W
,
a simple induction
W,r h
possible to
is
^z*->2.4,...(2r).(s-l)(s-3)...(s-2r+l). 2F (h-ls;l-sl
l) r+1
where the
If s
suffix
+1
+1
When
That
and
is to
is
say
l)is
so,
The
t
expression on the
left
zerof.
is
* It is to be
seems to fail at this he apparently overlooked the possibility of \ vanishing. The failure seems inevitable in view of the fact that J* 2 (z) + J _ n _ h (*) is an algebraic function of *, by 34. The +{ subsequent part of the proof given here is based on a suggestion made by Liouville, Journal de Math. iv. (1839), p. 435; see also Genocchi, Mem. Accad. delle Sci. di Torino, xxm. (1866), Comptes Rendiu, lxxxv. (1877), pp. 391394. pp. 299362
(1841), p. 7,
The
x (2)
1.
point, because
120
[CHAP. IV
and
that,
We
s is even,
s is odd,
vanishes,
when
2 A ny8 z- n
where
0)g
sums of powers of the roots, it appears that and that the equation (u, z) = is expressible in the form
(8)
M must be even,
u M + 2 u M~"
r=l
4 if
(1/*)
= 0,
in \jz.
l/'z,
When we
we
find that
m=0
a positive integer and, in the case of one branch at least, c does not vanish because the constant terms in the functions r are not all zero.
where n
is
And
2 cm z- m,n
wi=0
z.
When we
we
find that
the coefficient of the constant term in the result is c and so, for every branch, c must be zero, contrary to what has just been proved. The contradiction thus obtained shews that Bessel's equation has no algebraic integral.
4*74.
On
are
the impossibility
finite terms.
We
now
equation for functions of order v has no solution (except a null-function) which is expressible in finite terms by means of elementary transcendental
functions, if
2i> is
As
in 4-73,
we reduce
normal form
where x ( z )
= ~ 1 + P (P + *)/**
d
(log u)/dz
Now
write
t,
%'
* Goursat,
* + , + i-*k!>-a
Cours d* Analyse,
ii.
(Paris, 1911),
pp.
tacitly
assume the
4*74]
DIFFERENTIAL EQUATIONS
x iz)
is
121
Since
f order zero,
it follows
from
finite terms,
which
is
of order zero,
i.e.
it
integral.
integral,
If (2) has an algebraic integral, let the equation which expresses this t, as an algebraic function of z, be
(3)
a(t,z) = 0,
is
where S4
Since
(4)
an irreducible polynomial in
a solution of
(2),
and
z.
t is
we have
S4z +{ x {z)-t^s4t = Q.
t
As
satisfy (4).
more than two branches of t, and let three of them be called tlt t2 t^, the corresponding values of u (defined as exp jtdz) being uly v?, us These functions are all solutions of (1) and so the Wronskians
, .
dus
dz
du*
duj
dtu
du2
dz
2
dux
dz
dz
lf
,
dz
3.
dz
G C2 C
duz
Ua
Now
it is
n
and
t3
dus
rf
~d~
~ U* Ua
'
~"
^'
is
if it
were
zero, the
Hence G
0,
and so
UtU3
= dfcU -
2 ).
is
an algebraic function of z.
But
and therefore ux
is
Wl
=A
/^-^,
in 4-73,
an algebraic function of z. This, as we have seen t has not more than two branches.
Next suppose that t has two branches, so that S4 (t, z) is quadratic Let the branches be U V V, where U and V are rational functions, of z.
substituting in (2)
in
t.
By
we
find that
K)
Let
{V'
factorised so that
+ 4>UV=0.
V be
V=Az*Tl(z-a q y<i,
where
is
constant, Kq
and \ are
integers,
and
k,{
zero.
122
[CHAP. IV
member
of (5)
it
follows that
\z
(z
aq
first
member
of (5)
we have
(,),o.
.
Now
is
It
any one
of
them
greater than
it
must
Kq
T!/f q
=s
V/>
so that Kq is
or
tc
4, which
Hence
all
the numbers
are equal to
2.
Again, if
power in
we consider the principal part near V must cancel with the 1 in % (z), so
xJV
is rational,
oo
we
that
\ = 2 Kq
9
It follows that
is
z) is reducible,
which
contrary to hypothesis.
Hence
cannot have as
many
x
it
must be
rational.
in partial fractions be
7? -n, g
t= 2
where
If
A n zn + 2
r(*-o)*'
integers,
An
we
n assumes
positive
0.
we
find that
in (2)
If
we
we
see that
l/(z
aq )
cannot occur in
to a higher
first
and that
so that
Similarly, if
we consider the principal parts near and oo we find *=1, (A^y-JL-t-pip+l); X~0, 4,' = -l. Since p = v ^, we may take A- = p without loss of generality.
Y
that
It
u
Accordingly,
if
= z-Pe i?U(z-aq ).
9
we
replace
u by z~ p e u w
in (1),
we
must have a
z,
in
4'75]
DIFFERENTIAL EQUATIONS
123
the relation connecting
m (m 2p 1) cm 2icm-! (m p - 1) = 0,
and so the
series for
can vanish,
i.e.
unless
is
numbers +
if,
and
if,
and only
2v
is
Conversely we have seen ( 34) that, when 2v is an odd integer, Bessel's equation actually possesses a fundamental system of solutions expressible in The investigation of the solubility of the equation is therefore finite terms.
complete.
Some
338
340.
4*75.
On
the impossibility
By means
we can
brf
r az + f
with a view to proving that
It has
it is,
been seen
( 4*21)
reducible to
where
n=2q 2;
is
reducible to Bessel's
Hence
finite
is zero
is
equal
or to
4w 2w +
(i-0,
l
1,
2,
...)
Consequently the only cases in which Riccati's equation is integrable in finite terms are the classical cases discovered by Daniel Bernoulli (cf. 4'1 1) and the limiting case discussed after the manner of Euler in 412.
This theorem was proved by Liouville, Journal de Math. vi. (1841), pp. 1 13. It seems impossible to establish it by any method which is appreciably more brief than the
analysis used in the preceding sections.
124
4*8.
[CHAP. IV
The
seen
(
1*3) to
appearance in analysis of the general Bessel coefficient has been be in connexion with an equation, equivalent to Laplace's
We
shall
manner from
^ + ^ + ^=
solution in question
is
The
(2)
V
f denotes
f(z + ixcosu
n
-f
iy
smu, u)du,
in which
In particular, a solution
ek
J
(z
mu
^
by the equations
in
which k
If
is
m is
any integer.
we take
x = p cos
this solution
<p,
y = p sin
<f>,
becomes
eiipcog(u-*)
|
0**
J
cosmM ^ M=::e*z
m (v +
it
<f>)
dv,
= 2ekz
J o
is
gk
IT
(z
+ ix cos u + iy sin u)
sm mu
and
this
is
Jm (kp).
Again,
nates,
it is
if
found to become
d^V
dp'
*
ldV
p dp
1 p*
d*V
d<f>*
F
'
dz*
Monthly Notices of
the
R. A. S. lxii. (1902),
pp.
lvii.
(1902),
pp. 333341. t
The
simplest
method
See
Journal,
iv.
(1845), pp.
3342.
4-8, 4-81]
DIFFERENTIAL EQUATIONS
this equation of
125
which ekz
is
lcTV
Vd<p*
is
independent of
to
m where
2
factor of
is to be one- valued, it must be equal Consequently the function of p which is a must be annihilated by
<f>,
is
an integer.
dp 3
p dp
\
if it is
and therefore
line
it
0.
We
sin
m<b r
Jm (kp). r/
'
These solutions have been derived by Hobson* from the solution Maxwell's method of differentiating harmonics with respect to axes.
^*J
(kp)
by Clerk
Another solution of Laplace's equation involving Bessel functions has been obtained by Hobson (ibid. p. 447) from the equation in cylindrical-polar coordinates by regarding d/dz as a symbolic operator. The solution so obtained is
cos
.
sin
m+ 'V(pz)M>
is an arbitrary function but the interpretation of this solution when <@ involves m a function of the second kind is open to question. Other solutions involving a Bessel function of an operator acting on an arbitrary function have been given by Hobson, Proc. London Math. Soc. xxiv. (1893), pp. 5567 xxvi. (1895), pp. 492494.
;
;
where/(s)
4'81.
We
K }
shall
now examine
dx2
t
dy3
dz 3
~&
dt 2
'
in which
V=
J -IT
JO
sin v
+ z cos u 4- ct,
u, v)
dudv,
where
/ denotes
In particular, a solution
-r
where
f,
i
J? /
dU fJv
F denotes an
London Math.
arbitrary function of
Soc. xxn. (1892), pp.
u and
v.
* Proc.
t Math. Ann.
(1904),
431449. 342345. See also Havelock, Proc. London Math. pp. 122137, and Watson, Messenger, xxxvi. (1907), pp. 98106.
lvii. (1902), pp.
Soc. (2) n.
126
[CHAP. IV
the fact that
kc.
it
is
all
yfr) be the which the passes through the polar axis is the direction (6, <f>) and the plane y\r = .s-axis. The well-known formulae of spherical trigonometry then shew that
y, z)
be
and
let (m,
(u, v) referred to
new axes
for
cos
a>
= cos
<f>),
take the arbitrary function F(u, v) to be notes a surface harmonic in (u, v) of degree n we
Now
de-
Sn (U,
where Sn
is
V)
= Sn (0,
(a>, yfr)
<f> l
O), i/r),
a surface harmonic* in
of degree n.
We
Vn = e**
Since
(*
J
e*rc08 *- Sn (0,
<f>
to,
8n
is
(to, sfr),
we may
write
Sn (0,
<*> 5
<0,
t)
- ^ (0,
4>)
Pn ( C0S )
+
m=l
+ BnM (0,
(0,
<f>)
<f>)
sin mi/rj
Pnm (cos
o),
where 4
(0, 0),
A< w) (0,
and
Bn
< TO)
are independent of
-^r,
w and
^r.
we
get
332.
Now the equation of wave motions is unaffected if we multiply x, y, z and by the same constant factor, i.e. if we multiply r and t by the same constant unaltered so that A n (0, <f>) may be taken to be inand factor, leaving dependent f of the constant k which multiplies r and t.
t
;
<f>
is
is
A n (0,
is
<) is
a solution (independent of
of the equation of
wave
<f>)
motions, and
Hence
A n (0,
* This follows
from the
an invariant
for
changes of rectangular
axes.
t This
is
of k.
4-82]
is
DIFFERENTIAL EQUATIONS
n. If we assume it to be permissible any such harmonic, we obtain the result that
127
to take
A n (#,
to be
of the equation of wave motions*; and the motion represented by this solution has frequency kc.
solution,
To
justify the
assumption that
A n (0,
<f>)
may
n,
we
sin
md>. r
The
factor
is
*("*) -<++*'.
so that
if
this factor is to
it
must be a multiple
of
Jn + {hr))^,:
4*82.
the equations
of Mathematical
Physics.
It is possible to prove (or, at any rate, to render probable) theorems concerning Bessel functions by a comparison of various solutions of Laplace's
Thus,
if
we take the
function
e^t/o [k>s/(p*
+a 2
lap cos
<f>)},
we
It is therefore natural to
AJ
Assuming the
consideration
in p
is
(kp)
and
a,
we observe that the function under an even function of 0, and so Bm = and, from the symmetry ^4 m is of the form c m Jm (ka), where cm is independent of p and a.
possibility of this expansion,
;
We
thus get
Jo {kx'Xp*
+a
2
lap cos
<)}
If
we expand both
sides in powers of p,
<)
a and cos
m we
,
<f>,
get
* Cf.
p.
309.
128
[CHAP. IV
and
we
Jo
of which a
{W(ps + a*-2ap
cos
<p)}
= 1
m=
m<f>,
Again,
+
to
oc to
moving in the direction of the with frequency kc and wave-length 27r/k, we expect
(^) \K)J
where cn
is
iket
c n i*
=o
a constant
so that
eikrco *
=
(If)*
CninJn+i {kr)
n
Pn (cos
side,
e)
If
we compare the
on each
we
find that
n \-^ and
so cn
1T)
2+*r(n + f)"2.(n!r
we
ikrc se
= fiE) i
\krj =o v
will
(n
of
be given in
The
Hobson
to the case of
the equation
dx?
dx
'"
'
dxp*
c2 dtis
normal solution of this equation of frequency kc which function of r and t only, where
r
expressible as a
V(#i
a^*
. . .
+ xp
-),
9r
^
,
and
e
*
iket
<& ilp -*
(kr)l{kr)*^K
This
is
due
to
Neumann, Theorie
65.
z.
harmonics do not occur because the function is symmetrical about the axis of X This expansion is due to Bauer, Journal filr Math. lvi. (1859), pp. 104, 106. Proc. London Math. Soc. xxv. (1894), pp. 4975.
t
The
tesseral
4-83]
DIFFERENTIAL EQUATIONS
f
129
Hobson
of rank p
;
describes the quotient ^\{ P-i)(kr)l(kr)^fh %) as a cylinder function such a function may be written in the form
<$(kr\p).
By
Hobson succeeded
integral order
using this notation combined with the concept of jp-dimensional space, in proving a number of theorems for cylinder functions of
As an example
we
2
shall consider
an expansion
for
J {k v^r + o where
it is
2ar cos
<j>)
\
p),
xp by the equation xp r cos This function multiplied by e**ct is a solution of the wave equation, and when we write p = r sin it is expressible as a function of p, t and of
<f>
convenient to regard
<f>.
<f>,
<f>,
no other coordinates.
Hence
** J
is
{k
V(r2
+ a? -
2ar cos
<f>)
p)
dp2
that
is
dp
dx
'
to say,
by the operator
&
,
J>~1
r
(ff-2)cos4> d
r*sin<j>
d<f>
dr*
dr
r*d<p*
Now normal functions which are annihilated by this operator are of the form
"
(hr)ir->
pn (cos <p\p),
+ cpy-*P.
where
Pn (cos
<p
p)
is
- 2a cos
we
<f>
By
infer that
J \k V^r +
a*
2ar cos
1
<f>)
p)
00
1
~ (/fca^-^/fcr)^-
<f>
\p).
Now
(k?ar cos
<f>)
expand all the Bessel functions and equate the n on each side we find that
;
coefficients of
2"
2*"+ iP-
r (n + |p)
(
=
{2
An
+fc>-
T (n + ip
1
).
'
so that
*
A n = 2**-
+ %p - 1) T (|p Pn (cos
So
p) S3 C*
p_ * (cos
0).
130
[CHAP. IV
We
Jfr-,
(r
+a
An
all
valid only
when p
is
an
integer), will
be given in 114.
4*84.
Two
*
'
dx1i
dxf
2 9a;,
dx4*
c2 dt*
we
take
new
variables p, a, %,
yjr
xx p cos 2,
x3 a cos -^r,
Xi
x2 = p sin %,
the equation transforms into
a sin ^r,
'
{dp 2
p dp
p 3x
2
)
[do*
+
<r
'
d*
o*
dyfr\
c* dt*
A
where
is
J
3> is
any constant.
if
Further,
we
write
p=*r cos
so that(r, ^,
-^r,
0,
<r
= r sin <,
<)
into
O
Now
&V SdV
dr*
r dr
l&V
r* d<p*
c ot<t>-tfm<p
'
tV
dtp
r* 1
r cos <
&V +
9x
2
92
ld*V
'
r2 sin2 4*
W*
+ * c<)
c2
^'
e* (MX +"*
as a factor
are annihilated
3
by the operator
d_
dr3
r dr
182188; Proc. London Math.
Soc. (2) hi. (1905), pp.
111123.
4-84]
DIFFERENTIAL EQUATIONS
131
and since such solutions are expressible as the product of a function of r and a function of they must be annihilated by each of the operators
<f>
y
dr2
d
3 9
4X(X+1)
r*
'
rdr
l
9<
+ 4 \(\+l)+ (cot<t>-t*n<l>)^r \ ^ 2
3^>
** cos2 ^>
*
sin 2
'
where X
is
consideration.
The normal
now
easily verified to
be
of the form
(kr)-1
F (* - X,
X
^- + \ +
<f>
+ l
sin 2
*ox+*++**>.
J^ (kr cos
is
cos
<I>)
Jv (kr sin
<f>
sin
<I>)
2 a A (kr)~
J^ (kr)
cos*< sin"<
2JP,
(~^ - \ ^^ + X +
sin 2 <f>)
where the summation extends over various values of X, and the coefficients a* depend on X and <J>, but not on r or <j>. By symmetry it is clear that aA where
= 6 X cos*<I> sin" 2 F
.
(^-J^ -
X,
^^ + X + 1
1;
sin 2 <l> \
b K is
independent of
<.
X = Ha* +
and Bateman has proved that
")
+ .
(n
= 0,
1, 2, ...)
We
Bateman's
sin
proof,
which
is
we
J (kr cos
<f>
cos 4>)
Jv (kr sin
<f>
<I>)
by a
direct transformation.
CHAPTER V
MISCELLANEOUS PROPERTIES OF BESSEL FUNCTIONS
5*1.
In this chapter we shall discuss some properties of Bessel functions which have not found a place in the two preceding chapters, and which have but
all
obtainable by processes of a
We
some
indefinite integrals.
results
V>%
dz
(2)
[" 2-" +1
# (z) dz
(z).
To
^z^f{z)%%(z)dz;
let this integral
be equal to
z"+>
{A
+ B (z)
<&v+1 (z)},
where
(z)
and
B {z)
are to be determined.
The
z"+>f(z)
V, (z) =
jH-i {
I
>
(z ) < v (z)
+ A (z)
2l
^
z
(z)-A(z)<@ v+1
(jr)
(z)\
)
+ z^
In order that
take
B (z) <$
<*)}.
A (z)
A (z) =
it
B'
(z),
and B (z) may not depend on the cylinder function, we and then
/(s) .5 A' (z) +
2
^
z
A(z)
+ B (z).
Hence
(3)
follows that
** +1
\b"(z)
^~
B' (z)
J"
B (*)| # (*) dz
{B' (z) <@ v (z)
= *+
This result was obtained by Sonine, Math. Ann. xvi. (1880), p. 30, though an equivalent formula (with a different notation) had been obtained previously by Lommel, Studien fiber die BesseVschen Functionen (Leipzig, 1868), p. 70. Some developments of formula (3) are due to Nielsen, Nyt Tidsshrift, ix. (1898), pp. 7383 and Ann. di Mat. (3) vi. (1901),
pp.
bei, see
Whitehead,
340.
5-1, 5-11]
MISCELLANEOUS THEOREMS
is
133
(3),
The
an obvious consequence of
should be noted:
(4)
I
V+
ir
^ 0) ds
li
5*11.
The
312
'
J'_ (z)
r / \ T - Jv (z)Jv
(z)
I/7T 2 sin -
which gives
K
J
/"*
zJ*(z)
dt
2sini/7r
7T
.
'
(z)
/_,, (.g)
9 Kl)
)
2 sin
10g
*tt
'
(z)
and
similarly,
from
3*63 (1),
z
,ox
dz
_ir
Fy (>)
^
K
}
['
J
dx _ zYv\z)~
it
Jv (z)
(z)'
The reader should have no difficulty in evaluating the similar integrals which contain any two cylinder functions of the same order in the denominator. The formulae actually given are due to Lommel, Math. Ann. iv. (1871), pp. 103 116. The reader should compare (3) with the result due to Euler which was quoted in 1*2.
Some more
It is at
if
y and y
rt
d?n
then
j\P -Q)ynd^y^-v^.
*
Math. Ann.
520536.
134
[CHAP.
V
If
431
(17).
*^V> ^i,
and v
respectively,
we have
where
<f>
(z)
and
i/r
functions of
z.
This formula
<f>
is
As a
(z)
and
Z
yjr
(z) to
(7)
{{^-P)z-^^\^{kz)^
dz
(lz)dz
= ,fa(kz) d ^-^(lz)
^
dz
further in two special cases (i)fiv,
right.
is
= z {k^ +1 (kz) W. (Iz) - 1&* (kz) # r+1 (lz)} -(/*-) ^M (kz) <@ v (Iz).
The expression on the left simplifies
(ii)k=*l.
still
If
we take
,
fi
v, it is
found that
(8)/
(*.)
i. { u) d* =
'<ww>yww (Wi
by differentiating the expression on the
I,
This formula
may be
verified
when k =
for
the denominator
an application of
l'Hospital's rule
shews that,
when
(9)
I -*- k,
z<@ (kz)
J"
V (kz) dz^-^
#/ (kz)
The
right of (9)
Z
(10)
(A*)
#M+1 (kz)
5*11]
MISCELLANEOUS THEOREMS
135
(11)
(for)
#+,
(**)}
<**)}
LM (kz) dz = *
^
^
,
(fa)
f _M _x (fa)
+
the latter equation being obtained by regarding
function of order
e~'
t 'rt
+1
(fa)iU +1 (fa)},
as a cylinder
^ M (kz)
k=
I
/*.
To obtain a
is
in (7)
and
it
found that
i3)
/y <*.)*.
result of
(*)
- - fat^w^y^w^wt
ff,(fa)ff,(fa)
,
The
(14)
making
j/
-*- /*
in this formula is 8
+1 (fa)
| V, (fa) tf,(fa)
^ = g j^
-
-^|
(*)
(fa)| 3^ M+1 g-
^(fa)^M (fa)
2.
The
V^(,)0, V^ d
by \<f*(z)
^^ = 2f^
li
-^^(z)
z
respectively, subtracting
re-
OI&
placing z by kz.
As a
(is)
special case
we have
jj(kz)^ -g{/
w (fa)gKfa)-^(fa)gu
this result will
+1
(fa))+~j
|l
-(fa)L
An
alternative
method of obtaining
be given immediately.
Results equivalent to (11) are as old as Fourier's treatise, La Theorie Analytique de la Chaleur (Paris, 1822), 318 319, in the case of functions of order zero but none of the other formulae of this section seem to have been discovered before the publication of
Lommel's memoir.
Various special cases of the formulae have been worked out in detail by Marcolongo, Napoli Rendiconti, (2) in. (1889), pp. 9199 and by Chessin, Trans. Acad. Sci. of St Louis, xii. (1902), pp. 99108.
136
5"12.
[CHAP.
Lommel's
second method.
An
alternative
method has been given by Lommel* for evaluating some By this method their values are obtained in a
adding the two results
{<&> (z)
The method
consists in
^
so that
(p
[z
tf r+1 (z)
+9
(z)
(,)}
1
+ (p +
{zo S? M+1 (z)
/M
& v+
>
CO}
=*
[V, (z)
(z) <@ v
(z)dz
+ (p-p-v- 2)
+1
(z)
dz
= Z>
and then giving
special values to p.
{<&
(jf)
v (jr)
+ ^M+1 (0
+1 (,)},
Thus we have
e
(1)
(2)
J
^
if
^
)
(z)
*'
{z)
+l ( *>
<*>!
^()?r Wdf- f
~7^
v+
i^(f),(f) +
^(f)?rH (i)i,
As
(3)
^g
=
{, (*) + *V
(*)
<*)}.
rz
(4)
+1 r
(*) <fc
^_ W
+ *V+i (*)}
Again,
p be made zero,
it is
found that
type,
we get
rz
(5)
0*
+
)J
J~
(,) V (,)
M + + 2n) J
A*
- ^M () ^r (#) + 2 2 V,^ ()
* Math. Ann. xiv. (1879), pp.
^+
(#)
530536.
5'12-5*14]
MISCELLANEOUS THEOREMS
if fi
137
In particular,
(6)
= v = 0,
/'*.(#)*.(*)*
where n
1, 2, 3, ....
to
be no simple formula
for
%(z)<& (z)^,
Mag.
(5) xi. (1881), p. 217.
For a
I.
[Scientific Papers,
(1899), p. 516.]
5*13.
The
30
33,
may be expressible
in the
form
M {0 <)}W W {+ ()>+* <*)
A (*)
^M+1 ft
(*)}
f\,
{*(*)}
1
5*14.
A reduction
which
formula for
\' *<&*() dz,
is
5*1 (4),
him
*
1
{*+#/ (*)+(,*
>
**+
%%
(z) 'w:
(* dz.
partial integration,
+2
(z)
+ (* - *>
(^)j cte,
138
[CHAP.
and so
(fi
1)
J
V+
<&,'* (z)
dz
= |>+
W
2
(z)]
J'**
(z*
2
i/
Hence, on substitution,
(/*
1) [
V {? 2
1/
# (z) dz
1)
= |>+3 ^V
+
(*)
- (ji +
3
^+
(*)
#/ (*)]
{(/a
2 [
'
V+
(*)
l)2
M j"V+#
2
(*)
^V(s) ek
= |>+
- (/*+
Z
^+
^(*)
^ ^)]
3
-(p+3)j
By
(fi
z+*'@vi (z)-(ti
1) {
(/*
+ 1) -
v*\
J'rW(t)dz.
rearranging
we
find that
+ 2)
** #,
J*
(5) (2c
= (n+ 1)
(i/
-\
(jjl
1 )2 } [
V
2
#2 (*)
2
<2e
+ * |>+'
and
{*#.' (*)
1 )}
^8 (*)],
5*2.
Expansions in
shall
series
of Bessel functions.
We
now
discuss
2*7.
of such expansions
is
reserved
The
(1)
result of
27
to Y .l<**p!$ j^.1
is
which
is
valid
when
ji is
To
n=
ni
is a series of analytic functions which converges uniformly throughout any bounded domain of the 2-plane (cf. 3'13); and since
-^ {(\z)- J+ M (z)}
it is
v-m
s r(u + w)
n
-
,.
r(At+n +
-o
(2),
-
i)
1
J
Lm-o
*
(1877), pp.
124130.
5*2, 5*21]
MISCELLANEOUS THEOREMS
a constant.
139
the constant
and so
is
the' sum is
is
unity; that
to say
result
is
established.
it is
not
difficult to verify
z,
all
that when the expansion on the the coefficients except that of z* vanish ; but
result.
5*21.
series
of Bessel functions.
The expansion
(i)
($z)-'J (z)
= az)-r( v +
v
i-fi)
5
.to
Qi+2n)r(g+n)
nir(F + l-/*-)r(ir + n +
/j.
l)
T jM+2n
W
.
is
is
a generalisation of a formula proved by Sonine* when the difference v ft are not negative a positive integer ; it is valid when fi, v and v
integers.
It is
most
easily obtained
in the expansion
of i\zY~"J,(z) with the aid of 5'2, and rearranging the resulting double series, which is easily seen to be absolutely convergent.
It is thus found that
(*zY
*'
\z )
n/
TT\
*
l) pto
(fj,
_ 2
+ 2m + 2p)r(n + 2m+p)
/*-h*W
_ v
- I
W
/\
(-)Tfr + m + n)
V(p + n)r(v+l-fi)
;
v
".?.
,Q\r
by Vandermonde's theorem
If
and the
find that
result is established.
we put
v fj.
+ m, we
which
is
is
readily proved
by induction.
140
[CHAP. V
that, if
may prove
is
any
constant,
(8)
(**)-*(*)-*.?.,,#&
x 2F,(/x
+ n, -n;
+ 1;
h?) (fi
+ 2n) J^+^iz).
5*22.
It is evident that (z
with
3-21 (6) ;
we have
*
(1)
(,
+ h) -4.Jp y (z +h)] . J
km
tJm
{g
-* Jw Wz)]
Again, (z
is
analytic except
when
z+h = 0;
and
so,
provided that h
(2)
(,
<
,
\
we have
m-0
+hpj
W(z + h)}=Z
^ !*- [zJ
"*J ***
(V*)}
(2)
These formulae are due to Lommel*. If we take v \ in (1) and = ^ in we deduce from 3*4, after making some slight changes in notation,
i/
equation (4) being true only when J*|<i|*i- These formulae are due to Glaisherf, who regarded the left-hand sides as the generating functions associated with the functions whose order is half of an odd integer, just as exp {\z (t ljt)\ is the generating function associated with the Bessel coefficients.
Proofs of (3) and (4) by direct expansion of the right-hand sides have been given by Glaisher the algebra involved in investigations of this nature
;
is
somewhat formidable.
* Studien iiber die Bessel' schen Functionen (Leipzig, 1868), pp. 11 16. Formula (1) was given by Bessel, Berliner Abh. 1824 [1826], p. 35, for the Bessel coefficients. t Quarterly Journal, xn. (1873), p. 136 ; British Association Report, 1878, pp. 469470. Phil. Tram, of the Royal Soc. clxxii. (1881), pp. 774 781, 813.
5-22]
MISCELLANEOUS THEOREMS
shall
141
We
(5)
now enumerate
jv {W(i + *)} = a +
/ (z V2)
wi
(-) w (i^) ?
m=0
V
1,
y+w (^)
>
and, in particular,
(6)
- 2* 5 fcJ^if^J^ (*).
and then make k -*
If
we
divide (5) by (1
+ &)*"
(i/
we
find that
1)
TO=0
m!
(2),
provided that k
If
<
-*-
1.
we make k
lim
+ 0, we
find,
by Abel's theorem,
(
~ )W(
)W>
+*)}] i
m=0
w^
-
,/y _
w (^),
when
is
an integer. If v
./_
is
is convergent. The convergence is obvious not an integer, then, for large values of to,
m (*)
sm (to - *) 7t
l '
{1
(1/w)}
= _( 4f^sin^ +1
7TTO*
for
is
convergence
absolute.
is
R (v) > 0,
and
if
the condition
is
the convergence
Consequently,
when
R (v) > 0,
and
also
when
(9)
is
any
integer,
I
m=
if
- )m{h * )m
ml
Jv- m (z) = Q.
In like manner,
(10)
R(v) >
1,
and
also
Jv (zJ2) = 2-i X
m-0
kind in
when
v is
any
integer,
we have
r W!
Jv -m (z).
may be substituted provided that h < z (8)
j \ \
and
and k <
\ J
so that
(11)
(*
+ A)-*'
K,
M* + h)} -
2 N-L*l
(i2)
(*
da)
(14)
{*
vd + *>] = (i +
V(l
<-rl^>
Yv+m (zl
Y,
\z
+ k)} = (1 + jfc)-* 5
m=0
&* F_
TO!
lrt
(*).
142
[CHAP.
These may be proved by expressing the functions of the second kind as a linear combination of functions of the first kind by proceeding to the limit when v tends to an integral value, we see that they hold for functions of
;
integral order.
(11)
see that
^ for the
symbol
Y throughout.
of
formulae were noted by Lommel, Studien, p. 87. Numerous generalisations be given in Chapter xi. It has been observed by Airey, Phil. Mag. (6) xxxvi. 242, that they are of some use in calculations connected with zeros of (1918), pp. 234
These
last
them
will
Bessel functions.
When we combine
that,
(5)
and
(13),
+ k) by
X,
we
find
when |\-1|<1,
9. (\m) = X* 2
(
~ )m(V
(15)
m-0
", m.
^h ,
and, in particular,
when X
is
unrestricted,
d)
Bessel functions.
It
/.(x.)
H-fr'-^fr^ w
for
(8) is
may be
v is
(when
observed that the result of treating (14) in the same way as taken equal to an integer n)
that
(17) x
'
-(n-l)l
i-Jfm Yn . m {z).
alternative proof of the multiplication formula has bee.n given by Bohmer, Berliner Sitzungsberichte, xiii. (1913), p. 35, with the aid of the methods of complex integration
An
p.
108,
and
(for
formulae) Wagner, Bern MittheHungen, 1895, pp. 115119; 1896, pp. 5360. special case of formula (1), namely that in which v = l, was discovered by [Note.
Lommel seven
(1861), p. 356.
xxxvn.
The
i
co (r cos
1,
and evaluating
t)
by two
different methods.
is
fir 2ttJ_i|_
(r cos sin vs
M + r)r sin 0)
.
-\JQ-P>
-
d
a rsin.0
.
_W0-f)
\
_i
*r
(-rtrnw* n
(2i+l)!
m=o
(
)m , id$
^
TO =o
'
p. 88.
5*23, 5*3]
MISCELLANEOUS THEOREMS
<f>)
143
and the
is
5~
f
I
4bW J
v J
*W
/""
/
f
I
ir
=
If
^ ff
we
cos tfr)
we compare
and r2
these equations
sin 2 *.]
=l
with
and h replaced by
r2 cos2
5*23.
series
of Bessel functions.
5*22 (7), Lommel has deduced an interesting series of Bessel functions which represents any given Bessel function.
From formula
ft
If
fi is
we have
The repeated
arrange
it
we may
re
by replacing p by
m n, and
then we have
n)l
(v
+n+
l)j
a)
j,
| n.-^^ w = r^+1) M r + m+
1
(v
o^
m!
,
ft)
(v
1)
* +TO
'
Functionen (Leipzig,
and then
*, Fo (,)- Ji
,
(.)
log <*)
-tel>
C^> (^~^M+m W
*{V' (l +i)+V''(-/*)-'<Kn-M)K
and,
53.
An
An
(1)
(*
+ <)- 2 J^iftJ^z),
This formula
is
where z
|
<
{,
being unrestricted.
due
to Schlafli*
and
%\{z + t)=
to Soninef.
*
2
= 06
^(0/.(i)
( JWd.
xvi. (1880), pp.
due
78.
144
[CHAP.
It will first be shewn that the series on the right of (1) convergent series of analytic functions of both z and t when
a uniformly
|*|<r,
R^\t\^A,
numbers
v
where
r,
R,
When
m is large and positive, J ^m (t)Jm {z) is comparable m-V sin vtt {\Ry {r/Ry* V(
)
. .
with
series is
is
r/R)".
When m
large
comparable with that of the binomial and negative (= w), the general
term
is
comparable with
(-)"(*A)'(jAr)"
T(v + n+l).n\
and the uniformity of the convergence
the test of Weierstrass.
follows for both sets of values of
m by
is
=H 2
*
{
" m= ao " 1
J,--! (0
and
that
it is
d-a)J./~< >'-<'>- a
,
30
Hence, when z
|
<
\t \
the series
m=
2 Jv -m (t) Jm (z)
-oo
is
an analytic function
of z
and
which
is expressible
t
as a function of z
+t
F {z +
If
1),
then
we put
if
= 0,
F(z + t) =
Jy-nit)
Jm (z).
evident.
Again,
m in
l =
(1) be changed,
we have
(z),
+ t)= X
-oo
and when
(3)
this result
is
combined with
Y,(z + t)=
* Cf.
we
see that
X
1=-00
Y _m (t)Jm (z).
v
Modern Analysis,
5*3.
5*4]
MISCELLANEOUS THEOREMS
this is
145
When
combined with
(1),
The reader
(4)
when
<
|,
(5)
(6)
(<-*)= X
Of these
results, (3)
(Leipzig, 1868),
when
v is
was given by Lommel, Studien iiber die BesseVschen Functionen an integer while (4), (5) and (6) were given* explicitly by Graf,
;
142.
will
be
5*4.
the expansion
it
is
published
by
SchlafliJ.
||
(cf.
Ym (z)Yn (z).
satisfied
by the product of two Bessel functions, and solve it ( 541) obtain the expansion anew by direct multiplication of series.
in series.
We
shall
then
Given two
normal forms
if
/
,
differentiations
with respect to
z.
* See also Epstein, Die vier Reehnungsoperationen mit BeueVsehen Functionen (Bern, 1894), [Jahrbuch tiber die ForUehritte der Math. 18931891, pp. 845846]. t Ueber die Auswerthung bettimmttr Integrate mit Hillfe von Ver&ndetungen det Integrationsweget (Bern, 1877), p. 13. The authorities who attribute the expansion to SchSnholzer include Graf and
Oubler, Einleitung in die Theorie der BesseVschen Funktionen, u. (Bern, 1900), pp. 8687, and Nielsen, Ann. Sci. de Vltcole norm. sup. (3) xvin. (1901), p. 50 ; Handbueh der Theorie der Cylinder/unktionen (Leipzig, 1904), p. 80. According to Nielsen, Nouv. Ann. de Math. (4) u. (1902),
obtained some series for products in the Iserlohn Programm, 1862. % Math. Ann. hi. (1871), pp. 141142. A trivial defect in Schlafli's proof is that he uses a contour integral which (as he points out) converges only when B(/t+r + l)>0.
p. 396, Meissel
Proe.
||
Camb. Phil.
93100.
'
146
[CHAP. V
dz
and hence
Hence, in
(1)
= - 2Ivw' - 2Jv'w y'" + 2 (/ + /) y + (/' + J') y = (/ - J) (v'w - vw'). the special case when IJ,y satisfies the equation y + 4/y' + 2/>=0;
/'/
but, if
IJ,
it is
(2)
This
*
is
qr+w+wv+J-),
}
_ _ (/ _
;
^
(1),
Appell, Comptes
the form of the differential equation used by Orr Rendm, xci. (1880), pp. 211 214.
in connexion with
see
To apply these
to a
be reduced
it
normal form
(z) as
z
is
9!> v
simpler to
variable
by writing
dz~dO~*'
-^
so that
$%$ + (* - *) / (*) = 0.
satisfied
ft*
^ v*,
2
is
dd
that
is
y = 0,
to say
(3)
>
by
(z)
J (z)
is
where a
m
/*
and
4 (a
+ 2m !)( + 2m) ctn _i (a + ft + + 2m)(a + p -v + 2m)(a- p+ v + 2m)(a- fi-p + 2m) If we take a = /* + v and
1>
^-2^+"r(/*+i)r(i/ +
i)'
we
(_)w
w?o m!r(/* + v + m + l)r( A* + w+l)r(i/+w+l) and the other series which are solutions of (3) are obtained by changing the
signs of either
/i
Qgy,+p+m, Y(
/*
fl
+ y+ 2m + 1)
or v or both
and
v.
"
5*41]
MISCELLANEOUS THEOREMS
considering the powers of z which occur in the product
if 2/*, /M (z)
147
By
is
fi*
Jv (z)
it if
2v and 2
(~)m
(/* + v)
^ ,
then
r t .\ T u\- v J|lW,/rW
m w W
and,
(^Y ++im r (^ + + 2m + 1)
series,
we
find that,
when
2i> is
not a
'
W -wr /!r(2
2
=0 (w!)
2
+ m-Hl){r(j/ + m+l)}
(-)m (zYm (2m)\
'
when
>
is
(7)
J(s)J_(s)=
r (i/ + m + 1) T(- v + m +
1)
reasoning which resembles that given in 4'42, it may be shewn that (6) holds when v is half of an odd negative integer, provided that the quotient T (2v + 2m + l)/r {2v + i + 1) is replaced by the product (2v + + l)m
By
5*41.
by
This method has the advantage that special investigations, for the cases in which fi* = v* and those in which y. + v is a negative integer, are superfluous.
The
coefficient of (-)"*
(yY +v+sm
convergent series
n=nn\T(v + n + l)
(-)w (**)"+w
= n!
Y(v
_l r (/* + w!ir/
Cw
(-'-)w_n .(-/A-m)
fl
m!
T(/*
+ m+
p (i> + w +
l)m
l)
(n
+v+m +
is
used to
v,
sum the
finite series.
and
n\ (1)
and
(-)m
m]
and
.
(7) of 5*4.
----H--'
'
*~
S
148
[CHAP.
earlier writers; it
This obvious mode of procedure does not seem to have been noticed by any of the was given by Nielsen, Math. Ann. lii. (1899), p. 228.
series for
cos z and J {z)e,mz were obtained by Bessel, Berliner Abh. 1824, and the corresponding results for Jv (z) con z and Jv {z)smz were deduced from Poisson's integral by Lommel, Studien uber die BesseVschen Functionen (Leipzig, 1868), pp. 16 18. Some deductions concerning the functions ber and bei have been made by Whitehead, Quarterly Journal, xlii. (1911), p. 342.
The
{z)
[1826], pp.
38
39,
More generally, if we multiply the series for J^(az) and an expansion in which the coefficient of ( w orb" (fcY +v+am
m
,o!
Qitn2n A2n
J (bz),
v
we
obtain
is
and so
(2)
J, (az) Jv (bz) -
V2
'
i/
+1
fr/q')
and
series is
One
another
case of reduction
is
is
the case b
the case
b = ia,
provided
r/
v
w/
;
\
'
v
wt=0
r(|m
-
i)
- %
=0 w! T (v +
m + 1) r
(i/
+ 2w +
1)'
(5)
/fa*)/
fart- v
in (3)
a
If
(6)
we take a = el"*
ber (z)
+ bei
(z)
- 2
ptt
it***"*
3*8.
formula
(53).
5*42]
MISCELLANEOUS THEOREMS
(3), (4), (5)
149
(5)
The formulae
xv. (1906), pp. 490 497 and Monatshefte fiir Math, und Phys. xix. (1908), pp. 164170, from a consideration of the differential equation satisfied by Jv (az)Jv (bz).
Some
types
series
have been given, Quarterly Journal, xli. (1910), p. 55, for products of the but they are too cumbrous to be of any importance.
By
(7)
(2), it is
which 2v
is
The
and
Ym {z)Yn {z)
is
have been the subject of detailed study by Nicholson f; the following outline of his analysis with some modifications.
an
We
have
irJ, (z)
Yn (z) = 1
to
J* (,) Jw ()}
where
i/
is
to be
made equal
Now
l~
{
(~)r (^>t+,,+8r
T(p + v + 2r + l)
and
r=0 Lr:
We
n-l
oo
.
2 and X
have
* Proc. London Math. Soc. xxv. (1894), p. 06 8 je also Cailler, ilfm. d ia Soc. de Phys. de Geneve, xxxiv. (19021905), p. 316. t Quarterly Journal, xliii. (19i2), pp. 78 100. The expansion of Jq(z) Yq (z) had been given previously by Nielsen, Handbuch der Theorie der Cylinderfunktionen (Leipzig, 1904), p. 21.
;
150
[CHAP.
When
by n + r,
seen that
(1)
wJ,(b)Yu <*)%
9
r!r</*
n +* r
+ r+l)
l) r
{2 log (*)
2^r (p
+ n + 2r + 1)
l)
-^r(^ + n + r+l)-^r(/* + r +
right
is
a continuous function of
fi
at
/tt
=m
where
fi
= 0,
1, 2, ...,
and so the
(1).
series for
irjm
(z)
Yn {z)
is
obtained by replacing
by
m on
The
the right in
series for
Ym (z) Yn (z)
|_
for
dfldv
jn-m, v-n
in a similar manner.
The details of the analysis, which is extremely laborious, have been given by Nicholson, and will not be repeated here.
5*43.
Jv (z).
integral ( 2*6) for
A generalisation of
applying the formula*
Neumann's
Jn
(z) is obtainable
by
eo**-+Jo
to the result of 5
-
cos fr
41 ;
v)
0d0 = 2iL++M
that
R (jM + v) > - 1.
then evident that
It is
J(*)J(*) =
so that
when
J?
(/u.
/M (s) J (z) = -
TJfl
Cauchy
for
5-43-5-51]
MISCELLANEOUS THEOREMS
151
R n) > 1, then JM (z) Jn (z) = *i~L P'j^ (2z cos 6) cos + n) 0d0, T
(jj,
(fi
Jo
/*
and
unrestricted.
Formula
(1)
was given by
Schlafli,
p. 142,
are both
integers; the general formula is due to Gegenbauer, Wiener Sitzungsberichte, cxi. (2a),
(1902), p. 567.
5*5.
series
of products.
5 2)
(1)
is
(**r
r{/M
+ v+1 y
m-0
m.
v are
The formula
is
true if
if
fi
and
R(fi
+ v + l)> 1.
From
5*2
we have
^^
;
we multiply by
cos
(fi
v)0 and
to-0
2^+
ml
of
follows
result
by evaluating the integral on the left for other values may be established by analytic continuation.
is at
lxxv.
5*51.
Lommel's
series
An
expansion derived by
Lommel* from
the formula
i8
j>
so that
(*) <fc 2 J'xJK-i
*
p.
(*
+ 2) JV2 (*)1
ii.
The
;
532
MUnchener Abh.
(1870), pp.
632633
xiv. (1878),
152
[CHAP. V
Hence, by
(1)
we have
**
{./
_ x (z)
- J^ (z) J (z)} = 2
n-0
(i;
+ 2n) JV (*),
2
;
beginning of the
fluous.
by adding on terms at the on taking zero as the lower limit when R (v) > is superseries, it may be seen that the restriction R,(v) >
If we take in turn we have ( 3*4)
(2)
= \, v = f
results so obtained,
-= 2
n=0
(n
+ *)JV(*).
(3)
*.
by taking i>=
1,
(-r<n + *)/+(*).
while,
(4)
we
see that
\z* {J *
(jr)
+ J (5)} = I
is
(in
-0
1) An+x
(*)
2
-o
for it is evident that
J\ + (s)
a
^-
2 n=0
2 eJ\+n (*) = 2 2
00
e /"+ (*)/'+(*)
=0
n
t/"
= 2
t=0
Jv+n+i (*)}
= ivJ?(z)\z,
and so. when iJ(v) >0, we obtain a modification of Hansen's formula (2 namely
00
-
5),
(5)
2
=o
en
f*
fit
An
bound
Jv (x)
\,
will
be given in
found that
13 42.
By
taking
v \,
v 2
m=o
it is
, \ t e J n+i (*)
2 dt = - P sin** 1 "
tJo
f
|_
2 sin2 *]*
7T
t
2 f*
7T.I
_ ,cft
*
Jo
and
so
(6)
2
n=0
/'
n+J (*)
= -i(2*), T
where, as usual, the symbol Si denotes the "sine integral." This result is given by Lommel in the third of the memoirs to which reference has been made.
5-6]
5*6.
MISCELLANEOUS THEOREMS
Continued fraction formulae.
153
Expressions for quotients of Bessel functions as continued fractions are deducible immediately from the recurrence formula given by 32(1); thus,
if
/,_,(*)
Jv (z)
it is
*/_,(*)
\z-j{(v+m-l)(v\-m)}
\z
J + m+1 (z)/{v + m)
v
This formula
is
J v+m\ z )
'
J ^ (z)
v
2v/z
'
(1)
was discovered by
is
of
v.
An
that, if Q {z)
then
Jv+1 (z)
(4)
z
2(i/
z*
Jv (z)
/<
J*
z2
zJv+m+1 (z)
J, +m (z)
x
'
W
The
,k\
^+2 (*) _
'
Bessel functions in
these formulae
may
cylinder functions.
It
that,
when
m -*-
<x>
may be
neglected, so that
,x (b)
AA^L-^ll
J_i(*)~"
1
i^l\ V (v
+ l)}
$*/{(+ !)(+
1
2)}
* Berliner Abh. (1824), [1826], p; 31. Formula (2) seems not to have been given by the earlier writers; see Encyclopedic des Sci. Math. n. 28, 58, p. 217. A slightly different form is used by
Graf, Ann. di Mat. (2) xxin. (1895), p. 47. integral values of t Zeitschrift fiir Math, und Phys. n. (1857), p. 142; Schldmilch considered
v only.
J Studien
fiber die
BesseVschen Functionen (Leipzig, 18G8), p. 5 see also Spitzer, Archiv der (1858), p. 332, and Gunther, Archiv der Math, und Phy*. lvi. (1874),
;
154
[CHAP.
though it happens to and a rigorous proof of the expansion of a quotient of Bessel functions into an infinite continued fraction will be given in 9'65 with the help of the theory of " Lommel's polynomials."
be
so,
[Note.
the fraction
is
is that,
even though
pm /qm tends
;
to a limit as m--oo
it is
aw ^">
^m+1
this
may be
seen by taking
pm =m+amm,
q m =m,
a,=-l.]
The reader will find an elaborate discussion on the representation of Jv (z)/Jv -i (z) as a continued fraction in a memoir* by Perron, Munchener Sitzungsberichte, xxxvn. (1907), 504; solutions of Riccati's equation, depending on such a representation, have pp. 483
been considered by Wilton, Quarterly Journal, xlvi. (1915), pp. 320323. The connexion between continued fractions of the types considered in this section and the relations connecting contiguous hypergeometric functions has been noticed by Heine, Journal fiir Math. lvii. (1860), pp. 231247 and Christoffel, Journal fur Math, lviii. (1861),
pp. 9092.
5*7.
It
(i)
W -iin r a^
the
..i'1
J
>
M;+i;-^).
\ and
We shall prove
fi
this result for general (complex) values of v and z when tend to infinity through complex values.
If
1/8, fi
= 1/t),
(m -f
K
is
This
is
/,
=-,
[(1
+ rh) (1 + rv )\
,
a continuous function of 8 and rj and, if 8 r) are arbitrary positive (less than 2 z j"1 ), the series of which it is the (wi + l)th term converges uniformly with respect to 8 and 77 whenever both 8 ^ 8 and 17 ^ i For the term in question is numerically less than the modulus of the (m + 1 )th
;
numbers
r^,+\)
& W*., ^
is
>
+ 1 K^o),
>
uniform, the
sum
of the terms
is
a continuous
(1908), pp.
memoir 8588.
is
t Leipziger Abh. n. (1855), p. 252 ; see also a Halberstadt dissertation by F. Neumann, 1909. [Jakrbueh ilber die Fortschritte der Math. 1909, p. 575.]
5-7, 5*71]
MISCELLANEOUS THEOREMS
($, rj)
155
is
the
sum
/
that
is
to say
^o\r(v + l)
\B
r}'
m . m!r(i/ +
m+l)
and
5*71.
known that solutions of Laplace's equation, which are analytic near the origin and which are appropriate for the discussion of physical problems connected with a sphere, may be conveniently expressed as linear
It is well
Pn (cos0),
rn
these are
coordinates
Now
the cones approximate to cylinders, and the structure resembles the structure
associated with cylindrical-polar coordinates
(kp)^m<f>.
when r and n
is
small
in such a
way
that r sin
(i.e.
we must expect
The
Hansen's
limit.
of this type
is
limPn (cos^)=/ (4
result,
This
investi-
gated by Mehler, Journal fiir Math, lxviii. (1868), p. 140; Math. Ann. v. (1872), pp. 136, 141 144; Heine, Journal fur Math. lxix. (1869), p. 130; RayJeigh, Proc. London Math.
(1899), pp.
6164 Proc. Royal Soc. xcii. A, (1916), pp. 433437 [Scientific Papers, 338341 vi. (1920), pp. 393397]; and Giuliani, Giorn. di Mat. xxn. (1884), 239. The result has been extended to generalised Legendre functions by Heine
; ;
easier to prove
it
when n tends
would be
to infinity as
Pnm (cos 0)
(sin
m 0)/m
!.
3649.
156
[CHAP.
We
Pn (cos z/ri) =
modification that
sjFx
n,
+1
1
is
sin2 \z\n)\
we use the
inequality
|rin(f/)|*f|(*/n)|,
when
|
\z
$ 2 n
|
,
|
series
;
& (where B
is is
n,
n+
sin2 \z\n\
,F, (1/S
1/S6
+1
|
^$
2
1
2
!
),
an arbitrary positive number less than z _1 atid the comparison made when \n\>lj8 The details of the proof may now be left to the reader.
|
When
the proof.
is restricted to it is
for
Pn (coaz/n)
terminates, and
convenient to appeal to Tannery's theorem * to complete This fact was first noticed by Giuliani the earlier writers took for
;
limit.
between
+1
and
1 or not
x (between
n,
;
and
ir)
we have
sin" **/).
Pw- (cos -J =
so that
(2)
'
'
^_
& (-
n+ 1 m + 1;
(as),
lim n m
Pn~m (cos -) = Jm
_,
so that
(3)
lim n m
second kind
may be deduced
;
Qnm
~
in terms of
,
Pnm and P~ m
it is
(4)
w _.ooLsm
hm
,.
m sinri7r rn -^
(
m+
Km
(z).
This formula has been given (with a different notation) by Heine +; it is most easily proved by substituting the integral of Laplace's type for the Legendre
function, proceeding to the limit
(5) of
62 2.
* Cf. Bromwioh, Theory of Infinite Series, 49. t Cf. Barnes,' Quarterly Journal, xxxix. (1908), p. 109
the equation
1
is
2r(-m-n)
in Barnes' notation,
BinVlir
nnir
P~" Q n= r(l-m +
P-"*
?j)
r(l + i + n)
which is adopted in this work. + Journal fur Math, i/xix. (1868), p. 131.
5*72]
MISCELLANEOUS THEOREMS
slightly different from those just discussed, is
157
Another formula,
(5)
this
is
K
.
"/(il*)" /,(
it
);
may be proved by
.
formulae, namely
Pn (cos 6) = cos" %6 F
2
(-
n,
-n
- tan
\6).
[Note. The existence of the formulae of this section must be emphasized because it used to be generally believed that there was no connexion between Legendre functions and
Bessel functions.
in his Elementary Treatise on Laplace's BesseVs Functions (London, 1875), p. vi, that "these Bessel functions] are not connected with the main subject of this book."]
it
Thus
and
5*72.
completely different method of establishing the formulae of the last by Mehler and also, later, by Rayleigh this method depends
;
(cos
i sin
I TT.'O
V cos $) n d<j>
fifo
Since
n log
uniformly as n
-- oo
{cos (z/n)
+ i sin
(z/n) cos
<} -*-
iz cos
<f>
when
lim
Pn (cos z/n) - - f *
=J
(z).
Heine f and de Ball J have made similar passages to the limit with integrals of Laplace's type for Legendre functions. In this way Heine has defined
Bessel functions of the second and third kinds
results in 6*22
;
reference will be
made
(z).
to his
when we
Mehler has also given a proof of his formula by using the Mehler- Dirichlet integral
2 Pn (cos6) = If n<f>=yjr, it
; o
*<&4*L
V{2 (cos <p -COS0)}
may
be shewn that
-^
C0S
little difficulty
is
an im-
Various formulae have been given recently which exhibit the way in which
* Journal de Math. (3) i. (1875), pp. 384385; the formula actually given by Laurent is trioneouB on account of an arithmetical error. t Journal fur Math. lxix. (1868), p. 131. See also Sharpe, Quarterly Journal, xxiv (1890)
pp.
383386.
X Attr. Nach. cxxvm. (1891), col. 1
1.
158
[CHAP. V
infinity.
degree tends to
is
Pn -(cos0)
= (n + )"' (cos 0)- [Jm (x)
+ sin
2
\0 {^x
(x)}
.],
where x = (In
1) sin \0.
Other formulae, which exhibit an upper limit for the error due to replacing a Legendre function of large degree by a Bessel function, aref
(2)
Pn (cos
rj)
iir- 1
Qn (cos rj)
= ^(sec rj)
+ h) ^n v}
iY
{(
+ ) tan rj\]
4^i-\/(sec v)
+ 7r{R(n)
+ ^'
(4)
K
+
{(n
+ \) tanh f
V(sechg).e-(+*>*
ffl3
R{n)+$
; ,
and (4), f > the numbers lt 2 < t) < where, in (2), 3 are n may be complex provided that less than unity in absolute magnitude, and its real part is positive. But the proof of these results is too lengthy to be
%tt, and, in (3)
given here.
5*73.
by Hansen's formula as a limit of a hypergeometric function has led Olbricht j to investigate methods by which Bessel's equation is expressible as a confluent form of equations associated with Biemann's P-functions.
The
fact that
expressible
If
dz
V
is
z*
'
z~*Jv (z),
fa,
b,
z-Y
c,
v (z),
and compare the equation with the equation defined by the scheme
1
Pio,
l',
* Proc.
,
ff,
y,
z\,
-I
7.
London Math. Soc. (2) xm. (1914), pp. 220 221 eome associated results had been obtained previously by the same writer, Proc. London Math. Soc. xxxi. (1899), p. 269. t Watson, Trans. Camb. Phil. Soc. xxn. (1918), pp. 277308 Messenger, xlvii. (1918),
; ;
pp. 151160.
5*73]
MISCELLANEOUS THEOREMS
159
namely
d* y
Q- a -a
\
(
'
\-fi-fi'
l-y-yl
z-c
(b
)
dy
dz
yy' (
dz*
z-a
z-a
z-b
+ fifi'
aa'(a-b)(a-c)
-c)(b-a)
-a)(c- &)
z-b
(z
z-c
y
) )
-a)(z -
b) {z
c)
= 0,
we
a
while
b, c, fi, fi', 7,
= 0,
= v fi, + 1)
0,
= v fi,
way
that
2
y tend
to infinity in such a
2/*
fi
remain
finite (their
sum being
while
yy - $b fifi'
-2ifi,
y,
\
7'
We
2ifi,
fi,
limPJ
~*
v-p,
2
z\,
J
l-v-fi,
where
7,
-fi,
7,
- M + i V{0* + ) + tf
is
(
2
}-
0,
ifi,
fi,
oo,
J
limP<
"*"
v-fi,
y,
z\
)
i-v-fi,
-fi,
y,
as before.
is
scheme
for
(z)
0,
qo,
-4a&
0,
lim
pj
\v
a-\v,
fi-\v,
zA.
J
ilZl)
those which have
Note.
It
[-% v>
v+l-a-fi.
Olbricht has given other schemes but they are of no great importance and
now been
whose solution
may be
when the
zero.
invariants
g2 and g3
made
to tend to
CHAPTER VI
INTEGRAL REPRESENTATIONS OF BESSEL FUNCTIONS
6*1.
In this chapter we
and Bessel's integral ( 2-2). By suitable choices numbers of elegant formulae can be obtained
also be applied in
which express Bessel functions as definite integrals. The contour integrals will Chapters vu and VIII to obtain approximate formulae and
Jv (z) when
z or v is large.
It happens that the applications of Poisson's integral are of a more elementary character than the applications of Bessel's integral, and accordingly
we
shall
now study
which we
Poisson's
The simplest
of the formulae of
33
is 3'3 (4),
a single exponential under the integral sign, while the other formulae contain
circular functions,
shall
which are expressible in terms of two exponentials. We therefore examine the circumstances in which contour integrals of the type
r
e izt
J a
Tdt
it is
not of
of
z.
z,
The
result of operating
differential operator
V,, defined
b
in 3*1, is
as follows:
V,
izt
Tdtl
= z v+i
izt
lzt
T{\h
dt
b
(2i/
izt +1 1) iz j e
Ttdt
= iz"+
*
T (t* -
1)1
+ iz+*
e izt \{2v
1) Tt
- ^ {T (i 2
1)}1 dt,
Math. Ann. i. (1869), pp. 473 485. The discussion of the corresponding integrals for Iv(z) is due to Schlafli, Ann. di Mat. (2) i. (1868), pp. 232242, though Schlafli's results are expressed in the notation explained in 4-15. The integrals have also been examined in great detail by Gubler, Zurich Vierteljahrsschrift, xxxm. (1888), pp. 147 172, and, from the aspect of the theory of the linear differential equations which they satisfy, by Graf, Math. Ann. xlv. (1894), pp. 235 2C2 lvi. (1903), pp. 432 444. See also de la Vallee Poussin, Ann. de la Soc. Sci. de
and
K v (z)
140143.
6-1]
INTEGRAL REPRESENTATIONS
integration.
161
by a partial
if T, a, b
= (2v + l)Tt,
izt
T(t2 -l)
= 0.
is
a constant multiple of
I)" - *,
and the
t
latter
either so that
initial
it is
shews that we may choose the path of integration, a closed circuit such that e izt (t2 l) v+i returns to its
e
izt
value after
(&-
l) v + i vanishes
at each limit.
A
t
contour of the
first
type
is
=1
clockwise.
And,
if
we suppose
is
+ coi and
1,
1 counter-clockwise (Fig. 1
and Fig.
2).
If
we take
a, 6
1, it is
Fig. 1.
Fig. 2.
R(v +
> 0, and we
(t
2
function
the phases
and
1 to
1.
We
therefore proceed to
/(1+,
-1-)
z"
e (a - 1)"-* dt,
z"
eizt
(t*
- 1)--* dt.
at
* It is supposed that v has not one of the values \,\,\\ 1, and both integrals vanish, by Cauchy's theorem.
...;
162
[CHAP. VI
It is to be observed that, when R (2) > 0, both integrals are convergent, and differentiations under the integral sign are permissible. Also, both
v.
first
we
expand the integrand in powers of z, the resulting series being uniformly convergent with respect to t on the contour. It follows that
/"(1+.-1-)
00
zu
Ja
2
J*(P-l)'-*dt= 2
m=o
or
;m z Vrm,
an odd function of t according as m is even or odd; and so, taking the contour to be symmetrical with respect to the origin, we see that the alternate terms of the series on the right vanish, and we are
then
left
t"
J
J* (t* a
1)"-* dt = 2
= 2
on writing
u
is
t
;a
t* (t2
1)"-* dt
Jo
,-(1+)
'
w -o
(2m)
,
!
u"-i(u-l)-*du,
Jo
= <Ju;
u and u
vanish
when
To evaluate the
integrals
R(v + %)>0;
from
to
1
the contour
going from and on the second part, returning from where, in each case, the phase of 1 u is
1 to 0,
we have we have
zero.
*-* (u
1)"-*
Jo
du
= =
{-(-
. 2t COS
- e<"- '}
n
/
um -l (1 - )"- du
'
.
Jo
Z>7T
Now
(1+
are analytic functions of v for all values of v) and so, by the general theory of analytic continuation*, this result, which has been proved when (v + ) > 0,
v.
Modern Analyst*,
5-5.
The
it
when
+ )~-0,
Jo
which
is
++J Jo
(>,
+ n + i)um-*
{M
1)
v+ M -i.
then expressed in terms of an integral of the same type in which the exponent of
6-1]
INTEGRAL REPRESENTATIONS
for all* values of v,
oo
163
Hence,
r(l+, -1-)
J
(\m z v + *m Y( m
</ (z).
a.
1\
l)
m-o (2m)!r(j/+m +
is
j.
w - ^"ty /r
|
and
Next let us consider the second type of contour. Take the contour to lie wholly outside the circle 1 = 1, and then (t 2 l)" -i is expansible in a series of descending powers of t, uniformly convergent on the contour thus we have
1
m=o
(|
and
lies
between
it and + ^tt.
i')
Assuming^ the
/(-i+,i+)
J<*>i
we have
zy
TVA
A-m\ r(-i+.i+)
V)
J
oot
m -Q Wil (-
But
J
oo i
aa
exp ia
where a
is
Hence
,J
r
*><
'
1+)
*- <* - 1)-'
s * - ,-o
"
r(
v)
-h
1 1)
to express
T (2m
2i>
1) in
terms of
* If v
-J
is
way
u= 1.
/
'
|
t
its
To
justify the
A'.
tzt
dt
is
convergent
let
- l)" it follows that, wheu we independent of t, such that the remainder an integer We then terms of the expansion does not exceed e/K in absolute value when M ^ after have at once
value be
-^
number
e,
we cau
find
cot
aud
sum
Cf.
of
an
infinite series.
12-15.
Modern Analysis,
164
Thus,
[CHAP. VI
R (z) >
and v + %
is
Next consider
(-1+.1+)
eizt
{t*-iy-*dt,
J oo
of z which
is
exp ( - 1*>)
when
\ir +
and,
if
fyrr
+ co;
| J
arg 2 < 77-, the contour can be deformed into the second of the two contours just considered. Hence the analytic continuation of J- V (z) can be defined by the new integral over an extended range of values of arg z; so that we have
z
is
(3)
J-*(')~
giving
co
r^-\,>" n7V/'(-i+,i+)
f
\j?rm
.1Tl
X \-%)
f ociexp(-Ju)) J
**V-iY-*dt,
\ir
\ir + to tr
and
co.
By
for
a suitable value*,
we can
and
it.
3,
\/ /\
Fig. 3.
in
which
By
it is supposed that the radii of the circles are ultimately taking each straight line in the contour separately, we get
made
indefinitely small.
J-> {z)=
T{
K^n\)
ei
[/I
+
"
' sni(v ~
h)
(1
-^" *
i
Je * e-3rC4
1
J)
(1
- #)-\
dt
izt e -Hi>-l)
(\-t*y-l dt
e*t
e^C-V (l-Py-h dt
(1
+
* If
I
f*
e* **("- i)
- t*y~i dt~]
*-),
w be
I
sentation of
J- v (z)
an appropriate value (greater than any preassigned value of arg z may be obtained.
a repre-
6-11]
INTEGRAL REPRESENTATIONS
bisecting the third path of integration
t,
165
by
it,
On
and replacing
is
-t, t,
it
respectively,
we
J- V (z), due
for
Jv (z)
S
the formula
" f jo
J_ v (s)=
2{
mvn
e-'til
+ tty-hdt+f'cosizt + v^.il-fiy-idt],
Jo
it is
and,
if
found that
'V-Tb$&l[fl'*<*< 1
- f>rl
*-K'
it
rm,:l+ *r ~ >
*l
a formula which was also discovered by Gubler, though "Weber t in the case of integral values of p.
After what has gone before the reader should have no difficulty in obtaining a formula
closely connected with (1),
namely
(6)
*w- r -*k^/>-ir.~>.*
it is
in which
t2
vanishes
when
is
on the
real axis
i
on the
right of t\.
6*11.
Taking R (z) > 0, let us modify the two contours of shewn in Figs. 4 and 5 respectively.
Fig. 4.
Fig. 5.
By making
those portions of the contours which are parallel to the real xxxm.
(1888), p. 159.
* Ziirich Vierteljahrsschrift,
t Journal fUr Math, lxxvi. (1873), p. 9. Cf. Hayashi, Nyt Tidsskriftfor Math. xxni. b, (1912), The formula was examined in the case v=0 by Escherich, Monatshefie fiir Math,
(1892), pp. 142, 234.
und Phys. m.
166
axis
[CHAP. VI
we
0)
J.{z)
2ttiT(i)
d+)
e
izt
(t
l)"-i dt
+
J
e izt (t2
-1+ooi
iy-i dt
_J l+i
(2)
J _{*)-
2 7nT()
r(i+)
e
.'
rzt
(t
f(-i+)
I)""* dt
+
J
eizt (t2
-1+ooj
- 1)*-* dt
l+ooi
In the
first
result the
t
2
the phase of
1
is
to
be
many-valued functions are to be interpreted by taking at A and to be + ir at B, while in the second the
phase of
t-
at
and
is
it
at B.
in
t
1
To avoid confusion it is desirable to have the phase of t2 1 interpreted the same way in both formulae; and when it is supposed that the phase of
is
1
(3)
+ 7r at
is
by
J-.(*)1
2r(i)
/(i+) d+)
e
J 1+ooi
izt
(t
1)"-* dt
f(-i-)
e
J -1+ooi
izt
-i
+ e~ vin
(t
1)"-* dt
J
2
In the
1
(2),
has
On
(4)
3'61
we
see that
T(\-v).(\z
y[^
(5)
unless v
is
(1)
and
We
value
and
when
v has
an integral
HJ
(z)
lim
HJV
(z)
dt
and similarly
for
Hn
(z).
6-12]
INTEGRAL REPRESENTATIONS
167
(5)
As in the corresponding analysis of 6*1, the ranges of validity of (4) and may he extended hy swinging round the contours and using the theory of
Thus,
analytic continuation.
if
have
Z)V
*""
(6)
(z)
T(
*~rn?
f
" 1_)
(*
W*
dt '
while, if
(7)
gw(#)B E fl-^-y):
7r * * <.)
^(^-i)"-*^
*exp(-t)
lies
(7),
the phase of z
between
\ir + <u
and
H^ (z) when
s has
2tt,
and of
{z>
(z)
when arg
If co be increased beyond the limits stated, it is necessary to make the contours coil round the singular points of the integrand, and numerical errors are liable to occur in the interpretation of the integrals unless great care is taken. Weber, however, has adopted this procedure, Math. Ann. xxxvn. (1890), pp. 411 412, to determine the for-
HM
V
- z),
HW
V
- z)
with
HJ (z),
it
V V) (*).
The formula
2i*F
(z)
possible to express
Yv (z) in
terms of loop integrals, and in this manner Hankel obtained the series of 3'52 for Y(); this investigation will not be reproduced in view of the greater simplicity of Hankel's other method which has been described in 3-52.
6*12.
In the formula 6'11 (6) suppose that the phase of z has any given value between tr and 2tt, and define by the equation
arg z
so that
a)
\ir<fi<
+ @,
\tr.
Then we
shall write
ir
+ fi
/
as
describes the
contour
(1)
and
it
*,"(*)-
iTilt
^x*) -.Ur^-K
+ \iujz
has
its principal value.
vSe* I'-l**
t>
f o+i
in \ _ i
1
^)
if /3
du
Again,
be a given
\ir
<
arg z
<
f ir
+ 0.
168
[CHAP. VI
Similarly*, from
where
is
Sincef, by
by restricting
v so that
f 7r 4- /S < arg z <%ti- + /3. #_W (*) = "* # (*), it follows that we lose nothing R (v + |) > and it is then permissible to deform
;
oo
for the
integrals taken round a small circle (with centre at the origin) tend to zero with the radius of the circle J.
On
(3)
^-(i)
/3
'tf/.
r-^(l + )
.
where
may
ff.(,).(l.)
T?
_r
-fir
Jt
^^ (!_)
*,,
where
# may
R(v + $)>0,
+ <arg*<lir + .
The results (3) and (4) have not yet been proved when 2v is an odd positive integer. But in view of the continuity near p = n+$ of the functions involved (where n = 0, 1, 2, ...) it follows, as in the somewhat similar work of 6'H, that (3) and (4) are true when v \, f f .... The results may also be obtained for such values of v by expanding the integrands
,
,
z,
(3)
and
(4) ar_e of
Jv (z)
is
These
A useful
^*)~
H<*(,) "
r(y+1)r(|)
W
( 6) *
W=
t
^
z" 1
{
s^h,,
r"eW.
e
dtf >
To
+ 1 = 6"^
- u), ~ U( ~
- l = 2e
w (1
- iujz).
T{h - v)
is
2
vi,
u)V
~^ i+i^
hdu
12-22.
V
6*13]
INTEGRAL REPRESENTATIONS
169
J "W-r(
y +|)T(i)Jo
sin+^
cos-*
f**
.
**'
to
-*-
cos (
- vO + \6)
^ cntede
were
R (v + 1) > 0,
that the
expressions on the right are solutions of Bessel's equation which behave in the
is
when
fi
= v, has
89102.
from the aspect of the theory of asympby Brajtzew, Warschau Polyt. Inst. Nach. 1902,
The
nos.
1,
6*13.
integrals.
Some elegant definite integrals maybe obtained to represent Bessel functions To construct them, observe that, when z is positive (= x) and the real part of v is less than \, it is permissible to take a = ?r in 6*11 (6) and to take to = - \ir in 611 (7), so that the contours are those shewn in Fig. 6. When, in addition, the real part
is
of v
greater than
of
it is
manner
to
612)
taken twice while the second contour consists of the real axis from 1 to oo taken twice.
oo
^A-'J
Fig. 6.
<rS,\
r,^< l'^
" e-^-^) f>< (* 2
We
HP (w) = r
#<*> (as)
-.{?7fr^- (1
1)"-* dt,
=-
^~$;$^
(1
- e2 <"-i>)
te < (
- l)r~* dt,
611
(7)
by replacing thy t.
170
[CHAP. VI
given by 361
(7).
It follows that,
when x >
(2)
bj* (x) - r
"
/
v
iT{ i-
v)
r(i).($xy
f"
J i
11
'-ir*
- i)- + *
so that
w
(4)
,o\
w " r(i-.or(*).(iy
2
(x)
ain (xt).dt
(*
2
=-
f <*<*>*
.'i
r<*-iOr(*).tt*) F
(*-ir**
Of these results, (3) was given by Mehler, J/a*A. Ann. v. (1872), p. 142, in the special case v==0, while Sonine, Math. Ann. xvi. (1880), p. 39, gave both (3) and (4) in the same special case. Other generalisations of the Mehler-Sonine integrals will be given in & 621.
6 "14.
Hargreave and Macdonald.
it is
to
(6)
that
t2
lies
between
and
\n.
and
so,
when
v+i
is
a positive
integer,
we have
r(v+i)r(i)
(1+
When
+\
is
representation of HJV>
not a positive integer, the last expression may be regarded as a symbolic (z), on the understanding that/ (D) (e^'jz) is to be interpreted as
Afci
i
I
M /(it)dt.
*
J i+i
Consequently
(1)
(1)
(?)=
r(v+i)r(i)
,
(l+^)
(i
and similarly
(2)
#<
W - r(+J)r(j) m
+i^- 5Elti22L'
'
so that
<
3>
jr
<
4>
r'Vfi$fmO +Br**T-
6-14, 6'15]
INTEGRAL REPRESENTATIONS
from
(4)
171
The
unless
series obtained
it
terminates; the series obtained in a similar manner from (3) converges only
when R(y)>\.
The expressions on the right of (3) and (4), with constant factors omitted, were given by Hargreave, Phil. Trans, of the Royal Soc. 1848, p. 38 as solutions of Bessel's equation. The exact formulae are due to Macdonald, Proc. London Math. Soc. xxix. (1898), p. 114.
An
v, is
(4).
If
is
any
positive integer,
v
we
v) H W(z) = T ^~ 7rtr ;^
C?/
U+)
f
J
l
(<2-l)"--l(-).(l + Z)2)e*<fc
+ o
= (-)"r(W)(^
so that
(5)
p
(1+i)2)W
{ti
_ iy - n . iiztdtf
Bv ,_ r(.+t-.).) F (1 +i)J).
m^ H^
^
',
w)
kind, and so
(z)} -
^^ = r(y+rV+i) (K
result,
"y
^-"
This
v
proved when
;
R (z) > 0,
is
by the theory of
analytic continuation
it
when
in the
= r and
when
xvm.
(1882), p. 338
when
= n+%;
case
= n+%
of Bessel functions)
the result was given slightly earlier (without the use of the notation proof by Glaisher, Proc. Camb. Phil. Soc. ill. (1880), pp. 269271.
based on arguments of a physical character has been given by Havelock, Proc. London Math. Soc. (2) II. (1904), pp. 124125.
6*15.
Iv (z) and
iz,
v (z).
If
i
we take
|
we
find that,
when
arg z
<
\tt,
2tt.
-1
Fig. 7.
4
of the circles
If
we take
R (v + ) >
may be taken to be the contour of Fig. 7, in which the radii may be made to tend to zero. We thus find the formulaf #)r r( (* 2
T-" {2) =
*2^r<i)
(1
dt
- ""
+ e-3vni )
e~zt (1
- <2)"-* dt
* Ann. dt Mat. (2) i. (1868), pp. 239241. Schlafli obtained the results (1) and (2) directly by the method of 6-1. t Cf. Serret, Journal de Math. ix. (1844), p. 204.
172
[CHAP. VI
and of
Now, from
3'7l (9),
we have
(2)
and so
(
-T<MmL'*<
/_, (.)
to say*
-*r-
- irt
3)
is
_ /.
W = r <*-*?*"<* .c*"
v
((a
that
(4)
K
K
v
(,)
^j^ffi J"*-*
(*
- I)""* dt,
a result set by Hobson as a problem in the Mathematical Tripos, 1898. The and args < \ir. The reader will formulae are all valid when R (v + \) > find it instructive to obtain (4) directly from 6*11 (6).
j
6*16.
K
(
(xz).
positive
and z
is
v
77"'
611
(6)
may be
written
in the form
v> _ (xze )^xze n_ H
v
*
(^-l)-+i-
^T(^)
f J +00
\,
the integral, taken round arcs of a circle from p to <x> by Jordan's lemma. Hence, by Cauchy's
may be opened
zt
out until
it
becomes the
2
2
line
on which
R (zt) = 0.
If then
we write
(8) that
= iu,
the phase of
- (u /z ) -
1 is fl-
37
-*" r (v + ) _ = 2TQ)
(%xz)~ v
r xi z
'
)i,z
f
e~ xzt dt * (* -l)" +
2
r (V 4-
1)
00 (2z) v f*
_e e-" du
(i)
J->(^
^
when
\xz)a;
V (\)
1
] o
(u2
+z
v+ i'
)
valid
R (p + |) ^ 0,
>
0,
arg 2
< \tt
Basset f,
*
K (x)
as the limit of
Chemie,
was examined in the case i=0 by Biemann, Ann. der Physik und 130139. (Cambridge, 1888), p. 19. t Proc. Camb. Phil. Soc. vi. (1889), p. 11 ; Hydrodynamics, n.
The
6*16, 6-17]
INTEGRAL REPRESENTATIONS
173
a Legendre function of the second kind and expressing it by the corresponding limit of the integral of Laplace's type {Modern Analysis, 1533). The formula
for
Kn (xz)
j
is
zdz
jis
but there
an error
Among
some
on the right in (1) was studied by numerous mathematicians before Basset. these investigators were Poisson (see 6'32), Journal de V&cole Polytechnique, IX. (1813), pp. 239241; Catalan, Journal de Math. v. (1840), pp. 110114 (reprinted with
integral
Serret,
The
Mem. de la Soc. R. des Set. de Liege, (2) xn. (1885), pp. 26 31); and Journal de Math. vm. (1843), pp. 20, 21; ix. (1844), pp. 193210; Sohlonrilch, Analytischen Studien, n. (Leipzig, 1848), pp. 96 97. These writers evaluated the integral in finite terms when v+$ is a positive integer.
corrections,
(1841), pp. 65
Other writers who must be mentioned are Malmsten, K. Svenska V. Alcad. Handl. lxii. -74 (see 7'23) Svanberg, Nova Acta Reg. Soc. Sci. Upsala, x. (1832), p. 232 Leslie Ellis, Trans. Camb. Phil. Soc. vm. (1849), pp. 213215 Enneper, Math. Ann. xi.
;
Glaisher, Phil. Trans, of the Royal Soc. clxxii. (1881), pp. 792 815 J. J. Thomson, Quarterly Journal, xvnr. (1882), pp. 377-381 ; Coates, Quarterly Journal, xx. (1885), pp. 250260; and Oltramare, Comptes Rendus de V Assoc. Francaise ' xxiv. (1895), part n. pp. 167171.
(1873), pp.
;
360365
'
The
last
named
writer proved
.
/ cos xu du
o
(u*+z*)*
(~)w
~ iff
i
/e~ xs ^P\~\
\~^fr~)] p=1
~
The former
of these results
equation
cos
Jo
and the
latter is
xu du u*+z2 p ~
.
ire~ xz ^P
2zjp
'
6' 17.
fa J
eia Tdt.
It
may be shewn by
and
so the integral is a solution if T is a solution of Legendre's equation for functions of order v - \ and the values of the integrated part are the same at each end of the contour.
* Proc.
(1903), pp.
198206.
174
If
[CHAP. VI
and end at
taken to be the Legendre function Q_j (t), the contour may start is an acute angle (positive or negative) provided that z satisfies the inequalities
T be
T
it
is
possible;
is
but the
at
= 1
(when
not an integer)
makes
to 1 as a contour except in
the special case considered in 3*32; for a detailed discussion of the integral
in the general case, see 10'5.
We now proceed
First consider
oo
iexp (tui)
t
t is
=1
at which the
and expand
Q v -{t)
in descending powers of
t.
6*1,
that
}
(*)
P Kt) 7rl
,
xiexp(-iw)
#* Q.-i (0 dt,
and therefore
(2)
J_ r (,) = "*>
S* Q__j (I)
^f^
J o&iexp(-i<o)
<fc
vr 1
If
(3)
^f
4* P,_i (0
(ft.
Again, consider
R (t) = a,
that
it is
+ oo Hence -* oo i, we find
the integral
a multiple of
(l i
(z).
making
J
*
oo *
exp ( i)
The
relation, discovered
by
Schlafli, is
cf.
Hobson,
Phil.
Tram, of
the
(1896), p. 461.
6-2]
is
INTEGRAL REPRESENTATIONS
(z).
175
a multiple of Zf(2)
(4)
^
9
a,
<*)-
(5)
H
J5Tr
(z)
WJrm - W ra
'
v
From a consideration
I
of (1)
it is
**Q.-Mdt,
e'*
&- (*> dt
>
Schlafli's relation,
(#) from
^^
(l*\i
**<" + >
r(-l+)
(-M
4* P,_ x <fc; V * (0
The
of
+1
instead
is
is
In (5) and (6), arg (t+ 1) vanishes where the contour crosses the on the right of 1, and, in (5), arg (t 1) is ir at that point.
6*2.
We shall next examine various representations of Bessel functions by a system of definite integrals and contour integrals due to Sonine* and Schlaflif. The fundamental formula which will be obtained is easily reduced to Bessel's integral in the case of functions whose order is an integer.
We
integral
1
1 (v
+>
'
+m+
Ittx) _,
in
increases from
(|,) 2iri
- ir
to
ir
as
then
no + ) _)m
,t,J
m!
* at"
right; it is
This
is
an analytic function of z
z,
and,
when expanded
in
ascending powers of z by Maclaurin's theorem, the coefficients may be obtained by differentiating with regard to z under the integral sign and making z zero
after the differentiations^
Hence
(Moscow, 1870) ; Math. Ann. xvi. (1880), pp. 929. t Ann. di Mat. (2) v. (1873), p. 204. His memoir, Math. Ann. m. (1871), pp. 184149, should also be consulted. In addition, see Graf, Math. Ann. lvi. (1903), pp. 423432, and Cbessin, John*
* Mathematical Collection, v.
(1895), pp. 20
21.
Cf.
Modern Analysis,
532,
4-44.
176
[CHAP. VI
*W-^n '-pI'-sI*
,
and the
latter writer
|
was the
first
When
it
arg z
< \ir, we may swing round the contour about the origin
real axis.
On
< 2>
writing
\zu,
we then
1
find that,
\tr,
J' = 2S
= e w we
,
(p.
335).
have
j
r oo
+ B-t
e
first
z8inh '-
/., (z)
= =-.
is
dw,
by
Schlafli.
valid
\ir.
This
the
In this formula take the contour to consist of three sides of a rectangle, as in Fig. 8, with vertices at oo tri, tri, vi and x> + iri.
ni
-ni
If
Fig. 8.
4
w
-
we
write
+
l
iri for
w on
iri,
on the
(4)
lines joining
to
we
^^ J"
known
-t-z**t
dt>
valid
\tr.
If
if
we make arg z
true
~-
\tt,
the
first
is
continuous and,
{z) is
if
to be continuous.
So
(4)
is still
a pure imaginary
R{v)
is positive.
The
second memoir; in that memoir he obtained numerous de. te integrals by be an acute angle appropriate modifications of the contour. For example, if
(positive or negative)
and
if
6'21]
INTEGRAL REPRESENTATIONS
177
may be replaced by one which goes from oo (w yfr) i to + (V + yjr) i. By taking the contour to be three sides of a rectangle with corners at oo {ir ^) i, (ir ^jr) i, (ir + ^r) i, and oo + (tt + yfr) i, we obtain,
as a modification of (4),
(5)
J(z) =
efc-^^cos^tf-scos^sintf)^
Jo
e -"*sin>7r r~
_, ginh ,,^_,
if
JO
Again, if we take ifr to be an angle between and ir, the contour in (3) may be replaced by one which passes from oo {\tr 4 ^r)i to oo + {\tr -4- >fr) t, and
so
we
(6)
find that
J,
(*)
=1
+
provided that
|
gin (s^gjj
C08
^ - $ 1/W - 1^)
take
(ft,
arg z
is less
than both
positive
<ty
and
When
(7)
J?
(i)
>
and z
is
(= x),
^. we may
tt
^=
cosh
*
in the last
J (#)=-!
I? J
cos
(i/0
- x sin 0) d0 +
I
|)
*->* sin
(a?
- Ai/tt) dt.
TT J
Another important formula, derived from (1), is obtained by spreading out the contour until it is parallel to the imaginary axis on the right of the origin; by Jordan's lemma this is permissible if R(v) > - 1, and we then obtain the
formula
in which c
positive value
many
of
Sonine's investigations.
Integrals which resemble those given in this section are of importance in the investigation of the diffraction of light
(1899), pp.
;
by a prism see Carslaw, Proc. London Math. Soc. xxx. 121161 W. H. Jackson, Proc. London Math. Soc. (2) I. (1904), pp. 393414; Whipple, Proc. London Math. Soc. (2) xvi. (1917), pp. 94111.
;
6*21.
If
we substitute
62 (4)
for
Y
we
ir
(z)
find that
(z)
= cot vir
cos (vO
cos (v$
+ z sin 8) dO
/>
J o
Jo
/"oo
e- t
-ZBinht
Cf. Oubler,
178
Replace
[CHAP. VI
it is
found on re-
duction that
(1 )
(z)
=-
! "sin (z sin
0-v0)d0--
7TJo
vt ( (e TTJo
+ e~
yt
dt,
a formula, practically discovered by Schlafli (who actually gave the corresponding formula for Neumann's function), which is valid when arg z < \ir.
| \
By means
of this result
we can
evaluate
tjzsinhw-i-w
r<x>
.
+iri
dw>
9,
TTIJ _oo
%ir
for
we take
and
ni
Fig. 9.
write
t,
i0, t
+
is
7rt
for
we then
see that
the expression
_
J
equal to
Too
.
yt-zsmhtfa
+_
fir
/
eHzsiiiO-ye)
0 +
g vni _
7TI
foo
e -v-sinh
<ft,
7rl.'o
'"Vo
Jo
and this is equal to Jv (z) + 1 F (s) from formula Hence, when arg z < \tr, we have
|
|
(1)
combined with
62 (4).
(2)
{1)
(*)
A
1
feo+iri
e*
(3)
#<*>(*)=-
"V *""""^.
811
pp. 327
Formulae equivalent to these were discovered by Sommerfeld, Math. Ann. xlvii. (1896), 357. The only difference between these formulae and Sommerfeld's is a rotation
of the contours through a right angle, with a corresponding change in the parametric
variable; see also
(3)
xvm.
'{1911),
pp.
1-16.
By an obvious change
(4)
J5T.M (z)
of variable
we may
=A
|
u--> exp ^z (u
o
- -^ du,
(5)
HJ* (*)
-.
I
Wo
6-21]
INTEGRAL REPRESENTATIONS
179
the contours are those shewn in Fig. 10, emerging from the origin and then bending round to the left and right respectively results equivalent to these were discovered by Schlafli.
;
Fig. 10.
[Note.
There
is
no
difficulty in
v,
in view
cf.
6*11.]
proceed to modify the contours involved in (4) and (5) to obtain the analytic continuations of the functions on the left.
If
o> is
We
an angle between
tr
and
rr
such that
to
(6)
HW
r
(*)
= -.
1
/>
exp hz ( u - )
exp (-)*
(
du,
and
/
is")
(7)
^"W-sL,,.
^p{*.(-J]
left
du,
the contours being those shewn in Fig. 11 and Fig. 12; and these formulae
give the analytic continuations of the functions on the
Fig. 11.
Fig. 12.
to
and
o>
value
between* v and
|
I
ir.
would
180
[CHAP. VI
by replacing
why w \iri
it is
H^
v
(Z)
g^vwi foo+Jirt
re TTl J _ao-Jiri
i*cosh-vu>
fa
/"oo+Jiri
TTl
Jo
j'ao
eizcosh
v>
cogh vw dW)
(9)
(z)
= -^\
Wl
p\vni
^iri
e-izcoshu>-*w
dw
J -ao+tni
=
provided that arg z
| \
Qghrxi
:-
( Jtt*
e
-izcosh w
^h vW
^ Wj
7rt
Jo
< \tt.
by taking z
positive
Formulae of
(=#)
in (6)
and (7)
and
take
lemma
(to circles of
large
and small radius respectively) shews that, in such circumstances, we may = %-rr in (6) and to = \tt in (7). It is thus clear, if u be replaced by
that
HJti (x) =
p\vai
ie l ,
(10)
^7T?
foo
eixcoaht-H fa
J_oo e
=
<ft
OgJnrt
.
Too
/
Tt
ecosh t
cogh
<ft,
JO
^-r"TTl
/
(11)
H (x) = -r\ Wl
v
a;
- ixcoah -
f
itf
<ft,
J -oo
7o
>
and
1,
we have
cosh
i/tf .
Jv (x) =
2 r - j sin (x cosh 2 r
I
i>7r)
dt,
(13)
(x)
(cf.
cos (# cosh
vtt)
cosh vt.dt;
TTJo
and, in particular
/ta\ (14)
/, -.
6*13),
*<*>-; j, v^T)'
TT
,
.
2 fsinarf.<ft
(15)
f " cos #
<fo
when we
The
^'"iJ.W^U'
replace cosh
t
by
two formulae are due to Mehler, Math. Ann. v. (1872), p. 142, and Sonine, Math. Ann. xvi. (1880), p. 39, respectively ; and they have also been discussed by Basset, Proc. Camb. Phil. Soc. "in. (1895), pp. 122128.
last
slightly different
(1901), pp.
369384;
if
we
write
x=2
y/(ab),
xt=au+b/u, we
find that
(16)
f"
The reader
n (+)
/"
^= "A {2 J{ab)}.
from the formula
,.
rf
Note.
n i cos0)=as (
(1).
sin (
^^
+ )*
183192.
6-22]
6*22.
INTEGRAL REPRESENTATIONS
Integrals representing
181
Iv (z) and
(z).
The
due to
cussion of
Iv {) and
t)
Schlafli*,
function F(a,
his results
The
analysis of
62
is
/-<>- (
|
&L'-*(' + s)*
\ir,
*(*
f
1)}
du-
foo -Hri
(3)
/ (z)=er-.
(2)
^coshw-wr fa
ITTIJ aa-ni
The formulae
at 00
(4)
and
if
R(v) > 0.
in, tri,
in, 00
+ iri,
it is
found that
/,(*)*
ifV^'coBi/tfcW-^^f
ttJo
-***--cft,
jo
so that
/_.(*)-/,(*)
= ^52^ f V^'coshitf.cfc, T Jo
=
I
\tr,
(5)
jBT.,
(z)
e-"osh * C osh
i><
<ft,
From
we can evaluate
ezcoshw-vw
foo+irt
fa
\ir.
For
it is
/*oo+irt
/ iri
J
roo+irtl
escoshto-^afo^
^"n"* I
+
-0O-1T*
gzcoshw-rv, J
ATTIJ -OO-Trt
(../
JoO-lrt)
raoiri
= 5
ezeoshw-vw
Too
ATTIJ -a>- w i
gviri
^w + /^ (^
= 5.
e-*cosh*- dt
+ jv (^
(2) v. (1873),
pp.
199205.
(2) v. (1873),
pp. 199
201
this formula was used by Heine, Journal which reference was made in 5*72.
;
fiir
182
[CHAP. VI
and hence
(6)
2ttc
2i sin vtt
Again,
we may
(7)
K
1 j
(z)
= ^r
e -*cosht-,.t dt}
621,
exp-J
(8)
K, <*)
7r
^
I / Oexpt'u
\z (u +
-\\ du,
and
\tt
+ a>.
- " -1
f
/
e"" /_ (*)
valid
- e~ vwi Iv (z)
sinv7T /""expOr-ult
Jx")
J Oexp (-+)
exp -Us ( u
(
+ - )\ du
tt/J
this
is
when
for
^7r
+ eo.
The contours
respectively.
Fig. 13.
Fig. 14
Further,
tion in (8)
when
is
positive
(= x) and
may
;
imaginary axis
K* 0*0 = i r_ *"**
so that
- \ix
(v
dv,
(10)
(x)
= le-fc"* f
j'*
J -ao
v,
<
(*)=*J*'
<ri*sinht+,t
(fc i
6*23]
INTEGRAL REPRESENTATIONS
these results
183
From
(12)
so that
we
see that
.
2 cos | mr
K
^
/oo
(x)
e-**sinh t cos h vt
^
.
(13)
(x)
=
all
cos
(a;
sinh
t)
COS|l/7Tj
cosh
j/
dt,
valid
when x >
and
1.
may
be observed that
if,
in (7),
^ze t
we
find that
(15)
jr.W-K^^expj-r-^}^,
R (z2 > 0.
The integral on the right has been studied by numerous mathe) among whom may be mentioned Poisson, Journal de VEcole Polytechnique, ix. (cahier 16), 1813, p. 237; Glaisher, British Association Report, 1872, pp 1517; Proc. Camb. Phil. Soc. in. (1880), pp. 512; and Kapteyn, Bvll. des Sci. Math. (2) xvi. (1892), PP 4144. The integrals in which v has the special values \ and f were discussed by Euler, Inst. Calc. Int. iv. (Petersburg, 1794), p. 415; and, when v is half of an odd
provided that
maticians,
integer, the integral
I.
(Paris,
Journal fur Math. xxxm. (1846), pp. 268280. The integral in which the limits of integration are arbitrary has been examined by Binet, Comptes Rendus, xn. (1841), pp. 958
962.
6'23.
type.
The
integrals
f"
I
a?
sin
sin
P"
t
f*
dt,
a?
cost. cos
f.
dt,
Ja
Jq
in
of
which x > 0, 1 < R(v) < 1, have been examined by Hardy f as examples Du Bois Reymond's integrals
f(t)*
/, o
m t.t~*dt,
By constructing a differential equation
cos
in
which f(t)
oscillates rapidly as
* 0.
Hardy succeeded in expressing them in terms of Bessel but a simpler way of evaluating them is to make use of the results
*
of
^621,
6-22.
(1881), p. 182.
51.
184
If
f"
I
.
[CHAP. VI
replace
.
by x&,
dt
it is
clear that
(a?e*)
2 a?
sin
sin
f'
.
= x" __
f
I
sin
JO
_
1
sin (xe~ l ) e vt dt
.
in
fgitzcosht
g 2&ccosht
girfzsinht
g2tassinhtJ
git
fa
./
00
= - \af [Trie-*!! (2a?) - Trie* H^\ (2x) - 2er** Kv (2a?) - 2^"'* #_ (2a:)],
and hence we have
(1)
/
sin
sin
JO
%
t
"K2
--
?\ A = 4sin|i/7r
.
(2a?) [/ L
and similarly
(2)
lob
f
cos
r-
eft
"ITT"
,
,
(X (2a?) -
When
(3)
sin*sin^.^ = 7ry
Jo
it
t
(2a?)
+ if
(4)
r Cos<cos^.^=-iTrF
Jo
t
(2a?)
Jfir
(2a?).
6*24.
curious extension of Jacobi's formulae of 2*2 has been obtained in the case of J {x) and Jx (x) by Theisinger, Monatshefte fur Math, und Phys. xxiv. (1913), pp. 337 341 we shall now give a generalisation of Theisinger's formula which is valid for functions of order
where
If
-\<v<\.
is
it is
j9
<
X) =
flf)"
+
Now
2
(y
(l
_ g oxsinfl
- ix cos 0)
sml1
ft
'
(*jzl!*f* "^
2f
*'
where the contour passes above the origin. Take the contour to be the real axis with an indentation at the origin, and write z= tan < on the two parts of the contour; we thus
find that the last expression is equal to
exp(-cmcot<6) y ^
Jo
sin(.rcot<)
,
,.
,_..
+ .[**
J
o
- eXl (a
sin
;
(.r
cot
j.
^
<)
-^-r ^sin<p
d<b
</>)
sin
<*tan v
i.
W ^ .e-cot<j>\ ^sin0
2
.
=4
,
r**
I
sm
/i
j\
/i
f vir)
<f>
sin
.
(.r
tan
i<A)
rf<
* In Theisinger's analysis, a is
an even
integer.
6'24-6'3l]
and therefore
INTEGRAL REPRESENTATIONS
185
r(y+
(1)
^ J
2 +
/
A.) [
**
e -.in C08
(-p
cos 6) sin*"
0d0
/**"
Jo
/i
d>)
j\
.sin (a? tan Ad>) ,,, /i AJ cot8 " d> cos (4<m; cot <b - vn) r-^-.
vi
sm (j;cotd))
^~
^sind)
^-^
dd>
The transformation fails when v >, because the integral round the indentation does not tend to zero with the radius of the indentation. The form given by Theisinger iu the case v = \ differs from (1) because he works with 3-3 (7) which gives
r(-j)r(j)
(2)
Z-r a
{*)
2 6) sin
""2
6 cos 6d6
,,_.,,
'
+4
provided that \
f**
/
2 ?Z cot "
d>
-r-~
cW>
<v<
the integral representations
6'3.
2%e equivalence of
of
K
v
(z).
obtained in
Three different types of integrals which represent 6'15 (4), 622 (5) and 616 (1), namely
(z)
'
(,,
'i|frr
=
Jo
f
e~*
{t *~
iy ' h dt
e zcoaht
cosh vt.dt,
.
^T(J)
The
equality of the
first
by
Schlafli*; but Poisson proved the equivalence of the second and third as early as 1813, while Malmsten gave a less direct proof of the equivalence of the
We
6*31.
Scklq/li's transformation.
We
pp.
first give an abstract of the analysis used by 199201, to prove the relation
Schlafli,
Ann. di Mat.
(2) v. (1873),
r
which
arises
v (2),
and which
convenient
may
at
We have, of course,
first
R(z)>0 to secure
convergence, and
it is
to taket
-<R(v)<l.
(1837), pp.
* An earlier proof is due to Kummer, Journal filr Math. xvn. much more elaborate than Sclilafli's investigation.
228
242, but
it is
The
W.
R {)
either
186
[CHAP. VI
Now define
by the equation
Ji
'
(*-0 r
'
where x >
and then,
if
t=x - (x -
1) ,
we have
on expanding the
Replacing
x by
cosh
6,
we
see that
Ji
so that, by a partial integration,
(coah 6
-ty
v r($)
'
2"
= 7^
, r(l-i') Jo
tf-'w'MtfsuihdcM
dtdx
-w=i)l'!'"^r*r(i-r)Ji
J/
'
1}
*(*-<)*
1}
r(i-0 J, J.*
('
the inversion of the order of the integrations presents no great theoretical difficulty, and the transformation is established.
6*32.
Poisson's transformation.
The
-~^-
ft_
I>+*)-(2*)"
r cos(*M
)rftt
due to Poisson* Journal de VEcole Polytechnique, ix. (1813), pp. 239241. The equation true when |argz|<r, x>0 and R(v)> -, but it is convenient to assume in the course of the proof that R (v) > | and arg z < w, and to derive the result for other values of z and v by an appeal to recurrence formulae and the theory of analytic continuation.
is is
| |
If
we
replace
it is
sufficient to
prove that
,.
f coa(xu)du
4r(A)
,,
,,
6-32, 6-33]
INTEGRAL REPRESENTATIONS
is
187
Now
r
equal to
s"-i exp
Jo Jo
s 2
<v,..n )* + i +*2
.
dtdu=
Jo Jo
/OO
o
/"OO
I
{exp (
- 2) cos #m dv)
.
exp ( ss2 )
ds,
Jo
of the integrations*.
when we
write
Now
and so we have
y o
r(v + ^ )
/o"^Spi ==ir(*
result.
'"
IeXp{
/o"'
-^- i
<i8
v(2z])"J
It is evident that
for modifying
r
is
is
jf - a2 .*~ B
dx.]
/;
6*33.
Malmsten's transformation.
in proving that,
R {?)>
-i,
1Y X)
- idt
ah
not so direct as the analysis of 6*31, 6*32, inasmuch as it involves an appeal to the theory of linear differential equations. It is first shewn by Malmsten that the three
expressions
*d^-V v - l) dx- XZ
and that as x-*~
that
2 <*
^ d
9
'
+ oo
the third
is
(**)
while the
R (v) > 0.
first and second are bounded, provided and third expressions form a fundamental system
S- <*->-*'-*
t
Bromwich, Theory of Infinite Series, 177. Svenska V. Akad. Handl. lxii. (1841), pp. 6574. X The reader should have no difficulty in supplying a proof of
A'.
* Cf.
this.
188
[CHAP. VI
and the first is consequently a linear combination of the second and third. In view of the unboundedness of the third as a?-^+oo, it is obvious that the first must be a constant
multiple of the second so that
(cos xu)
du
C,
/;
where
V is
independent of x.
To determine
so that
r()r(i)
^cryv)
Z**
'
2z*T(v+$)
follows,
when R(v)>0,
if
we use the
duplication formula c a ?
Gamma function.
consequence of Malmsten's transformation
;
An immediate
gtt
.'
'
is
that
7o
expressible in finite terms
for it is equal to
(*+82 r
is
n-i
(2^)(2n-m-
1)
*
(2s)"-*(-l)! m=0
m.'(ra-ro-l)!
This method of evaluating the integral is simpler than a method given by Catalan, Journal de Math. v. (1840), pp. 110 114; and his investigation is not rigorous in all its stages. The transformation is discussed by Serret, Journal de Math. vnr. (1843), pp. 20,
21;
ix. (1844),
pp.
i.
193216;
[Collected Papers,
(1889), p. 313.]
6*4 Airy'8
integral.
The
integral
cos
(<*
Jo o
act)
dt
which appeared in the researches of Airy* "On the Intensity of Light in the neighbourhood of a Caustic" is a member of a class of integrals which are
The integral was tabulated by Airy by quadratures, but the process was excessively laborious. Later, De Morgan f obtained a series in ascending powers of ar by a process which needs justification either by Stokes' transformation (which will be explained immediately) or by
expressible in terms of Bessel functions.
Tram. Camb.
402.
/:
but this
is easily reduced to the integral- given above. t The result was communicated to Airy on March 11, 1818
;
see Trans.
(1849), pp.
595599.
;
48.
6'4]
INTEGRAL REPRESENTATIONS
satisfies
189
as!
**-<>,
and he
values of
As was
observed by Stokes
;
to Bessel's equation
cf.
= 3. The expression
first in
was published
Vereins in dnruibruck,
xxm.
15,
and
later
by
617.
Subsequently Hardy, Quarterly Journal, xli. (1910), pp. 226 240, pointed out the connexion between Airy's integral and the integrals discussed in 6*21, 6*22, and he then
(.10*2
10*22).
To evaluate
Airy's integral J,
we observe
exp (***
that
it
ixt) dt.
\r.
Now
The
consider this integrand taken along two arcs of a circle of radius p with
p,
-*-
oo
hence,
(<*>
xt) dt
co
expert
Jd
exp (?
ixt) dt
Zjooexpfiri
Ti
=\
~
I* {^
Jo
the contour of the second integral consists of two rays emerging from the
origin
rays.
is
on these
Now,
may be shewn
that
exp(-T +
/,
* Trans.
^a!T)dT=
2
w=0
; -
T*exp(-T)<*T,
.'o
166187.
pp. 329349.] See also Stokes' letter of May 12, 1848, to Airy, Sir G. O. Stokes, 160. Scientific Correspondence, n. (Cambridge, 1907), pp. 159
Memoir and
+ For other occurrences of these functions, see Rayleigh, Phil. Mag. (6) xxvm. (1914), 609619; xxx. (1915), pp. 329338 [Scientific Papers, vi. (1920), pp. 266275; 841349] on stability of motion of a viscous fluid; also Weyl, Math. Ann. Lxvm. (1910), p. 267, and, for
pp.
is
toe. cit. p.
150180.
Series, 176.
190
[CHAP. VI
and so
|
cos
(<*
Jo
+ ~
art) eft
'?
m =o
ml
Jo
(- t*) dr rm exp r\
/
"*
This
is
U-omir(m + |)
De
Morgan.
+ **
mt.iti!r(m + |)J
the series on the right
When
we
is
to be taken to be positive)
(i)
I/-,
d$) + a $%)]
(2)
6*5.
Kummer's formula
of 442.
- T (2m - s)
at s
{izy
is
- 2m + r, where
is
= 0,
1, 2,
.
. . . .
This residue
(iar)
lm+r/r
!,
so the
sum
of the residues
( ) m zim e~ iz
J ^ Z)e
if
27ri\ntoJoo
2.m\r(p + m +
1, 2,
aS
l)
'
It
may be
verified,
by using
Now suppose that R (v) > \, and choose the R (y + s) > $. When this last condition is satisfied r ( 2 m - g> 5 .m!r(i/ + w + l) =o2 m
2r'l
contour so that, on
the series
it,
is
r <-*)
wt l.l
l..
r(-)l> + j + t)
559562
;
694700, 792799.
t Mellin has given a summary of his researches, Math. Ann. lxviii. (1910), pp. 305 337. X Gamb. Phil. Tram. xx. (1908), pp. 270 279. For a bibliography of researches on inteRraU of this type, see Barnes, Proc. London Math. Soe. (2) v. (1907), pp. 59 65.
'
6*5]
INTEGRAL REPRESENTATIONS
If therefore
191
we change the
order
-_<*
l0+)
r(-)r( y + +
|).(irr<to
The only poles of the integrand inside the contour are we calculate the sum of the residues at these poles, we
at 0,
1, 2,
When
find that
so that
(1)
^
is
( ^ )e
- =
_(ML- F
i
(I
i; -2iz),
which
(2)
Rummer's
relation.
{
In
like
M*) *= r *+ x yK ( +
2"
l;
2iz).
These formulae, proved when B(v) > , are relations connecting functions of v which are analytic for all values of v, and so, by the theory of analytic continuation, they are universally true.
It is also possible to represent Bessel functions
by
integrals in which no
exponential factor
is
involved.
To do
this,
we
r(-*->r (-)(!)'+
qua function of s. It has poles at the points
5
= 0,
is
1, 2, ...;
- v ,-v+l,-v + 2,
(~)m (^Y +Sm
....
The
residue at s
iri
sin vnr
'
ml r
(v
+m+
1)
= v + in
iri-"
is
sin V7r
ml T
(v
+m+
v
'
1)
so that
(3)
7re-4<" +1 >'rf
H (z) = - 2^.
v
T (-
- s) T (- s) (^iz)"^ ds,
irew H
w ( z)
start from
+ oo
\
When
arg iz
<
<
\tt
may
oo i
and end at
oo
i.
If
we
we
find that
c+ l
(5)
7re-*<" +1 "*
r ( _ _ ,) r (- *) (ii>)' +M ds,
Infinite Series, 176.
-c-aoi
* Cf
Bromwich, Theory of
; ; ,
192
provided that argiz
|
[CHAP. VI
(6)
vJW H
|
o)
( z)
= -L r
frm]
e+x>%
r (_ _ S ) r (- s) (- \%zy*~ ds,
is
-c-oot
exceeding jR
(v)
is parallel
There is an integral resembling these which represents the function of the and the argument first kind of order v, but it converges only when It (v) >
of the function
is positive.
The
integral in question is
'.-K,$g&*>
and
it is
way
the reader
when
is
large
is
0(\s |~* -1 )-
6*51.
By
Gamma
function
we may
write
m UJ
Consider
now
the integral
1
_ *l.
Ziyir
in
I"
J _aot
r(-8)r(-2p-8)r(p + 8 + i).(2igyd8,
periodic in
which the integrand differs from the integrand in (1) by a factor which is It is to be supposed temporarily that 2v is not an integer and s. that the path of integration is so drawn that the sequences of poles 0, 1, 2, ... 2v, 1 2v, 2 2i/, ... lie on the right of the contour while the sequence of poles v \, v~\, v \, ... lies on the left of the contour. In the first place, we shall shew that, if arg iz < f ir, the integral taken round a semi| \
circle of radius
for, if s
= pe**,
v
<x>
'
'
and,
by
1
Stirling's formula,
~ pe
i9
log (2iz)
^ log (2tt)
and the real part of this tends to co when ^7r<6<\ir, because the dominant term is p cos 6 log p. When is nearly equal to |7r, sin sir is comparable with \ exp [pir sin 6 and the dominant term in the real part of the logarithm
|
j)
is
.
p cos 6 log 2z
|
and
this tends
p sin $ arg 2iz pcosd log p + pd sin 6 + p cos 6 - 2p7r|sin &\ arg iz < \w. to - <x as p -* oo
if*
j j
6*51]
INTEGRAL REPRESENTATIONS
s times the integrand tends to zero all along the semicircle,
if
193
Hence
to pass
the semicircle
is
and so drawn so as
is
not an
- ^rJLi^/tr j
may be
contour.
Pr
-oof
(-
r (-
2*
- ) r (* + + j)
^y
ds
The
residues of
r(-s)r(-2v-s)r(v+8 + i).(2izy
at s
and
= 2v + m
are respectively
7T
ein2inr
_
'
tt_ r (sin2*"7r
v + m + \) (2iz)-+m mlV(2p + m + l)
.
and hence
_(2z
p r(-s)r(-2v-s)T(v +
+ l).(2iz)ds
+
sin 2v7r
'
It follows that,
(2)
when
j.argiz
< f ir,
B.*to-
rn~' mi r>-Mr
rrrt
x ["* r(-)r(-2v-)r(i +
J -oe
f).(2w)<fo
>
and
similarly,
(3)
*,(*)--
i*->n)
cog
(,.)
^2^)"
7T*
*r
r(-)r(-2v-*)r(i/+s + ^).(-2wyrf*.
The restriction that v is not to be an integer may be removed in the usual manner by a limiting process, but the restriction that 2v must not be an odd integer cannot be removed, since then poles which must be on the right of the contour would have to coincide with poles which must be on the left.
CHAPTER
Approximate formulae for
ill
VII
Jv {z).
In Chapter
cases
z1
is
argument z, multiplied in some by log z. These series are well adapted for numerical computation when since the series not large compared with 4 (v + 1), 4 (v + 2), 4 (v + 3),
.
converge
such values of
z.
But,
initial
when
is large,
the series
approximate values of Jv (z) and Yv (z). There is one exception to this statement when v + \ is an integer which is not large, the expressions for J v (z) in finite terms ( 3*4) enable the functions to be calculated without difficulty.
;
The
is
when
is large.
There are
gation
large.
really
the investi-
when v is large is very different from the investigation when v is not The former investigation is, in every respect, of a more recondite
it is
vm.
before Poisson 's -J"
(x),
was
/ 2
\T
COB
[
U)
<
12
.3 2
"^>-t -2l(8^
+
(
l2
.3 8 .5 2 .7 a
4!(8*)'
-"j
3 2 52
+ sin(*-i7r).j
when x
is
l2
IT
^-3^
l2
..
to be regarded as suggestive
from his method of obtaining the dominant term) and ingenious rather than convincing.
is
* Iticerche sulla convergenza delta serie che serva alia soluzione del problema di Keplero (Milan,
1817).
An
1*4.
t Journal de VEcole Polytechnique, xn. (cahier 19), (1823), pp. 350 352 ; see l 6. An investigation of Jv (x) similar to Poisson's investigation of J (x) has been constructed by Gray and
-
38.
7-1]
It will
ASYMPTOTIC EXPANSIONS
-
195
;
be seen in the course of this chapter that Poisson's series are asymptotic has been proved by Lipschitz, Hankel, Schlafli, Weber, Stieltjes and Barnes.
this
It must be mentioned that Poisson merely indicated the law of formation of successive terms of the series without giving an explicit expression for the general term such an
;
W.
R. Hamilton*
for
(cf.
1*6).
(x) is
is
now
usually
Jn (x)~(
(
COS(#
(4n2 - I s) (4rt2
f
.
2UMf
,
,
,
~ +(
4n2 -
(4n*
- 3 ) (4>n* - 5
2
(4n 2
7 2)
_
Y\
.
4!(8*)<
-sin^-^Tr-i^.j-yj^
(42 -l 2
(4n 2
-l 2)(4n2 -3
)(4n2 -5 2)
gy^^
L +
...JJ
These expansions for J (x) and Jx (x) were used by Hansen for purposes of numerical computation, and a comparison of the results so obtained for isolated values of x with the results obtained from the ascending series led Hansen to infer that the expansions, although not convergent, could safely be used for purposes of computation .
Two years before the publication of Jacobi's expansion, Plana had discovered a method of transforming Parse val's integral which placed the expansion of / (x) on a much more satisfactory basis!". His work was followed by the
||
researches of Lipschitz**,
first
asymptotic expansion of Jo
tion
;
with the aid of the theory of contour integraLipschitz also briefly indicated how his results could be applied to Jn (z).
for
Jv (z)
and
v (z),
* Some information concerning W. R. Hamilton's researches will be found in Sir George Gabriel Stokes, Memoir and Scientific Correspondence, i. (Cambridge, 1907), pp. 130135. + Ermittelung der absoluten Storungen [Schriften der Stermoarte Seeburg], (Gotha, 1843),
pp. 119123.
is
t Astr. Nach. xxvm. (1849), col. 94. [Get. Math. Werke, vn. (1891), p. 174.] obtained by making the substitutions
Jacobi's result
^/a.8ln(*-iii-J)=:(-l)*" (
in the form quoted.
a note by Niemoller, Zeitsehrift fiir Math, und Phys. xxv. (1880), pp. 4448. della R. Accad. delle Sci. di Torino, (2) x. (1849), pp. 275292. expansions of Jv () and IV (z) by If Analysis of Plana's type was used to obtain the asymptotic JVlcMalion, Annals of Math. vin. (1894), pp. 5761. ** Journal fur Math. lvi. (1859), pp. 189196.
See
II
Mem.
+t Math. Ann.
I.
(1869), pp.
467501.
196
[CHAP.
VH
Yn (z)
memoir
Functionen (Leipzig, 1868), just before the publication of Hankel's and the researches of Weber, Math. Ann. vi. (1873), pp. 146 149 must also be
mentioned.
was investigated (and proved to be this result was reproduced, with the addition of the corresponding formula for / (z), by Kirchhoff f and a littleknown paper by MalmstenJ also contains an investigation of the asymptotic
of
(z)
Kummer*
expansion of
(z).
(x),
(x),
(x)
and Kq{x) has been made by Stieltjes, Ann. Sci. de VEcole norm. sup. (3) in. (1886), pp. 233 252, and parts of his analysis have been extended by Callandreau, Bull, des Sci. Math. (2) xiv. (1890), pp. 110 114, to include functions of any integral order; while results concerning the remainders when the variables are complex have been obtained by Weber, Math. Ann. xxxvu. (1890), pp. 404416.
Inst, polyt.
/ (z) and Yv (z), when seem to be most simply carried out with the aid of integrals of Poisson's type. But Schlafli1[ has shewn that a large number of results are obtainable by a peculiar treatment of integrals of Bessel's type, while, more recently, Barnes** has discussed the asymptotic expansions by means of the Pincherle-Mellin integrals, involving gamma-functions, which were examined in 65, 6*51.
is
7'2.
Asymptotic expansions of
shall
H^ (z) and i/
|
(2)
(z) after
Hankel.
We
now
;
|
by Hankel ff.
6*12 (3),
namely
valid
|7r + j8<
args <
|7r
+ jS,
provided that
The expansion
+ $iu/z) v -l in descending
powers of
z is
(v-j)iu
"*"
2z
+ (,,-fr)(-f).(^ +
270
*"'
* Journal fUr Math. xvn. (1837), pp. 228242. f Ibid, xlviii. (1854), pp. 348376. t K. Svenska V. Akad. Handl. lxh. (1841), pp. 6574. See the Jahrbuch Uber die Fortschritte der Math. 1907, p. 492.
Ibid. 1905, p. 828. IT Ann. di Mat. (2) ** Trans. Camb. Phil. Soc. xx. (1908), pp. 270279.
II
vi. (1875),
pp.
120.
tt Math. Ann.
i.
(1869), pp.
491495.
7-2]
ASYMPTOTIC EXPANSIONS
197
but since this expansion is not convergent all along the path of integration, we shall replace it by a finite number of terms plus a remainder.
For
1
all
we have*
{1
J
+ 2z)
-J -^TWz)
=0
+ (p-l)\{2iz)
tf
I
;
dt
2iz)
It is convenient to take
so large that
satisfies
R {v p \) ^
any
positive angle 8
which
the inequalities
||^|7T-S,
The
effect of this choice is that,
|arg*_(|7r + )| $ 7r-$.
when
>sin8,
|arg(l-^)
<7T,
2iz
for the values of
t
and u under consideration, and so P~i < e'l 'wi (sin 8)*<"-p-*> (i _
~Y~
= Ap
say,
where
A p is independent
of
z.
On
term,
+ liu/*)*"*
we
1L1 + jKp
J'
where
2?p is
a function of v,
p and
8 which
is
independent of z.
Hence,
(1) v 7
when R {v - p - f) <
v '
and
R (v + J) > 0, we have
(#->)]
'J
,
when
angle
;
is
such that
function
The formula
(1) is also
valid
when
R (v - p |) >
this
may be
seen by
* Cf. Modem Analysis, 5-41. The use of this form of the binomial expansion seems to be due to Graf and Gabler, Einleitung in die Theorie der BesseFschen Funktionen, i. (Bern, 1898) pp. 8687. Cf. Whittaker, Modem Analysis (Cambridge, 1902), 161; Gibson, Proe. Edinburgh Math. Soc. xxxvm. (1920), pp. 6 9 ; and MaoBobert, ibid. pp. 10 19.
t Cf.
Modem Analysis,
2-1.
198
[CHAP. VII
supposing that
R {v p J) >
if
R(? q \) <0;
contained in
[ ]
in (1)
is
then
it
may be
expressed as
terms
sum
p + 1
terms
is
therefore
(z~ p).
i
Hi* (.)-)'.----
g;<i^|^ +
tr
(*-,)]
is
provided that
R (v + |) >
we
now given
by the inequalities
+ 28 % arg z ^
25.
If,
following Hankel,
write
{4>v*
I s } {4* 2
-3
2
}
.
..
{tv*
- (2m -
l)2 }
2 2m .w!
(4)
H,
W = (A)*e-<-^g |^ +
o
<)(*-*)]
HP (z) ~ f A) V^-w
m>
t
-=-.<f'
(6)
Since
361
sponding functions of order v, that the restriction that the real part of v exceeds - \ is unnecessary. So the formulae (1) (6) are valid for all values
of
v,
when z
is
Sir.
HP
(z)
~ (A)* ^*~-w
.f, (i
+ ,
_^
.
(8)
(^ + v>i
(8)
_ v _ _1_
2-ir
y
arg z
of which (7)
valid
when
2-tt,
and
when
<
< ir.
7*21]
7*21.
ASYMPTOTIC EXPANSIONS
Asymptotic expansions of Jv {z), J- V (z) and
the formulae of 7
2,
199
v {z).
If
we combine
in terms of
{2gym
(-)w -(^,2m
/ i \ v i -sin^-Ji/w-iir).^
+ iy
J2zj^i
.(f,
(2)
n W ~(iy
i sin(z \ vtt
i x v $ir). 2.
(-)w
2w)
m-0
i
(2zfm
(-)'*('.
/i
2w+l)"
'
^.(ij^^w-wyt^-)
n
only),
2m +
1)"
'
^y w+1
and
(5)
T.(.)~(
T)
Ln(*-W-i).jS o
+ cos(^-^7r-|7r).^
o
(2 , )m
v (
^ )2m+1
|
J.
|
<
tt;
and the error due to stopping at any term is obviously of the order of magnitude of that term multiplied by \\z. Actually, however, this factor \\z may be replaced by ljz2 this may be seen by taking the expansions of HJ$ (z) and {z) to two terms further than the last term required in the particular v combination with which we have to deal.
;
(2)
(z)
and
(z),
but,
when R(p)% ^
the integrals
from which the asymptotic expansions are derived are those which represent {2) .ff (,) _ (z) and -y (z). This difference in the mode of treatment of J, (z)
and
{z) for
is
formula (1)
* Cf.
> \.
;
Sheppard, Quarterly Journal, xxm. (1889), p. 223 Searle, Quarterly Journal, xxxix. The error appears to have originated from Todhunter, An Elementary Treatise on Laplace's Functions, Lamfs Functions and Bessel's Functions (London, 1875), pp. 312 313.
(1908), p. 60.
1 l 1
200
[CHAP. VII
grating
e***
and
f
J
+ oc
of J (z) was obtained by Lipschitz* by interound a rectangle (indented at 1) -with corners at + 1 Cauchy's theorem gives at once
8 )-*
2
eizt (1
-1
f V-">*
.'o
-* (2
rct>
+ tu)-* du
M -i (2
- l
o
e-(f+)z
- tu)-* du = 0,
but in order to
it
seems necessary to use a method which involves at some stage the loop integrals discussed in Chapter vi.
It
may be
m v (-)
initial
involved in equations
.
are as follows
2
(v,
2m)
) (4i/
-3
m-o
(2*)*
2!(8*)
+
(-)
.
(4i/
(4y
(y,
2m + 1)
2+ 1
4^ l!8s
(4g
(4i^
mt
(2^)
3!(8s)T
- 32) (4 - 5 8 ) +....
notice that
p. 67.
The method by which Lommel endeavoured to obtain the asymptotic expansion of Y n (z) in his Studien, pp. 93 97, was by differentiating the expansions of J v (z) with respect iX> v but of course it is now known that the term-by-term differentiation of an
difficulties.
be noticed that Lommel's later work, Math. Ann. iv. (1871), p. 103, is free from the algebraical errors which occur in his earlier work. These errors have been enumerated by Julius, Archives Nderlandaises, xxviit. (1895), pp. 221 225. The asymptotic expansions of Jn (z) and Yn (z) have also been studied by McMahon, Ann. of Math. vni. (1894), pp. 57 61, and Kapteyn, Monatsheftefiir Math, und Phys. xiv. (1903), pp. 275282.
A novel
in recent years
In such investigations the dominant terms of the exThis fact combined with the the of Bessel functions that theory forms only a trivial part of consideration has made it seem merely to mention in question desirable the investigations and more by Hardy . Wigert the recent papers the work of Voronoif and J
divisors of numbers.
* Journal far Math. lvi. (1859), pp. 189196. t Ann. Sci. de Vticale norm. sup. (3) xxi. (1904), pp. 207'86b. Math. Kongresses in Heidelberg, 1904, pp. 241 245.
459534
140.
Journal, xlvi.
(1915), pp.
263283
Proc.
London Math.
pp. 125.
w
7*22]
7*22.
ASYMPTOTIC EXPANSIONS
Stoked phenomenon.
for values of z
201
The formula 7-21 (1) for Jy (z) was established and arg z < 7r. If we took arg z to lie between between tr and tt) we should consequently have
| |
such that
lies,
arg ze~ ni
1ci
V* T
-i
,-i
m v (-)
x
(*.
2m ) 2m +
)
5
m-o
(-)"*
(v,
1)"
\&ze
so that,
when
<
arg z
<
27r,
}
*w~ **<(-)
and
[cos <,
+ i^)j
v2 ;r
/.
2 , Vj^h V/^
this expansion
(1).
is
721
We
shall
now make a
The expansions
in
which
2rr,
the function
HP (z) has
W
(1)
is,
\irtj
for
TO -o
(2iz)m
HP
i
{z),
namely
HM(z)co(--)
,irz)
e-i-l<>-i*
X m.
it.
("'
w)
'
(2iz)m
27r
To obtain an expansion
3'62 (6),
namely
HP (zer*) + e HP (ze-),
iw>~)'.^^.sj$
VtTS/
TO _o
(2t^) TO
tt
]
difference between
The expansions (1) and (2) are both valid when < &rg z < them has the asymptotic expansion
2COBV7T.
now the
()'*<#*"+* 5
Kir* J
m-o
<7^J2>, m
(2t*)
iz
which multiplies the series, this expression -is z is large) than the error due to stopping
j
202
at
[CHAP. VII
(e~ iz z-P-l)
(1)
any
when
which
we
and
(2),
occurs
its
remainder.
we have
where the constants c,, c2 have values which depend on the domain of values assigned to arg z. And, if arg z is continually increased (or decreased) while \z\'\% unaltered, the values of c x and c have to be changed abruptly at various 2 stages, the change in either constant being made when the function which multiplies it is negligible compared with the function multiplying the other constant. That is to say, changes in c, occur when I (z) is positive, while changes in c2 occur when I(z) is negative.
It is not difficult to prove that the values to be assigned to the constants
Ci
and
Cl
g2 are as follows
C3
= $a**"4) c, = i**+iM
[(2p
1)
7T
1)
tt],
[2P
7r
<
arg 2
where p
is
any
This phenomenon of the discontinuity of the constants was discovered by Stokes and was discussed by him in a series of papers. It is a phenomenon which is not confined to Bessel functions, and it is characteristic of integral
functions which possess asymptotic expansions of a simple type*.
was discovered by Stokes in (March?) 1857, and the discovery was apparently one of those which are made at three o'clock in the morning. See Sir George
Gabriel Stokes, Memoir and Scientific Correspondence, I. (Cambridge, 1907), p. 62. The papers in which Stokes published his discovery are the folio wing t: Trans. Cavib. Phil. Soc. x. (1864), pp. 106128; xi. (1871), pp. 412 425; Acta Math. xxvi. (1902), pp. 393 397. [Math, and Phys. Papers, iv. (1904), pp. 77 109 ; 283298 v. (1905), pp. 283-287.]
The
by Stokes.
7*23.
Asymptotic expansions of Iv
7*2 (5)
(z)
and
{z).
The formula
3*7 (8),
which connects
(z)
and if
(l)
(iz),
-(2z)
* Cf.
~nr +
( 6-4).
2i(8^
-J'
and
^
Bromwich, Theory of Infinite Series, 133. t Stokes illustrated the change witb the aid of Bessel functions whose orders are
the latter being those associated with Airy's integral
7*23, 7*24]
ASYMPTOTIC EXPANSIONS
\
203
tt.
,
And
= e* vni J
<
(e
-* z) shews that
<o\
(-) w ( y
m)
r'+(,,+ *"''
(",
"0
provided that
f 7r.
On
er*"*
,
Jv (e* ni z)
|
TO
shews that
(y,
<*\
e-'-<"+*>"
m)
(2tt^
to(2^r'
z has a value for
provided that
(2)
is,
and
(3)
when
\tt and
|7r
investigated.
section were stated explicitly by Kummer, Journal far Math. xvn. 228242, and Kirchhoff, Journal fur Math, xlviii. (1854), pp. 348376, except that, in (2) and (3), the negligible second series is omitted. The object of the retention of the negligible series is to make (1) and (3) formally consistent with 37 (6).
p. 135,
The formulae are also given by Riemann, Ann. der Pkysik unci Chemie, (2) xcv. when v = 0. Proofs are to be found on pp. 496 498 of Hankel's memoir.
(1855),
A number of extremely interesting symbolic investigations of the formulae are to be found in Heaviside's* papers, but it is difficult to decide how valuable such researches are
to be considered
A
x
f
remarkable memoir
cos
ax dx
.
tre~ a
1
jo (1
.M!
x[(2a)n
is
+ nC
(n
+ l).(2a)n- + n C.
1
(n-rl)(n+2).(2ay-*-r...]
obtained
(cf. 6*3).
f
This formula
cos
is
ax dx
.
ire~ a
Jo (1
+x
2
)
n+1
2 :W+1 .?i!(
\2a
1
_1
|>]
,
the
[ ]
is
to be replaced
by (n)_ TO and
this, in
Malmsten's
notation,
means
l/{(w+ l)(+2)...(n
m)}.
It will
is
7*24.
bei(z).
From the formulae obtained in 721, 7*23, the asymptotic expansions of Thomson's functions ber (z) and bei (z), and of their generalisations, may be written down without difficulty. The formulae for functions of any order have been given by Whitehead^:, but, on account of their complexity, they will not
* Proc.
Royal Soc.
lii. (1893),
pp.
ii.
(London, 1899).
My
thanks are due to Dr Bromwich for bringing to my notice the results contained in the latter work. t K. Svenska V. Akad. Handl. lxii. (1841), pp. 65-74.
X Quarterly Journal, xlii. (1911), pp.
329338.
[CHAP. VII
Russell*; he found
n ^ K
'
'
(9\
V
'
ker (*)
_ ex P(-^) c OS o
V(2*/tt)
\
(kei(s)
sin^
13
128*4
;'
where
. .
a (z)
z ~ -r= +
'
25
-v/2
8^V2
7r
384^ V2
*
'
^^~V2~8~~8772"T6V~384*V2
The ranges
|
R(
25
+ ""
in the case of (1)
arg^
< \ir
and
arg z
< |tt
These results have been expressed in a modified form by Savidge, Phil. Mag.
(1910), p. 51.
(6) xix.
7*25.
Hadamard's
which
is
result
importance
is
due to
that
it is
expansions, so that they become convergent series together with a negligible re-
mainder term. The formulae will be stated for real values of the variables, but the reader should have no difficulty in making the modifications appropriate to complex variables.
We
u,
take
first
when
> \. When we
replace sin
\B by
we have
I' ( * )=
uMml?
2
(2a?)" e*
J
i
a"' ,An'' em
.
i>+i)r<i).Jo
2
(2a?)" e*
- v)m
m!
-I> + 1)1^)^0
We may
C\\ {)
Wa +2m eXp(
,~ , 2Ma?)rfW '
is
io
convergent.
T (t\ K{a!)
(J
v )m
rzx
fv+m-hf-tfif l dt e
V(27r) TOto
'
7*25, 7*3]
ASYMPTOTIC EXPANSIONS
x,
205
y
is
(v
+n + \,
r( + $)
2x)
ana
tt
+v >
is
*' w
H, 1+ <>-U)T>+tfJ.' "^( s)
WxJ r(v+t)
so that
OT
m!(2#) w
and
similarly
(3 )
^w
-( i) r
;
^- M
\
is
From
i^K^ ^
)+o
;
>
these results
it is
of the first
obvious.
7*3.
Formulae for
the
In
7*2
we gave an
(z) are of the same order of and v magnitude as the first terms neglected. In the case of functions of the first and second kinds, it is easy to obtain a more exact and rather remarkable theorem to the effect that when v is real* and x is positive the remainders
the asymptotic expansions of
v
Hw
(z)
are
numerically
less
than the
first
terms neglected, and, by a slightly more recan be proved that the remainders are of the
Let us write
i).f;--{(^ir-(-r }*=^->'
*
We may
take
v^Q
206
so that
[CHAP. VII
(2
(2)
[cos (x
v)
+ cos (x + \mr
tt)
Q (x,
v)\
Now I(v)
we may take
B to be \ir since
1.
p be taken
J,
there exists a
number
0,
2^
<*-"> f !?Y" + *'<*-"> f 2!\* Y~* _ ~ TOto m! V2tW "^ (2^)! ^2**/
'
we have*
2* (-*> / (2p)I
+
_,_
1
2x)
|
Y~* + fi
j.
**)
-2 Y~* "
O
"v
(WW
(2m)!
to
UW
JLY" 4.
M* W
'
where 6
^1
and, since
is
obviously real,
- 1 ^ $ $ 1.
It follows
on integration that
P( *'* )=
and since
mto
"
f
9
(2,n)!(2*)-
(2p)!(2^r(, + |)J
^
P
du
>
e~
W +2p-* cZw U
e~tt u+w-* dw
= T (/ + 2p +
(x, v) does not we see that the remainder after p terms in the expansion of exceed the (p + l)th term in absolute value, provided that 2p>v \.
From
the formula
(i
V-2W
we
of
j.
*U
Y~*
1
'
find in a similar
in the expansion
1 )th
These results were given by Hankel, Math. Ann. I. (1869), pp. 491 494, and were reproduced by Gray and Mathews in their Treatise on Bessel Functions (London, 1895), p. 70, but small inaccuracies have been pointed out in both investigations by Orr, Trans.
Camb. Phil.
Soc.
In the case of
(x)
* This result
(1859), pp.
in a rather different
manner by
7-31]
ASYMPTOTIC EXPANSIONS
p
207
and
(i V
+ iX' = * (-lJLz2!k ( u y
h
2x)
mZ
ml
\2x)
and,
when
p>v \,
may be
written
(p-D!
where
<
<$
1,
and
so,
on integration,
*.()where
This
()V
v
v -\(
,t
(2ar)
(2a )"pJ'
f
$ #2
is
<$
when p >
- \.
P (x, v)
is
and Q (x,
v)
is, of course, the fact that (1 + %ut/x)"-P-l is positive and does not oscillate in sign after the manner of (1 + \iutjx) v -P-* liut/asy-P-*. (1
the reason
why
secured
7*31.
(x),
(x)
and
(x).
The results of 7 3 were put into a more precise form by Stieltjes*, who proved not only that the remainders in the asymptotic expansions associated with J (x),Y (x) and (x) are numerically less than the first terms neglected,
but also that the remainders have the same sign as those terms.
Stieltjes also
it+.
It is only to
all
the terms
(x), but his result is complicated and we shall not reproduce be expected that I is intractable because in the dominant expansion have the same sign whereas in the other three asymptotic expansions the
(..)
examined I
terms alternate in
sign.
P (*' 0) =
Q (X
'
2^r
~ UX
~ UX
U~k K 1
+ ** M) "* +
(1
** M >"*J
~'
]
du
>
0)
2J^So
^
1
{(1
+ url " (1 ~
hiu)
du
j'w) -i
/'*
by
d<f>
_
</>'
"".'o
*
|iw
sin 2
Ann.
Sci. de
also been
pp. 233252. examined by Schafheitlin, Jahresbericht der Deutschen pp. 120129, but he appears to use Lagrange's form for the
(3) in. (1886),
it is
remainder
theorem when
inapplicable.
208
[CHAP. VII
[*"{!
($u*Bm*<i>)r-1
is
Now,
obviously,
Jo
l+Ksin
and
1
;
2
</>
Jo
K%
*'
where 8
lies
between
and hence
\ (a + \iu)-i + (i
- \wr*} - 1 -
If
we multiply by
integrate, it is evident
that
(1)
P(*.0)-l-g^ s ^,+
..+(-)
(2p-2)!(8a>)*-
+ ^ rUi
and jj is any positive integer result which had to be proved for P (a?, 0).
where
i.y.y...(4p-iy
(2p)!(8<r)v
<
0,
<
1,
(zero included);
and
this
is
the
i ((1
fi"
6 *^&
ittsin9
dd>
we
find that
m
(2)
n/ n\ c^,o)_-
la
(ai),
i'-
3*- 5'
1
8! (8*)
"
l'.ff.5...(4p-8y (2p-l)!(8)- 1
and p is any positive integer (zero included) and result which had to be proved for Q {x, 0).
where
< 6t <
1,
this is the
In the case of
K
and replaced
(1
(a)
+ $w)~ J by -
i,
t ^ie
follows the
result of 7*3.
By an ingenious device, Callandreau* succeeded in applying the result of Stieltjes to obtain the corresponding results for functions of any integral order ; but we shall now explain a method which is effective in obtaining the precise results for functions of any real order.
* Bull, det Sci.
Math.
pp. 110114.
"
7-32]
7*32.
ASYMPTOTIC EXPANSIONS
The signs of
209
with
(x)
and
(x).
been seen that Jv (x) and Yv (x) are expressible in terms of two functions P(x, v) and Q{x, v) which have asymptotic expansions of a simpler type. We shall now extend the result of Stieltjes ( 731) so as to shew that for any real value* of the order v, the remainder after p terms of the expansion of P (x, v) is of the same sign as (in addition to being numerically
It has already
less
thanf) the (p
(x,
holds for
on
which these conditions Q enable the theorem to be stated in the following manner:
.
that 2p
>
- \:
a corresponding result
lay
restrictions
In
P {x, v) and
7 3,
Q {x,
v),
the
remainders
are of the same sign as, and numerically less than, the first terms neglected.
By
we have
p {x
>
v)
2T^ry)/
{{1
* iu)v
~h
+(1 "
(1
iu)v
~ i]
du
'
Q{x, p)=
2iT^Y) \1
;
"""' {{l
+ ** M) ""*
~ * iuy ~ i] du
'
and, exactly as in 7 3
we may shew
that
Ji
(1
trAtz^itr
\&ni).
wi=0
)P(i
ur~7
(2p~2)?
o
- typ
"
{(1
dt.
The reader will see that we can establish the theorem if we can prove that, when 2p > v \. the last term on the right is of fixed sign and its sign is that of
It is clearly sufficient to
1
[
shew that
{(1
is
\\ _ t?P-* U
expression
+ \%vty-** - (1 - liuty-v+l)
equal to J
-
dt
is
iut)
- e~ x
<1
- i M 1 d\ dt
'
=r7K-
1
f {2p-v + $)JoJ 9
I "( 1
e~x d\dt
sin ($\ut)dtd\.
As
in 7-3
we may take
v^O
t This has already been proved in 7*3. t Since isin^Xwt) |^$Xuf, the condition
infinite integral.
2p>-b
210
[CHAP.
t;
VH
t)w~*
is
between
and
1,
|(1Jo
t)
2?-*
sin ($\ut) dt
( sin (l\ut) dt
> 0.
Jo
is positive,
Since
T (2p v + )
we have succeeded
+
in transforming
2F-T^Ti
1 (1 C ~ O^"*** K
W-*
- (1 " ImO'-*^} dt
into an infinite integral in which the integrand is positive, and so the expression
under consideration
is positive.
That
is to say,
i{(i
+ 4y-*+(i-iwy-}
"
mto
(2wi)!
it
(2p)!
( 7*3)
where
that in
these circumstances
6 ^
|
1.
Consequently
^1 we at once
;
P (x, v).
v) follows
i + 1 r. - (i
_ })-*) =
-g
<->"-^ -;^ft">""
(
The
lies
analysis fails
when
\<v<\
if
of
(l$m) r -*
between
{{l+\iu) v -\ + (\ -iiw)"
*}
has the
and
manner, {(1 + %iu) v ~h (1 - \iu) v ~ i}/t has the same sign as $(j-, v) have the same sign as the first terms in their
still
true
is
true for
Q (x,
v)
when
7*33.
Some
by the remainders
in the asymptotic
expansions of H,
have been given by Weber*. These inequalities owe their importance to the fact that they are true whether z and v are real or complex. In the investigations which we shall give it will be supposed for
(z)
(2
and
>
(z)
simplicity that v
is real,
though
it will
,,
7-33]
ASYMPTOTIC EXPANSIONS
There
is
211
of
v.
no further
loss of generality in
We
= r,
sideration,
we
shall
suppose that
have*, by
1r^v \.
If v
- \ > 0, we
612 (3),
j(iy y- -'-'.r e
,
-.,
>->f 1
y-'
2\i e'G-W-1")
\TTz)
;iy
If
w
r(- +
i)
/ *oxp {-
(l
j.
(l
- ^*)} - A*
-
_^p
1)
^ v<
we use
// (z) = (2/*) ( +
It is thus (1)
(*)
- f>, (z)
on the right.
HM (z)^G\ (%Trz)-i
e*(-W-
and similarly
(2)
H (z)^0\ ($ir*)-*e-*fr-*"-W
where
(3)
("<t)
The
results
and v (z). more refined inequalities from them in the following manner
'
may be called Weber's crude inequalities satisfied by Hv (z) By an elegant piece of analysis, Weber succeeded in deducing
{1)
Take the first p terms of the series involved in Hankel's two expansions and denote them by the symbols 2 (z p), 2 2 (z p), so that
(1)
( '
u>
(z
v)
V
m=0 dz
(-)m -(v>)
2 m
(z
o)
V (W)
Yd
dz*
Z%
+
z>
z " {z,p)
-z*{-2izy-i
We
,i,
^+B ^^
(
(2)
^ 1 + x (x^O)
212
[CHAP.
Vn
where
A (z)
and
of z so chosen that
(A' (z) H v v
(z)
= - (*x#)-* e^M
/^^
9
A' (z)
= * <*)r* *<- W
^
P
Jfr
(,),
and so
il (#)
h 2 :^'f
H (Z +
t) dt,
t)]p
where
for
it
follows that
2, m
p)
- {-4#
(1
>
(*)
+ ##,
,2)
-**P-toP)J. \^t)
F*F+0|5
-*-
*
+ oo
,
By considering
is
it
not
difficult to see
A=l
and
B = 0.
in the forms
Hu
y
(z)
\ttz)
(z)
J*-*"-**) {2 r (z; p)
e -*- W-i*> {2<*>
+ Rp
<
H,
where the remainder
= (~\
(z;p) +
Rp %
Rp w may
j,
" ***
'
{V
'
P)
\JTV
\z
\-2i(z
+ t)}*
so,
dt
Sinee
R (z) ^ 0,
we
we have
inequalities,
see that. the modulus of the last integrand does not exceed
2*-p 7T"1
.
(r3
+ t*)-* lp+1)
|
Hence
|
Rp w
1,
^ 2 1-* G*p
(v,
p)
f
.'
(t*
+ )-*<p+ dt,
and
so,
when p >
we have
(4)
IVI<l(y)l^f|^r
I-p
and similarly
r(Jp + J).|(2*)r and it will be observed that in Weber These are the results obtained by the relative values of v concerning made the analysis no hypothesis has been results obtained by from the differ and p\ in this respect Weber's results
;
K^w
,p;i
other writers.
7'34]
7*34.
ASYMPTOTIC EXPANSIONS
Approximations
to
213
remainders in
When
expansion
term in
it
is not very large*, the asymptotic not well adapted for numerical computation because the smallest (with the remainder after the smallest term) is not particularly
may
be sufficiently large
for
the
An
by
Stieltjesf;
we
(x)
method
K (x) and state the results which were obtained by Stieltjes by applying the
method
to
and
(x).
We
so that
7*31 to
\- e
~x
~mi du
V
7T
2
m = oJo
JO
*-?-(lu V
M*
m Bin* 0) '
d0dn
JO
That
is
to say,
n"^"(-i-^^ds*."
.'( 'o
*(1
+ |usin
0)
where
D Rp=~i\
at f"
f*
.'o
t'.'o
w*(l
(0,
w,
+ usm
,
r~-^rd0du. 2
0)
is
Now
when x
the value of
is
m for which
m)j(2x) m
2x
large
(x),
we choose p hp +
;
so that
x=
where a
is
<r,
and then
.
Rp=-i
e _1
tt*Jo Jo
m*(1
..,
$usm +\
:
2
-
0)
m d0dn.
up
to a
Now, as u increases from to qo \ue~* H increases from (when u = 2) and then decreases to zero so we write
,
;
maximum
$-*
= e~
~S\
to oo
we
write
sin 2
= ef.
J
(x),
* The range of values of .r under contemplation for the functions from about 4 to about 10. t Ann. Sci. de VEcole norm. sup. (3) in. (1886), pp. 241252.
(x)
and
A' (x) is
214
[CHAP. VII
plane
;
The domain
found that
and
it is
R
where
a o,o
2f^r
7T*
f-
pM)
2o- 2
a r>s
J -oo J -oo
? v>}dtdV
)
lr%=0
= i
2,o
^j,
ao,2
= i>
dominant terms of
of
are
so that
It is easy to verify
by
Stirling's
theorem that
K
(2xy
term omitted.
'
W
if
'
is
first
In
like
manner
Stieltjes
proved that,
in 7-3, then
(3)
{>
~Jlo
(2*)
'
where
(5)
(6)
provided that
is
x,
and t
is
defined to
be x
p.
Results of this character are useful for tabulating Bessel functions in the critical range;
some similar formulae have been actually used for that purpose by Airey, Archiv der Math, und Phys. (3) xx. (1913), pp. 240 244-; (3) XXII. (1914), pp. 30 43 and British Association
Bromwich, Theory of
8*3.
will
be
proved in
7*35, 7'4]
7'35.
If
ASYMPTOTIC EXPANSIONS
Deductions from Schafheitlin's integrals.
replace a by 2 tan
in the formulae of
f*
]
215
we
p/
(2x) v+ i
1
sin-* 6 cos (v
cos 2 " +1 6
sin"-*
sin
(i;
- V) 6
(v
+
+
i)
n,
(2x) v+ l
A
[**
o
- 4)
(v
f).
cos 2 " +1
612.
P (X,
Q (*,
V)>0,
*)
(_<<
3)
>
o,
(!<"<!)
Q(ar,i')<0.
(~\<v<\)
that
An
is
is
we can prove
that
Q(x,v)/P(x,v)
an increasing function of x when
it is
a decreasing
function of #
when
<
i>< f
For we have
Q'(x,v)P(x,v)-P'(x,p)Q(x,v)
(
(2x) v+ l
2 )
ft* ft*
where
F (e
>
*>
l^^
(cos0cos<)
:
<
tan *
*>
e sin ("
*)
<*>>
so that
!(, *)
+ *<*
0)
g=|^|^
;
tan
</,)
sin (i
-,,)(-
*).
If we interchange the parametric variables 0, $ in the double integral and add the results so obtained we see that, when | < v < f the double integral has the same sign as \ - v and this proves the result.
,
7"4.
Schlafli's investigation
its
In a memoir which seems hardly to have received the recognition which importance deserves, Schlafli* has given a very elegant but somewhat
elaborate investigation of the asymptotic expansions of the functions of the third kind.
The
integral
formulae from
which he derived these expansions are although Bessel's integral is not so well
for
Ann. di Mat.
memoir
is
recognised
120. The only standard work on Bessel functions is the treatise by Graf and Gubler.
which
216
[OHAP.
VH
(z)
large
and
compact form.
Schlafli's
that,
on
it,
the
phase* of
+ l/)
is
He
it
;
being
(I)
and
and
it is
supposed that r
positive
and a
is real.
Let us
first
ir
write
,
+ pe a
where p
is
positive
and 6
is real,
and then
+ pe *)
1
negative,
and
is
consequently equal to
its
conjugate complex.
Hence we have
(1) {L)
sin(a
+ 2fl)
sm(a + 6)'
U
variable
sinfl
sin(a
<f>
+ 0)
such that
= 20 + a - 7T,
.
and then
' ( v2)
/^
w=
as
<f>
- <b)
+
<p)
,,
(u
- l)
u
TT e *.
re
~=
T7 cos | (a
- r sin JX cos i/ a ~ 9) f( +
2
,
.x
9)
(ir a) to (ir a), u traces out a contour emerging from the origin at an angle {ir a) with the positive real axis and passing to < a < 2v. infinity at an angle (ir a) with the positive real axis, provided that
Now,
varies from If this restriction is not laid
infinity
more
than once.
We
value
a,
shall
now lay
this restriction
on a
and then the contour is of the type we give a> and arg z the same
as is permissible.
It follows that
2i sin vrr
^bC'-'M-DI-J*
<f>
where u
*
is
defined in terms of
by equation
(2).
this section with the
The reader
viii.
"method
Chapter
is
7*4]
ASYMPTOTIC EXPANSIONS
<f>
217
Changing the sign of is equivalent to replacing u by 1/u, and so, replacing the expression on the left by its value as a function of the third kind, we have
(8)
e*""
H (re'>-**>) = TV"' +
.
i*")
exp
\reu (u
+ 1)1
^~to
d<f>.
From
as
<
oo
- re u (u so that
t is
iy/u
t,
positive
when u
is
du
~u
dt
it is
must be
n. +u-. = <*+j
-f
,jr?(.t-}W
Jo
J
Hence
W
Now
1
'
'"
p- 1
27rirM"
(f-iy +
(-)p
tP
W^)
'
it is
evident that
(-)
g>
" w = (C- l)""* (r+r + (- l)v (re-)P {(?- 1)' + #/(re*)} (f - 1)' + #/(re*)
where
ft>
is
any
exceeds both
R(p )
and
R ( v $).
and observing that
On making
j_
h+ w+
,
1+
2-Tri]
*
-
^
2),
a*
I>-m + ).(2m!)
I
I> + m + i)
= m\(v,m)
(2m)!
*
we deduce
that
H.
(<-.-.)
) [ <=^> + V>]
<?-#>-* "+p-*
where
(_)p
ddt
"*
, .
2mV(2V); 27rtV(27r)io
d
dtp cos
J i
sin 2
(- V)^ (re{(r1
1)'
+ ?*7(ia )}
a + cos ft
_ ~
sin
ft
(1
W.
B. F.
218
First consider
I
[CHAP.Vn
" +*>-*dZ
When p
circles
(c-iy^-M^-iy+w^M is so large that it exceeds both R (v ) and R{v ^), we take the
ton]
when the
and small
to zero.
tend to
oo
and
them tend
If
now we
write
=e ,ri (l-a;)Joo
on the two rays (which are
1
r (+, l/t*+,
all
we
find that
i+)
+p-i
d
(-)* cos
vir
1
ton)
(-l)*p->{(f_l)*+#/(re*)}
f
aP-* (1
- xY+*-*dot;
'
tx(l-x)/(reia )
Fig. 15.
Now
the numerator of the integrand is positive (when v is real), and the modulus of the denominator is never less than 1 when \nr< a< tr\ for other
values of a
it is
sin
Therefore
where
is
is 1
or
cosec a
according as cos a
is
negative or positive.
When
complex,
(7)
it is
Vk
cosi/tt
\(R(v),p)
|
cos
(i/tt)
(2r)*>
7'4]
ASYMPTOTIC EXPANSIONS
finally,
219
Hence,
(8)
when
*.<,
w.
$v~ [s fcg^>
>
| |
+ * fcggtfj
is
where
fied
0j
according as I(z)
|.
positive or
and p + $ >
has
its real
R{1 -te(l - x)jfrd*j\ > when ^(e^^O, we part positive when v is real and l(z)^0.
by
iz in (8)
If z be replaced
we
find that,
and,
when
v is real,
(i)
(ii)
The
(II)
We
ire<-(M-2
to be zero.
+ l/tt)
As
before,
we
write
u=
+ pe**,
therefore equal to its conjugate
(1).
and then reu />9 **/(! +pei9) is positive, and complex, so that we obtain anew equation
preceding analysis by writing
<p
We
= -(20 + a)
(-iy =
u
rsin'ft
so that
(10)
rCfa
'
sin(a-)sin(a +
4>)'
Now,
angle
as
varies from
to
a,
origin at
emerging from the and passing to infinity at an provided that a lies between ir and it.
6'22 (8)
if,
The contour is then of the type specified for formula missible, we give a> and argz the same value a.
It follows that,
as
is
per-
tt,
K
where u
(11)
is
(re**)
= cos vir
u-"' 1 exp 1
- ^re* (u +
;
-) f
jt d<t>,
defined as a function of
<
<f>
by equation (10)
1
and therefore
(/*'<-*>)
d log u
1)
220
[CHAP. VII
and hence,
now
t
= reu (u-lflu,
we
find that
have consequently expressed a second solution of Bessel's equation in a and the analysis its asymptotic expansion can be deduced (z), the final result being that, when proceeds as in the case of v
form from which
;
We
{1)
fa,
pz 1
H 0) = (}
|
e-*
<*-*'*-*>
(m)
(v,p)
L-o(2i*)"
|
(2**)M
argz according as I{z) ^ or I(z) > 0, and .ft (02 ) ^ when I(z) ^ 0. If v is provided that v is real and p + 1 > u complex the form of the remainder has to be modified, just as in the case of (8).
where
2
|
sec
|
functions of
v, it is
,
must exceed |
Poisson's type.
unnecessary in this investigation to suppose that R (v) as was necessary in investigations based on integrals tof
7'5.
The asymptotic expansions of functions of the third kind follow immediately from Barnes' formulae which were obtained in 6*5, 651. Let us consider " t " '~p r (- s) r (- 2v - s) r ( v + s + j) (- iizy ds
,
/.
ao
ivp
= (- 2u)-*-*
If arg
|
|
["'
J -TCi
( iz) ^ f it
we have
r(-s+v+p)r(-s-v+p)r(s-p + i)(-2izyds\
I/:
1
< j"
J -coi
'
ds\,
is
and the
last integral is
integral of all
0{(-2i*)-'-p}.
But, by the arguments of
6'51,
the
first
integral
is
2iri
times the
it is
sum
at
equal to
(-)"'
* Trans.
273279.
7'5, 7*51]
ASYMPTOTIC EXPANSIONS
j
221
and
so,
M
and
- 1Fr(,
:yS +
i
' ti)
^-H
investigation of
The
HJ
may be
constructed by replacing
by - i throughout.
The reader should notice that, although the determination of the order of magnitude of the remainders by this method is transparently simple, it is not possible to obtain concrete formulae, concerning the magnitude and the sign
of the remainders, which are ultimately supplied
7'51.
It does not seem possible to obtain asymptotic expansions of the four products Jn(z) J,(z) in which the coefficients have simple forms, even
when
ifM
<2>
fi
= v. The
(z)
H^ (z)
term
(z) (z) and which no simple expression the leading terms in the two expansions are
is
H^
2 eg**ri0+>+i).ri
trz
2/^+2^-1
4dz
...}.
expansions
j;(i)/P (#)+r,(i)F,(i)
and
for
J, (z)
simplest
(z)
- FM (z) J (z).
is
The
by Barnes' method,
we should examine
the integral
is to be chosen so that the poles of T (2s + 1) lie on the left of the contour and the poles of the other four Gamma functions lie on the right of the contour and it is temporarily supposed that /*, v and fiv are not
the contour
no double
poles.
The
integral is convergent
provided that
arg(t>)
< f 7r.
222
[CHAP.
VH
to enclose the
lie
sum
m + JO* + v) is
ggH+ww
~~
(jj.
It follows that
^D<^ +1 r (H-H r
>
ft
?--)
xr (^- a r (-^-^) (
)
)s,<fo
sin (ji
+ v) tr
sin
(/*
sin
(ji
- v) ir
| )
J (z)
(,g)
fi) w
+ v) ir
sin (^
2lnT^ ^
C
)J' (^)+
(^
)F ' (
2 COS J
(/LI
+ V) IT
.[{/(*)/,(*)+ r,(*)F.Cr)}
tl
{z)J (z))l
if
By
| |
writing
|
for i
\
throughout the analysis we deduce that, less than $ w, i.e. if arg z\<ir, then
|
both
(J (*)
r (=*-) r(-^-*)cos*7r.(|*)<fc,
7*51]
ASYMPTOTIC EXPANSIONS
223
and
(2)
-
M . t +>))r [|J">WJ.W+F,(,)r.(.)
+ tan i 0 - ")
F () /, (*))]
-BD^^er-'-) "^-)
xr(J- s) r (-^-) sin, w .(i,)-<&.
These results hold for all values of p and v (whether integers or not) provided that, in the case of the former p + v and p p are not even integers, and. in the case of the latter p + v and p p are not odd integers.
We
(1)
now
left
of
and
manner of
7*5.
We first
take
to be
tegrands on the
left
an integer so large that the only poles of the in= -p-\ are poles of Y (2s + 1) ; and then
raoi-pl
J
-x*
-xj p J
(when either integrand is inserted) is equal to 2wt times the sum of the residues at the poles between the contours. Since
, x.-p-J J
-xt-p-J
^ ^ ^ ^ ^_^
when arg z <
j i
we deduce
(8)
.
ir,
are
/*
(z)
F (*)]
7T
.=o
"
(2m +
1) !($*)*+
ipi-rini^-Oir-^l
and
+1
'
+^ -- +
1,
g"
2'"W
2
4
2
At +
\
|/
+1
/A-I/
+ 1 I/-/A+1
2
1-/A-1/
2
'
_1\
?/*
~7rzcosi(/*-v)7r"
2'
224
[CHAP. VII
formula reduces to
(5)
JS(*)+7S(s)~2- 5 {1.3.5...(2m-l)}^,
and, in particular,
(6)
Jo
(,)+I.
<*)~~J& o
I.
(2m)l(&)-
Formula
formulae
(3)
(5)
more general
were stated by Orr, Proc. Camb. Phil. Soc. x. (1900), p. 99. A proof of (5) which depends on transformations of repeated integrals was given by Nielsen, Handbuch der Theorie der Cylinderfunktionen (Leipzig, 1904), pp. 245 247 the expansion (5)
and
(4)
is,
pp.
494498.
It is not easy to estimate exactly the magnitude or the sign of the remainder after any number of terms in these asymptotic expansions when this
method
of J,2
(z)
is
used.
An
alternative
method of obtaining the asymptotic expansion 13'75, and it will then be possible to form
such an estimate.
CHAPTER
8*1.
VIII
The subject of this chapter is the investigation of descriptive properties, including approximate formulae, complete asymptotic expansions, and infunctions and the probe examined are of primary importance when the orders of the functions concerned are large, though niany of the results happen to be, true
;
perties
which
will
We
approximate formulae with the aid of Kelvin's* "principle of stationary phase." And finally, we shall examine the contour integrals discovered by
Debyef
when
secondly, to obtain
numerous
inequalities of
varying degrees of importance, and thirdly, to obtain asymptotic expansions of Bessel functions in which both the order and the argument are complex.
In dealing with the function J (as), in wfiich v and x are positive, it is found that the problems under consideration have to be divided into three
classes,
according as xjv
is less
made
The
and
trivial
(),
when
v is large
z is fixed,
may be noticed
by applying
here.
Stirling's
It is evident,
3*1,
that
^(*)~exp{v+vlog(^)-(v+|)logv}.[co+^+^ +
where
p. 369.
...l,
lii.
1/J(2n)',
(1899),
For physical applications of approximate formulae for functions of large order, may be consulted: Macdonald, Proe. Royal Soc. lxxi. (1903), pp. 251 258; lxxii. (1904), pp. 5968; xc.A (1914), pp. 5061; Phil. Trans, of the Royal Soc. cxx. A 144; Debye, Ann. der Physik und Chemie, (4) xxx. (1909), pp. 57136; (1910), pp. 113 March, Ann. der Physik und Chemie, (4) xxxvii. (1912),, pp. 2950; Rybczynski, Ann. der
[Note.
the following writers
* Phil. Mag. (5) xxiii. (1887), pp. 252255. [Math, and Phys. Papers, iv. (1910), pp. 303306.] In connexion with the principle, see Stokes, Trans. Comb. Phil. Soc. ix. (1856), p. 175 footnote.
[Math, and Phys. Papers, n. (1883), p. 341.] t Math. Ann. txvu. (1909), pp. 535 558; MUnehener Sitzuvgsberichte, xl.
[5], (1910).
226
[CHAP. VIII
Phyrii und Ckemie, (4) xli. (1913), pp. 191208 Nicholson, Phil. Mag. (6) xix. (1910), Love, Phil. Trans, of the Royal Soc. ccxv. A (1915), pp. 105 131 Watson, pp. 516 537 Proc. Royal Soc. xcv. A (1918), pp. 8399, 546563. The works quoted all deal with the problem of the propagation of electric waves over the surface of the earth, and are largely concerned with attempts to reconcile theoretical with experimental results.]
8*11.
The approximation
obtained by Carlini
is
the
;
first
term of the
subsequent terms
281284,
(1)
if
we
define a function
^
us>
...
in the following
may
+^-^(i-^.W=o;
u
^ dz
2
transforms into
* [it' (*)
{u (s)} 2 ]
+ zu (z) -
i/
(1
- 5 ) = 0.
(z) is
If
now we assume
v,
expansible in a series
of descending powers of
thus
u (z)
vu +
it,
+ v^/v + u
3 /v*
...,
where
i*o,
i, Ma,
v,
by
substituting in (2) and equating to zero the coefficients of the various powers
of v on the
left,
we
find that
,{V(i
-)}/*,
s
^=20^).
Ui
'
"2== 3
Ws=
iu
_
s
64^ + 560a
+ i28(i-^)V
9
32(l-5 )
it is
2 7
Hence, on integration,
found
(cf.
1*4) that
J'u (z) dz -
v {log
1
1
+
(
^
(
_ 2, +
)
V(l
*)
2
- l} - i log (1 - z*)
2
+ 353 _ 24i/|(i_^
2
\
j
4>z
+ z*
|
16r> 2
(l-5 )3
3
+ 57607
+
3252
16
2 88-g4
23 25s
13-g8
128^(1
-5 )
2 6
'
8*11, 8*12]
227
Hence we have
(3)
( pz)=s
<?*y ex p y
&T (v + 1) (1
- * ^ *)* {1
4>z*
exp (- F*>
V(l
- *s )}'
where
<*>
+ z*
1__
!6
J
J
5760^1
32**
(i_*.)*
is
the
sum
of the
when
lies
between
and 1 and v
is large.
It is stated
(3),
namely
j
is
;
(ve)
*pW(r-/
(2irv)i(l-**)i{l+V(l
-*)}''
due to Duhamel but a search for the formula in Duhamel's writings has not been and it seems certain that, even if it had been discovered by Duhamel, his discovery would have been subsequent to Carlini's.
successful,
Note.
(5)
may
also
.
= seen a) '
y/(2trv
' , '
tanh o)
where
(6)
^=^^(2 + 3secha)-^^(4secha+sech*a)
- gsLrJ (16 - 1512 sech8 a - 3654 sech4 a - 375 sech8 a)
-C
Ma
4
+ ....
8*12.
Jv (vz) when
may
such values of z, Meisself obtained two formal solutions of Bessel's equation and, if we write z = sec /3, the reader
;
will see,
in
81 1
(5),
A tM.
i.
363368.
228
[CHAP. VIII
where*
(1)
P, = ^(4sec*,8
+ sec</3)
+
s
^^
(32 sec8
288 sec4
+ 232 sec
+ 13 sec8 )
8
+ 4242 sec
(2)
10
+ 103 sec
12
Q,
= ,(tan-/3)-^^(2 + 3sec
I
/3)
5 yg
^ (16
- 1512 sec 8
3654
sec*
- 375 sec
0)
+
to far,
3 225(30i;
....
< 256
+ 7872 sec2 3 + 18 9!200 sec4 + 4744640 sec6 + 1914210 sec8 /3 + 67599 sec )
'
10
To determine
Jv (vsec 0)
;
we take
to tend
%tt),
and we
(3)
infer that
JT. (v sec 0) =
J(~~)
e-^+^-K
e- p-- iQ-+ini
(4)
#<"
(i/
sec
0)
= J(r^!~)
/
'
It follows that
(5)
J, f> sec
/S)
^/
(^~
- p" cos (Q
- p " sin
(Q,
- ^r\
(6)
F, (v sec 0) =
^/(~^)
- ,
where P and Q are defined by (1) and (2). It will appear subsequently ( 8 41) that these formulae are actually asymptotic expansions of Jv {y sec 0) is any assigned acute angle. and F (y sec 0) when i> is large and
#
of
Formulae which are valid for small values of 0, i.e. asymptotic expansions valid when z and v are both large and are nearly (z) and F (z) which are equal, cannot easily be obtained by this method but it will be seen in 8*2 that, for such values of the variables, approximations can be obtained by rigorous methods from Schlafli's extension of Bessel's integral.
The reader
8*2]
Note. form
(7)
229
and
(6),
which
may be written
in the
Wh6re
(**-*) )*>
(v/x),
had been obtained two years before the publication of Meissel's paper by memoir on Physical Optics, K. Dantke Videnskabernes Selskabs Skrifter, [Oeuvres Scientifiques, L (1898), pp. 421 436.]
L. Lorenz in a
(6) vi. (1890).
The procedure of Lorenz was to take for granted which has been proved in 7*51,
that, as
-* Jf *'
t*L
Til + 8'
y2
y2
-*,
"I
-J
-JLTi
*#!_
-*T*
J
'
tf*
SL'._L_
dx
-nxM*'
(1),
which
is easily
to prove that
for
Subsequent researches on the lines laid down by Lorenz are due to Macdonald, Phil. Trans, of the Royal Soc. ccx. A (1910), pp. 131144, and Nicholson, Phil. Mag. (6) xiv.
(1907), pp.
ix. (1911), pp.
is closely
697707 (6) six. (1910), pp. 228249; 516537; Proc. London Math. Soc. (2) 6780; (2) XL (1913), pp. 104126. A result concerning Qv+1 -Q which connected with (8), has been published by A. Lodge, British Association Report,
;
8*2. The principle of stationary phase. Bessel functions of equal order and argument.
The principle of stationary phase was formally enunciated by Kelvin* in connexion with a problem of Hydrodynamics, though the essence of the principle
is to be found in some much earlier work by Stokes f on Airy's integral ( 6'4) and Parseval's integral ( 2*2), and also in a posthumous paper by Riemann %.
to find
an approximate expression
for the
2irJ
which expresses the effect at place and time (x, t) of an impulsive disturbance at place and time (0, 0), when /(to) is the velocity of propagation of two-dimensional waves in water
corresponding to a wave-length 2n/m.
The
by Stokes
* Phil. Mag. (5) xxra. (1887), pp. 252255. [Math, and Phys. Papers, rr. (1910), pp. 303306.] t Comb. Phil. Trans, a. (1856), pp. 175, 183. [Math, and Phys. Papers, n. (1883), pp. 341, 851.] t Ges. Math. Werke (Leipzig, 1876), pp. 400-406.
230
and Rayleigh
of values of
[CHAP. VIII
and wave-velocity suggested to Kelvin x-tf(m), the parts of the integral outside the range (ft -a, /* + ) are negligible on account of interference if /x is a value (or the value) of m
which makes
^[{*-*/(i)}]=0.
In the range (ji-a, /a+a), the expression m{x-tf(m)} three terms of its expansion by Taylor's theorem, namely
is
first
M {*-<f (m)} + 0. (m
and
it is
0*)}
(m-tf,
W2
"n/[-Mm/"(m) + 2/'(m)]'
cob
{*//' 0)
then
+ !*} Ar
a,
oo
and
+ oo
due
to
be seen from the foregoing analysis that Kelvin's principle is, effectively, that in of the integral of a rapidly oscillating function, the important part of the integral is that part of the range of integration near which the phase of the trigonometrical
is
function involved
stationary \.
It has subsequently been noticed J that it is possible to give a formal mathematical proof of Kelvin's principle, for a large class of oscillating functions,
integral formula
will
due to
for
Dirichlet.
be adequate
the applica-
let
ybea function
i
jo
Let F{x)be a function of x which has limited total fluctuation when x^O; Then, ifl<fi<l, of v which is such that vy -* oo as v -* oo
.
v*
a?- 1
F (x)smvx
dx-~
(+ 0)
tr~ l sin t dt
.
Jo
and,
if0<fi<l,the
sines
may
now be used
to obtain
at =/*;
an
This
is
for a
maximum
e.g.
t A persistent search reveals traces of the use of the principle in the writings of Cauohy. See equation (119) in note 16 of his TMorie de la propagation des Ondes, crowned Sept. 1815, Mem. prisent&s par divers savants, i. (1827). [Oeuvres, (1) i. (1882), p. 230.]
t Proc. Camb. Phil. Soc. xix. (1918), pp. 4955. Bromwich, Theory of Infinite Series, 174.
8*2]
231
This formula,
large
and
positive.
J9 (v),
r(*)
2^.3*7T^'
This formula has been investigated by means of the principle of stationary phase, comparatively recently, by Nicholson, Phil. Mag. (6) xvi. (1909), pp. 276 277, and Rayleigh,
Phil.
Mag.
10011004
[Scientific
Camb. Phil.
4248.
From
J( v) =
and obviously
tJo
1 f'cos
[v
(0-
sin 0)}
d$-^^ Jo
ff
( V-ie-fsinh*) dt,
smw
Hence
"
e
^ {t+Anh() dt
[v
["^t dt=Q ^i v)
d$
J
Now
let
<f>
()
f T JO
1 'cos
(6
- sin 6)}
(1/v).
= sin 0, and
f 'cos { v (0 Jo
then
I'
- sin 0)\ d0 l
cosv<l>
Jo 1
-cos
d<f>.
But
and hence,
it
im
_i_
-cos0
2
6*
(0, v),
*-ol
if
<f>*/(l
cos 0)
['
I
Jol-costf
6 r ~ -= r 6* Jo
|
/*
...
* cos v<bd<b
,.
r ^
'
=S
ra)C Si7r
and then
It
still
this result
- cos 0)
1
to establish
<# (1-costfJ
where
so that
_ <fr~*sinflg(0)
3(1 -cos 6)*
'
o ^ * v '
2(1-008^ _
sin ^
3 $ _ 8m 0)
^(0)=0,
gr(,r-0)=+a>,
[Oeuvret,
(1)
proof by
232
THEORY OF
By means
of
BjSSSEL FUNCTIONS
it is
[CHAP. VHI
possible to obtain
and
it
may
also
be proved that
(3)
J/W-lr^ + o (*"*);
is
an associated formula
(4)
F.(,)~-?
2*7Tl/*
The asymptotic
in 8-42.
8*21.
The
It will
now be explained how this a more complete form) by Meissel the theoretical cesses employed will be investigated in 8*42.
;
been used by Cauchyf and Meissel $ to Jn (n) when n is a large integer. expansion was obtained by Cauchy and (in
justification of the pro-
Jn (w)
let
=-
cos {n (0
sin
0)} d0,
us write
sin $*
it
m=0
and
\q =
1,
\ = $,
\2 = TlW
^* =
26&00>
^<
~ T7 llooo
>
It follows that
Jn (n) - -
f( I
(2m +
1)
d0.
When n
is large,
upper
/n(w)~- 2 (2m
7Tt=0
+ l)*. O .'0
\
t*" cos
* BeeProc. Camb. Phil. Soc. xix. (1918), pp. 4248. t Comptet Rendu*, xxxvm. (1854), pp. 990998, 11041107. [Oeuvret, (1) xu. (1900), pp. 161164, 167170.] t Astr. Nach. cxxvn. (1891), col. 359 362; cxxviu. (1891), col. 145154. Concerning formula (1), Meissel stated "Sohon vor dreissig Jahren war ich zu folgenden Forme! gelangt."
8*21, 8*22]
233
where
is
and hence, by
(1)
I I A T (fm + i) Jn (*) ~ 7T (l ) \/
in _
,m+}
cos
(*m +
*) .
when
(1).
m is large;
and
this approxima-
The approximation is obtainable by the theory developed in the memoir of Darboux, "Sur l'approximation des functions de tres grands nombres," Journal de Math. (3) iv. (1878), pp. 556, 377416.
function of
fails
to
dirn and
t=(12rir)*, where
r +1, 2, 3,
...
and near*
**=
* + -41'
By the theory of Darboux, an approximation to X m is the sum of the coefficients of fr** 1 in the expansions of the two functions comprised in the last formula ; that is to say
that
Xw ~2.(36,r)i
*-M-<.->
(2w + l)!(12ir)*" +
,
'
1
2r(2m+)
3^r(|)r(2m + 2).(127r)*m
and
so,
by
Stirling's formula,
(2)
Xm
~
;
(18)*r(|)(m+J)*(12ir)*
This is Meissel's approximation by Cauchy, loc. tit., p. 1106.
8*22.
to
J
i.
(v sec 0).
The
by Rayleighf
to obtain
J (v sec 0) where
is
As
in 8'2
we have
J, (v sec 0)
= - (cos [v (0 - sec
sin 0))
d0+O (1/v),
= 0.
in-
and
sec
sin
is
stationary (a
minimum) when
<f>,
sin
tan +
so that
<f>
decreases to zero as
to
tr.
increases from
t Phil. Mag. (6) xx. (1910), p. 1004. [Scientific Papers, v. (1912), p. 620.] % See also Macdonald, Phil. Traru. of the Royal Soc. ccx. A (1910), pp. 131144; and Proc. Royal Soc. lxxi. (1903), pp. 251258; lxxii. (1904), pp. 5968.
234
[CHAP. VIII
sec
sin 0)}
r/-0
dd
f-/5
Jo
+ tan/5-]
dd
and 0*
Hence,
i/^
as
0-/8.
V(2 tan 0)
<f>*
(d0/d<f>)
0^0 ^ tt,
it
cos \v (0
- sec
J*
sin 0)}
d0 ~ 2
cos (v (<
+ - tan 0)}
//
2 <f>tan)
^i/
tan 0,
and
so
(v / "\
n\ (1) v '
The formula
/0 \ (2) '
is
ti
o\ sec r'/ 0)
V(i*"rtanJ)
( 777
(1).
The reader
expansions
8*12 (5),
To complete the
we have
to
shew that 0*
(d$/d<f>)
has
Now
<f>
(dO/chf))
2
,
is
.
equal to
'
But
.,
^
'~
cos
ft
cosecd
(1
ir, and then increases steadily since k(0)=O wben 0=ft<r, it follows that h! (0)^0
when
0^0 <ft
|
and
ft^d^JT.
Hence
point (a
yjh
(ff)
is
< 6<
|
minimum)
it
when
proved.
< 6 <
<6<
since
is
8'3]
8*3.
235
the method of by Debyef to obtain integral representaof Bessel functions of large order from which asymptotic expansions are
If,
readily deduced.
(w) dw,
v is supposed to be large, the contour is chosen so that it passes through a point w at which /'(w) vanishes and the whole of the contour is then determined by the assumption that the imaginary part of f(w) is to be constant on it, so that the equation of the contour may be written in
[
|
in which
the form
If(w)=If(w ).
are real
To obtain a geometrical conception of the contour, let w = u + iv, where u, v and draw the surface such that the three coordinates of any point
;
on
it
are
u,
v,
Rf(w).
If
absolute
all
Rf(w) = z, and if the z-axis be supposed to be vertical, the surface has no maxima or minima except where /(w) fails to be monogenic; for, at
other points,
dv 2
du*
The
points [u
R/(w )]
the plan of a curve on the surface which goes through one of the passes on the surface. This curve possesses a further property derived from the equation of the contour for the rate of change of the contour of integration
;
f(w), at any given value of w, has a definite modulus, since /(w) is supposed to be monogenic and since If(w) does not change as w traverses the contour,
;
it
follows that
is
as rapidly as possible
that
is
to say, that
the curve
it,
characterised
is
so chosen that it is
any point of the steepest curve through that point and on the
its direction, at
surface.
It may happen that we have a freedom of choice in selecting a pass and then in selecting a contour through that pass our choice is to be determined from the consideration that the curve must descend on both sides of the pass for if the curve ascended, Rf(w) would tend to + oo (except in very special
;
cases) as
left
if
R (p) > 0.
* French "Methode du Col," German "Methode der Sattelpunkte." t Math. Ann. lxvii. (1909), pp. 535 558; Miltichener Sitzmigsberichte, xl. [5], (1910). The method is to be traced to a posthumous paper by Biemann, Werke, p. 405 ; and it has recently been applied to obtain asymptotic expansions of a variety of functions.
236
[CHAP. VIII
oscillate rapidly
The contour has now been selected* so that the integrand does not on it and so we may expect that an approximate value of
;
:
8*2)
from the physical point of view, we have evaded which occur with any other type of contour.
The mode of derivation of asymptotic expansions from the integral will be seen clearly from the special functions which will be studied in 8*4 8*43,
8*6, 8'61
convenient to enunciate at this stage a lemma f which will be useful subsequently in proving that the expansions which will be obtained
;
but
it is
be analytic
when r ^ a
\
8,
where a
> 0,
>
and
let
F(t)= I am ^ m^~\
when
|
^ a,
let
F (r)
<
lTe*
r
,
where
and
b are
the
positive
numbers independent of
when t
is positive
and r > a.
Then
asymptotic expansion
is
1 am T (m/r) v~ m lr m=l valid in the sense of Poincare when \v\is sufficiently large and
e- VT F{r)dT~
/,
where
is
It is evident that, if
can be found
such that
whenever t
whether r ^ a or t
j
J
> a and
;
therefore
e- vr F{r)dr=
2
\.
m=l JO
e^a^^^dr + RM
where
l^jfl^f
\e~T
K^i^^dr
Jo
x
provided that
R (v) >
b,
which
is
the case
when
analysis
remains valid even when b is a function of v such that compared with v. We have therefore proved that
f JO
and
so the
lemma
established.
an account of researches in which the contour is the real axis see pp. 1343 Burkhardt's article in the Encyclopadie der Math. Wits. n. 1 (1916). t Cf. Proe. London Math. Soc (2) xvu. (L918), p. 133.
* For
1350 of
8*31]
8*31.
237
It has
been seen in 6*2, 6*21 that the various types of functions associated
On
we
shall
explained in
In accordance with the principles of the method of steepest descents, as 8*3, we have first to find the stationary points of
x sinh w vw,
qua function of w,
(1)
i.e.
we have
a;
cosh
w v 0;
consider,
and it is at once seen that we shall have three distinct cases to in which xjv is less than, greater than, or equal to 1, respectively.
sider these three cases in turn.
(I)
We con-
(2)
x v sech
is
a,
w = a + 2mri.
It will
sinh
w vw
is zero,
/ (x sinh w vw) 0.
Write
w = u + iv,
= 0,
or
where
u, v are real,
and
this equation
becomes
cosh u sin v
so that v
/ox (o)
v cosh a = 0,
v cosh siny
a
.
coshw =
,
The contour
We
and ir, correspond pairs of values of u which are equal but opposite in sign and as v increases from to n, the positive value of u steadily increases from a to + x
;
.
To
values of v between
* The contours investigated in this section are those which were discussed in Debye's earlier paper, Math. Ann. lxyii. (1909), pp. 535 558, except that their orientation is different; cf. 6-21. t The effect of taking stationary points other than a would be to translate the contour parallel to the imaginary axis.
238
[CHAP, vin
is unaltered by changing the sign of v and so the contour is symmetrical with regard to the axes the shape of the part of the contour
The equation
between v
= tr
and v = tr
is
shewn in
TTl
Fig. 16.
>
-77
Fig. 16.
If
it is
in the figure)
As
mi
since,
to oo
+ iri,
t decreases from
+ oo
to
+ oo
and
by
6*2 (3),
IlTlJ ao-vi
we have obtained a curve from which we can derive information concerning Jv {x) when x and v are large and x\v< 1. The detailed discussion of the
integral will be given subsequently in 8*4, 85.
iri give information concerning a second but this problem is complicated by Stokes' phenomenon, on account of the two stationary points on the contour.
oo
to oo
(II) (4)
When x/v>l, we
/3
such that
and
w cos /8 = 0,
are
w= i#.
take the stationary point
if$,
When we
we
obtain
is
I (sinh
so that, replacing
/
why
+ iv,
5\
cos h
u=
+ 1^_J^ __
(v
ft)
cos
ft
8*31]
239
Now,
between
sin ft
and
tr,
the function
has one
minimum
(v
zero
for other
values of v between
sin ft
when
= ft.
They
are infinite
when
v is
or
ir.
The shape
is
as
shewn
in the upper
is
real
directions indicated
to oo
by the arrows. As
+ iri,
r decreases from
+ oo
\
to
oo
and so we
^^
ni
S^3
-"""*^
-ni
Fig. 17.
have obtained a curve from which [ 6'21 (4)] we can derive information concerning (#) when x and v are large and x/v > 1. The detailed discussion v of the integral will be given in 8*41, 15'8.
{1)
If
stationary point
ift, we
curves shewn in the lower half of Fig. 17, and the curve going from
oo
oo
to
7rt
{i)
in 8*41.
The two
integrals
<
240
(III)
[CHAP.Vm
to be con-
may be
from
(I) or
equal to
sidered are v
(6)
and
cosh u
18.
v/sin
v,
Fig. 18.
We
H,
ll)
(v)
and
{i)
(p)
by considering the
curves from
oo
to oo
iri,
Jv (v)
is
obtained
from the curve which passes from oo mi to tion will be given in 8'42, 853, 8*54.
Geometrical properties of Debtee's contours.
+ vi. The
detailed investiga-
8*32.
An interesting result which will be found to be important in dealing with zeros of Bessel functions ( 15*8), and which is also used in proving certain approximate formulae which will be stated in 8*43, is associated with the second of the three contours just discussed (Fig. 17 of 831).
oc
to
+ jrt
is
positive
is
from
smh u -T- =
.
du
dv
sin 8 v
But sin(; /3)secv-(i>-/3)cos/3 has the positive derivative cos /3 tan 2 v, and hence
follows that
sin
it
v cos
has the same signt as v /3. Therefore fur the curve under consideration, dv/du
since v /3 and u
is positive.
which
3=0,
the slope
is
on the
left
of the origin
and
^3
f This
is
8*32, 8*4]
241
we
write
Wr) = ^
and then
it is sufficient
to prove that
3^'*(*>)-iP(t;)+l>0.
Now
2*'
left
W
:
[*" ()(v)
3
+ ()]
{sin'p+3 cos2 } cos 0+ sin 2 v sin
,
- [(*-)
0- 3 cos v sin
(i>
- /3)J.
v, But
.
/o\ o (v - 0) cos 0+
x
'
8in 2
vsin-3cost>sin(v-/3) -' i
sin2 r +3 cos* v
--5
s
*.
T~\%> an<*
^ ^ nce
8i
(v)
it is positive
when =0,
it
it is
when
<v<
ir.
Therefore, since
- 0,
follows that
3^'()ip()+l
has =/3 for its only minimum between This proves the result stated.
v0 and
t>=r
and therefore
it is
not negative.
8*4.
the results obtained in 8'31 we shall now obtain the asymptotic first kind in which the argument is less than the order, both being large and positive.
From
We retain the notation of 8*31 (I) and it is clear that, corresponding to any positive value of t, there are two values of w, which will be called wx and Wj the values of wx and w% differ only in the sign of their imaginary part, and it will be supposed that
; ;
l(w )>0,
1
J(>8)<0.
We
then have
where x v sech a.
Next we discuss the expansions of /i and w% in ascending powers of t. Since t and dr/dw vanish when w = a, it follows that the expansion of t in powers of w a begins with a term in (w a)* by reverting this expansion, we obtain expansions of the form
;
Wl
*
_= i
Wa _ a=
2
=0
(-)'n+1
ni+1
-^
T*<"+1 >,
558.
in this section
and
242
[CHAP. VIII
and, by Lagrange's theorem, these expansions are valid for sufficiently small
values of Ti.
f(Q+.o+)/d t i;A
am
dr
>
~2wJ
2vi]
Ut/t*< +1
dw
Ti(m+i)
*
The double
fractional
powers of t and a single circuit round a in the w-plane corresponds to a double circuit round the origin in the T-plane. From the last contour
integral it folio ws that
am
is
o)
in the expansion of
T-j(m+i)
m ascending
wa=
t
powers of
w a; we
coefficients a m .
Write
= - sinh a (cosh
c,
where
= | sinh a,
coefficient of
in the expansion of
is
the
a (m),
= c -* (wt+1)
ai(w)
a l(TO)
/
\
= Co-H-,|_^.|},
(m + l)(m + 3) m + 1 .Ca f _ <v *i*n|__ + s
i.
<,j.t>
2!
a,(m)
(1)
= c -Hw|-||
Co
(m + l)(m + 3) 2dc,
2s 21
.
'
c2
(ro+l)(i + 3)(wi
2*. 3!
+ 5)
c^
'c*
(m +
1)
(m +
(m+l)(m + 3)(ra + 5)
2s 3!
. '
W 3^
c
8
3) /2cj c8
CoV
(to
+ 1) (m + 3) (wt + 5) (m + 7)
2*. 4!
c^)
'cA
On
substitution
coth
a},
(2)
| I
= a, (2) = - (- J sinh a)"* {$ - ft coth* }, 8 a, a, (3) = - (- 1 sinh a)" {-& coth a - ft coth a}, coth'a 4 fljfr cotha}, 4 - a* (4) - + (- i sinh a)"t T ^ {
ff
8*4]
243
Now
when t
|
j
^= 2
a, m
T"-
small
and since
-7it
aw
= cosh o cosh w,
to zero as r tends to
follows that
+ oo
and
so
in the
lemma
dr)
\dr
"
o^IVm + l )
when x
is large.
{(a/,
Since arg
-o)/t*}-- \ir as t
-*-
0, it follows that, in
($5),
the phase of a
= + ^7r,
and hence
>
Jr ( y secha)^
v(2
^ tenha) J
!
-^^--. (||;taDha)TO
where
(4)
The formula
when
by taking a
oo
to
oo
+ irt,
. ,
is
v , r,(,secha)~-
K-tnha)
V(iffytanha)
^-T^T -(^vtanhard^-w^ldr,
qua function of the
r(m+i)
(~)m A m
The
complex variable t, should be noted. These singularities correspond to the points where w fails to be a monogenic function of t, i.e. the points where dr/dw vanishes. Hence the singularities correspond to the values a + 2ntri of w, so they are the points where
t = 2mri cosh
a,
and n assumes
all
integral values.
;
It is convenient to obtain
(W( To)
a formula for dw/dr in the form of a contour integral be a pair of corresponding values of (w, r), then, by Cauchy's theorem,
'
if
fdw\ = J_ /"('+) dw dr Jl_ f (*+> dw_ tt 2tri J dr t-t \drJo 2irt J where the contour includes no point (except w ) at which t has the value t
244
8*41.
[CHAP. VIII
In 8*4 we obtained the asymptotic expansion of a Bessel function in which the argument was less than the order, both being large ; we shall now obtain the asymptotic expansions of a fundamental system of solutions of
Bessel's equation
when the argument is greater than the order, both being large.
;
We
it is
any
two values of
oo
+ tri
it
and
it will
We
then have
HJm (v sec 8) =
where x =
third kind.
It is thus found that
v sec 8.
is
wi
XT
\
Jo
[dr
~-
-r- \ dr,
dr)
The
analysis
now proceeds
by
except that a
replaced throughout
exactly on the lines of 8*4 %8, and the Bessel function is of the
Jo
\dr
dr)
of o
,
m=0
that
is
xm+*
of ( i sin )""*,
we
obseive that
i8)/r} -* + \w
as t -* 0,
a
and so
= el7V(sin/8).
Consequently
(1)
H
like
{1)
(v sec
8)
we
*JQvrr tan 8) M =
find that
T ()
1,
L-*l
'
*
. *
(| pi tan )*
In
w=0
T(i)
In these formulae, which are valid when /8 is a fixed positive acute angle and v is large and positive, we have to make the substitutions
:
(3 >
j
^2 = Th
and
(2),
+ ih cot2 y3 + $& *,
we
find that
If
(4)
we combine
(1)
J (v8ec8)rsj
\pwtaa 8/ L
_i_
^
*>
^
i
*
n
=o
T(i)
(i/tan/S)
m ? (-)
r(2m + 8)
Atn+r
"I
8'41, 8-42]
(5)
(
245
F(i/sec)~
/
w r(2m l) ilw, / \*T sin(>tan-i//3-7r) a a x v ?'.- 2 (-) =-r + *-'.,-. 3/ Vi^rtan^/ L r<$) (|itanJ* =o
*.
'-
-cosCi/tan^-iz/g-iTr)
*
'
|
)W =o
0-)"-r(2m + f)
r(i)
(^tan/9)2 ' +1
by the principle
8*42.
The formulae which have been established in 84, 8'41 obviously fail when a (or ) is small, that is when the
argument and order of the Bessel function concerned are nearly equal. It is, however, possible to use the same method for determining asymptotic expansions in these circumstances, and it happens that no complications arise by
supposing the variables
to be
complex.
#.(*),
HJ*(i)
where z and v are complex numbers of large modulus, such that \z v\ not large. It will appear that it is necessary to assume that z v *= o (z$), order that the terms of low rank in the expansions may be small.
is
in
We shall write
v
= z(l-
),
and
it is
We
(1)
then have
H^
v
(z)
= .j
exp
\z
(sinh
w w) + zew] dw,
is
w w
is real
We write
t=
w sinh w,
Y
and the values of w corresponding to any positive value of t will be called w and wt of which wx is a complex number with a positive real part, and w2 is a real negative number.
,
We
(2)
then have
^-
exp (zews )
^l dr.
246
[CHAP. VIII
its*,
and hence we
exp (zeWi)
-j-i
*"
= t~ 3 2 m T Jm
b.
m=
exp (zewt)
^p = t-3 2
is sufficiently
e <"*+
bm t*
small.
To determine the
coefficients bm
1
/-(0+.0+.0+) 0+.0+) /(<>+,
we
observe that
/dw
rf
<+!)
_j_r(o+)
~
As
is
6wt J
6XP (
<' +1 >
powers of t; a triple circuit round the origin in the T-plane corresponds to a single circuit in the w-plane.
It follows that bm is equal to Je* ,m+1,w * multiplied
by the
coefficient of
wm
in the expansion of
exp (zew)
{(sinh
w w)/w*}~l
("*+1 >.
&i(w), b2 (m),
...
so that
It is easy -to
shew that
(b,(m) = &*"**,
bx
(m) = &<*+*&,
w.).a- {-*}.
(3)
w=
61(m+l)
tT--V-|'
--
o*{m)
o
|
24
12Q
504()0
j.
= 6*"+m (e*),
8-42]
so that*
247
5 B
(4)
= l, - fa % (m) - \ <?z*
(e*)
(e)
4()-e* B (es) = *V - ^,
t
^5
*3
+ **,
4.(0)
[3,(0)
We
then have
exp
-^ ^=
(jreto,)
ft<0)-T7*flfanr.
^jb.}
TO (e*) t**
t
ttT
ir~*
2
=0
ei<+u 6* <M+1
>
ezp(#e^)^*-iT-< 2
\
e^
m+1
(IT
.
^Q^m+li Bm (e2)T^n
lemma
TO=0
satisfies
of 8*3.
of 8'3 that
(ez) S
(5)
H^(z)^^^i^^Bm
h *{*)
9
mUni + l)^.^^,
and
similarly
(6)
AJ^-i( ^
+
5TO(
^)sini(m + i )w .r|i^i|)
We
(7)
j^z
F,
^Lh
Bm{ez)
^ Hm+l)v '~m^
821
(2), it is to
(8)
W ~-^2
(->.^Min4(m + l).5jg^).
f
From
<)
m is large,
^m(0)~ r() (w + i)J(127r)Jw
,
for
Bm (ez).
The dominant terms in (7) were obtained by Meissel, in a Kiel Programing 1892; and some similar results, which seem to resemble those stated in 8'43, were obtained by Koppe in a Berlin ProgrammX, 1899. The dominant terms in (8) as well as in (7) were also investigated by Nicholson, Phil. Mag. (6) xvi. (1908), pp. 271 279, shortly before the
The values of B (0). Bt (0), ... B w (0) were given by Meissel, Astr. Naeh. czxvii. (1891), 359362; apart from the use of the contours Meissel's analysis (of. 8*21) is substantially the same as the analysis given in this section. The object of using the methods of contour integration is to evade the difficulties produced by using generalised integrals. The values of B (tz), Bj (cz) and JB9 (ez) will be found in a paper by Airey, Phil. Mag. (6) xxxi.
*
col.
(1916), p. 524.
t See the Jahrbueh liber die ForUchritte der Math. 1892, pp. 476 178. I See the Jahrbueh itber die ForUchritte der Math 1899, pp. 420, 421.
248
[CHAP. VIII
|
We
The
arg z
<
\tr,
which
(8), is
removable.
qua function of t, are the values be a monogenic function of t, so that the t 2 singularities are the values of r corresponding to those values of w for which
singularities of the integrand in (2),
of t for which
(or
w)
fails to
drjdw
= 0.
They are
Intri,
where n assumes
all
integral values.
It is consequently permissible to
less
swing the contour through any angle w than a right angle (either positively or negatively), and we then
analytic
obtain the
continuation of
(l)
(z) or
H
v
(z)
V < arg z < \ir t\. By giving v suitable values, we thus find that the expansions (5) (8) are valid over the extended region
w -
7r <
arg z <tt.
If we confine our attention to real variables, we see that the solution of tbe problem is not quite complete ; we have determined asymptotic expansions of Jv (x) valid when x and
v are large
and
(i)
xjv
<1,
(ii)
x/v>l, (iii) x- v not large compared with #*. But there (i) and (iii) and also between (ii) and (iii), and in these
|
a?
is large.
In these transitional
regions simple expansions (involving elementary functions only in each term) do not exist.
But important approximate formulae have been discovered by Nicholson, which involve Bessel functions of orders + J. Formulae of this type will now be investigated.
8*43.
The
8*42 in the
n,
manner
Jn () = and,
cos (n0
x sin 0) d$,
it
when x and n
is
principle of stationary phase ( 8*2) that the important part of the path of
integration
is
small
\ 0*.
x and n under
-x0 + %x0) d0
OS -
"cos (n0
-x0 +
\x0*) d0,
und Phys.
ir J a
* Phil.
Mag.
pp.
247249
(3)
239250.
8*43]
249
It follows that,
and the
when x<n,
(1)
Jn (x) ~ - {-^^-} *1 [
a?
>
and,
when
> n,
(2)
^(^^Ij^iOp^H.^,
{2
for
Yn (x)
when
a;
> n was
it is
also found
r.(.)~-p^(V-i-^-
The chief disadvantage of these formulae is that it seems impossible to determine, by rigorous methods, their domains of validity and the order of
magnitude of the errors introduced in using them.
remedying this defect, Watson* examined Debye's integrals, and discovered a method which is theoretically simple (though actually it is very laborious), by means of which formulae analogous to Nicholson's are obtained together with an upper limit for the errors involved.
With a view
to
is
the following
its
argument
may be
J
where
t
(v sech a)
~
1)
is
rx>-\-ni
/
e~XT dw,
= sinh a (cosh w
positive
expanded in ascending powers of w, Carlini's formula is obtained neglecting all powers of w save the lowest,
all 0, Cauchy's formula of 8 2(1) powers of w save the lowest, \wi
.
is
similarly ob-
tained by neglecting
first
is
desirable to
w3 cosh a,
iirst
may
transitional region.
The
we
\e-**dW,
where
* Proc.
r
Camb.
is
= - \ W* sinh a - W
cosh
a,
t This w.
B. F.
250
[CHAP. VIII
positive
and the contour in the plane of the complex variable is so chosen that t is on it. If U + iV, this contour is the right-hand branch of the
W=
hyperbola
tftanha
and
+ Jtf^F
2
,
It therefore has to
r<x>
I
e~ ZT
oo
dw
is
e'^dW.
-ni
J ooexp( Jirt)
These integrals
differ
by
i/:+/;H-^K
and so the problem
|
is
for
{d(w W)fdr)
will
j.
And
which
d (w dr
and so
<
Sir sech a,
ifj>j:H-i*
Hence
<
6-7T
~
ZTTl J ao-ni
e-*r
dw = J_l
ATI J
,
e-*rdW
ex p (- Jiri)
+ -l,
V
where 6l
j
<
1.
To evaluate the
6*4),
integral on the right (which is of the type discussed in modify the contour into two lines starting from the point at which
W = tanh a and making angles + %ir with the real axis. If we write W= - tanh ^e ini on the respective rays, the integral becomes
4-
e*
ni
- \ v%e$ ni tanh2 a}
a)
|
dl-
exp
powers of tanh 2 a and integrate term-by- term and we get on reduction easily justified
in
7ri tanh a
[7~_j (1/
Jv {v sech a) =
IT
75-
VO
exp
{v
(tanh a
-f
3 ^ tanh a
a)} if
exp
[v
j (
i>
tanh3 a)
a)},
-I-
3#i
i.-
-1
(tanh o
whore d
]
<
1.
This
is
(1).
8*43]
It can be
251
whether \v tanh8 o be small, of a moderate size, or a smaller order of magnitude (when v is large) than the approximation given by the first term on the right.
large, the error is of
in
is less
We
then have
H (v sec 0) =
where
T
gvi(tanj30)
:
foo+Hir-p)
Tn
e~*r div,
J -co-ip
= - i sin
(cosh
w - 1) cos
is
(sinh
it.
w - w),
positive on
The
/
where
r
e-^dW,
= -\i W* sin
cos 0,
and the contour in the plane of the complex variable is such that t positive on it. If WsU+iV, this contour is the branch of the cubic
is
r*)
=
oo
oo t'tan/9
foo+t(ir-/3)
I
exp^Trt.
It therefore has to
e-^dw
is
e~XT dW.
The
and
it
when*
UiT
\ir,
then
d(w- W)
Hence
it
<
127r sec 0.
follows that
-.
TTIJ
e-^dw^.
-*-ip
irXJ- oo-i tan
fi
e -xr
dW + zl^ v
where
& <
j
1.
To
evaluate the integral on the right, modify the contour into two lines
meeting at
real axis.
W=
tan
Kir
and
ir
respectively to the
On
W= *
-f
- i tan
+ %&\
If
/3
of
/3
is
of Debye's contour
which was
proved in
252
[CHAP. Vin
it is
powers of tan a
j3,
W=f
=
iri
[-**
J_j
(4 v tan8
+ e**' Jj (^ v tan
3
#)]
3
7re<rttan
^ exp (- J i* tan
it is
3
#j<'>
(^ tan
).
On
(5)
(tan
(6)
F (i/ sec )
= | tan /3 sin
{i>
{*/
- tan - )} [J. j - Jj] + 240 (tan - tan - )} [J"_ j + J] (tan - tan -)}. [7"_ - </j] + 240 M
3
.
2 /i/,
where the argument of each of the Bessel functions J>. on the right is %v tan 3 and and are both less than 1. These are the more precise 2 3 forms of Nicholson's formulae (2) and (3); and they give effective approximations except near the zeros of the dominant terms on the right.
|
|
8*5.
Descriptive properties* of
integral,
a?
< 1.
which was obtained in 8*31(1) to represent an asymptotic expansion of the function. v But the contour integral is really of much greater importance than has hitherto appeared for an integral is an exact representation of a function, whereas an
(v sech a)
The contour
was shewn
in 8*4 to yield
And
is
it
when x =
1)
Jv (vx)
when
= log {reie
so that
u = log r,
},
= d.
selected,
x sinh w w is equal to its conjugate complex, and the path of integration flexion in the real axis. Hence
1
is its
own
re-
roo+n-t
(vx)
= rJ
i __
rw
I
e"
,a
dw
gv(as8inhww) fly^
ttJo
*
The
150174.
8-5]
253
20
r
so that
sin
and,
when
this substitution is
,
made
2
(w x sinh w)
a 2
is
log
e + y/ifr-X* sin 0)
-.
x sin
.a a at cot y (tr
.
a;
sin 2 0).
*m
(1)
J {vx)=^{'<r'>de,
(a procedure
which
is
easily
found that
l \ Ti J (vx) =
w
f
e- *<.*<
fib*
ttJo
x */(&* x* sm* 0)
- Ja
fl- a'Binflcosfl .
dv.
This
is also easily
raa+iri
.
JJ (vx) =
(x*inhu>-w)
sinh
^
it
<
3>
m^'^
w
so that
(5)
4^.,-^a^
fl-g'sinflcosfl
^-M)*^-^' ^
6
n
F(0,x)>F(O,x)>F(O,l) = O;
d
V(<* \
and
also
ia\
Next we
(7)
+ x>).
To prove
we
shall first
i
shew that
O aPsmO cos 6
..,
.
t It is
This function will not be confused with Schlafli's function defined in 4-15. supposed throughout the following analysis that O<.c^l,O^0<ir.
254
It is clear that
[CHAP. VIII
= s'(\-x*)<<J(\+x*),
g(Tr,x)=\
so that, if
<J(\+x%
0,
or
tt,
^(1+x
O
).
If,
its
when
1
had a value
between
2
and
O
ir,
then
a;
cos
2fl
2
_ (0 - x
O 2
si n
cos
o
f_
(tfo'-^sin ^)*
(0o -a?sin 2
)i
'
and therefore
g {0, x)^g (8
so that, no matter
x)
its
+#
a
Hence
^W-*w^^
1
sin
cos
2
V(l+a )
and
so
whence
x)
+ * 0* V(l - *").
To prove
observe that
dF
e f
'
X)
>
\/(0*
- " sin
is
0)
and integrate
obvious.
From
these results
we
are
now
J (vx) and
JJ (vx)
qua functions of
Thus, since
OV
IT JO
(5), it follows
that
JJ {vx)
is
of v.
Also, since d {**<*>
JAva:)\ = _ 1
n
(
SF{0> x)
OV
_ F ^ 0> x)}
*-*>. *>-"*.*>
d0 <0,
7T.' o
(5), it follows
e"
that e vF({t x
'
>
Jv (vx) is a decreasing
F,0,x)
'
(vx).
8*51]
255
Jv ( vx) ^
<
so that
Jo
exp {- i v0- V( 1
- a )} dd
8
eF(0,x) roo
exp {- i/0\/(l
p -vF(0,x)
- x*)\ d0,
(9)
<L(vx)^
(1
is
x*)l
^(Zttv)'
The
last expression
approximate expression
is
( 1*4,
811)
for
always in error by
* positive values of
result for
0*
The corresponding
and replace G(0,x) by
JJ {vx)
2
is
derived from
(7).
Write
- a? sin
=(#,#),
for brevity.
Then
2xj;{vx)=
fi
/V-.*
fir
I
dG ( >^ {G(0,x)}-id0
-vF(0,x)
7T
expi-^GyVa+a^.G-MG
.'o
e -F(Q,x) fao
IT
Jo
(vx)
Texp {- J^(l+^)j
^ -'*<. )
2 a? ) (
G-idG,
and so
(10)
xj;
factor
+ *)VV(2 w).
is
^(1
remarkable.
(11)
J
+ V (l
-W-
than
its order.
8'51.
Lemma
concerning F(d,x).
We
(1)
shall
now prove
dF{0,x)
lemma
when
y,^^
is
The lemma
will
(vx).
in error by excess for sufficiently
large values of
t Cf. Proc.
London Math.
256
If
[CHAP. VIII
sin2 0)
= H{0, x), we
dF(0,x) dH(0,x) d0 d0 /
is
a non-decreasing function of
that
is
i
to say that
(\-0cot0)2 +
-a?sm*0
a? sin
is
cos
a non-decreasing function of
0. is
The
(0 - x* sin
- 1 - sin 0) (1 - x*) + 2 (0 cosec - cot cosec - % sin* 0) (1 - ar ) + 2a- (1-0 cot 0) (0 cosec - cos 0) + sin 0(12 2 2
2 2
a,-
) ],
and
To
we
1
first
observe that,
when
< 6 ^ *r,
d+sindcosd-2^sin
sin
d>0,
> 0,
positive differential
(iii)
2(cos0-0 -1 sin0) 2
(ii)
sin 6 (6
- sin
6 cos 6),
and then
2
sin 2 6
)
+ cosec
(sin
- 6 cos - J sin 3 6) ^ 0,
cos 6 - sin 3 6)
cosec2
0-l-$sin 2
(5
= 6 cosec2
sin 2 0)
+ cosec 6 (sin
>
0>
shewn that
(}><>
where
the. variables are
understood to be
0.
and
x,
differ-
It is
now
obvious that
ifiM-*{S>*
and,
if
we
to 0,
we
get
when = 0, F>(0,x)H{0x) _
if (0, #)
>Q
and the truth of the lemma becomes obvious when we substitute the value
of
(0,
8*52]
8*52.
257
We
is fixed,
shall
and
now prove a theorem of some importance, to the effect that, if x ^ x ^ 1, then Jv (vx)/Jv (v) is a non-increasing function of v, when
is positive.
be valid only when S^.r^l, (where 8 is an some expressions introduced in the proof contain an x in their denominators; but the theorem is obvious when ^ x < 8 since e vF > z Jv (vx) and e -vF{o, x)/jv () are non-increasing functions of v when x is sufficiently small moreover, as will be seen in Chapter xvn, the theorem owes its real importance to the fact that it is
will
>
x in
the
neighbourhood of unity.}
It will first
be shewn that
(1
To
we
and,
when we
differentiate
OV
7T
L\/^
W
> <*
<* *))-*
>
or
' e vn$,x)
de
"
\\C((l -rVh
dFie X ^
IP/A \in(A
^W^1^L
A,)
x e-J?ie,x)d0,
if
we
Hence
follows that
where
by using the inequality F(yfr, x) > ^(0, x) combined with the theorem of 8*51.
'
258
Since CI
that
is
[CHAP. VIII
is
to say,
we have proved
J
so that
( vx
'
\dJv (vx)
1,
we get
~dJv (vx)
so that
,"|*
dJv (vx)
dv
r -I J v (VX)
*Z
Since
(vx)
and
(v) are
in the form
(2)
^{J
v.
(vx)/Jv (v)}^0,
to
and
is
which was
(vx)jjv (v)
8*53.
If,
Properties of Jv (v)
and
J '(v).
v
for brevity,
we write F(0)
in place of F(0,
1),
so that
(1)
F (0) ^ log +
J
v
^
v
^n*
B),
(v)
and
J '{v)
The
first
5
4^/(9 V3)
and we
is term in the expansion of F (6) in ascending powers of shall prove a series of inequalities leading up to the
s
result that
F (0)1
is
a non-decreasing function of
0.
We
shall first
shew that
To prove
this
we observe
6-*
that
cot *)/*}
sin* 0)
*"(*>
IP-* ~ ^~ V(^ -
^
*
Vi &2
,,
_ siR2 ,.
'
* It is to
be understood that
JV (v) means
the value of
the particular
value
v.
^
FUNCTIONS OF LARGE ORDER
d^
1
8-53]
259
and that
j
dd
d_
f
-0 cot 0) _ ~
fr~~~)
a
fl
cosec 2
fl
+ flcotfl-
2
'
p
"'"
vW-sin
0*
fl))
d&\
Hence
it
2
'
follows that
d (F'(0)}
<** I
#- | - * (* - sin
0-sin0cos0
2
W <*
//te
C0Sec2 *
0/1
+ ecot0 ~
x (0 + sin
v 2>
cos
- 20* cot 0)
by inequalities proved in
Consequently
(3)
851.
0F"(0)-2F'{0)>Q,
to say
that
is
^ [OF'
(0)
3F (0)} > 0.
to
If
(4)
we integrate
this inequality
from
we get
0F'(0)-SF(0)>O,
F(0)/03 should be a non-decreasing function of
0.
and
It follows that
*m
\
9V3'
and therefore
(v)< -
exp j-
(*
"SIalways in error by
is
<
ii."
9lp |-
T(i)
2*3**'
so that Cauchy's approximation for
/
(v) is
excess.
An
(5)
- sin 2
8)
F'
{&)
0.
The truth
of this
may be
left in
the form
%
{&*
{BF'
{$)
- ZF{fi)\
in
positive
(cf.
8-51).
is
Jv
( vt )
dt
~ J- - T
see
ZVtiV.
Mag.
(6)
260
8*54.
[CHAP. VIII
It has already
been seen
v.
( 8'5)
decreasing functions of
It will
and JJ (v) are now be shewn that both v*Jv {v) and v*Jv '(v)
that the functions
v
(v)
v.
To prove the
dv
first
result
we
observe that
dj^ioi _
Sir J
r
o
|_
e vFi e )de
^r
tJo
f
./
F(d)e-.F W dd
'
o
Jo
oTT
fl
'
(0)
_ 3F(0)}
e~' F
^ dd
>0,
since the integrated part vanishes at each limit
positive.
and
is
is
= r(J)/(2*S*ir)- 0-44731.
noted that
2/8 (8) =0-44691.
may be
J
To prove the second
o7T
(1)
= 0-44005,
result,
find that
dl'
|_
JO
>o,
by
v*Jv'(v)
is
an increasing function of
v.
Hence
(2)
v*
{*i
It is to
be noted that
J
8*55.
'
(1 )
= 0-32515,
4J6
'
(8)
= 0-38854.
A
is
theorem which
slightly
W(V)} +
15
as v-*-<z>
not possible to deduce these monotonic properties from the asymptotic expansions. If, /(*)~ </>(), and if ${) is monotonic, nothing can be inferred concerning monotonic
8-54, 8-55]
261
853,
U,
place of
0,
vKTAv\
Ti/
lv
to be
>/r
in the first
in
we
find that
ir
J o J o
where
n,
(0,
t) =
r (0> </(*-
si,,'
0)
uT
yT '
KY "
>
^^
;
{* (*) - *<*)) -
^#
by 8'51. The function fi x (0, i/r) does not seem to be essentially positive (cf. 8*52) to overcome this difficulty, interchange the parametric variables B and yjr, when it will be found that
^V^-sin'^^-s^
+
^ + VrgJnVrC0B^-2gin'Vr U
^V(^
2
-sin2 ^)
_
'
"
fl
+ flsinflcosl-2sinl
&)/(&- sin* 0)
l5r
J
vr '
vr/;
- sin
0)
and
&F'(0)
term in the sum on the right are both positive or both negative; and, by 851, 853, the second and third terms are both positive. Hence fl! {6, yfr) +- llj (yfr, 6) is positive, and therefore
the factors of the
first
x
'
dv
Jv {v)
262
8*6.
[CHAP. VIII
and
(8'31 8*42) by Debye in connexion with J (x) where v and x are large and positive were subsequently extended* to the case of complex variables. In the following investigation, which is, in some respects, more detailed than Debye's memoir, we shall obtain asymptotic
The
(z)
|
when
where a and # are real and 7 is complex. There is a one-one correspondence between a + ifS and vjz if we suppose that ft is restricted to lie between^ and 7r, while o may have any real value. This restriction prevents z/v from lying between 1 and 1, but this case has already ( 8'4) been investigated.
The
HJV
(z)
TTIJ
e-^ w
>
dw,
H
where f(w) =
(z)
=-
e-'/M
*.'-
dw^-~\ mJ
is
_.+
*/(* dw,
w cosh 7 sinh w.
at 7,
and we
If(w) = //( 7 ).
If
+ iv, this equation may be written in the form (v - fi) cosh a cos & + (u a) sinh a sin - cosh u sin v + cosh a sin /8 = 0.
replace
we
by u
The shape
{(u
(a,
#)
is
2 (u
- a) (v - 0) sinh a cos /3 =
yS),
i if +
creases as
where the arc tan denotes an acute angle, positive or negative Rf(w) inw moves away from 7 on the first branch, while it decreases as w moves away from 7 on the second branch. The increase (or decrease) is steady, and Rf{w) tends to + 00 (or - x ) as w moves off to infinity unless the curve
has a second double-point t.
* Mnnchener Sitzungsberickte, xl. [5], (1910) ; the asymptotic expansions of Iv were stated explicitly by Nicholson, Phil. Mag. (6) xx. (1910), pp. 938943. f That is > say 0<yS-cw. X As will be seen later, this is the exceptional case.
'
(x)
and
A" (x)
8-6, 8-61]
263
If (i) and (ii) denote the whole of the contours of which portions are marked with those numbers in Fig. 19, we shall write
S,v
(z)
f
.
-*/'">
(i)
dw,
TTI
e?fM dw,
(il)
and by analysis identical with that of 8 41 (except that i# is to be replaced by 7), it is found that the asymptotic expansions of Sv w {z) and 8^ (z) are
given by the formulae
( i)
s.{.)~
s' a,
.
m
V ;
2)
W ~ ^_
\
/7
^
TO
s ~
r( ",+ * )
T ()
j)
o
'
(h
A"
tanh 7) m
'
yHW
tanh y))
r(t + s i
am
r (i)
_ ^ y tanh 7)w (
= arg z + arg ( i sinh 7), and the value of arg ( i sinh 7) which lies between \tr and \ir is to be taken.
where
arg ( | i/7ri tanh 7)
(i)
Fig. 19.
The values of
A A
,
1}
A.2}
...
are
1
(A =l,
(3)
n) remains to express v w (z) and (z) in terms of v and to do this an intensive study of the curve on which
It
Sv n)
(z)
and #
( -'
(*)
//()- //(7)
is
necessary.
8'61.
The equation
(1
)
(v
/8) cosh o cos y8 + (u a.) sinh o sin fi cosh u sin v + cosh a sin /8 =
are current Cartesian coordinates and
is
0,
where
(u, v)
<
/S
<
it.
unaltered by
a,
we
which a ^ and since the equation is unaltered and it # are written for v and #, we shall also at first suppose that 0</9^^7T, though many of the results which will be proved when y8 is an acute angle are still true when /9 is an obtuse angle.
when
it
264
[CHAP. VIII
<f>(u, v).
be called
v,
Since
= sinh
.
o sin
p sinn
.
u sin
it
follows that,
when
u,
one value of
u,
and so
the equation in
<f>
(u, v)
= 0,
is infinite
whenever v
is
a multiple
When
<v<
7r,
we have*
,
<f>
<(+ oo ,v) = oo (a, v) = cosh a {(v /3) cos /3 sin v + sin /3} ^ 0,
<f>(- oo
v)
oo
in u,
< (u,
v)
= 0,
a.
is less
is
greater than
/3.
By
it is
(u, 0)
= 0,
0, and
<f>
(u, it)
= 0,
a,
is less
than
to
+ oo
or to
ir
than
or just
greater than
of the curve
the equation
<f>
(u, v)
therefore roughly as
between
and
ir.
\
)
<z
y-2iri
Fig. 20.
^
TTi
When
is
/3,
d(j>(u,
v)/du vanishes at
= a,
minimum
at
value
2 cosh a sin
13
(1
/Q cot y8 a tanh a)
u o. There
are
now two
1
yS cot j8 o tanh a
4> (a,
is (I)
*
Since
r)
has a
= .
8-61]
265
The domains
y=a
ift for
which
is
0^#<7r)
complex
numbered
1, 4,
5 in Fig. 21
in the
for the
viz
TTl
-1>
7b
1
66
5
la
3
2
6a
Fig. 21.
Fig. 22.
(I)
that
When 1 ft cot ft a tanh a is positive, (a, ft) is essentially positive, = ft. The only possibility therefore the curve after crossing the real axis goes off to 00 as shewn by the
<f>
20.
<f>
(II) When 1 ft cot ft a tanh a is negative, the equation ( a, v) = and ft 27r, for has no real root between cos v). o<f>( a, v)/dv = cosh a (cos Therefore has a ( a, v) single maximum at #, and its value there is so ( negative, that and $ 27r. a, v) is negative when v lies between
<f>
<f>
Also
(f>
(u,
/S
2tt)
<j>
has a
maximum
at u
there
is
negative,
(u, v)
= /S 2<rr
as
in,
Next we have
line v
what happens
= + ir.
<f>
Since
the line
<f>
(a, v)
positive
when
^ ft,
u~a;
(u,
also
266
[CHAP. VIII
which go
off to
and
when u >
lie
a,
infinity
as shewn
When
i.e.
when
= 7r
passes off to
any of the domains numbered 1, 2 and 3 in Fig. 21, it is = 2v ft, and so the curve after crossing ac + iri as shewn in Fig. 23 by a broken curve.
Fig. 23.
We now
numbered 6a
in the
domain
In such circumstances
and
u
^ ( a,
v)
has a
maximum
negative.
The
= 2tt ft, the value of ( a, 2ir ft) being = tt, consequently remains on the right of
=a
Now
( a,
(/3,2rr-0),
let
+ ft,
4tt
(4>Tr
+ ft,
6ir
- ft),
. .
the
first
becomes positive be
(2Mtt
Then
positive,
<f>
(u,
2Mir +
2ir
ft)
has a
and so the curve cannot cross the line left, and consequently goes
-oo +(2M+l)iri;
it
this, for
meet a horizontal
points.
8-61]
267
When
-oo +(2M+l)Tri,
where
is
tanh o +
{(M
1) tr
/9}
cot
is positive.
and $ (z), and thence we can express these integrals in terms of (z) and (z) when (o, fi) lies in the domains numbered 1, 2 and 6a in v v Fig. 21 and by suitable reflexions we obtain their values for the rest of the complete strip in which < yS < ir. The reader should observe that, so far as the domain 1 is concerned, it does not matter whether /S is acute or obtuse.
{1
>
(i)
If
is
l-o tanh a +
is
{(M
1)
ir
} cot
fi
positive
when
cot
is positive,
and
if
is
positive
when cot#
is
Sv
(1)
(z)
and Sv
(z)
are as follows
Regions
End-points
S0>( 2 )
1,3,4
2,6a
5,
00
oo
iri,
,
76
- oc
HW(z)
2J
(z)
oo
66 7a
oo
-oo,
2Niri, oo +- iri
cc
+ (2M+l)iri
e-Mvniffv V)(ze-Mni)
Regions
End-points
SyW (z)
1,2,5
3,
oo + iri, oo + ir,
oo
oo
#<*>(*)
iri
7a
oo
oc
2J.(z)
Zevwi
4,66
+ 2ri,
+
(2J/
J_ v ( 2 )
( ze
6a
76
-ao
+ l)ri,
oo
oc
ao + rt,
2Niri
e-Wvwi ffJP)
Nwi)
From
when v and z are both arbicomplex numbers, the real part of z being positive. The range of validity of the expansions can be extended to a somewhat wider range of values of arg z by means of the device used in 8*42.
solutions of Bessel's equation can be constructed
trarily large
268
The reader
pass from
[CHAP. VIII
oo
1
to oc
+ iri
and from - oc
prove that, in the critical case /3=r, the contours + ni to oo , so that the expansions appropriate to
the region
are valid.
differences
Note.
The
8*7.
Jn (nz).
extension of Carlini's formula (811, 8'5) to Bessel coefficients in which the argument is complex has been effected by Kapteyn* who has shewn that, when z has any value, real or complex, for which z- 1 is not
An
(1)
\Jn(n*)\*
exp{w\/(l--g2 )}
{l
+ V(l-* )} n
a
2 This formula is less precise than Carlini's formula because the factor (2jrn)* (1 -s )* does not appear in the denominator on the right, but nevertheless the inequality is sufficiently powerful for the purposes for which it is required %.
To obtain the
Jn {nz) = ^.
in
r
J
exp { \nz
w
,
(t
1/t)} dt,
is
circle of radius e
where u
is
a positive number to
be chosen subsequently.
If
we
write
eu+i8 we
,
get
[n [\z
Jn (nz) Now,
if
f exp
|
id}] dO.
- e- u e~
i0
-u- id]
on the contour,
it is
clear that
\jn (n*)\*M:
But
if
- pe ia
where p
is
positive
and a
u e-*)
is real,
\z (eu eu>
is
\p
this attains its
w {e cos (a
and and
*
maximum
value
tan 6
its
coth u tan a,
+
sin2 a)
value
is
then
s p V(sinh u
- u.
Ann. Sci. de VEcole norm sup. (3) x. (1893), pp. 91120. when z approaches the real axis it follows that the t Since both sides of (1) are continuous sign may be given to the inequality is still true when * - 1 is positive: for such values of z, either radicals according to the way in which z approaches the cuts.
X See Chapter xvii.
' '
8-7J
269
Hence,
Jn (npe
im
v/(sinh 2
+ sin8 a) nit].
may be
by
as small as
this
We
now choose u
method.
The expression
2 p V(sinh u
sin3 o)
is
has a
of u,
when u
the equation*
sinh u cosh u
\/(sinh 2
-f
sin* o)
With
this choice of u it
may be proved
s
"8
that
2 V(l
and,
sinh u cosh u
(cosh 2
g2la ),
must be taken
in
by taking z
to be real,
it is
the ambiguity.
Hence
2 {1
V(l
%
!
u=
e2"
.
|
*,
and
so
z
exp
\/(l
log
T+
-z) _
2 V(sinh a u
+
I
sin2 a)
e
3*
exp
I
\/(l
-z)
a
- e**
= 5V(l-z')-
sinh 2 m
sin2 a
sinh u cosh
w
sin 2 a)
= p \/(sinh 2 u +
and
it is
u,
j
j
now
clear that
Jn (nz) $
I
An
as both
is
that
Jn (nz)
1 so
long
'.
and
I
s exp
-v/(l
z')
is satisfied,
To
write as
before z
= pth,
_
a)
"~
V(sinh 2 w
+ sin
The previous
when
z*)
2
1'
then
*
2 p \/(sinh a
+ sin
a)
u = 0.
This equation
is
270
It follows that
[OHAP. VIII
P*
sin*
u).
As u
increases from
to
T1997
...,
to 1
and p dea
It
* ex PV(l-^2)
|
+ V(i-* )
$1
origin.
This curve
Fig. 24.
The domain
it
in
which
J*n (nz)
is
shewn
in Fig. 24
will
When the order of the Bessel function is positive but not restricted to be an integer we take the contour of integration to be a circle of radius e* terminated by two rays inclined + it arc tan (coth u tan a) to the real axis. If we take 1 = ev on these rays, we get
1 1
, ,, T \W)\<X'+\\) % **l-*P
,
\amvir\
/""
\/(sinh^ + sin^)
, - vV d
"1
j^/l +
[
s
|
I
2!L!^|
w
\\
I
r
J u
e xv{-v(v-u)}dv\
)
and so
|Jr(v*)|< \l
+
is
l/it
This value
CHAPTER
IX
The
definition
of Neumanns polynomial
On (t).
is the discussion of certain polynomials which occur in various types of investigations connected with Bessel functions.
The
The
first
Neumann's *
investigation of the problem of expanding an arbitrary analytic function f(z) into a series of the form 2an n (z). The function n (t), which is now usually
called
Neumanns
l/(t
polynomial,
is
en
Jn (z)
in the
expansion of
z)
W
From
(<)
* <*)+
Jn (z)On (t).
shall derive an explicit expression for the function, and it will then appear that the expansion (1) is valid whenever \z\< \t\. In order to obtain this expression, assume that z < t and, after expanding l/(t - z) in ascending powers of z, substitute Schlbmilch's series of Bessel coefficients ( 27) for each power of z.
j \ j j
we
- =- + t-z
1
2
s=1
<
** +1
7*
1
fr
s
m=0
(,\a.
v
*=1 &
v
=
(s
+ 2m).(s + m-l)l
nl
-
(,i
Assuming
*
for the
Math,
series.
314.
Neumann's procedure,
15, 33 see also Journal fur assuming the expansion (1), is to be given subsequently ( 9 12) and to solve it in
;
after
t In anticipation of 1611, we observe that the expansion of an arbitrary function by substituting for l/(t - z) in the formula
. ,= f(z)r
is
obtained
/(*+)/()"
(
si/
1st.
Lomlardo,
(2)
xv. (1882!,
224225.
272
[CHAP. IX
series
we
is
effect
we thus get
(<i(-D 2"- 2"1" 1 * n^ , n_ m+1
V
i r- = 7 c
*
n/(')+
S
n=l
n.(n-m
m=<\
=0
m!
1)1)
J
/.(*)
r/x
by the equations
n u\
o,()
^"i,V ml(i^
-iyt
^"n.in-m-l)!
-
{n>l)
(3)
It is easy to see that
(4)
en
O(0*-^iT
|i
2(2n
2)
+ 2i4t(2n-2)(2M _ + ...|, 4)
and the
We
l/(*
have now to consider the permissibility of rearranging the repeated series for
- 1).
A sufficient condition is
.i
t
i
*/
.
U-o
. \ , +2> \'/ If
should be convergent.
(4),
To prove
that this
is
we observe
that,
by
2' 11
we have
m=0
'MI
m=0
(*+
1)!
+ <3(J|*|) -{exp(i|*P)}/(2m)!
<(il*l)*e*P(il*l
)-
Hence
s
_J1_
(+2m).( + m-l)l
,
,,
||
_,.,_,*
||eipQi|)
'
||(|*|-|*|)
The absolute convergence of the repeated series is therefore established under the hypothesis that \z\ < \t\. And so the expansion (1) is valid when \z\< \t\, and the coefficients of the Bessel functions in the expansion are defined by (2) and (3).
It is also easy to establish the uniformity of the
1
1
^ R,
9*1]
ASSOCIATED POLYNOMIALS
these inequalities are satisfied, the
x
273
When
sum
( /)*
^_
(8
+ 2m).( + m-l)!
"!
exp
.lo^^lio
*"
(*
(fr*) \
exp
(frr )
+ 2m)!
z
Jand
t,
fl-r
Since the expression on the right is independent of convergence follows from the test of Weierstrass.
was called by Neumann a Bessel function of the second now used (cf. 353, 3*54) to describe a certain solution of Bessel's equation, and so it has become obsolete as a description of Neumann's function. The function n (t) is a polynomial of degree n + 1 in 1/t, and it is. usually called Neumann's polynomial of order n.
The function
n ()
is
\n m
/k\ (5)
or \{n
\) m
/*\
for
in (2), according as
* *4
n.fjw
a
+ ma
s
1)1
Q'O-ijSt dn-m)!^^ w n* n ( -2 1 _ J.
r=
1
"
*~
)
i
(n2
-
-4-
-2 )(w
2
-4)
'-
+ ...
(n odd)
(6)
n (t)
j JE o (1
__ 1Jiara=s
3
'
nnW (t\-"4 mv t
(5), (6)
w T (j n
+ j m) r <>8
j ( t w ) w
'
ran-|m + l).(|Om+1
it is
The equations
By
(8)
(9)
lO,(0k*.(n!).(*|*!> ^xpCil*!^
enOn (t) = l.(n\).(lt)--*(l
|
+ 0),
(n>l)
where
<;
[exp (
*
|
- l]/(2 - 2).
Sa n Jn (z)
;
From
these formulae
series
it
n (t) is convergent
when z is outside the circle of convergence of the latter series. a n Jn (z) n (t) does not tend to zero as n -- oo and so the former series does not converge. Again, it is easy to prove that, as n -*- oo
absolutely convergent
whenever the
lan (zjt)n
and,
Jn <*)
(t)
^
(2n
{l
<-)}
By analogy
Qn (t), which
is
such that
t-z
r.
-=, n=o
+ l) Pn {z)Qn (t).
Cf.
Modern Analysis,
% 15*4.
274
[CHAP. IX
and hence it may be shewn* that the points on the circle of convergence at which either series converges^ are identical with the points on the circle at which the other series is convergent. It may also be proved that, if either
series is uniformly convergent in
t,
so also
is
the other
series.
is
when
J
<
1
1
\,
it
follows
by
differentiation! that
(10)
(-)g.(p r -^ y
(t-zf + * +1
+ g)!: = T y_ 2
w=0
dV
'
may be
o.(0i/.
o
8 5
,
192/*
1920/**.
The
Otti,
coefficients in the
polynomial
On (t),
for
=0,
1, 2, ... 15,
4, 5.
9*11.
satisfied by
On (t).
l)
n (t)
,
We
(1)
(2)
(3)
shall
(n
- 1)
+ (n + 1)
1
W _,(0
- 2 (n \~
2n --
n?r
,
(>1)
(n^l)
(<),
The first of these was stated by Schlafli, Math. Ann. in. (1871), p. 137, and proved by Gegenbauer, Wiener Sitzungsberichte, lxv. (2), (1872), pp. 33 35, but the other two were proved some years earlier by Neumann, Theorie der BesseFschen Functionen (Leipzig, 1867),
p. 21.
them, but give in their place an investigation by which the recurrence formulae are derived in a natural manner from the corresponding formulae for
Bessel coefficients.
Taking
<
1,
observe that, by
n
91
(1)
and
2-22 (7),
(t-z)l
7i=0
Jn (z)
n (t)
=1= 2
n=0
if 2fr n is
2 M COS \niT
Jn (z),
is
theorems that,
convergent, so also
2&/n,
Sbjri*
is
absolutely convergeut.
\i) in.
(1881
2), p. 160.
Modem
Analysis, 5-33.
9'11]
ASSOCIATED POLYNOMIALS
275
and hence
z
I
n=0
en
Jn (z)
n (0
=I
=0
en
=X
since tO (t)=l.
If
now we
Jn (z)
to
modify
we get
{Jn- X
(z)
2
If
=0
en
Jn (z)
n (t)
=2
n=l
+ Jn+1 (z)}
[tOn (t)
- cos
mr}/n.
it
we
notice that
is clear
on rearrangement that
(t)
Jo (z) {0
- tO, (t)} + J
I
- \t0, (t) + }
Now
all
regard
zasa
variable, while
left
remains constant if the coefficients of do not vanish, the first term which does
;
not vanish can be made to exceed the sum of all the others in absolute value, by taking \z\ sufficiently small. Hence all the coefficients vanish identically* and, from this result, formula (1) is obvious.
To prove
(2)
and
/d
\dt
+ dz) t-z~
d\ _i
'
and
so, \z\
being
less
than
j
t
j
we have
n' (t)
2
By
Jn (Z)
+2
we
en
Jn
'
(z)
(t)
= 0.
left
find that
-J
(z)
(t)
(jr)}
On (t)
= - J (,) 0, (*) - 2
that
is
to say,
J. CO {Oo' (0
'
(t)
n+l
(0 "
n _, (0}
= 0.
On
*
Jn (z)
on the
left,
just as in the
proof of
This
we
(3).
if
then
a convergent power series vanishes identically, 3-73). Ttie argument is valid here because the various series of Bessel coefficients converge uniformly throughout a domain containing z = 0.
all its coefficients
vanish
(cf.
Modern Analysis,
276
[CHAP. IX
and
(2)
1)
1
)
we
(4) (5)
0 it) = (n - l)tOn
n (t)
(
(t)
'
+ n sin
\mr,
= - n + 1 ) tOn
(t)
+ n sin 2 \mr.
If
(6)
(7)
be written
(n
for
{d/dt), these
formulae become
2
- 1) (^ + n + 1) n (0 = n \tOn-* (t) - sin %rnr}, (n + 1) (^ - n + 1 On (t) = - n {tO n+1 (t) - sin |wtt}
)
0_(<)
this
n.
= (-)n On (0.
(1)
With
values of
(7)
n
all
integral
9'12.
The
(t).
From the
(*
(7), it is clear
that
+ n + 1) (* - n + 1)
=-
Tit
-j
(^
+ n + 2)
2
n+ , (0
+ n sin \mr
8 2
\n-jr)
+ n sin
\nir
l)2
n (t)
(<
sin2 \nir.
is
y = On (0 +
(1)
^(0,
which
is
sometimes convenient
form
where lere
(w even)
( odd)
*
t
(1871), p. 138.
ASSOCIATED POLYNQMIALS
differential equation is to observe that
277
f
and so
z?
+z
enw2,7n {z)
* (0
'
t^z t-z
(t-zf
(t-zf
00
00
Now
and hence
Therefore
t^J^ (z), z= 2
00
fin +
(2 + 1)
J2n +
l (z),
+ z = t2
n=0
2 en,gn
(t)
Jn (z).
+ 1* - *} On (t) - V ()] ^ 0.
equating to zero the coefficient of Jn (z) on the left-hand side of this identity, just we obtain at once the differential equation satisfied by On (t).
9*13.
It has
Neumann's contour
been shewn by
O n (z).
Neumann*
Jc
(1)
Om (z) On (z) dz = 0,
J m (z) On (z) dz =
0,
(m = n and
mn)
(2)
(ma n>)
,
(3)
f Jn (Z) On (Z) dz c
= 27TlA;/n
where k
is
number
number
of negative circuits.
The
larity of
first
(z)
is zero.
The
grand
is
manner
V m Jm (Z) = 0,
Jm (z)
respectively,
./
U (z)
be written in place of
W iifO|<f)i-, W ^),
BesseVschen Functionen (Leipzig, 1867), p. 19.
*U' (z) + zll{z) + (m2 - w2 ) zJm (z) On (z) = z*g n (z) Jm (z),
* Theorie der
278
[CHAP. IX
and hence
[z
Jo
Jo
z*gn (z)
Jm (z) dz.
The integrated part vanishes because U(z) is one-valued, and the integral on the right vanishes because the integrand is analytic for all values of z and hence we deduce (2) when 2 n 2
;
Two
(4)
corollaries,
1
due to
/(o /> +
/
)
Schlafli,
Math. Ann.
m.
J* {x+y)
)
/"<-* +
/
(5)
-
first is
The
obtained by applying
and
00
(3) to
(1),
namely
Jn {*+y)=
2 /, +p (i)/.j(y), p 00
variable.
9*14.
It
Neumanns
integral for
On (z).
(i)
o, (.)
shall
J"
+**+
= /;
^i+i:
2
^
;
+ *
^du.
We
(2)
equivalent formula
On (z) =
is
U Jo
/*>
exp ia
[{t
+ V(l +
|
)}
where o
\ir
on writing
u/z,
the truth
A
(3)
is
0 (z) = \\
(2)
w {e *
Jo
To prove
we observe that
/"ooexpia
racexpia
O
and
so,
(s)
=
Jo
e-*dt,
0,(z)=
Jo
te-*dt;
911
<f>
(2), it follows
that
we may write
exp ia
gt n (t)e~ dt,
0. <*)-["
Jo
where
(4)
<f> +1
(0 -
2t<j> n (t)
- <_, (0 = 0, =
*.
and
(5)
*
<M0=1>
*!<)
Theorie der BtsseVtchen Functionen (Leipzig, 1867), p. 16; Journal filr Math, lxvii. (1867),
p. 312.
9*14]
ASSOCIATED POLYNOMIALS
solution of the difference equation (4)
<f>
279
The
is
n (t)
where
and
The
A B=\
established.
who wrote
<p n
(D)
(1/z)
where we
(0
*"*** dt>
9*1 (1),
*=o
When
<
s
j
we have
= IT\ 2 9
2" '">
.
oo (n=-<* V.n=
1
Jn (0\ e-du,
j
if
p be so chosen that
u
z
It follows that
is justifiable; it
(such that
|
<
|),
fu + J(u i 4-z )\ n
2
n=N+\
can be made arbitrarily
3^-all
by taking
N sufficiently large J
now
\uJ{u*+z*)\^2(u + \z\),
andso
Jn{0
r\^^p^l\
J
2
|
e -u
du
^J^ fV(+|,|)
z
I \
-*
1*1"
J LI
<|(C/*)"|exp{|| + iim.
*
Math. Ann.
xvi. (1880), p. 7.
280
Therefore, since
|
[CHAP. IX
<
I I
|,
we have
|
r-
.-l+i J.
{uJW+f^jn() (z)
*=
left
If
*
*P
{!!
+*
<f}
'
\*\{\'\-\i\rby taking
can be
made
arbitrarily small
N sufficiently large
Hence, when
<
j,
we have
/(f)
.'0
= 2
w = - ee
{w+^+f!)} e~"rfu _
*
y)
tti
'
by the equation
\*
0 (,) = j"
and
it is
1.
+
n
so defined,
is
n+
When
term,
it is
the integrand
is
expanded*
in powers of z
9*15.
Sonine's investigation of
Neumann's
integral.
An
of expansion of l/(a
z)
is
Neumann's formula
difficulty.
91 4 can be
(w)
derived without
as follows
Let
so that
yjr
<fr
w;
and, if
yfr
= x,
let
w = ^ ()>
Let
Zn
and
A n be defined
1
by the equations%
z =
Then
it
f<+>
ti)
** m
CL
k> ^=/o
= Zn A n
n =-a>
din
*-"*{*(*)}"<**
Suppose that
curve
for
C surrounding
-fa
on a closed
on a closed
* Cf.
t Mathematical Collection (Moscow), v (1870), pp. 323 modified slightly, but the symbols ^ and ^ are his.
382.
J This
Wist.
ii.
is
781784.
9-15, 9*16]
ASSOCIATED POLYNOMIALS
z.
281
curve
Then
+
2iriJ
1
1
oo
/*oo
y%n
gg(w)-
dwdx
\Jc
J
-" x
Jc
w-^{x)
gZ^(W>) a*
2iriJ
=
provided that
Jo
e (z
;
dx =
l/(o
z),
B(z)<R (a)
and the
result
is
established
if it is
assumed
(w)
\ (w
l/w),
vf (x)
= x V( +
2
1)
and then
n= oo
= i KJ(*){4 n + (-)A_n
=
}.
Since
] <*#,
we
at once obtain
Sonine notes
{jp.
Neumann's
integral.
328) that
Jn (z)~(hj) n ln\, *nOn (a)~n\($a)-*, and in the later part of his a\ of the expansion so that lfta-z) converges when J z memoir he giyes further applications of his general expansion.
\
<
9*16.
(z).
The
(-^n^On (*),
which
is
On (z),
the
does not converge for any value of t except zero. series after the method of Borel, in the following manner
"summed"
n.(n + m-\)\t2n
_
_1
(n
+ $).(n+m)l
2n +
nlomlo
1
"
t*
n Ion=m
_1
(2w)
2
~ 2 JE
(2m+l)l
2"* + 2
<* + 1
(1
(l+<^
2m + 2
)
"(**)*" +
(1
-'
2
)
2m+1
'
w=
(i*)
- <a
l+<
*
(-).m! tm
*(l-<*)m=o{*( 1 -< , )} m
Nieuw Archief voor Wiskunde
pp. 49
55.
10
282
[CHAP. IX
^
e~"du
tot
>
an d this integral
is
convergent so long as
(1
1'*)
z\t is
not negative.
There
is
no great
2
n=o
t
n(
nt
On (z)
|
is
an asym-
may be
(z).
of the theory of
this result.
9 '17.
It
is
(m).
Neumann's
by
n (nz)
by
J n (nz)
obtained
in 8-7.
We
have
1 n < n *) = 2i^+i
r*
J
z*)} ]
e~nw dw,
^
I
~
i
["
|
[{,
y/(vfi
+ *)} e-] | dw
!
,
j
J o
is
Now
is
[w
+ V(w + z )} e~ w
2
V(l
(!)
-z ),
and
so
On (nz) ^
|
JL_
|
Lt^^p
is
Jl {"
+ VK + z>)} e-\.\dw\,
is
greatest at the
is
is
and both
w=
and
w = + oo
(2)
Vd-*
Hence, since a contour joining the origin to infinity can be drawn when (2) is and since the integral involved in (1) is convergent with this contour, it follows that, throughout the domain in which (2) is satisfied, the inequality
satisfied,
(3)
is satisfied for
O n {nz)<.~
z
i
V(l
j"" 1
and
this is
9-17, 9-2]
9'2.
ASSOCIATED POLYNOMIALS
283
If
z
we expand
z)
in
by the expansion
we
find
on
rearrangement that
zv
* z v+s
2 "+*
(v
+ s + 2m).T (v + s + m )
(i/
"
J iio
no greater theoretical
* -
+ n)
1"
(v
+ n - m)
A
;
~~^.
Jv+n {z)
\
2m, and
it
presents
We are
A ntV {t),
defined by
the equation
whenever
and when
z\<
(2)
\t\,
we have
t
. =2 A Z =0
M
n>v
{t)Jv + n
{z).
difficulty in
+ n-VAn+^M + iv + n + VA^.it)--*
2" (v
1
[
J t
An
(t)
+ n)
\(v
+ nY -1}
t\\\n +
1"
(v
+ \n- \)
Sinz *W7T,
\)
(5)
(i/
+ n) tA n_ 1<v (t) - (n +
+ n - 1) 4,,,, (0 2"(v + n)(v + n-l)r(v + ln-$) , ,, ., I^ 8in2l = ( y + w _l)^' i". n> () + rfln + i)
1) (v
v
(6)
(1/
+ w) ^+1 , (<) - ( + n +
1) (2
+nn)(i
1)
A n>v (t)
,
4^)2T(+l)/t
Wiener Sitznngsberichte, lxxiv.
(2),
130.
284
[CHAP. IX
is
which
A n>v (t)
is
a solution
)
(R\
d*y
J>-2vdy
where
(9)
^(
.-A
r(J
L__.l
WW
+^ _I ^+n (0also
The
is -4 ,,()
Of these
proved that
(10)
_.
J
where
Cn v (z)
is
is
this
formula
easily proved
origin.
Neumann's polynomial
izt
1
I
/"< 0+)
Lttx J
(t) e
dt
= in cos
n arc cos
z)
The
(12)
following formulae
may
also
be mentioned
v
A m v (z)A n
c
'
(z)dz
= 0,
(m = nandm^n)
(13)
(m2 ^fen2 )
(14)
I
.'
Jv+n (z) A n
n = 0,
(z)
dz =
2irik,
c
1,2,
. . . ,
where
G is any closed
first
contour,
and k
is
number
number
of negative circuits of
The
the second
is
= 0,
V+n
{*-
An
(z)}
whence we
find that
(m - n)
(2v
m + n)
Jo
z~ v
~v
0.
9*3.
Schlafli's
polynomial
O n (t) was
investigated by Schlafli.
properties,
In view of the greater simplicity of some of its frequently convenient to use it rather than Neumann's poly-
nomial.
9'3]
ASSOCIATED POLYNOMIALS
polynomial
is
1 \
285
(1) (2)
Sn (t) = 2
S.()-0.
<n {n (n
_m_
A)
(iO-"
+MW
,
(n
>
1)
On comparing
(3)
we
and
(2),
we
(4)
+!
latter,
x
(t)
+ _! (0 - 2n~ Sn (t) = U~
1
- 1) #n _ (0 - 1 (n +
this
1) +,
(0 = nSn
'
(t)
cos2 fnw.
we multiply
(5)
result to (4),
we get
Sn^{t)-Sn+1 (0 =
(4)
25/(0.
may, of course, be proved by elementary algebra by using the definition of Sn (t), without appealing to the properties of
(5)
The formulae
and
Neumann's polynomial.
The
(6)
and, with this definition, (4) and (5) are true for
integral values of n.
The
(7)
is easily
by
Schlafli,
Other forms of the recurrence formulae which may be derived from (4)
and
(5) are
(8) (9)
If
we write ^
for
(d/dt), these
formulae become
2 cos^wtt,
(10) (11)
it follows that
(S
+ n) Sn (0 = tSn ^ (0 -
(^
and so
(12)
Sn (t)
is
du - + t-~ + (< - n
d?v
8
)y =
2 sin 2 /wr
+ 2n cos
\rnr.
Math. Ann.
m.
(1871), p. 138.
286
It
[CHAP. IX
:
(t)
= 2/t,
16/t
3
,
&(*)-4/*,
S3 (t) = 2/t +
5
S
5
,
(t)
1
;
series,
) '
'
t-
2 (n2
l2)
2 (wf -
(w 2
-3)
2
polynomial (<)> for n = l, 2, ... 12, have been calculated by Otti, 14 Otti's formulae are reproduced (with some obvious errors) by Graf and Gubler, Einleitung in die Theorie der Bessel'schen Funktionen, II. (Bern,
The
coefficients in the
1900), p. 24.
9*31.
Formulae connecting
the
Schlafii.
have already encountered two formulae connecting the polynomials of Neumann and Schlafii, namely
We
\nSn (t) = t0n (t) - cos \ntr, 8n- (t) + S (t) = 40n(f),
2
l l
of which the former is an immediate consequence of the definitions of the functions, and the latter follows from the recurrence formulae. A number of other formulae connecting the two functions are due to Crelier * they are easily derivable from the formulae already obtained, and we shall now discuss
;
When we
find that
(1)
93 (3) and
either
93 (8)
or (9),
we
(2)
Next, on
(3)
summing
(5),
we
find that
Sn (t) = -2
<(i-D
2
w=
(t)
S'n^mr-i(t)
+ Bin*lnir.S
(t),
and hence
(4)
*
Sn (t) + Sn _
pp.
6196.
9-31, 9*32]
ASSOCIATED POLYNOMIALS
287
Again from
4
93 (7) and (5) we have {<?_, (0 + O n+J ()} = n _ 2 (t) + 2Sn (t) + S n+2 (t) = (Sn_ 2 (*) - Sn ()} - {Sn (t) - Sn+2 (t)\ 4Sn (0 = 2S' n_ (t)-2S\l+1 (t) + 4 Sn (t) 4Sn" (0 + 4^(0,
1
!
so that
(5)
(t)
+ O n+1 (t).
911
(2),
This
Again, on
(6)
summing formulae
of the type of
we
2
find that
(t),
On (t) = -2
I
5/1=0
0'n -2m-! 0)
and hence
(7)
n (t)
0^
(t)
= -2 "%
m=0
(t)
0,
(t)
{t).
9*32.
The
(1)
peculiar
Sn (z) = Tr
m= n
Graf and most readily proved by induction it is obviously true when n = 0, and also, by 363(12), when n=l. If now the sum on the right be denoted temporarily by n (z), it is clear that
This formula
is
;
<f>
<n+i (Z)
<n-i (Z)
~ (%n/z)
2
<f>
n {Z)
m+1
m= n-1
Ym (Z) - 7T
n+1
Fn+j (z)
2
m=n-l
Jm (z)
Jm (z)
m=-n
(z)
2 m= n+l
Ym (z)-7rYn_
(z)
m= +l
-(2n7rjz)Jn (z)
2
m=n
Now
modify the summations on the right by suppressing or inserting terms at the beginning and end so that all the summations run from n ton; and
we then
vanish.
<f>
Ym (z)
n+1 (z)
+ <_, (z) - (2n/z) n (z) = 7rJn+1 (Z) Yn+1 (Z) + Y_ n ^ (Z)} - 7T Fn+1 (z) {Jn+1 (z) + J_n_, ( Z)\ - -rrJ^ (z) n (z) + F_M (Z)} + 7T n_ (z) {Jn (z) + J_ n (z)\ _ - {l + (_ l){ \Jn _, ( Z ) Yn (z) - Jn (z) Fn _ (,)}
<f>
{
= 4.2 -1 eos
^wr,
is satisfied
by 363(12); and so n (z) satisfies the recurrence formula which by Sn (z), and the induction that n {z) = Sn (z) is evident.
<f>
<f>
(1893), p. 138.
41.
288
9*33.
[CHAP. IX
Sn (z).
If
exp
ia
On (e)
=h
.'
[{t
V'(l
+ t*)} n +
\t
and integrate by
parts,
we
find that
eft
_ cos'^wtt + " 5
Hence
it
00
_^ 2zJ
i"
\t
+ \/(l
4-
P)}
V( 1
+ * )} w
2
_rf
V(l+**)
follows that
Schlafli,
p. 146, in
the form
fundamental in
which we
shall
now
give an outline.
We
write temporarily
Tn =
and then
so that
[t
+ V(l + *2)} n -
\t
- V(l + ?)}'\
y"+! _ ot 4T
and therefore
7'
C
IT
*
'
+ 2 + 2*+...+2e'
the continued fraction having w elements. It follows that quotient of two simple continuants^ so that
Tn ]Tn
x
is
the
Tn+ l= K(2t,2t,...,2t)n
Tn
the suffixes
n,
K{2t,2t,...,2t) n
^
and since
It follows thatj
we have
Tn = 2</(l + t*).K(2t)n_
* Comptes Rendus, cxxv. (1897), pp.
;
1 ,
% Since
all
the elements of
421423, 860863 Bern Mittheilungen, 1897, pp. 6196. 494502. tbe continuant are the same, the continuant may be expressed by
9*33, 9*34]
ASSOCIATED POLYNOMIALS
fee
289
and hence
exp ia
(3)
Sn (z) = 2
Jo
K(2t)n _ e~^dt,
1
From
9'34.
Schlafli's expansion
of Sn (t
+ z)
as a series of Bessel
coefficients.
We
shall
now
when
<
1
1
Sn (t 4- z)
(1)
Sn (t + z)=
m= 00
I Sn _n (t)Jm (z).
is
The simplest method of establishing this formula for positive values of n by induction f. It is evidently true when n = 0, for then both sides vanish
1,
when n =
is
equal to
S, (0 Jo (*)
+ X
{Si-m (t)
Jm (*) + S^ (0 J_ m (z)}
2 {Sm^W + S^ityJ^-z)
*=1
= 2Oo (t)J
=
2
{-z) +
I e m Om (t)Jm (-z)
=
by 91(1) and
2/(*
+ z) =
S,
(t
+ z),
93 (7).
truth of (1) for Schlafli's polynomials of orders
Now,
if
0, 1, 2, ... n,
Sn+i
+ z) = S^ (t + z)- 2Sn
'
(t
+ z)
2
S'n _,n (t)Jm (z)
m=
X
2
Sn-r-i(t)Jm
(z)-2
wi= 00
30
m= *>
and the induction
is
established
result that
S'(t
*
+ z) = ^SH (t + z).
;
141
is
left to
9-1).
The extension
on the proof
by
9*3 (6).
290
[CHAP. IX
The expansion was obtained by Schlafli by expanding every term on the right of (1) in ascending powers of z and descending powers of t. The investigation given here is clue to Sonine, Math. Ann. xvi. (1880), p. 7 ; Sonine's investigation was concerned with a more general class of functions than Schlafli's polynomial, known as hemi-cylindrical functions
< 10-8).
When we make
(2)
use of equation
n (t
when
<
1
1
\,
+ z)=
= 00
(1872),
n (t+z) in ascending
powers of
$ 9-11 (2)]
(3)
2P
^r-)==
(t),
is
when \z\<
\t\),
(4)
Sn (t-z) = On (t-z)=
I Sn+m^Jmiz),
(5)
Wl =
i O n+ m (t)Jm (z).
oo
9*4.
The
definition of
Neumanns
'polynomial ft n (t).
The problem
formulae of
Neumann f by
the
z)
given in 91, is the expansion of l/(F z2 ); and the function ft n () will be defined as the coefficient of n Jnz (z) in the expansion of l/(tfa z*), so that
C1)
ir-r, t*-z
= # (*) Oo (0 +
n=0
<*) fli
(0
W
tft
<*)
n. (0
= 2 c/.'Wfl.^
To obtain an
panding
*
ke
<
*
I
aQ d, after ex-
l/(s
z})
in ascending powers of
141
z,
Math. Ann.
142;
[Jahrbuch
Math. 18931894,
t Leipziger Berichte, xxi. (1869), pp. 221256. [Math. Ann. in. (1871), pp. 581610.]
9-4]
ASSOCIATED POLYNOMIALS
given by
291
( 2-72).
a 2
Neumann
As
in
we have
z 2*
-s ~,=oM+a
series
by writing
ns
for
this
rearrangement
ment
in 9*1.
is
V>
(3)
Un
W ~ 4 ,r
(n
- *)! (2s)! (i
we
M+a
'
(n>l)
n i (o-i/'.
the order of the terms in (2)
find that
On reversing
<*)
nW ~4 mto
"nW-i+oir+o ^3.4 2
**
-
(n>l)
'
14na
1.2 4n'(4n-2')
/L
(5)
/6
Also
A*
(6)
@44
2.(2w-l)(2n-2)
Hen*
2"
+
)
'
+ 1.2...n.(2n-l)(2n-2)...nf
where
.= ~
2 n-l
2/i
@ = (2n-l)(2n-3) 2n(2n-2)
4
'
2n(2n-2)(2n-4)
0 =
Since
(8)
(2m-1)(2w-3)...1 2n(2w-2)...2
211
),
< Bm <
1, it is
that
n it)
<*
2*
j
- 2- 2 (n 2 exp !) (
2 2
2
|
and,
when n > 0,
en
(9)
nw (0 2W <- "|
(n!)a (l
+ 0),
where
0\
S {exp|*| a
- l}/(2n- 1).
292
[CHAP. IX
easy to shew
difficulty in verifying
Kapteyn*,
ao)
-(')- s
5/r^(a^)Qn
(t).
9*41.
911
(2)
and
(3) are
n'(t) lln{t)
= n -^
**<*>
n-1
n+
2n 'Q n-l'
(2)
(2/<)rV(o
(2/0
= *n (0-**MO,
(3)
'
(t)
= - 2n, (t) + 2n
(t).
9*11 (1).
Neumann f
is
9*4 (1),
and
that,
by Hansen's expansion of
2 Jo ()
J' (z)
2*5,
We find by
t,
to
z,
that
-2t/(t*-z>Y= 5
n=0
a/(ff
Jn*(*)fV(0>
j; to
flo
- *y = 2 j
(#>
(0
J'n., (*)
=* I
n=l
{/*.,
to -/Vito){n(*)-n. (*)}/
it is
On comparing
*-
these results,
clear that
2
en
+ ito}.{(0-n(o}/n=a
left
On
Jn
(z)
on the
and equating
it
to zero
(cf.
91),
*
we
Ann.
m.
(1871), p. 606.]
9-41, 9*5]
9*5.
ASSOCIATED POLYNOMIALS
293
Q n (t).
in ascending powers of z and replace each power If we expand z of z by its expansion as a series of products of Bessel functions given in 55, we find on rearrangement (by replacing s by n 2m) that
gfl
+v
* gH + v + S
2*i+ " + *(
t/
M+ j+i
J v+ ig +m
(z)
=0
i=0 (,m=0
*n swi + i
&
(/A+y + n)r(/i,
+ ^w-m +
m!r(/tx
1) r(i/
i/
+ |n-m + + + w-2i + l)
l)r(/A
+ y + M -w)
it is
supposed that z
|
<
|,
We
Bnillt (t),
defined by
the equation
m
(-
1)
R ;**(*) -n
W
,
'
This polynomial was investigated by Gegenbauer*; it recurrence formulae, none of which are of a simple character.
It
(2)
satisfies various
may be noted
that
,;
0,0
(0
= **" (*)
functions in ascending
The
on
series
record.
(3)
t-"J
J
(2t sin
<f>)
B2n +
dt
1;
>v (t)dt
= 0.
(4)
2^. J
t~ >
J, (2t sin
/
<f>)
Bm
M>
(t)
In the
1
_ 2* + "((j + v + 2n)r(fi, + l)r(fj, + v + n)8m''<b n\T(fi + v+l) x aFa (-n,/ji + l,fj.+ v + n; f* + $* + $, \f- + l v + special case in which ft = v, this reduces to
r(o+)
2 *J sin
<f>).
(5)
t-*Jp {2tam4>)Bm
v, v
(t)dt=2*(v+n)r( v )sm*<l>Cn*(co82<t>).
<f>
This formula
specialised by taking
(2),
equal to tt or tt.
(1877), pp.
218222.
294
9*6.
[CHAP. IX
RmiV (z).
Jv+m {z)
linearly in terms of
Jv (z)
and
to
J ^{z)\
and the
polynomials in \jz
We
proceed to shew
how
The
result of eliminating
(z)
equations
{2 (v
(z)
0,
(p = 0,
1,
. . .
m - 1)
=0.
seen to be
-2z- 1
(v
1,
^ + m(2),
0>
+ m-l),
-2z- l
1,
(v
+ m-2),
0,
0,
1,
0,
0, 0,
(z),
0,
0, 0,
Jp('),
Jv .
1
-22- 1
1
("
+ l)
(z)-(2v/z)Jy
0,
By expanding
Jv+m (z)
is
in cofactors of the first column, we see that the cofactor of unity; and the cofactor of (-)"*- 1 Jv {) is
i.
-2?- 1 (v + m-l),
1,
0,
1,
0,
0, 0,
-23~ 1 (v+m-2),
1,
-2-i(i/+i-3),
0,
0,
0,
_2*-J( + l),
1,
0,
-22-1
The
cofactor of (-)m
~1
of the last
The
and
it is
cofactor of (-) w,
is
(z) is
also of degree
in
v.
The
which
unity
;
suppressing the last row and column of the determinant by is equivalent to increasing v and diminishing by and so the cofactor of (-)m ~ ./_, (z) is (-) mr " 1 iC-i.r+i (z).
effect of
Hence
follows that
(z)
J_
v
( z)
R m_
+1 ( z )
= 0,
Math. Ann.
108116.
9*6, 9*61]
ASSOCIATED POLYNOMIALS
295
that
is
to say
(1)
(z)
It
Rm
(z) is
2*rl(v
+m~
1>
>-2z->(v+m-2)-2z-*(v + m-3)-...-2z->v-
-R,,(z) was denned by Lommel by means of equation (1). then derived an explicit expression for the coefficients in the polynomial by a somewhat elaborate induction ; it is, however, simpler to determine the coefficients by using the series for the product of two Bessel functions in the
The function
He
way which
It
will
be explained in
9'61.
A n . x (2),
2
2)
such that
'.<)-
-2
'
'^- r"
M*
<*.
where
[cf.
962
(8)]
^n_i
(2)
BH (2) - A n (2) 5B _
(2)=
2K^...(2n-2f2n
'
It should be noticed that Graft and Crelierf use a notation which Lommel's notation ; they write equation (1) in the form
differs
from
Ja + >
W-^CT
<*>
9*61.
The
series for
Lommel's polynomial.
(z),
satisfies
the same recurrence formulae as J,+m (z); and hence the analysis of
96
also
= J.
(z)
Rm, v {z)
results.
J+m (z) J_+1 (z) + (-)m J- -m (z) /.-1 {z) = Rm (z) {Jy (z) J-^ (z) + /_ (z) /_, (z)}
,
2 sin vtt
Rm,v{z),
TTZ
* Cf. Chrystal, Algebra,
11.
(1900), p. 502.
t Ann. di Mat. (2) xxm. (1895), pp. tionen, n. (Bern, 1900), pp. 98109.
4565; Einleitung
131163.
t Ann. di Mat.
pp.
296
[CHAP. IX
by
541, we have
j v+m {z)
j_ +1
X
w(z)
opir{v +
m+p+
+
l)r{ _
+ 2+p y
n)
1
J- v - m (z) Jv -
= 2
n~l
n\T(v- m + n + 1 V (v +
)
_ S
n =o n\
(~) (-
-)n
{\z)- m+tn
T(
m+n
1)
r(y + n)
+ i
when we
that
(jo
P =o(r/i
replace
??
in the
(|*y+g+* (-v+*- P + i) + p + l)!r(-i/ + 2+p)r(i/ + w+p + l)' last summation by m-f jo + l. Now it is clear
!
mw
(m-\p+l)\
and
so,
+ l)m+p+i = (m + 2p + 1) = (m + p + p\(m+p+\)\ p\
2)p
we
find that
2 sin vir
__
(-) m+w (-
m + n) n {\z)- m+m-
_ smvir
^~
<* (-)n (m
Mr
'
> \m
in
infer that
<* (-) {m
- n)\ r (p + m n!
w) {\z)- +
(i-2)!
-
r(i/
-
n)
But the
that
(\ n
drZ\~ m +-n
R m>v (z) is
so,
and
for
m (z), by means of a determinant, shews a continuous function of v for all values of v, integral or not by an obvious limiting process, we infer that (3) is a valid expression
original definition of
R mv (z)
When
T(-
is
a negative integer
it
ma}'
T(v
in part of the series.
+ m-n)
y
y
-n + l)
T(v+n)
'
r(-v-m + n + l)
The series (3) was given by Lommel, Math. Ann. iv. (1871), pp. 108111; an equivalent result, in a different notation, had, however, been published by him ten years earlier,
Archiv der Math,
und
355.
An
(4)
mentioned, was
(*)
Ann.
di
Mat.
9'62]
ASSOCIATED POLYNOMIALS
Pochhammer
.
297
In the notation of
(5)
(cf. 4*4,
4*42),
;
we have
z*).
Rm (z) = (v) m {% z)~m J\ (\-\ m, - \ m v,~ m, l-v-m; Since R mv {z)/z is a linear combination of products of cylinder functions of
+
2
{(i;
2
2
}
m and v 1, it follows from 54 that it is annihilated by the operator + m y + (v~ l) ^ + + my -{v- \Tf] + 4s (^ + 3^ + 2) where ^ = z (d/dz) and so Rmv (z) is a solution of the differential equation [(* + m) (* + 2v + m -2)($-2v- m) (^ - m - 2)] y (6)
orders v
O* -
}(i/
+ 4^-^(^+l)3/-0.
equation equivalent to this was stated by Hurwitz, Math. Ann. xxxm. (1889), p. 251; and a lengthy proof of it was given by Nielsen, Ann. di Mat. (3) vi. (1901), pp. 332 334 ; a simple proof, differing from the proof just given, may be obtained from formula (5).
An
9*62.
We
Rm<v (z),
which were
published by
Lommel
in his
memoir of 1871.
In the first place, 9*6 (1) holds if the Bessel functions are replaced by any other functions satisfying the same recurrence formulae and, in particular,
;
(1)
Y^^R^^iz),
whence
(2)
it
follows that
Yv+m (z) J (,) - J^ m ( Z ) Y^ (z) = R m {z) Yv {z) J _, (z) - Jv (z) Yv ^(z)}= - 2R m<v (z)l(7rz).
,
961 (2), take m to be an even integer replace m by 2m, and v by v - m. The equation then becomes (3) Jv+m (z)Jm+1 _ v (z) + J_ v _ m (z)J_m _ i+v (z) = 2 (-)m sin vir. R^m v - m {z)\{TTz\ and, in the special case v = , we get
Next, in
;
,
(4)
(z)
that
is
(5)
j Wi ( + ./._
is is
2 5 vr~v-)i{*-w W - t^w=o
|(ra
This
order
In particular, we have
()
J>
M + y_ W - 2
(i
+ ),
298
Formula
(5)
[CHAP. IX
it
at that time
by a
followed by a
As
special cases of
96 (1) and
9"61 (1),
we have
cos z
/
.
Jm+i (*)
]
(J
# w_
li
(z),
(-)'"
[
J-m -i (z)
2 \4
cos
z.Rm ,i (z) + ^
2 \*
sin *
.
i^
lf
(*).
By
(8)
we deduce from
(4) thatf
R\
if,
(z)
Finally,
in 9*61 (2),
m by the
v
and then
replace v by v
(9)
m, we
get
-m-i (f) = 2 (-)" sin vTrR%m+Ai _m (z)/(ire).
(1901), p. 23, is that
An
if
by
n
Nielsen,
v+
we have any
z,
algebraic in
we can
at once infer
fm (z)J
2
m (z) = 0, where
the functions
fm (z)
are
identities
/, (*) i2m _!, +
2 fm {z) m=0
(*)s0,
in
(*)
= 0,
Jv -i(z)lJv ()
is
not an
340, that this result leads to many interesting expansions in series of Lommel's polynomials; some of these formulae will be found in his Handbuch der Theorie der Cylinderfunktionen (Leipzig, 1904), but they do not seem to be of sufficient practical
9*63.
Jv+m 0) = J*
replace
(*)
Rm, v (?)
m and
we
by
m+1
and
Jv +m (z),
see that
+ Rm+1,,-1 (*)} =
{/",
(*)
+ /,-(*)}
(1871), pp.
Rm,u(z).
Math. Ann. n. (1870), pp. 627632. t This result was obtained by Lommel, Math. Ann.
iv.
115116.
9-63]
ASSOCIATED POLYNOMIALS
299
it is
Divide by
(1)
J ^ (z), which
v
is
apparent that
Rm-1,,+1 (*)
+ Rm+l
^- Rm
}
v (z).
To
and
we
replace
m in 9 62 (2)
-
by
w+1
m 1, and use the recurrence formula connecting Bessel orders v + m 1, v + m and v + m + 1 it is then seen that 2 iC->, (*) + R m+ (s) = ^^1 Rm> (z\ (2) z
;
functions of
i,
(1)
and
(2),
R m-i
(z)
(m +
_
Rm, v (*)
Again, write
962 (2)
in the form
\z-i F_,
-z-*-*R mtV
7T
(*)}-{*--'
Y9m (s)}[r-*J w
(z)},
and
differentiate
We
=
deduce that
~
(4)
R'm, v (^)
z
(2),
Rm,
,
(*)
and
so,
by
(3), (1)
and
(5)
R'm>
(s)
= i2 TO
Z
,,
(z)
+ Rm (z)
lt
(z)
(6)
=
z
Rmt
lt
(z),
(7)
R'mA Z) = ~
T"" Z
RmA Z
+ Rm+hv-i (z ) + Rm-h* ( z
)-
of these formulae were given by Lommel, Math. Ann. IV. (1871), pp. 113 due to Nielsen, Ann. di Mat. (3) VI. (1901), p. 332; formula (2) has been used by Porter, Annals of Math. (2) in. (1901), p. 66, in discussing the zeros of Rm> v (z).
The majority
may be used
to define
is
Rmv (z),
JLv
m is zero or a negative
of m,
we
R%v\ z) =
that
(8)
4,v(v
+ l)
1>
~i
p 3 )= ^i,A
.
0<
y(z)=l,
R- m
we compare
(Z)
If
we
find that
(10)
Rm,y(z) = (-)" R-r^^y (z) = - R-.m^ v+m+1 {z) = (-)" Rm ,.y-m +i (z).
the functions of negative parameter are defined by equation
(9), all
When
the
formulae (1)
300
9'64.
[CHAP. IX
Lommel
polynomials.
It
is
which has been discussed by Crelier*. The relations were obtained by Crelier from the theory of continued fractions.
shews that Jv+m (z) and Rmv (z), qua functions same recurrence formula connecting three contiguous functions; and so a repetition of the arguments of 9*6 (modified by replacing the Bessel functions by the appropriate Lommel polynomials) shews that
First observe that 9*63 (2)
satisfy precisely the
of
in,
( 1)
Rm+n
(z)
it
\Z)
Kn
i,
v+m+i
\Z)'
**t, v
\Z) -tim.v+i
z)
~~ -^"in+i.v z \ )
Rm-i,v+i \ z )
v+1 \ z )
Rm, v\ z
unaffected by changing
when
m into m=
m-
lt
v+1 (z)
1,
96)
= 0.
+ n,
More generally, if in 9*63 (2) we had we should have similarly found that
R,n,
v
replaced
m by m n
and v by v
and so the value of the function on the left m 1. It is consequently independent of m is R n - i<v (z\ we find from 9*63(10) that
(3)
is
;
unaffected by changing
and since
its
value
when
m into m=n
Rm>
(z)
Rm
-.
n+ i +n (z)
t
1,
\Z),
Replace
(4)
m and wbym-1
v
and n +
and
it is
,v
found that
J-
Rm-\
(z)
(z)
Rm-n-2, v+n+1 \ z )~
If
we
in place of n
and
be-
n>
(z) R m -P - v+p+i (z) RPt (z) rt m _w _j v + n+i \z) Rm, v (z) [ Rm-p-2, v+p+i (z) Rm-n-i, v+n+1 ( z ) ~~ Rm-n-l, v+n+i ( z ) Rm-p-^ v +p+i \Z)] Rm, v \Z) Rn-pi, v+p+i \ z
Jt
i
)>
by
(3).
If we
we obtain
(5)
p. 143),
,
ni
Rp
136
\Z)
Rn-mi, i/+m+i \ z )
)'
Ann. di Mat.
et seq.
Math Ann
-
Iv<
1871 )'
P-
U5,
9'64]
ASSOCIATED POLYNOMIALS
is
301
;
This
it
the same
z.
that
is
(7)
similar
result
may be
by
If
we
eliminate
Jv+m -i (z)
between
the equations*
"c+n \Z)
^Jy+p \Z)
we
/
find that
Jv +p (z) R n -m-i,v+m+i (z) ~ J +m \Z) L-ftn-m, v+m \Z) -ftp-m-i, v+m+i \Z) -Hp-m, f+m {Z) Rnmi, v+m+i (*)] *= J+m (Z) Rp-n-i, f+n+i (z)
'
is
m,
n, p,
by
0,
m, p m, v + m
It follows that
(
8)
+m+1 (*)
= 0,
where
The last two formulae seem never to have been previously stated explicitly, though Graf and Gubler hint at the existence of such equations, Einleitung in die Theorie der BmeVschen Funktionm, n. (Bern, 1900), pp. 108, 109.
[Note.
If
we eliminate J^-i
i
z)
()
Rm (*)
1.
p+
1 (*),
Rm-2, v + 1
(*)i
find that
1 (*)>
by
- m, we have
(')
By
that
we deduce that
()
-Jv -l () R-m-h v +
(*),
say that the equation 9*6 (1), which has hitherto been considered only for positive values of the parameter wi, is still true for negative values.]
* It is supposed temporarily that result is symmetrical, this restriction
m is the
may be removed.
p ; but since the final See also the note at the end of the section.
302
9*65.
[CHAP. IX
Lommel polynomial.
We
(1) V '
shall
now prove
that
lira
<
*-.
**)^.^(*> = j
r(i/
+ m + l)
(,).
This result was applied by Hurwitz, Math. Ann. xxxm. (1889), pp. 250 252, to discuss Jv (z) when v has an assigned real value ( 15*27). It has also been examined by Graf, Ann. di Mat. (2) xxiii. (1895), pp. 49 52, and by Crelier, Bern " Mittheilungen, 1897, pp. 9296.
the reality of the zeros of
From
961
(3)
we have
l)
r(v + m +
n= on\V{v
'
Now
write
(m n)\T
so that
(v + m n + 1) _ (m-2n)!I> + m + l) =
Q ( ' n)
>
(m, n)
7^
(m
(i/
+ ra)(i' +
w)(ra r-P
1)
...
(m - 2n +
v
m-l)...(i/
-.v
+ m-n +
|
1) TV 1)
>
If
now
numerator of 6 (ra,
numerically less
> N.
\0{m,n)\<l,
Hence, when
n> N,
and
m > 2N,
value,
while,
lim
(m, n)
1.
Since
is
(->;(*'>+"
absolutely convergent,
lim
it
theorem* that
hZY+"
is
e <m, n)
= i
established.
=o!
ru++i)
,
rr is
it
$z)" +m
to its limit
is
/^
+ 1 (2)/r(
+ m+l)
also uniform in
* Cf. Bromwich, Theory of Ivjinite Series, 49. f An arbitrarily small region of which the origin excluded from this domain when R (v) < 0.
is
9*65, 9*7]
ASSOCIATED POLYNOMIALS
the theorem of Hurwitz
it is
303
From
fraction for
Jv-
(z)/Jv (z).
For,
when
Jv (z)
0,
we have
^# =
by
lim
r^t^l
L^-l/I^M-l,
m_
,
lim
963 (1). On
R* v+m-Az)
^i,+m(*)
=2
_
2z
l
1
(v
+ m)
we
find that
^Hi^-2^
ttm,, + i (z)
(i/
+ 1) z- 1
(i/
1 2) r-
. . .
-2
+ m) z~*
and hence
(2 )
^W-J
v
(z)
2(v
-...
last
This procedure avoids the necessity of proving directly that, when m-*-ao, the element of the continued fraction
Jv {z)
may
be neglected
;
2( + l)*--...-2>+)*-iis
Jv + m (z)
p. 52.
the method
(2)
xxnr. (1895),
9*7.
In order to discuss properties of the zeros of Lommel polynomials, it is convenient to follow Hurwitz by making a change in the notation, for the reason that Lommel polynomials contain only alternate powers of the variable.
Accordingly we define the modified
equation *
Lommel
polynomial gm>v
l)*"
*
(z)
by the
m
(2)
(1)
so that
n ^'
M-^ W- :
tt
ro-
C ntB
(-)*r(
Rm
,r +1
(z)
By making
(3)
formulae
[
0+i,r (*)
(4)
(5)
= (" + m + l)gm>v (z) - zgm_hv (z), #,+,, r-i (z) = vgm (z) - zg^, +1 (z),
,
IF*
bTz
,- 1 (z),
[J
(6 )
^m ^ (,-m-i gmv
gm
,
<,
m+1 _, (*)
+1
gm+1<v (z),
0<
9-63 (4)]
(7)
- gm+2t (z) gm _
1%
(z) = z m g
[A
* This notation differs in
304
[CHAP. IX
These results
it will
9'6
964.
(3) is of
2>
The result of eliminating alternate functions from the system some importance. The eliminant is
(f
(z),
where
(v
+ m + 1) \{v + m) (v + m + 2) 2z).
We
(v
+ 2)gi<v (z) = ca (z)g2>y (z)- (v + 4>)z'g0tV (z), (z) = c4 (z) gitV (z) - (v + 6) *g%w (z), (v + 4) g
St
(8) (v
= Ca, (z)
<7M)
(z)
- {v + 2s + 2) z g^ {z),
3
t
(y
9*71.
2.
> 2,
the
We
shall
now
1 > v > 2,
a ^ real; and also that they are in which case one of them is negative.
is
when
of Sturm's functions.
97
be the case are (i) the combined with (ii) the theorem that
zeros alternate, it
is sufficient
a monotonic function of the real variable z, except at the where the quotient is discontinuous. denominator, zeros of the
We
where
have
f^ %9
,,
(z)
*M - j dz { {gtm-^Kz))
%
w(
= - 389**, *-*,
and from
B9m-i.m-8 so that
9 (z),
m > 1,
5KK2IW,*W_8
is
expressible as a
sum
of positive terms
when
> 2.
* Math. Ann. xxxin. (1889), pp.
254256.
9-71, 9-72]
ASSOCIATED POLYNOMIALS
305
The monotonic property is therefore established, and it is obvious from a graph that the real zeros of g^^, (z) separate those of g2mv (z).
It follows from Sturm's
any
number of alternations
of sign
g^^
(z),
end of
quotient gr^
i.
(1918), 96.
The arrangements
when
oo
0, oo
are as follows
2m
2m- 2
2m -4
00
+
+
+
+
~ (~)m l
(
+
+
_)m-2
...
...
+ +
00
(->*
...
The upper
negative
;
+1
is
positive or
of this Table.
9*72.
p< 2.
Let
be
less
than
2,
and
let
inequalities
-2s>v>-2s-2.
It will
now be shewn
that*,
that,
1, 2,
gsm ,(z)
t
Provided
each case,
m is taken to
any) of
+ 2m
is positive.
It will first be
(if
g^^z)
alternate
see Proc.
London Math.
(1921), pp.
266272.
306
[CHAP. IX
clear that
= ^2m-2, zg*m-i, + 9*m+i, v-i( z )} - 9*m, ( z ) l^am-i, 1(2) ~ S'am-i, -(*)} = (1/ + 2m)0 tM -i,r() + gmi~2, v {z)g*m,v-\(z )-g*m-i,v-i(z)gvmAz) = (l/ + 2m)5r 2m_ (^) - ^nrWft.rtW-lW
{
1>
>o,
provided that
i>
+ 2m
is
positive
and z
is
negative.
is
is
evident.
The
now shews
of functions
g*m,v(z), g*m-2, v(z)>
g+t, *( z )>
.g*,
*( z )>
-g**-*A z)>
form a set of Sturm's functions.
+g*-4,v( z ),
(~Ygo,v( z)
The
when
is
oo
are
(-)*,
....
+,+,-,+,
...,
and
When
is
zero, the
signs of the
functions are
, ,
-.,
+,-,+,
,(-)*,
the upper signs being taken when 2s > v > 2s 1, and the lower signs being taken when 2s l>v> 2s 2; there are s and s + 1 alternations of sign in the respective cases. Hence, when 2s >v > 2s 1, #,() ^ as no negative zero but when 2s 1 > v> 2s 2, gzm,v(z ) nas one negative
;
zero.
The theorem
stated
is
therefore proved.
9*73.
Positive
2.
As
inequalities
- 2s >
It will
v>
2s
2.
between
now be shewn*
that when v
but that,
lies between.
lies
when v
has
Provided
that, in
mis so
large
that
m+v is positive.
;
* This proof la of a more elementary character than the proof given by Hurwitz paper cited in 9*72.
see the
9-73]
ASSOCIATED POLYNOMIALS
first place, it
307
In the
9tm,v{z) cannot have more than the specified number of positive zeros. when v lies between 2s and 2s 1, the signs of the coefficients of
1, z,
For,
...,
zm in
ffsm v (z)
,
are
+, -, +, -,
...,
(-);
and since there are m 2s alternations of sign, there cannot be more than m 2s positive zeros. When v lies between 2* 1 and 2* 2 the corresponding set of signs
is
_ _ _
and since there are
than
_ _
a.
( V-
m 2sl
more
w 2s
positive zeros.
Next, we shall prove by induction from the system of equations 9*7 (8) many as the specified number of positive zeros.
When
v lies
between - 2* and
the coefficients in gVtV (z) have no +) and so this function has no positive zeros.
1,
2s
On
ao )
;
=-
and so gu+2lV (z) has one positive zero, aM say and, by reasoning already given, it has no other positive zeros. Next, take <7+4,i,(s); from 9*7(8) it follows that its signs at 0, a M + ao are +,, + hence it has two positive zeros, and by the reasoning already given it has no others. The process of induction (whereby
, ;
we prove
function)
now
evident,
and we
infer that
gm
y (z)
has
m 2s positive zeros,
2,
and no more.
Again, when v
-
2s
),
the coefficients in
and
On
#+4,r(0)
< 0,
5T +4(K (+ 00 )
=+
CO
and so g4a +4t ,(z) has one has no other positive zero.
positive zero,
it
By
we prove
in suc-
gv+8tV (z),
...
have
2, 3, ...
positive zeros,
and in general
m 2s 1
v<
positive zeros.
By combining
that ni
such that
2.
- 2s > v > - 2s -
CHAPTER X
FUNCTIONS ASSOCIATED WITH BESSEL FUNCTIONS
10* 1.
(^) investigated
by Anger and H. F.
Weber.
We
inte-
gral
and Poisson's
The
first
function to be examined, J,
It is defined
by the equation
J (z)
-=
(1)
7T J o
f 'cos (v$
- z sin 6) d6.
when
v has the integral value n.
Jn (z)
an integer, the two functions are distinct. A function of the same type as J, (z) was studied by Anger*, but he took the upper limit of the integral to be 27r; and the function J,(z) is
v is not
when
v.
A
(2)
similar function
was discussed
later
E (z)
In connexion with this function reference should also be made to researches by Lommel, Math. Ann. xvi. (1880), pp. 183-208.
It
may be
Jo cussed by Anger is easily expressible in terms of J (2) and E, 2tt 6 in the right-hand half of the range of integration, we get
2m
1
/"2"-
cos {?6
- z sin
6)
dis-
for, if
we
replace 6
by
*T
/"2t
/
cos(v0-2Sin0)cW=
fn
/
*w J
2
If' cos(2vir-vd+zain8)dd
I
2,1t
cos
vit
(z)
+ sin
vit cos
vn
E (z).
Vv
to give Anger's
v)
sin v jt,
10*12), but as
76.
Ev (z).
10-1]
ASSOCIATED FUNCTIONS
in ascending powers of
:
309
z,
write %tr
<f>
for
vdd$ =
cosm
<f>
sin (^vtt
+ v<f>) d(f>
JO
J-Jir
2 sin \vir
-'o
it
sin ^yn-
by a formula due
to
Cauchy*.
In like manner,
Sin TO
0COSl>0d0=_-=7-;
But, evidently,
J(*)=- 2
Tr
v
;
/c m =o (2m)! Jo
:,
sin2m
sin vddd,
so that
(3)
J r (^) = cos^7r 2
jTT
\:l
co
Tx
(_)m/l^)m+i
=r.
+ Sill &V7T 2
and similarly
T Ti
(4)
B(^) =
sinii/7r
2
TO -
'
i/
+ l)
1
vVi/
i
-cos*i/7r
2 =-
5--
( AZ) ~
._ sini/7r
inr
I"
z*
|_
+ -i^
z^
z6
2
"1
(2
-^)(4
-i/
)~(2
-y )
2
"'J
+
(6)
sin
w["
L
12
7T
-;2_ (l 2 -'2 )(3 2 -va ) + (l 2 -i'2 )(3 a -i/2 )(5 a -y2 )~'
.2?
1 s E,(,)= -i:^
VTT
^[l-^L+ r 2 -i>
"
^
2
i/
,-...1
*"j
z*
cos virY
tt"
z
2 2
3
2
[l 2 -!/2
~ (l - v ) (3 -
+ (l
'
memoir) in a
17,
The formula corresponding to (5) was given by Anger (before the publication of his letter to Cauchy which was communicated to the French Academy on July
1854
*
;
135.
M&m.
Cf.
Modem Analysis,
p. 263.
310
[CHAP.
be apparent subsequently
Z3
iy
(
( 10*7), it is
convenient
to write
(7)
So,,
(*)
p^ ~ (p _ y
z2
yi
a ^2
2
3a _
^+
18
_ ^ (32 _ ^(52 _ ^ ~
2 2
i/
2?_
(8)
_,,(*)
1 =-- +
z*
_ ^) ~
a(2
_ *) (4 ,
+
)
'
"
we have
,
.
J,
(z)
sin vrr
v sin vir
s 0t v (z)
*_,, (*),
(10)
E (z) =
/
^
l+cosj/ir
,
So,
(^)
i/(l
cos i/7r)
*-i. , (z).
It is easy to
(11)
cos (z sin
6)d6 =
dd
v sin vir
s_hv
(z),
Jo
(12)
Jo
sin vd
cos (z sin 6)
= - v (1 - cos
vjt)
s_i
(z),
(13)
sin v 6
sin (z sin 6)
d6
= sin vtt
s0< (*),
Jo
(14)
f cos Jo
Mir
I
i>0
sin (* sin 0)
d0 =(1
+ cosi/7r)
.
s0> (z),
(15)
cos
v<f>
= v sin ^irr
= cos i/7r.s
,
S- h {z),
Jo
/in-
(16)
.(*).
Jo
Integrals
namely
COS
sin
(ne-cos8)dd,
Sitzungsberichte, lvii. (2), (1868), pp.
611620.
(vO-zsin
equal to
6)
-r=-,
that,
when
v is real, it is
\it
rj
Jn (z),
where
tf n
is 1,
or -
11
is positive, zero,
or negative.
10*11.
Anger s functions.
It is evident
(1)
101
/
J (z)
J_ (z) =
(2)
J (s)
- J _, (s) =
A.
gin ftvir
v<f>
cos (z cos
f*
I
cos
v<f>
d<f>,
1011, 10-12]
ASSOCIATED FUNCTIONS
E (2;) + E_ (.s) =
a
311
(3)
T T
Jo
.'0
cos
v<f>
sin (s cos
<f>) cUf>,
(4)
E (3)
E_ (s) =
^^
cos
v<f>
cos (2 cos
<) d<f>.
J, (z)
so that
(5)
J (z)
and similarly
(6)
sin vir
.
The formulae
(5)
and
(6) are
10*12.
The recurrence formulae which are satisfied by the Weber have been determined by Weber.
It is evident from the definite integrals that
_! (z)
Anger and
+ Jr+i (z)
- J (z)
=-
(cos
0--J
cos {v0
- z sin 0) d0
\* ^-{sm(v0-zam0))d0
irz J
d0
_
and
B_, (s)
2 sin vir
irz
+ E, +1 (z)
2i>
= _
It is also very easy to prove that
z sin
0)}
d0
these results
we deduce the
t / \ +J r+1 (0) =
.
eight formulae
(z)
/ \
t / \ J,-! (*>
t J, Z
2l/
2 sin
w
TTZ
(2)
(3)
(4)
312
[CHAP.
(5)
(6) (7) (8)
-"\
r (*),
(^ + v) E (z) = Z E_, (z) + (1 - COS 1/7T)/7T, O - v) E (z) = - ^B+ (z) - (1 - cos i/7r)/7r,
i
for z(d/dz).
- v>) J (z) = O - v) \z J v _! (s) + (sin w)/ir} = z(^ + l v) J v _! (*) {v sin vji)\ir = z J v (z) + (z sin i>9r)/w - (v sin tnr)/7r,
2
so that
(9)
v.J.(,)-^~.
also
We
have
(^
- v*) E (z) = (*-v) {*_! (z) + (1 - cos i*r)/w} = Z (^ + 1 - v) E,., (2) (1 COS V7r)/7T = Z E (2) Z (1 + COS I/7r)/7T (1 cos v7t)/tt,
J/
1/
so that
(10)
v >B , ( ,)
(9)
p.
iJ!
7T
<'')""
.
7T
Formulae equivalent to
(1879), p. 47, respectively
;
and
(10)
formula
(9)
earlier
by Poisson
(cf.
10"1).
10*13.
Anger and
H. F. Weber.
It is evident
(1)
J, (*)
E (z) = 7T
By means
is
of this result, combined with formulae obtained in 6'2 622, it possible to express numerous definite integrals in terms of the functions of
Bessel,
6*2 (4)
we have
(2) V '
{
l
J (z)
- Jv Or)},
jarg.r|
valid
when
= |7r,
provided that
f"
J
e"
-"inht dt =
-^ \Jsini/7r
(z)
- J- (*)},
10*13, 10*14]
so that, (4)
ASSOCIATED FUNCTIONS
(2)
313
when we combine
and
(3),
Jv (*)} - \ ir {E (z) +
J (z)}
F (a)},
(5)
itf
d<
- %w
{E (*)
+ Yv (z)\.
but
The
integral [
e-zcosht cosh
Jo
does not appear to be expressible in a simple form; its expansion in ascending powers of z can be obtained from the formula of 6*22 (4), 2siny7r M -z^ht sinh vtdt} I_ v (z) + Iv (z) = - PVcose cos v d6 + ( e
7
7TJo
7r
'O
but, since
[*
-
..
cosm
j
(-)m sinv7r
* co8 ***
- V(v + m)
(5) are
*
'
Fl {- m>
+ "2~
m_
;
i/
-f
ra
2-'
"V'
6*21,
(2)
nugatory when v
is
an
integer,
but from
rent Jo
z * inht
dt
(7)
I" e- M
Jo
- zainht
dt
= \ (-)n+1
The
associated integrals
08 tC l" e-* *(x cosh t)dt (xaiuh t)dt, ("e-^ am 8m v Jo Jo xx. Journal, (1885), p. 260. have been noticed by Coates, Quarterly
Various integrals of these types occur in researches on diffraction by a prism Whipple, Proc. London Math. Soc. (2) xvi. (1917), p. 106.
see, e.g.
10' 14.
It follows
from
asymptotic expansion
of
J v
(z)
when
\z\ is large
args| <%tt,
it is
Jo
out this investigation we shall first expand cosh sinh i/t/cosh t in a series of ascending powers of sinh t.
To
carry
i;/cosh
and
* See Anding, Sechsstellige Tafeln der Besselschen Funktionen imaginaren Arguments (Leipzig, Takeuchi, Tdhoku Math. 1911) [Jahrbuch Uber die Fortschritte der Math. 1911, pp. 493494], ard
11
W.
B. V.
314
If en
[CHAP.
manner of
7*4,
(
M+,l/+)
ui
i/M *
so that
cosh vt COshy<
cosh*
<
(+, i/+,
i+)
^ r j>-i 2i*
gm sjn h t
2*p
2^7
L-o
(?-l)2m+1
p sinh *
2
*
2
(f-l^-iftf-l^-^sinh
]dr*}
Now
(i+)f*+m-*d
$7rt J 27rtJ
r(i/
+ m + f)
r(m-t-.^
'
(^-l)am+1
r(|i/-m+).(2m)!
OT _ (-) cos^7r
^)r(m+|-^)
(2m)!
7T
and,
if
we take p
shewn
/(+, l/t+,l + )
so large that
to be that
1
in Fig. 15 of 7 '4,
fp
1
+ * v ~irf^
2m
- 4fsinh 2 }
7r
~
Jo
1
+ 4<c (1 - x) sinh
'
If v
and
may be
<2p)I
where
dx <
1, since 1
+ 4#
(1
x) sinh <^ 1.
2 j/)
It follows that,
when
R (p + \ ^
^ 0, we have
(2m)!
,
COSh
COSh
I/*
COS V7T
7T
m=o
+ ,, (zmi+M<irW (2 sinh
(2p)!
r]
For complex values of v and < this equation has to be modified by replacing < X ^ 1 by a less stringent condition, in a way with which the reader will be familiar in view of the similar analysis occurring in various sections of Chapter vn.
the condition
Similarly
we have
u h
_ u -* = _L
f( + ,l/M+)
2-iri-J
z *
k-w
1 J - _L_
\
-1/4
d,
10-14]
so that
ASSOCIATED FUNCTIONS
315
sinh vt
~ 2iri J sinh 2t
1
+ ,i+) 1_ r(u+,ilu
$.#
( - I) - 4f sinh
2
"as J
whence
sinh
vt
it
r L:
1
(t-ir
+a
"I
,*.
follows that, if
we take p
so large that
R (p + 1
= sin^irn- p7T
(-)w
cosh*
|_t
2 ^^
(-)^( P +
it
ui y)r( P + i-^)
(2p
+ l)!
- )" r(m+ *
1
sinh
'
+1 1
On
follows that
< o
cosh
* r^.a~qgJg:
.
J
2
;^
2m+2
(ro+
"^
& y)
sinh vt
sin^Tr * (-)OT r (w +
2
27T
+ jQ r (m + 1 -
OT= o
(f2)
If
i;
is real
and z
is positive,
of the
same sign
as,
and
is
numeri-
the (p
l)th term
(2)
when p
(3)
I2
is
so large that
R(p + l
$v) ^ 0.
It follows from
_
,
1013
v
and
1
combined with
(1)
J r (*Wr (*) +
sin vir
trz
-"2
(I 2
~ v*) (32
v (22
a
L.....1
2
sin vir \v
trz
|_i
y(2-y)
?
r, 1
+
2
,
- v ) (4a - y )
?
(!*-*){-*)
;
"'J'
1
/ox (2)
- r (*)~-F(#) x v/n
/
+ CQBinr
^
[z
l
2
-"
2
ii-
-J
2
-cosier Vv
irz
y(2 2
-y )
y (2 2
'"
v*)
z*
(4
- i/ ) _
8
"I
z*
"J
These results were stated without proof by Weber, Zurich Vierteljahrstchri/t, xxiv. 48 and by Lommel, Math. Ann. xvi. (1880), pp. 186188. They were proved as special cases of much more general formulae by Nielsen, Handbuch der Theorie der Cylinderfunktionen (Leipzig, 1904), p. 228. The proof of this section does not seem to have
(1879), p.
Since the only singularities of cosh vtjcosht and sinh vt/cosht, qua functions of sinh t, are at sinh t = i, it is possible to change the contours of integration into curves in the tf-plane on which arg (sinh t) is a positive or negative acute
angle; and then
(1)
we deduce
in the usual
|
manner
(cf.
and
arg z\<ir.
316
10*15.
[CHAP.
argument.
We
when
j
shall
viii,
(z)
be adequate to obtain
f*
/
e T M-*Bnh
fa
TTJo
As
in
Chapter
viii,
we write
v
where
(I)
^ ft ^
it
and 7
is
We
first
Wo
in which
t
""Jo
it is
cosh 7 + sinh t steadily increases from to 00 as shall take this function of t as a new variable t.
It is easy to
increases from
we
shew that
is
when
where n
positive
is
number
{
imaginary
iff),
is
none of these values of r is a real cosh y does not vanish, and, when y is a pure the singularities are on the imaginary axis and the origin is not one of
for,
them
since
not equal to
The expansion
is
T
where
m=
(ft
,
'
a-=2^j
f<
0+ >
^^
1
~ &S J
If
(0+) <fc
.jjm+i
and so a is the coefficient of I ft in the expansion of t-2-1 in ascending powers of t. In particular we have
a _
~l+cosh7'
8,3
ai
_ ~
2(1
225
cosh 7 )4
'
* ~ 24 (1 + cosh 7
cosh2 7
cosh 7
'
_
From
~"
54 cosh 7
we
are
4-
now
in a position to write
down
the expansion
this section.
10-15]
ASSOCIATED FUNCTIONS
is
317
This expansion
valid
when
|
v/z is positive
j
it
has, so far,
been established
to cover the
arg z
<
\tr,
but,
by
may be extended
<
ir.
Next, we consider the modifications caused by abandoning the hypothesis that cosh 7 is real. If we write t = u + iv, the curve on which t is real has
for its
equation
#+
cosh
u sin
0.
The shape
of 8'61.
u sinh a
Since 4>
(u, v) is
sin
a,
we
first
Since
it
follows that,
w,
value of
when
one
(m, v)
=
is infinite
has, at most,
two
real
whenever
v is a
multiple of
it.
When
and,
oo
v)
= oo
4>
(+
oo
<1>
v)
= oo
when v = - rr,
the value of
<J>
the
maximum
{
value of
(u, v),
qua function of
u, is
at
= a,
(u, v)
then being
I
- cosh a sin fi
If this is negative,
- a tanh a 4- (it - ft) cot ft}. = has no real root, and so the (u, fi the equation = meet the line v @ ir or (by symmetry) the line
<I>
-rr)
= ir j3.
Hence provided that the point
(a, /3) lies
in
v)= lies as in Fig. 25, bered 1, 2, 3 in Fig. 21 of 861, the contour when a is positive contour the continuous curve indicating the shape of the
<
(it,
Fig. 25.
and the broken curve the shape when a increases is marked by an arrow.
is
318
[CHAP.
lies in
when
(a,
0)
any of the
domains
I, 2, 3.
(or, /8) does not any of these domains. To effect our purpose we have to determine the which passes through the destinations of the branch of the curve 4> (u, v) =
lie in
origin.
first the case in which a is positive and /3 is acute. The function maxima at v (2n + 1) ir $ and minima at v = (2w + 1) tr + fi, tt) is now each minimum being greater than the preceding; and since (a, positive when v is is greater than positive, it follows that <p (a, v) tt.
Consider
has
<S>
(a, v)
<f>
v = 7T. and similarly it cannot cross the line u = a below the point at which v tr. The branch which goes downwards at the origin is therefore confined to the strip a < u < a until it gets below the line v 2Kir + tt #, where
is
- a tanh a +
{(2K +
v
1 ) ir
- 0}
,
cot J3
>
0. it
The curve cannot cross the line line u = a and goes off to infinity
Hence,
if
(2/'+l)
acute,
7r+j8,
and so
crosses the
= 2Kir.
is
positive
and
is
we
get
TTJo
7m=0
while, if a is negative
and
is
acute,
we get
1
w. + .
(3)
6
ttJo
- (2m)l
a.
1
'
*m-o
J-*
By combining
If,
we
obtain the
however,
is
obtuse and a
is positive,
- a tanh o and
{(2L
is
+ 1) ir +
j3\
cot
/8
> 0.
Hence,
if
is
positive
obtuse,
we get
IT
Jo
~ m=0
and
/8 is
while, if a is negative
obtuse,
we get
(5)
if
r'--^
S.
'
10*15]
ASSOCIATED FUNCTIONS
these results with those obtained in 8*61,
4, 5,
is
319
By combining
we obtain the
66 and
76.
of practical importance,
we shall
An approximate
formula for
am when
-~
(II)
(1854), p. 1106.
integral
If j
00
grt-tnnht fa
The
is
the change in
1, 4,
numbered
ir
'
where a^'
is
7,
so that
_,_
2
(1
l-cosh7'
-cosh 7)*'
,_
+ cosh 7
24 (1
- cosh 7)'
This expansion
fails to
be significant when 7
is
when 7 was
nearly equal to
write
iri.
To
we
= z{\ e),
r = t sinht,
after the
method of 8*42.
00
If
e,t-zinht
dt= -l
ttJ
=s
If -00
e" e~ tzt
dt
7rJo
dr
-T-
dr
__Lj~V
oir.'o
6**+*
m =o
<->"
B m (-ez).(~.y*-*dT
i;"-<">
and hence
O)
l fv-,.,b((ft
TT'Q
result equivalent to this
1 37T
1
m=
r
(f'"+t> (m+l)
(if)*
p. 313.
320
10*2.
[CHAP.
The
If s
integral considered
in the following
<f>,
by Hardy*
= sinh
then
and generally
is
even or odd.
Now
so that
write
r(,
+ 2a
2 2 2
T
T
(t,a)
= t* + 3at
4 6 s
Then the
(1)
Cin ()
(2)
flK (a)
(3)
Ein (a) =
Tn (t, a)} dt
may be shewn J that the first two integrals are convergent when a is But the third integral positive or negative) if n = 2, 3, 4 (whether real fairly obvious that Ein (a) is indeed and it is complex is when o converges
It
;
an integral function of
a.
an even integer, the three functions are expressible in terms but when n is odd, the first only is so expressible, the function of H. F. Weber. the other two involving
When n
is
of Bessel functions
we observe
is
real
combination
+ iSin (a) =
* Quarterly Journal, xli. (1910), pp. 226240. + The sine-integral in the case n=3 was examined by Stokes, Camb. Phil. Trans, pp. 168182. [Math, and Phys. Papers, n. (1883), pp. 332349.]
ix. (1856),
J Hardy,
loe. cit.,p.
228.
10*2, 10-21]
ASSOCIATED FUNCTIONS
when taken round an
321
arc of a circle of
By
with centre at the origin (the arc being terminated radius with complex coordinates R, Re*11"), tends to zero as R-~cc
.
by the points
And
therefore
/ooexp(}irt/)
a)} dt
= e W
;
Jo
where r = te~^ln and the last integral is an integral function of o. The combination Cin (a)-iSin (a) may be treated in a similar manner, and the
result is then evident.
10*21.
To
and then,
in the integrals,
we
find that,
by
621
(10),
Gin (a)
+ iSin (a) = n
2a* f
J o
= iria* wthat
is
****
1/n
(2a* w)>
to say
nsi
{e*>
/w)
we equate
real
and imaginary
parts,
we have
^(.)- E
3=S^ 5
jlJ-*(W-)-^(^}.
In a similar manner,
Ein (a) =
so that,
(3)
2et* f
00
(u,'n)du,
J o
by
622 (5),
#iw (a) = (2a*/n) # (2a*).
These results have been obtained on the hypothesis that a is positive; and the expressions on the right are the integral functions of a which reduce to Cin (a), Sin (a) and Ein (a) when a is real, whether positive or negative. Hence,
equations
are
still valid,
'
"
322
[CHAP.
and n
is
/9, we
see that,
when
ft is
positive
even,
then
<*>
Ci <"
" 2n
^ + '
)
(2/Si0)1 '
(5)
J^
/ra)
(6)
^g^
(9) that,
It follows from
431
when n
and
is
first
two functions
it is difficult
by differentiating
10*22.
To
when n
is
odd,
we suppose temporarily
that o
is
Cin (a)
= 2a*
00
du
Jo
2ftin)
= 2a*cosOWn)
n
That
(1)
is
to say,
CTn(g)
2nsin(^7r/n)
{J_ 1/n
(2a*)-/1/n (2a*)}.
see that,
<-
n& W
(J-'
in 10*21,
we
when
/9 is
positive,
+ *' <**"
by the operator
annihilated
^+
for all real values of
* It
a.
(-)n 2 n
loe. eit., p.
integrals.
10-22]
ASSOCIATED FTTNCTIOHS
;
323
we
find that,
Next we evaluate Ein (a) when a is positive making the usual substitution, by 1013 (4),
Ein (a)
=
-
u) cosh (u/n)
du
Hence the
any value
(3)
is
series
is
Ein (a)
n cos (i7r/r?)
7T
[
J
g
r (m + 3 _ i/n) r (m + f + 4/f|)
(-)^ttmw
+ sin;(ir/n)
and hence
(4)
it follows
|m =
w!
T(m + 1 - 1/n)
(-)m amM
)
'
that
jj^ +
na on
-2
|
Ein () = wofK""3'.
is
Next consider Cin (a) + iSin (a), where a positive. From 10"13 (4) we deduce that
ta'n (a)
temporarily assumed to be
+ % Sin (a) = n
=
2a* f"
.'
(tan
.**
, { n sin (7r/w)
l
,
- J-1/n (- 2a*i)}
amn
7ra*<
n+1 U"
+ f - */n) r (m + f + /n)
(2o*)
1/n
- -
/1/B (2a*)},
and therefore
(5)
Sin (a) -
7ro* (w+1)
"
<*
mn
wcos(Ww)
2wcos($7r/w)
whence
(6)
it
follows that,
when # > 0,
Sin (- 0)
^ ^^
r (m + f _ i/n) r (^ + | + */)
W
)}>
324
[CHAP.
and hence,
(7)
-1 - n
an
~2
j
Si n (a)
= - not
<
n -3
>
3.
be noticed that
Sin (a)
+ (-)*<+ Ei n (a) =
^^
n sin (717/1)
different
{sin
(**/)
+ (- 1 )*<+}
1 )+>} J
= a,
and a and
/9
are real.
The formulae
his
in
10*3.
Cauchy's numbers.
In connexion with a generalisation of Bessel's integral which was defined (see 10*31), it is convenient to investigate a class of functions known as Cauchy's numbers.
The
typical
number, N-n,k,m,
t
is
defined by
Cauchy*
as the coefficient of
in the expansion of
/
1\* 1\*/
IV
and
in ascending powers of
t.
It is supposed that n, k,
(0+
1\*
l\ m
(i)
JU*--5RJ
2tt
,
<~'H)
\*\ \e- ni0 Jo
cos (^rriTT
(")
6d6
dt
+ (-)m eni0
cos*
sinw
6d6
7T
2+*
7T
Jo
N-n^m
zero
if
n+k+m
is
odd or
a negative integer.
473475, 510511; xn.
(1841), pp.
9293; xm.
(1841),
682687, 850854.
10-3]
ASSOCIATED FUNCTIONS
(1) it is seen that
325
From
(2)
#-*,*,
= (~)m
Fn,k.m
= (-)"-* &n.k,m-
These results, together with recurrence formulae from which successive numbers may be calculated, were given by Bourget*.
are
(4)
to
>
+ 1/0* (t - l/0m - *- (* + 1/0*" (t - 1/0 + 1-~* (t + 1/0*- (* - l/0~ " m_1 - t-n ~ 1/0* (* - I/O" (* r- (* + 1/0* (* - i/0m - p- (* + 1/0* (* - V0
1
By means
number
is
ultimately expressible in
N-n
r
t
t,<n -^ -,o,m-
due
to Bourget,
owes
its
It follows that
^.*--2^TTi)/
2w(ro
rn
K)
/
i
dA'-i)
d*l
V
dt
l)J
*/
by a
partial integration.
On
t
(5)
(m +
1)
and similarly
(6)
(k
x.
(18956),
k-$,
pp.
12, are
(7)
-^-n,k,
2 Cr m= r=Q
t
m>
(8)
*.
m-,-
and
(4).
^-.
This
is
*.
2 m" r=0
(
r
)
Cp - r m Cr
.
3354.
326
10*31.
[CHAP.
The function
<i)
Jn<k (z)
is
defined
by the generalisation
{*
of Bessel's integral
J,.
is
W = ^ P *~
is
))'
where w
an integer, and k
a positive integer.
It follows that
A..
* (*)
f*
I
2,r
ex P {~
* (
nB ~ z sin
0)\
(2
cos
W de
>
and therefore
(2)
/.*(*)
= - r(2cos0)*cos(n0-ssm0)d0.
T Jo
The function /,*(*) has been studied by Bourget, Journal de Math. (2) vi. (1861), pp. 42 55, for the sake of various astronomical applications ; while Giuliani, Oiornate di Mat. xxvi. (1888), pp. 151 171, has constructed a linear differential equation of the fourth
[Note. An earlier paper by Giuliani, Giornale di Mat. xxv. (1887), pp. 198202, contains properties of another generalisation of Bessel's integral, namely
i'
cos (nd
If
<8>
of (1) in powers of z,
we deduce from
;
10*3 that
/,*(*>
= I
=o
i\U,*, m ^Jf ml
and
it is
(4)
Jn
10*3 (2)
(z)
,
= Jn (z).
Again from
(5)
(6)
and
',
and,
if
we take k
1 in this formula,
(7)
These results were obtained by Bourget; and the reader should have no
difficulty in
(8)
-A*+2 (*) =
~ A*+i
(*)
(?)
- /+,,*(*)},
(10)
10-31]
ASSOCIATED FUNCTIONS
differential equation is
;
327
The
by Giuliani
thus
V n Jn *(*)=ir
f"
.'o
du
C
0)} (2 cos
0f d$
2k
7T
(n
Jo
cos 0)*"1 sin
ed6
cos (n0
./
- * sin 0)
7T
<W
{(2 cos
0?-* sin
0} d0,
and so
+
'
and using
G& + *) {Vn Jw
.
"*
W + 2kzJ ^ + ^Jn
n k
'
W -*(*-!)
a
}
</.*
CO.
Giuliani's equation
J\
t (z)
+ (2k + 5) zJ'\ k (z) + {2z* + (k+ 2) - n8 J\ k (z) + (2fc + 5) zJ' n>k (z) + (z* + k + 2 - n2 ) Jnk (z) = 0.
by Giuliani that
It
was
also observed
(12)
*" (2 cos
0)*-
5
n=0
*n J*n,k(z)<sos2n0
ao
+i t
=o
this is verified
right.
M+1
(cf.
A somewhat similar
(1883), col. 1
function J{z\
is
v,
k) has
8.
This function
defined
by the
(13)
J(z;
j<
it
,,
q-Jto
..
*+V= {y +_%\*+l + 2y
2* /* + (*)
whence
(16)
follows that
J(z; v,*)=
=t (i/ + 2m-2)(i+2m+2)*
328
10*4.
[CHAP.
Now
an integral resembling Poisson's integral. This function is called Struve's function, although Struve investigated* only the special functions of this type of orders zero and unity. The properties of the general function have been examined at some length by Siemonf and by J. Walker*.
Struve's function
H(4
-
of order
v, is
(1
>
<*>
r$m
(1
f.
8in
" *
"
provided that R(v)
r^hfel) .C
> - |.
we have
By
so that
(2)
H"^ )= J
is
values of
v,
whether
R (v) exceeds $
tegral function of
It
is
or not.
H (z) is an
integral function of v
is
also
an
in-
easy to see
[cf.
211
(5),
3121
(1)] that
< 3)
H'<*>=r<#rW)< 1+
ii*# P AIit -i},
{T
!
>'
where
w
and
j
+f
is
+f
I,
*>
I
+f
|>
^
I
+$
(3)
* Mim. de VAcad. Imp. des Sci. de St P4tersbourg, (7) xxx. (1882), no. 8 ; Ann. der Physik, xvn. (1882), pp. 10081016. See also Lommel, Archiv der Math, und Phys. xxxvi. (1861),
p. 399.
J The Analytical Theory of Light (Cambridge, 1904), pp. 392 395. The results contained in and (11), are there given.
'
'
))
104]
ASSOCIATED FUNCTIONS
obtain recurrence formulae thus
329
We can
H dz ^
and similarly
"
5 W = Jo
(-)TO (2u
2" +2m+1
d
<
t*
" (
* )}
_ S =
,
.o
2" +
- +1 r (m + f
(-)m (2m
+ l)g2w
)
r (
to
+*
=
,r_i
sF*"* 5 r (to +
1)
,'
r ( v + to + 4
=
On
(5)
2*r( + f)r(iT*~
H +l( *
''
)-
comparing these
results,
we
find that
HU
H^
(*)
+ H +1 () =
- H, (5) +
z
Y^\Wl\)
'
(6)
(,)
p (7^|yf(J)
(7) (8)
(z),
In particular we have
(9)
^ {^
(*
2
(,)}
= *H
(z),
{H
(z)}
= | - H, (*).
(z)}
(8),
2
we have
{*!!_,
- v ) H (*) = (* - v)
so that
H (z)
satisfies
(10)
V.H.
Ii
(a)
W - r(
^
> J.
(t)
(z)
The function
to
bears
xlii. (1911),
p. 218.
This function
L ()-
r(. + | )P(r+(l+t)
when R
(p)
g
330
10*41.
It
"
[CHAP.
H (z).
H (z)
fails
was noticed in
when
R(v)^ \, because
integral.
We
Let us take
(V
Jo
iy-i sin
zt
dt,
t* 1 vanishes at the point on the right of t 1 at which the contour crosses the real axis, and the contour does not enclose the point
= -l.
If
we suppose
r(i+)
(t
2
that
R (v) > \
.
ment
1)"-* sin zt
1
dt
2i cos wr
(1
2 )"-*
sin zt
dt,
Jo
Jo
yr
Jo
v - D-* *. ^
. ;
and
all
Both sides of this equation are analytic functions of v for all* values of v so, by the general theory of analytic continuation, equation (1) holds for
v.
values of
From
(2)
this result,
combined with
6*1 (6),
we deduce
(1+)
ei
that
(z)
+ <*) =
lie
^rVi)*^ /
'
lr
dt
and
a> be any acute angle (positive or negative), between \ir + a> and \ir + o>. We then deform the contour into that shewn in Fig. 26, in which the four parallel lines make an angle a> with the imaginary axis. It is evident that, as the lines
To transform
let
the phase of z
move off to infinity, the integrals along them tend to The integral along the path which starts from and returns to 1 + oo ie~ iu is equal to ffw ( z ) and on the lines through the origin we write t = iu, so that on them
parallel to the real axis
zero.
;
(P
It follows that
(z)
+r (y + 1) r
e (1 +^)"-*^,
|} j o
is
The
isolated values J, f
|,
...
then an
undetermined form.
10-41]
ASSOCIATED FUNCTIONS
1
331
2
it
has
its
principal value;
>
)
and hence
er^il
r(i) /t
+ u*y-*du.
This
which
is
z for which
ir
<
arg z
< ir,
will
|
ptotic expansion of
H (z) when
is large.
Fig. 26.
was obtained by J. Walker*, who assumed that R(z)->0, so that a> might be taken to be zero. In the case = z/ 0, the result had previously been obtained by Rayleighf with the aid of the method of Lipschitz ( 7 "21).
result equivalent to (2)
R(v)> \,
If,
as in 6*12,
we
replace
a>
by args /S,
it is
may be
(4)
H.().y.(, )+
<*)
r(|
j;
e-(i
+7.)
<fe,
where
\tt <
/8
<
\ir
and \ir +
(3.
H (z) when
we
define
H (z)
(5)
and use
by ze*.
* The Analytical Theory of Light (Cambridge, 1904), pp. 394395. t Proc. London Math. Soc. xix. (1889), pp. 504507. [Scientific Papers,
m.
(1902), pp.
4446.]
332
If
[CHAP.
where x
is positive,
we
see that,
when R(v)<%>
we deduce
that
(6)
L (*) = /_ (*) -
r(y
+yr
( |)
siu (** )
^+
tt2 )"" i
<**'
p. 219, in
which i>=0.
10'42.
H (z) when
\z\is large.
We
lating
now obtain an asymptotic expansion which may be used for tabuStruve's function when the argument z is large, the order v being fixed.
shall
(z)
investigated in Chapter vn, it follows from 10*41 (4) that determine the asymptotic expansion of
/ooexpi/3
/
to
t***\-*
As
in | 7*2,
we have
(p-.l)!*w
V
28.
1
'
We
for
take
so large that
22 (v p ) < 0, and
which
|/3|^tt-S,
|args-/3|^7r-o\
which
7T +
28 ^ arg z ^
ir
We
so that
then have
(
J
1
^sinS,
arg
--
<
tt,
(l
\
^V" "
= Ap>
say,
where
A p is independent of
on integration that
z.
It follows
where
B.
=f
v*
l
;*
e *dtt
(z-w).
10-42, 10-43]
ASSOCIATED FUNCTIONS
that,
333
We
(1)
deduce
tr
and z
|
is large,
j
H (,) - Yv (z) +
^ ^
may
r
^-^
{z
^
may
be
provided that
removed.
This asymptotic expansion
( 2)
b.
w - y.
(.)
n + ^S*r +
(z
"^-
It may be proved without difficulty that, if v is real and z is positive, the remainder after p terms in the asymptotic expansion is of the same sign term neglected, provided that as, and numerically less than the first ~& established This be by the method used in 7 32. 0. may R (p + \ v)
p.
The asymptotic expansion* was given by Rayleigh, Proc. London Math. Soc. xix. (1888), 504 in the case v Q, by Struve, Mem. de I' Acad. Imp. des Sci. de St Pe'tersbourg, (7) xxx. (1882), no. 8, p. 101, and Ann. der Phys. und C/iemie, (3) xvn. (1882), p. 1012 in the case v = l; the result for general values of v was given by J. Walker, The Analytical Theory of Light (Cambridge, 1904), pp. 394395.
If v has any of the values \,\,
...,
then
(1
+ u /z ) v ~l
2
is
expressible as a
It follows
a finite form.
is
when
H (z)
expressible in terms
Hj(*)=(^)*(l-cos*),
(3)
.-(s)
10*43.
K)-V*~0vm, which
represent Struve's function
large.
We
H (z)
As
shall
now
|
when
usual,
we
shall write
we
which cosh y is real and positive. The more general case in which cosh y is complex may be investigated by the methods used in 86 and 1015, but it is of no great practical importance and it involves some rather intricate analysis.
*
/;s(-r:r.
see Rayleigh, Phil.
Mag.
(6)
vm.
206211.]
334
[CHAP.
type,
we
dw
where t =
w cosh 7
7
is
log (1
-f
2
).
It is evident that t,
so that, since
qua function of w, has stationary points where w e iy equal either to o or to ifi, two cases have to be considered,
the stationary points
(I) e a ,
which give
rise to
(II)
e*
is less
Accordingly
(II) in
(I)
we
than
1,
and
which z\v
greater than
1.
When 7
is
increases from
to 00
e*
a real positive number a, t is real when w is real, and, as w -2a to e~a cosh a log (I + e t first increases from ),
,
.
then decreases to
cosh
a log (1
.
+ e**) and
finally increases to
00
In order to obtain a contour along which t continually increases, we suppose that w first moves along the real axis from the origin to the point e~, and
angles,
then starts moving along a certain curve, which leaves the real axis at right on which t is positive and increasing.
To
make a
w sinh f
where
,
i\
it],
e~ a
= sinh f
has for
and are
real.
is real,
its
equation
17),
and
it
We now
V) -
tj)
sm ^
whose corners
arc sinh
to
1,
arc sinh 1
\tri,
\tri.
,
As tges from
increases from
A, F(^,
i/)is
and
this steadily
to 1.
* Except
when a=0,
in which case
it
has a
triple point.
10-43]
ASSOCIATED FUNCTIONS
f
is
335
When
on AB, F(g,
rj) is
equal to
V2
and
arc tan
i>
cosec
nj
17,
increases from
to
%-rr.
Note.
To
*J2
dt
{r
aroten
=
<v(i+2<*)
t,
li+^
7 ^'
because
-=
-5-
When f is on BC, F(g, ij) is equal to ir sech , and this increases steadily from tt/V2 to 7T as f goes from to C; and finally when is on CO, F(^, rj)
is zero.
Hence the
the rectangle
rj) is
equal to sech
a,
OABG,
OA
and so the
The contours in the w-plane for which a has the values broken and continuous curves respectively.
0,
\ are shewn in
Fig. 27
by
Fig. 27.
is real,
consists of the
part of the real axis joining the origin to e~* and a curve from this point to and, as
00
powers of t
is
l-
Tm
336
then
[CHAP.
* m!&
^
2.
1=0 *
and hence, by 10 4
#
(1),
we have
(1)
h.(.)~-<J,(.) +
T7
^^ 2=&.
>
6,-1,
(II)
61
=2coshy,
is
&2
= 6cosh2 y-,
6,
= 20 cosh8 7 - 4 cosh 7
r
is real
When 7
to 00 as
a pure imaginary (=
i0),
from
to 00
~"
{TOT*) 3FJ
dT
that
H,(,)^Fy (,) +
j
r(y + i)r(i)
jS
o
r>
provided that arg z\<\ir. This result can be extended to a somewhat wider domain of values of arg z, after the manner of 8*42.
From the corresponding results in the theory of Bessel functions, it is to be expected that these results are valid for suitable domains of complex values of the arguments.
In particular, we can prove that, in the case of functions of purely imaginary argument,
(3)
1* v
v
I |
(vx)~Iv (vx)
when
is large,
arg v
i[ -4^w w(i +
i
(
*>}j
10*44.
H (z)
and
B B (z).
we can deduce from when n
When
(z) by a polynomial in z; and n (z) differs from 10-1 (4) that in 1/z. is a negative integer, the two functions differ by a polynomial
10-44, 10*45]
For,
ASSOCIATED FUNCTIONS
is
337
when n
a positive integer or
zero,
we have
n +m
and
BW
and
therefore, since
^w
gi(ml)tri/l^\nm
'
say
n \ (i)
In
(2)
like
^= 1 e E-^>
manner, when
/., _ (~) B -(*)
^ ^i
<n "
1>r
,,i
is
1
a negative integer,
'
n+1
Ic*'""
r (n -
m - j)
(js)-"*-^
r(m + |)
H"w( ^
10*45.
TAe
si^ra
of Struve s function.
interesting result that H(a;)
is
We
positive
shall
positive
when
a?
is
and v has any positive value greater than or equal to \. This result, which was pointed out by Struve* in the case v = 1, is derivable from a definite integral (which will be established in 1347) which is of considerable importance in the Theory of Diffraction.
To
we
J,
integrate 10*4(1) by
we
[*"
.I.
d cos (x cos 6)
..,-,)jgi*~edB
'*?
cos
0dd
Jo
*Y ^"p
V
1 1
- (2v - 1)
*- 2 -'?
= ( i*?
^0,
1 (i>
i^iV ./ + )r()
1(
f'sin2
cos
- cos (x cos
-9)1
<W
Mem.
de VAcad. Imp. des Sci. de St Ptersbour<j, (7) xxx. (1882), no. is the natural extension of Struve'- proof.
8,
pp. 100101.
The
338
[CHAP.
;
When
have
than
J,
and,
when
i/
= J, we
completely established.
A comparison of the asymptotic expansion which was proved in 10*42 with that of Yv (x) given in 7*21 shews that, when x is sufficiently large and positive, H (x) is positive
K (x) is not one-signed when v \ and that asymptotic expansion of H (x) is if v
>
<
.
for the
(ir)
F -
r(,+|)rd)
according
that Struve's function
is
or
/2\*
U>
sin
^-ii-)
x when
v
>\
The theorem
functions
;
and Bessel
Chapter
vn shew
values of x,
10*46.
If
{/o()-l(*-"f
j^r'dt
and choose the contour to be the imaginary +ttan |0, we find that
axis,
j {h (x) - Lo (*)} = f
and so
(1)
cos (* cot
<f>)
^^
, >
7 (*)-I*(jf)
=^ J
*cos(#tan<}i)logcot(^)
(
a formula given by Theisinger, Monatshefte fiir Math, und Phys. xxiv. (1913),
If
p. 341.
we
replace
x by *' sin 6,
multiply by sin
6,
and
integrate,
we
find,
on changing the
t
so that
(2)
E, (* tan
<f>)
*'
| f
{/ (x sin 6)
-L
1
6d6
X
,
is also
due to
The presence
of the logarithmic factor ensures the convergence of the integral round the
indentation.
10-46, 10-5]
10'5.
ASSOCIATED FUNCTIONS
339
Whittaker's integral.
The
integral
a*
I
izt
v _i(t)dt,
which
is
It follows
from
*
61 7
(1)
V, L*
<$*
P_x
(t) dt\
=-
lira
0*e* (1
- P) P;_j(0]
=
If
cos
7T
* vir .z*e
-&
.
e") in ascending
powers of s and
integrate term-by-term +
found that
.
Ji
(2)
2*1
.,_
(2r)m .ro!
The formula
we
write
and then
the series
it is
(2), or
for
Legendre functions
2ii-"z^e- iz
(3)
W_
v
(z)
+ -
W ^{z)
v
2v
z
W
,
'
(Z)
j(2Tr)r(-v)T(%+v)\'
2ii- v z~le- iz
'
(4)
W_
(z)
r+1 (z)
= 2W
(z)
^2ir)r^-v)r^ + v)
2i*~ y z*e~ iz
),
5)
r(t . t)r(i + ,)
2i*~ y zi e~ u
.
(6)
iw
An
asymptotic expansion of
W (z)
may be
obtained by
( 7 21).
London Math.
198206.
may
*"
be
verified;
1
and we
4
can prove
that
P^ (t)dt= p
1,
by expanding
(4
- *. 4 + "
- 4')
in ascending powers of 1
340
[CHAP.
-355-/,
Now
it is
e"
M p-<-'>*-^-/
-
M *-.>*
known
that*, near
<=
i;
1,
iV^(*)-t*i(*
i
-** +
8
/l
- \m
.*
(ml?
VY)
>
x log
|
\T~) ~ 2^ (m + 2 ) + ^ (w ~ v + ) + ^ ( + * + $)}
and since
we
(7)
W,(#)~ *#(*)
g-i <*+**+*)
COS V7T T
(if,
m)
+ ^(w + ^ + y) ^r(m+
Some
noticed
l)~log2^-^7rt'}
functions which satisfy equations of the same general type as (1) have been by Nagaoka, Journal of the CoU. o/Sci. Imp. Univ. Japan, iv. (1891), p. 310.
10*6.
Tn (z).
kind, of
The reader
integral order,
may be
%
+
(n-m- 1)1
"*
j
WM M
,
=0
?.
^(^)r
<
10
<*'> ~ * <ro +
>
- * < + m
The
series
which has
(1)
nrYn (z) = 2
n (z)
10*6]
ASSOCIATED FUNCTIONS
341
where
(2)
Tn (z) = -
"2
^-'--f^1 (^)-
n+2wl
w =o ra!(n + ro)!
and
(cf.
3*582)
<3>
g'<')-.!, ( ;i(,7m)
<< + ' +
l
>-< 1 "-
The
pp. 142
Un (z) have been studied by Schlafli, Math. Ann. m. (1871), though he used the slightly different notation indicated by the equations
Sn (z)=-2Gn (z),
more recent
investigations are
The
(4)
function
Tn (z)
is
definite integral
Tn (z) = -
0) sin (z sin
nd) dd.
To
(-ra*r*~ 1 r ( ,\ - [l I jfnW ~La m >riw _ i r(m+i+6)r(n+m+i-6)J e o _j_ra | (-)w (|^)w+wt /+> (i + ty***
-&r* [&>-*-*
i_
(n
i
J,.o
,
2w)!
/
J_,
(o+)
m+1+e
*
(-)w (jg)w+2OT
(i
nie
(- 1> sin
(n
.
fl)
w+2w
(0
- tt)
+ 2m)\
where
It follows that
Now
3
_ ttt . t m>in~i
ta"
(n
+ 2m)\
=
and
so
(n even) (n odd)
}
(
(sinh
-8in#
+ em(e-r)+izsin} <&
4269. Lommel'a
* Bern Mittheilungen, 1898, pp. 156. t Einleitung in die Theori-e der Bessel'tchen Funktionen, iv (Bern,
treatise, pp.
1900), pp.
77
342
If
is
[CHAP.
which
The corresponding
integral for
Un (z)
!
is
un ^ W = -fi
-and
so,
[I
r (i +
from
6*2 (4),
we deduce that
(5)
0)
d$
+ (-) n
" e-nc-*nhf
<ft.
Jo
Jo
10*61.
Tn (z)
and
Un (z).
From
we
see that
Tn_
=-
7T J o
I "(h
{2 cos
- 2n/z} d0
=-
8
irz J o
- n0)} d0
4 =cos hmr 2
z
Jn (z), '
v
Thus
(1)
Tn.
(z)
Again
and so
(2)
Tn
'
(z)
=7T
cos (s sin
(9
- n0) d0,
Jo
r-iW-ZV+ito-MV^).
these formulae
it
From
(3)
(4)
follows that
2
C* (*
(cf.
\nir
2
+ 2/ (s),
- 2Jn (z),
\nir
and hence
(5)
1012) we
Jn (z).
10*61, 10-62]
ASSOCIATED FUNCTIONS
343
for
Jn (z),
With the aid of these formulae combined with the corresponding formulae Y(s) and Sn (z), we deduce from 10*6 (1) that
(6) (7)
(8)
(9)
(10)
Un_, (z) + Un+l {z) - (2ft/*) Un (z) - (2/z) Jn (z), Un _, (z) - Un+l (z) = 2 Un (z) - {21z) Jn (z), (Sr 4 n) Un (z) = zUn ., (z) + 2Jn (z), (*-n)Un (z) = -zUn+1 (z), [cf. 358(1), V n Un (z) = -2zJn+l {z).
'
358(2)]
from the
T_ n (z),
ir
equivalent of
106 (4).
(z)
If
we
replace
by
in the integral
we
find that
T_ n
=vr
f "(Itt
7r
6) sin (z sin
nO) dd
Jo
=
and so
(11)
2 [*
I
it
Jo
(1
- 0) sin
(z sin
n0 + mr) d0,
(z).
T_ n (z) =
(-)"+*
Tn
We
it is
now
define
U-H (z) by
U-n (z)
(12)
(-)
{
10-62.
Series for
Tn (z)
and
Un {z).
- Jn _ im (z)}
We
(1)
shall
Tn (z) = I I
10*6 (4).
{Jn+2M
{z)
from
2 7r
= Z
sin
2m0
Tn (z), and then integrate term-by-term. This procedure needs justification, since the Fourier series does not converge uniformly near
=
To
and
0.
= ir,
is
values of
and
Since the
when
I
8
..
% $ ^7r-o\ we can
...
M
2
ro=l
(*tr-d)I
sin2m0| |<c,
following
is
pp.
of the constant
344
[CHAP.
of 8 between
values of
M exceeding m^.
Ljt-02 2 m =i
=
i
ye
_
.
i1* sin(2i/"-H)f J
9
t
'
_J_
sin
.
a=
/"(if+J)T sin
/
Jjt
eta:
^r,
/sin
is
it is
if this
\Tn {z)
i
| *>i
siu
(z
sin 8
n8) dd
I
",=iyo
21
,
<
-{/:
/r*aiK--i
|
!!
^i-'---'-
where
5 is the upper
bound of
sin (z sin 8
n8)
Since
2(AB + e) Bis
that*
an
infinite series
2 mn (z) = 2 T
(* sin
J
2m6
, sin (z sin n
.
.
- w0) dd
= 2
m=l
''*
{t/n + fc)
/n-2m (^)}
is
established.
be remembered that
defined
3*581) as a
by the equation
and
as a
that, in
Un (z)
power
series given in
106 (3).
10*63.
Graf's expansion of
Tn (z + 1)
as a series of Bessel
coefficients.
Tn (z + t)= 2 T^^Jmiz),
Wl= on
* This expansion
was discovered by
Schlafli,
Math. Ann.
m.
(1871), p. 146.
10*63, 10'7]
for,
ASSOCIATED FUNCTIONS
it is
345
evident that
J_
771
\
'
(\ir
ffl
=
by using
? ["(!.
;
_ 0) I
(0) sin
{*
sin
- (n - w) 0} d0,
under the integral sign is uniformly convergent, the order of summation and integration may be changed, and the
21
result is evident.
The
plicated
pi-oof of
;
it
the formula given by Graf, Math. Ann. xliii. (1893), p. 141, depends on the use of the series of 10*62 combined with 2*4.
for
is
more com-
Un (z +
t).
10'7.
sM) (2).
which includes as special cases the polynomials z0 n (z) and Sn (z) of Neumann and Schlafli, was derived by Lomniel, Math. Ann. IX. (1876), pp. 425 444, as a particular integral of the equation
function,
(1)
V v y = kz+\
fi
where k and
are constants.
It is easy to
z*
+i
z*+*
(2)
[> + If - S
,
{(/T+
If.
1/
2
}
[i>
+ 3)* - v*\
]
'
air
=0 (i> -$V + m=
(-)m
(^r
(\p
|),+i
+ P + |)m+i
wt
r^-^+m+Dr^+^+w+i)
ks^ v {z).
For brevity the expressions on the right are written in the form
The
ft
function sMt 0)
is
is
is evidently undefined when either of the numbers an odd negative integer*. Apart from this restriction the general
solution of (1)
(3)
In
like
evidently
y -#,(*)
manner
the general solution of
+ ***(*)
is
(5)
y = *-*<-!>
*
.
{<$ v
()
+ *V (2)}.
and
v is discussed in 10*71.
The
F.
12
346
[CHAP.
parameters."
y=A(z)J
where
(z)
+ B(z)J_ v {z),
v
Jv (z) A'(z) + J.
we
see that
v
(z)
B'
(z)
On
using 3*12
v
(2),
A (z) =
'
2 sin
^ V Jw
f J
\_
(z) dz,
B(z)
= - ^~i^2 sin
vir J
VJ
(z) dz.
Hence a
solution f of (1)
is
it
sin vir
i>,
whether
= \1ctt
lY, (z)
[V J
(z)
dz
- Jv (z)
\* z*
Y v (z) dz\
numbers fi v + 1 have positive real and (7) may be taken to be zero. If we expand the integrands in ascending powers of z, we see that the expression on the right in (6) is expressible as a power series containing no powers of z other than ^ +1 z +3 z***, .... Hence, from (3), it follows that, since neither of the numbers /jlv is an odd negative integer, we must have
It is easy to see that, if both of the
parts, the lower limits in (6)
, ,
(8)
,^ r
j o
it
an integer
but
if
we
we
find that
(9)
8^A*) =
this formula
and in
making
an
integer.
Pochhammer's notation
l)
( 4*4),
(10)
^ %(,)
tl)(^> +
|i'
+ f; -i*
2 )-
not an integer.
The
of
z,
generalisation of this result, obtained by replacing zn+i in (1) by an arbitrary function was given by Cbessin, Comptes Rendus, cxxxv. (1902), pp 678679 and it was applied by
;
10-71]
ASSOCIATED FUNCTIONS
associated function
347
The
Mi (2)
is
of (1) in the form of a descending series. solution and investigate its properties.
We now
10*71.
A particular integral
(1)
is
>
= kz^
- 1)* - v (fl + P
j (/
- 1)* _
,/>}
|(
- 3)* - V>
;
a-.].
if it
it is
This series, however, does not converge unless it terminates but a solution of 10*7 (1), and it will be called kS^iz).
terminates,
The
series terminates if
in
ft
v
//,
is
if /*
v is an
no other
= v + 2p +
K
'
(-rr(ij*-iF+|)r(jj + ii' + l)
m=o
r(i
l)r(i;
+ m+l)
When
fi
v=2p +
l,
the function
vanishes,
and
so,
when
fi
v is
(2)
2'~ir(i
''-*"
+ * )r(i '' +
Sln V1T
.
i ''
+ l)
V) IT
.
[COS %(fJL
v) IT
J_ (z)
COS |(/Ll +
Jv (z)].
Since both sides of this equation are even functions of v, the equation is true also when /x + v is an odd positive integer, so that it holds in all cases in
which
of
Sllv (z) has, as yet, been defined. We adopt it as the general definition $M( (z), except that, when v is an integer, we have to use the equivalent form
S^(z) = *,(*) +
shewn
2"" 1
(3)
r (/* - I v
4-
\) r(|A*
+ * +
|)
v) ir
.
cos % (fi
;
(*)].
in 1073 that S^iz) has a limit negative integer, i.e. when sMi (z) is undefined
sM)V (a)
when fi+
and
so,
v ov
fi-v
of Lommel's
and S^
v (z), it is
348
[CHAP.
defined
when
either of the
of significance
yields, in fact,
by means of which S^iz) is numbers ft v is an odd positive integer, is still when the numbers fi v are not odd positive integers. It an asymptotic expansion of S^ v (z) valid for large values of
the variable
10*72.
z.
Recurrence formulae
satisfied
by Lommel's functions.
It is evident
from
107 (2)
that
**'Wthat
is
fr
+ iy-*
8
to say
(1)
W.., (z) = **
it is
+1
- {i> + 1) -
:*}
M (4
Again,
TZ^ S
so that
*' "
^=
(/*
(2)
and similarly
(3)
(z)-(v/z)s^ v (z)
= (fi-v-l)s
we
ll
_ltV+1
(z).
On
(4)
( 5)
= (/M + v-l) s^ _j (z) - (fi - v - 1 ) $_i. m-i (*)> 2s'M (0) = (/i + v - 1 ) v-j, .-i (*) + (/* - v - 1 ) Sp_ ,+1 (z).
lt
The reader
deduce from 10*71 (2) that the functions of may be replaced throughout these formulae by functions of
;
so that
(8)
(9)
(10)
= **+1 - {(fi + 1) - #M( , (*), S^ v (z) + (*/,) SM, <) - 0 + * - 1) flMlPU4 (#), 6\, (#) - (v/z) ,(*) - 0 - v - l)8^ v+l (*), (2*/*)^ (*) = (,* + /- 1)^^_ ,,_ (^)-(m-i'- l)iSU.r*(*). 2S'^(z) = (M + v -l)S^. (z) + <ji-v-l)S^+1 (z).
(*)
2
1/
These formulae may be transformed in various ways by using (1) and (6). They are due to Lommel, Math. Ann. ix. (1876), pp. 429432, but his methods of proving them were
not in
all
Lommel's functions Sfl>v (z) when fiv is an odd negative integer. The formula 1071(2) assumes an undetermined form when fi v or fi + v
10'73.
is
an odd negative integer*. We can easily define Sv-i,(z) by a repeated use of 10*72 (6) which gives
in terms of
n\ (1)
*
*-i
<?
J.
o
(z) is
>*+*.,(*)SMl
integer.
2^(- J
of
(-fi^
,)m+1 ( v
-p)tn+1
2^! (1-,)/
which
ix
Since
an even function
v, it is
-v
is
an
odd negative
10*72, 10*73]
ASSOCIATED FUNCTIONS
349
We
next define
Sv-h ,(z) by
(2 )
^w^,[
x>
.-Cifey+ i)]fi
The numerator (which is an analytic function of when /* = i/ 1, and so, by L' Hospital's theorem*
near
/*
= v 1)
vanishes
Now
L
it is
pW
3/*
_U,-i
-i-iw n I U + 1,
.> +
(-)"(^)"
(p
l)ir( + m + 2)
+ 1) - ^ (m + 2) - yfr(v + m + 2)}.
[^
and
{2m+i
r_ x
follows that
s^y-i'-rwsJrlfSli)
x
{2 log
\z - ty (p + m +
1)
yjr
(m +
for
1)}
- 2"v
ir
V (v) Y (*),
a negative
(where n
is
and
this formula,
is,
is
integer,
when
when
=n
To discuss
-VWwhichgives
'
Sr
Since
-'.o(*)=^[p{^ +1 -^ +S.o()}]
Mss
_i
()-* {r (J, + f
)}Jo
^t^
.
(z)
- sin &**
()},
on reduction, that
<8L,,o(*)=5 *
* Cf.
(4)
350
10*74.
[CHAP.
From the descending series given in 1071 (1) it is evident that Neumann's polynomial On (z) is expressible in terms of Lommel's functions by the equations
(1)
0.im (z)
= (1/*) S^ n i, Sn (z)
is
0^ 0) = {(2m +
1)/*}
S0tWl+1 (z),
and
Schlafli's
polynomial
flU (*)
similarly expressible
by the equations
(*).
(2)
= 4 S-i. (*)
#J
;
(z) dz,
z"
(z)
dz
in terms of
Lommel's functions
.
thus we have
dz
d_
{z
Jv (z)
zl ~>
5M_
*-*
]>
_ 1 (*)}
dz
S^^
(z).
On
eliminating
find
8^^-^z)
1072 (6), we
(3)
by integrating that
(ii,
P ** Jv (z) dz =
j* z
(z)dz
p and
i>
so that
Neumann's or
l
Schlafli's polynomials,
thus
(6)
f z9g^
f Sf
(*)
dz = Z*
(*) -
#_, (*) Om
(*)J
(7)
9Bl+ x (*)
dz = * {#+,
and
(6) are
(1876), pp.
contained in Lommel's paper, Math. Ann. ix. were given by Nielsen, Handbuch der Theorie der Cylinderfunktionen (Leipzig, 1904), p. 100, but his formulae contain some misprints.
Of these
425444;
and
(7)
It should
it is
10*74, 10*75]
ASSOCIATED FUNCTIONS
is
351
The
/n
(8)
.
2r
r(y + m) v+ 2m Q
..
2 v+1
T(v+m + l) v+2m+l
III
.
It follows that the most general case in which the integral (5) is expressible in terms of elementary functions and cylinder functions is given by the formula
(0)
/'
^r-wm
(*)
dz ~
ml z*- v tvT (y + m) I
>M-2m \ z ) -2mi,v\ 2; )
+ 2m-l
(
_
The function denned by the
series
&,+tm-l (z)
v
A^
{z)
+ 2m
r(v+2m+i)
r(v-i)
;
,+l,lw
has been studied in great detail by W. H. Young t this Function possesses many properties analogous to those of Bessel functions, but the increase of simplicity over Lommel's more
general function seems insufficient to justify an account of
them
is
here.
The
integral v
f* I Jo
J
,
(t)
'.
t(z t)
an integer) by H. A. Webb,
Messenger, xxxin. (1904), p. 58; and he stated that, when /=, its value is On (z). This is incorrect (as was pointed out by Kapteyn) ; and the value for general values of v is J
when
It (v)
>
and arg ( - z)
|
<
ir.
10*75.
We
a)
when
|
shall
that,
when
ft
fll<
^. w ~^-[i-^-y-" + tz
j
--, i-...],
is
large
27rtJ_ooi- i> +i
'
sins7r
The contour
poles of the
*
is
to be
to
Gamma
Gegenbauer, Wiener Sitzungsberichte, lxxiv. (2), (1877), p. 126. t Quarterly Journal, xian. (1911), pp. 161177. % Of. (rubier, Zurich Vierteljahrsschrift, xlvii. (1902), pp. 422428.
Proc.
London Math.
59118;
cf.
352
[CHAP.
<
ir,
and
it
may be
seen without
difficulty that it is
(a*-2?).
It
Gamma
function
when
R with
p \, on
R -*- oo
provided
that
sum
of the
residues of
2
at the points
0,-1,-2,. ..,-<_p-l),
1,2,3,...,
When we
we
find that
+
so that
_-i (-yr(i-^-i-iir+m)r(i-^-|y + w)
sini>7r
x [cos i 0* -
v)
/_ (*)
(^~2p),
and
so,
by
1071
(2),
and
10*8]
10*8.
ASSOCIATED FUNCTIONS
Hemi-cylindrical functions.
353
Functions
(1)
Sn (z) which
combined with
(2)
B.W
'(*)
will
be called hemi-
S (z)
is
Sn (s)=/(D).S
where
fn (?)
satisfies
combined with
It follows
by induction
(cf.
91 4) that
tt
],
and therefore
(3)
If
(4)
w, it is
S_ (*)=(-)"Sn
tt
(*).
To obtain an
alternative expression to
(3),
put = sinh
t,
and thenf
(n even)
(re
t Mi)-
jcoshntf
{-sinhn*
odd)
1+ 2! ? + ~4i
]
*
fc '-.
..
(even)
odd)
,,
re
,.
re (re
-jjf
Hence
2
I
l2)
^-31
(re
So <*)
S" (#) |j
*
(5)
Sn (^)=
+ \, n(n2 _ 12)
ft
re
Cre
22 )
'
S (,)+...
(re
even)
It is to be noticed that
but
8n (z), Un (z)
It should be
n (z) Tn (z) and E (*) are hemi-cylindrical and H(s>are not hemi-cylindrical functions.
y
functions,
354
10*81.
[CHAP.
We
(!)
shall
now
Sm (z + t) = 2 J n (t)Sm _n (z);
n=
is
the expansion
at the point
z,
valid when z + 1 lies inside the largest circle, whose centre is which does not contain any singularity of the hemi-cylindrical
inside or
Take as contour a circle G with centre z such that S on the circle. Then
(f) has
no singularity
,, ( , +0
ZlTl] c
ifiM %z
t
a-z)jn (t)^dc
The
Sw (s + t)=
i
cn
Jn (I)
B* (0 On (t- s)
= $
Jn (t)fn
(-^8m (z)dz
-.!,^>/-(-a)/-(5) (')*
But
so that
it is
2)},
is
obvious.
S (z) denotes a function of a more general type than a hemi-cylindrical function, namely one which merely satisfies the equation Sn, 1 (z)-8n+l (z)^2Snf (z),
without satisfying the equation
(z)
=S
'
(z),
we
still
have
f*
V di) Sm W = Sm_n
(1) is
still valid.
( z)
+ ( ~ )n &m+n ( ^'
We
53, 91,
xvh. (1880),
xn. (1880), pp. 48. See also Konig, Math. Ann. pp. 8586.
310 340;
ibid.
10-81, 10-82]
10*82.
ASSOCIATED FUNCTIONS
355
The
1(2)
*U (*) + F
v+1 (*)
- (2v/z) Fv (z) =
%gv
(z)/z,
where fv {z) and g (z) are given arbitrary functions of the variables v and z, form an obvious generalisation of the pair of functional equations whereby cylinder functions are defined. It has been shewn by Nielsen* that the
functions / (z)
and
it
+ 9 (*) = - (*).
is
f (z )
equivalent to
(^
1(4)
It is
now evident
<*>*
(z)
a v (z)]
so that
= - z*F, (z) - z/3^ (z) -&-p) av (z), V V F (z) = - z$ v _ (z) ~(**-v) (z).
x
or,
Again
<*
*2 )
v (jr)
- <* + *) [- zFv+l (z) - fi m (*)] = - z*Fv (z) + za v+1 (z) -(* + p) &, (z).
V,F,(z)
We
(5)
= zv,(s),
-(*-p) (z),
- (* + we
v) /9 (z).
where
|(6)
1(7)
zm v (z) = - *_,
z*r y (z)
(z)
= + zap+1 (z)
On
(8) It
vr v (z),
condition
/_, (*) +/ +1
(*)
now has
to be
the exist(8) to
To prove
this,
we assume
be
Ann. di Mat.
5159. 4953.
356
[CHAP.
and
is
(7),
we
solve (5)
by the method
of variation of parameters.
(9)
The
solution
1-k
(m)
= / (*)
L - jy
v (t) r (t)
dtl
where a and b are arbitrary constants; and c and dy may be taken to be independent of z, though they will, in general, depend on v.
remains to be shewn that c and dv can be chosen so that the value of given by (9) satisfies (1) and (2), or (what comes to the same thing) that it satisfies (3) and (4). If (3) is satisfied, then
It
{z)
*/,_, (z)
|c - tt
J
(z)
F (z) r (z)
+ zY _
that
is to say,
(z)
I
<*_,
+ |tt
l"
*/"_! (z)
+ zY^
But
Tz
since (6)
(z)
[Jv (t)
r (*)
- J^ (t) r^ (t)}
dt\
+ a (z) = 0.
it is
- V. (z) a (jr)} ;
r (z)
<& (s)
- r r _, (z) V^ (z),
and so (3)
is satisfied if *
(z)
|^-d-i + iw r^by
3'63-(12),
(^^
(^)-^(^)^(^)l*|
+ a,(z) = 0,
this condition,
reduces to
zj_, (z)
- c_, + \tt [ Fr _, (a) &_, (a) - F, (a) a (a)]} + * F_, (*) {d - d _i - \ir \JV_ (b) &_, (b) - J (b) a, (6)]} = 0.
[c
X
it is sufficient
to choose c
and d
\(11)
c - c_, - - tt F,_ (a) &_, (a) - F, (a) a, (a)}, d, - <*_, _ | {J^ (6) _, (6) - J, (b) a (6)}
x
10*82]
ASSOCIATED FUNCTIONS
will
357
have no
same two
then
throughout) are
(4).
satisfied,
v {z)
given by
(9) is
a solution of
known*; and so the condition (8) is a sufficient, as well as a necessary, condition for the existence of a solution of the given pair of functional equaas
tions (1)
If,
and
(2).
as z -* *>
/(#)-0
where B >
0,
(**-),
,
5r,(*)
=
,
(**-'),
then we
may make a -* oo
may be
(v)
b -* x>
and we have
written
(12)
F. (z)
= J9 {z)
|ir,
+ tt
(J
+
where ir^v) and
7r 2 (i/)
Y, (z)
(v)
- |tt
rJ
v (t)
v. (0^4
Note. Some interesting properties of functions which $ati*fy equation (2) only are to be found in Nielsen's earlier paper, Ann. di Mat. (3) v. (1901), pp. 1731. Thus, from a set of formulae of the type
F_
v
(*)
(z)
- (2v/) *T () = 2g y ()/,
it is
(13)
Fv+n (z)=F
{z)Rn
(z)-Fv _ l (z)Rn . l
v +i(.z)
n-l
+ (2/)
the
first
and the
* An account of various memoirs dealing with such equations Math. Soc. (2) n. (1904), pp. 438469.
CHAPTER XI
ADDITION THEOREMS
11*1.
It has
been proved
( 4*73)
and
obvious from the asymptotic expansions obtained in Chapter vn that they are not simply periodic functions, and, a fortiori, that they are not
it is fairly
JV (Z + z) as an algebraic function
do not possess
of
,/
(Z) and
(z).
That
is
There
coincide
are,
as addition theorems.
;
In the case of functions of order zero the two classes for functions of the first kind is
<f>)}
indicated in 4'82.
is
The simplest
due
to
Neumann f,
depends on a transformation of Parseval's integral another proof is due to Heine J, who obtained the formula as a confluent form of the addition theorem
for
Legendre functions.
11*2.
We
(1)
shall
now
J
we
(*0
write
zr
= V( + * - 2* cos
2
0),
and
*
all
of
1893)
see
(1885), pp.
Schwarz' edition of Weierstrass' lectures (Berlin, 3342, and Forsyth, Theory of Functions (1918),
Ch.
xiii for proofs of the theorem. t Theorie der BesseVechen Fuvctionen (Leipzig, 1867), pp. 59 70. XHandbuch der Kugclfunctionen, i. (Berlin, 1878), pp. 340343;
cf.
5*71
and Modern
Analysis, 15*7.
xvi.
R. Accad. di Torino,
11-1-1 1-3]
ADDITION THEOREMS
359
We
which
is
and
a,
We
defined
by
the equations
ts
sin
ol
Zz cos
<f>,
ta cos
= z sin
<f>,
and
it is
()
= jj-
*
f
exp
{i
- z cos
0) sin
+ iz sm
<
cos 6)
dd
* [* j 27T J _ (
m=-x
JM (Z)e w l
)
Jm (Z)\
Jm (Z)
"
mie+izain
<*
6)
dd
= J_
%
X
mi <+> - 8in<>
m=
Jm {Z).Jm {z)e*,
<*>
the interchange of the order of summation and integration following from the uniformity of convergence of the series, and the next step following from
the periodicity of the integrand.
we group the terms for which the immediately obtain Neumann's formula.
If
values of
differ
only in sign,
we
for Beasel functions of order | were obtained by Clebsch, Math. lxi. (1863), pp. 224227, four years before the publication of Neumann's
11
3.
Neumann's addition theorem has been extended to functions of arbitrary The extension which seems to be of more immediate importance in physical applications is due to Graf*, whose
order v in two different ways.
formula
(1)
is
*<)
{^-^Y - J_J'+<
pp. 142
Z)Jm
{Z)
*"*'
ze f*
and
are less
than \Z\.
*
Math. Ann.
xliii. (1893),
Get. 1896,
360
Graf's proof
is
[CHAP. XI
based on the theory of contour integration, but, two years after it was by G. T. Walker, Messenger, xxv. (1896), pp. 76
80 this proof is applicable to functions of integral order only, and it may be obtained from Graf's proof by replacing the contour integrals by definite integrals.
argJ?=a, we have
1
HI
= 00
there
is
no special
difficulty in
integration *.
Now
write
(Z - ze-* )t
where, as usual,
-st
= vru,
z
and
sr
it is
is
-*-
+ Z when
all
For
admissible values of
z,
is
it
now an
acute angle,
positive or negative.
u- contour to start
This determination of
renders
ao
= arg m.
We
then have
by
6*2 (2);
and
If
we
by the equations
-/r,
Z z cos = r cos
<f>
z sin
<f>
= vr sin
yjr,
where
-*yfr
as z
--*-
we
obtain the
relation indicated
by Fig.
may be
written
(2)
*./.()- 2
<f>
m= oo
and
^r,
we have
r
Series, 176.
11-8]
ADDITION THEOREMS
it
361
whence
(4)
follows that
/(r)Ti/^r=
sin
!= oo
Fig. 28.
If,
in this formula,
we change
we
3-54 that
(5)
r(t!r)^in/r=
sin
#j=
and
so
(6)
^o*)V= sin
(5)
ifi= oo
^(^^w^^
Bin
was given by Neumann in his treatise in the special case i>=0; see Ann. xlv. (1894), p. 276; ibid, xlvii. (1896), p. 356. Some physical applications of the formulae are due to Schwarzschild, Math. Ann. lv. (1902), pp. 177 247.
The formula
If
it is
we
replace Z, z and
in these equations
by
iZ, iz
and
irsr
respectively,
apparent that
I. ()
(7)
(8)
K.(v)"!*v+- 5
TO= 00
stated
1881), pp.
by Beltrami, Atti
The
(9)
3?,(cos^= 2
m x
(-y^M-W^mW,
(~r^ +wl+1 (Z)Jm+l (z),
and |^|<|Z|.
(10)
9,(w)mnv+- I
Z=rcos^r,
where
= rsin^r
For the physical interpretation of these formulae the reader is referred to the papers by G. T. Walker and Schwarzschild it should be observed that, in
;
an integer and the only functions involved are kind, the inequalities se*** < Z need not be in force.
is
| I
|
362
11 '4.
[CHAP. XI
of
111
is
differentiated
Jn (m) _
K
Jm+n(Z) Jm+n(z) dn
em+n
COS
(w +
tt)
<f>
ttr
~,t
Zn
zn
d(cos<}>)n
This formula was extended by Gegenbauer to functions of non-integral order by means of the theory of partial differential equations (see 11*42); but Soninef gave a proof by a direct transformation of series, and this proof we
shall
now reproduce;
it is
sr
is
We
MV(C+A)} _ * (-hhyjv+P
and replace f and h by
in place of
m
}
Z2 + z
2
and
2Zz cos
</>
respectively
if
we
write
Jv (vr)/"
for brevity, it is
found that
(Zz cos
pi
<f>)P
Jv+P W(Z*+z*)
(Z*
p-o
+ )*+>"
> <j
_ "
-) zp+i cosp
Jv+p+q (Z)
Z'+*
'
pro ,to
^.p\ql
_ S
k
q\
+ p+2k
29
T( v + p + k)
T(v
Z
and so
fl
Z k\{q-k)\
(-) (v
+ p + q+k + l)
Jy+p+*(Z) Z
| 2 I
p -o 9 -o*-o
<f>
with
2 2 i
jki
1)
(1875), pp.
616.
2223.
11*4, 11-41]
ADDITION THEOREMS
363
q=0k=0
k=0 n=0
k)
and so
rt
_ | | |
To * - o
(-)*+" ( v
+ p + 2k) T (v + p +
z**** cos*
<f>
J,+ p+2k (Z )
*t
2*+ p\kln\r(v +
p\k\
p+2k + n + l)
>
Z'
^
p-oa-u
(~)k 2- +P(v+p
~ (-)* 2"+"t 2t (y
'(-)* 2" +w 2* (v
Z"
2
z*
= S ^
k-om-tk
^ J^^Z) J,+m
Z"
z"
<fr
(z)
|
TO
<if"
+ m) T (y + m - k) cos"*- *
2
J, +m ( ) J, +m (*)
-o*-o
(ro-2*)!*!
<*'(-)* 2 w-2*r(i/+m-A;)cos", - 2*<
z"
xt
where, as in
of (1
3*32,
<f>
2a cos
Cm " (cos denotes the coefficient of am in the expansion + a )~" in ascending powers of a. We have therefore obtained
</>)
2
the expansion
(2)
^ip = 2' T ()
W
is valid for all
2(v + m)
m-0
z,
-^^ '^# Cw
It
'
(cos
<f>),
which
values of Z,
2,
and
d>,
and
with the
exception of
0,
1,
is
we have
.
= . I (m + 4)
^>
This formula
227
it
is
also
given by Heine, Journal fiir Math. lxix. (1868), p. 133, and Neumann, Leipziger Berichte, 1886, pp. 75 82. The formula in which 2v is a positive integer has been obtained by
Soc.
61,
An
Phys.
extension of the expansion (2) has been given by Wendt, Monatshefte 125 131 the effect of her generalisation is to express
fiir
-"- sin 2?
<j>Jy + ,, (or)
somewhat complicated
determinants.
11*41.
The formula
(1)
^P =
2"
r (v) (-)- iv + m)
-^& %^ <V
(cos
<f>)
may be
established in the
11 "4.
This formula does not seem to have been given previously explicitly,
364
[CHAP. XI
though
in this section.
11*4,
|
the formula
is
true only
when
\z\ is so
ze i^
\<\Z\ are
22
satisfied; but, in
proving the
convenient
first to
|2Z*cos<|<iZ 2 +
are satisfied.
\z\<\Z\
the form
We
5*22 (2) in
p=
which
is
It is
J^>(?)
= I
(- Zz cos
<f>)P
^+
z, ri{v+p)
j_ v y( Z , + ^2)}
v v v 2* 2* 2*
v
(-) p+1 (v
+1 {v (-)p+^ (~)p (y
%
00
qZ k =
2n p \k\(q - k)\V
<f>
J- V-V ^k (Z)
? _ v t
t
+p+
"",-0*0.-0 =
00
v+*+ cosP <f> J-^-p-a (Z) zP 2k - n) z v - p - 2A; 2fc) T (- / 2k) 2k+M ' p\k\nlT(l-v-p-k) 2*+"p!ib!fi!r(l-v-i>-A) 2
<ft
=
?o *?o
co
J-r-.p_*(Z)
"
<ft
Jy+p+2k (z)
*
*'
+ m ) f ( y + m - k) COS "*-*
(Z)
~ Jo
J-"-w
"
;
^^ cm
Z
(cos
#,
|2^cos<!<|Z2 + *2 |,
\z\<\Z\.
lies inside
Now
is
the
|
m. and
<f>)
'
2
Hence the given
(^J
series converges and represents an analytic function of z provided only that ze{* \<\Z\\ and, when this pair of inequalities is satisfied, J- v (w)/-sr" is also an analytic function of z.
|
* Cf. 5-22.
11-41]
ADDITION THEOREMS
is valid
365
through the
Hence, by the theory of analytic continuation, (1) whole of the domain of values of z for which
\ze*\<\Z\.
If in (1)
we
)
replace v
by
v we
find that
(2)
^^if
2"*r(-v) 2
Again,
values of z
we combine
domain of
now under
consideration,
(3)
^P =
generally,
TO
2-
T ()
+ m)
%^- %^ Cm
}
(cos *),
and
so,
Tlt=0
If in (3)
we make
<7
i>
-*-
(cos
<f>)
= 1,
(v) (v
(m + 0)
we
find that
(5)
F,(r)=
2
=o
6 wl
Fm (Z)Jm (^)coswi0.
The formulae (1) and (2) have not been giveu previously; but (3) is due to Gegenbauer, and (5) was given by Neumann in his treatise (save that the functions Ym were replaced by the functions F<m >). The formula (3) with v equal to an integer has also been examined by Heine, Handbuch der Kugelfunctionert, i. (Berlin, 1878), pp. 463 464. Some developments of (4) are due to Ignatowsky, Archiv der Math, und Phys. (3) xvm. (1911), pp. 322
327.
If
we
replace Z, z and
by
iZ, iz
and
its in
the formulae of
11*4
and
this section
(6)
we
find that
= 2" r 00 5
(-)-
{v
+ m)
OT =
%^- %& CV
}
(cos 0),
(7)
!=) = 2T (v)l
o
(-. (v 4 m)
^W ^^
Z)
C, (cos
0),
(8 )
J^ =
157; while
2-r( y
)J
(7)
(,
o
+ m)
^/ ^^Cm
-(cos0).
Of these formulae,
pp. 156
(8) is
(6)
and
due to Macdonald, Proc. London Math. Soc. xxxn. were given by Neumann in the special case v=%.
(1900),
366
[CHAP. XI
in the case z
of 11*4 and of this section are of special physical importance If we change the notation by writing ka, kr and for Z,
and
<j>
we
become
sin
'
k V(r
</(r2
,=o
v?
cos
k ^(r
V(r2
-
e xp {' A;
5 (-r(+j)i=^^i^>i>,(cos),
m=0
Va
V?'
yXr
'
v (^ +
+ - 2 r
2 ft
cos ^)
77-
= Z (2m +1)
=o
^ (cos V).
V^
These formulae are of importance in problems in which pulsations emanate from a a from the origin, in presence of a sphere whose centre is at the origin. Cf. Carslaw, Hath. Ann. lxxv. (1914), p. 141 et seg.
point on the axis of harmonics at distance
The
by Gegenbauer, and
7r,
we have
(12)
If
TZTW~~
= \ir, we
(z
()
w -
M
}
'
&
*'
'm\P(2v)'
<b
have
(13)
If
^^"
and
r
-o
*"
'
'
Z= 2,
= 0,
df is
taken to be
JVi
,
2 2 *r(i/)r(i/
+ i) 2
r (2j/ +
7> ra
/x
a formula already obtained (5*5) by a different method; in this connexion the reader should consult Gegenbauer, Wiener Sitzungsberichte, lxxv. (2),
(1877), p. 221.
More
7
generally, taking
Z=z,
.
y.i.W _ 2
(2s sin 9)*
loc. cit.
Gegenbauer,
11*42]
ADDITION THEOREMS
it
367
Again,
.
if
R{v)>
[=0 _
2i
\,
(mp)
"- 1
sin 2"
J
<j>
Cm " (cos
<f>)
Cp ' (cos
<f>)
d<f> <
~ p)
and
so,
provided that
R (*/)> J,
y + ^-2^cos^
"
< C0S
*> Sm "
<f>
_ 7T T (2l> + Wi)
2"-1 .mlT(v)
+W1
(g) J,+w
z*
(2)
Z"
simple proof of this formula t, in the special case in which and the cylinder first kind, was given by Sonine, Math. Ann. xvi. (1880), p. 37. Another direct proof for functions of the first kind is due to Kluyyer, Proc. Section of Sci., K. Acad, van Wet. te Amsterdam, xi. (1909), pp. 749755. An indirect proof,
functions are functions of the
w=0
depending on 12-13(1),
pp.
is
Sitzungsberichte,
491 502.
[Note.
An
Sitzungsberichte, lxxiv. (2), (1877), p. 127, is that, if of integration, then (cf. 9*2)
(
+)
(18 >
2Tj
^^-^^ rf
^W-(^-)^9^^(cos0).
by taking
Special cases of this formula, resembling the results of 9-2, are obtainable equal to or r.]
11*42.
theorem of 11-4
is
transformation.
is
equation
dz*
2i>
+ 1 do
z
dz
+ z*
1_
d*Q
dtf
2v cot
J~ fy + o =
<f>),
dQ
in the
form
Q=
where
2
m=0
Bm
is
independent of 0, and
tfTO "(cos0) is
a polynomial of degree
in
cos0;
it
follows that
ra
|a02
+ 2 " cot
tf
* Gegenbauer, Wiener Sitzungsberichte, ucx. (2), (1875), pp. 433443, and Bateman, Proc London Math. Soc. (2) iv. (1906), p. 472 cf. also Barnes, Quarterly Journal, xxxn. (1908), p. 189; Modern Analysis, 16-51 and Proc. London. Math. Soc. (2) xvn. (191S), pp. 241246. t Formula (16) has been given in the special case ,-=0 by Heaviside,
;
(London, 1912),
p. 267, in
368
is
[CHAP. XI
a constant multiple of
and so
)
of am in the expansion of (1
differential
2a cos < + a2
" ".
Cmv (cos may be taken to be the coefficient And then Bm qua function of z, satisfies the
<f>)
,
equation
ozl
z
vz
so that
Bm is a
multiple of
From
considerations of
symmetry Gegenbauer
so that
inferred that
Bm qua
function of Z,
is
a multiple of
Z~ v Jv+m (Z),
Q=
where bm
z
is
a function of v
in
^
>
;
lV(cos<),
coefficients of
m Zm co8m
<ji
Q and
in the expression
A
seems
first
kind.
11*5.
The formula
(1)
e*cos*
^0 |
(2 n
+ 1) in Jn+i (z)
it
n (cos
tf>)
who obtained
(2 )
the expansion
e fcco8*
=2 T(i/) i
m=0
(v
^ + m)imJ z
^Cm
-(co8<j>);
Bauer's result
is
obviously the special case of this expansion in which v\. -*- 0, the expansion becomes the fundamental expansion
deducible from the expansion of 11 41 (4) by and making Z -*. co it is then apparent from -11*41 (9) and (10) that the physical interpretation of the expansion is that it gives the effect due to a train of plane waves coming from infinity on the axis of harmonics in a form suitable for the discussion of the disturbance produced
Gegenbauer's expansion
is
multiplying by Z" +i
origin.
A
in
expanding ^"etecos *
52
we thus
==
> =o
t'
n cosn
<
2 v+n (v
2,
+ n + 2k).r(v + n + k)
*
TT
<Ji>+n+sk\Z)'
,.
w!
4=0
355367;
lxxiv.
(2),
(1877), p. 128;
and
lxxv.
(2),
(1877), pp.
904905.
11-5]
If
ADDITION THEOREMS
we rearrange the repeated
m =o*=o
series
369
and
this is
Gegenbauer's
result.
e**
=
2"
2"
T (v) 2
m=0
(*)
(v
+ m)-^r^-Cmr(cos4>)
>
4)
e -*cos*
5
m=0
(-)*
(>
+ m)
"
+m
*y
Gm" (cos
z
<f>),
(5)
cos(*cos<f>)
2I
2
m=0
(-)m .(v
+ 2m)-^^Oim (cos<f>),
(6)
sin(*cos<f>)
= 2"I 2
i=0
(-)" (y +
.
2m +
(s)
1) r(if
"+2"? 1 ( **
0Wi (cos
ft),
(7)
+ o
.
n-0
m,i
(8)
jV~*fi,()^-
* r(>^)r(gr(, + )^ j^w.
The
last is
different
a generalisation of Poisson's integral, which was obtained by a method in 332. It is valid only when R{v)>-\.
first
of Gegenbauer's memoirs
Equation (1) was obtained by Hobson, Proc. London Math. Soc. xxv. (1891), p. 59, by a consideration of solutions of Laplace's equation in space of 2* + 2 dimensions, 2v+2 being
an
integer.
A
cos
<t>
more general set of formulae may be derived from (2) by replacing -1 by cos cos + sin sin cos ifr, multiplying by sin 2* i/r, and inte<f>
<f>'
<f> <f>
The
<f>'
integral*
ifr)
Cm' (cos
<f>
cos
<f>
+ sin
<f>
sin
cos
sin2 "
-1
yfr
dty
(cos *)
_
which
/
?^!M Cl
z
Oj (cos
is
valid
when
cos
ft'
-2-|r
Cf.
i
to-0
'^^)%Mc^(cos^)^ \Av-\-m)
A
(2),
(cosf),
370
[CHAP. XI
and
/Q v (9) '
Jv-\{z sin
(2;
/V sin
<f>
T^ exp rL COg
,.
<X
cog J/1 J
V(2^T m?
The
result is true for all values of
v,
r(2^
+ m)
~^ Cm"(cos^)C -(cos^).
TO
when
by analytic continuation.
case
This result was given by Bauer, Miinchener Sitzungsberichie, v. (1875), p. 263 in the v=\; the general formula is due to Gegenbauer, Monatshefte fur Math, und Phys. x. 192 see also Bateman, Messenger, xxxni. (1904), p. 182 and a letter from (1899), pp. 189
Sci.,
te
<f>'
equal to
<j>
or to
\n
and,
if
we put
<'
equal to
-,
multiply by eiZcos
' l>
sin 2 "
<f>
and
integrate,
we
find that
(10)
-^ I "j^^z sin
<l>)e
iZc0ii
't
>
sm z ''<f>d<f)
- )m r (" + w )-(" + 27n) m!
<?* +
= 2" WK J(2n)
so that the expression on the left
**(*) Jy + Zm(Z)
2"
m=o
is
Z"
'
in the case v
\.
11 '6.
Bateman' s expansion.
We
(1 )
shall
now
\z Jp (z cos
cos
<E>)
Jv (z sin
<f>
(f>
sin <)
00
= cos*
x
x
<f>
cos"
O sin"
..
sin" 4>
2
n-0
(-)"(/*
+v + 2n + 1) J^+H-zn+i (*)
1
.Tro> +
+ i)(r(, + i))'
;
r'<-^" + <I>),
'
1; ' +1;
""'
^ (-
n,
/u,
+v+n+ 1
/j,
+1
sin2
which
values.
is
Some
of the results of 11*5 are special cases of this expansion, which was
discovered by
Bateman* from a
We
wave motions examined in 4'84. proceed to give a proof of the expansion by a direct transformation.
xxxm.
(1904), pp.
* Messenger,
Soc. (2)
m. (1905),
pp. 111123.
11-6]
It is easy to
ADDITION THEOREMS
deduce from the expansion
-
371
\zJ
IL
{z cos
'5
<fr
cos <$)
<I>)
(-)
w,
(^y
4
+2m+1 (cos6cos<t>>t+2m
AX
<1>
sin" 3>
"
cos2"1
,
,,
x x
-U/*
-ol
+ v + 2m + 2n + 1) ^- n\T{v tr>/ + l)
,
,>.
/mh-2to+m-i(*)
2^1 (- w,
p + v + 2ra + n + 1
sin"
<
; 1/
+1
i'
= cos*
X
cos*
<t>
sin* 3>
(2)
( (-)
=ot
x
2
1
cos2"1 &.r(fi
F (m-n,/A + i'4-m + w + l;
. .
+ l;
sin2 <sin2
<&)U
i'
"
<f>
sin* <P
r(/i
n + l)
W.'+sn+i 1*7
/x
x^
(-n,/*
+ + w + l;
i/
fi
+ 1, v + l;
cos2 <cos2
<I>,
where jp4 denotes the fourth type of Appell's* hypergeometric functions of two variables, defined by the equation
We now
formation
we assume
that
(fj.)
>
0,
though obviously
may
a consequence of the following analysis, in which series are rearranged, and a free use is made of Vandermonde's theorem
The transformation
(-n,
is
cos3
+ + n + 1; fi+l, v + l; cos $ cos <>, sin sin 1 = "jf (~ n >*+* (/* + " + " + V+ cosM ^ cos^^ s n a s in- <D
2 2
2
*.$
/*
i/
</>
<P)
<j>
r -o-o
slr\(jjk+l),(v+l)r
"r-o.-o
^ (-nX^ +
(-)W^y>
t\(s-t)l
- (-)W-*<S>
!(/*+!).-
This transformation has not been previously noticed to exist except in the special case in which #=#, see Appell, Journal de Math. (3) x. (1884), pp. 407 428 ; some associated researches are due to Tisserand, Annates {Mimoiret) de V Observatoire (Paris), xvni. (1885), inm. C.
f-
372
[CHAP. XI
<S>
+ v + n + 1 ),+, (-)+" sin* ft sin8" l)r(t-r)\(r + S- t)l (u-r)\ (fl+ 1 ),+,_ "r^O -0 <- r (~ Vh (ft + y + w + l)r+< (-)'+ sin ft sin8" = S I I (u - r) !(/* + l)r+,_ -0 -0 r-0 ,-t-r r !( + !)r (t - r) \{r + s W+M 2" = I 5 i (- )t Q* + y + n + l)t (y + < + tt + !),.< (-) sin ft sin r!(i/ + l)r (-r)!(tt-r)! + l)n _ /-o-0r-o ((- ), Q. + + + 1), + (--f 1), a0_ ain, - i 3ipM ^ w! *!(*/ + l) <-o-o = V *2
+t (^
8'
<S>
tt
8f
<S>
(//,
= (-)w |*** ) V ^
)
(-n,/*
+ * + n + l;
+ 1;
sin'ft)
x/i(- /m n, v + n + 1; v+1;
= (~)n
(
sin8 <E>)
*i\-cos^<S>.^
(--n,
A*
+ + n + l;
i/
v
/i
+ 1;
sin8 ft)
F (-
1
+ + + l;
+1
sin8 <).
Hence we
\z /M (z cos
ft
cos
<l>) /
(# sin
ft
sin 4>)
'
am' *
irfr + l)r( y + l)
'
^^
<*>
"
+1
sin2
^)
x/jf-w./t + v+B+l;
+1
sin 8 <),
is
evident.
CHAPTER
12*1.
XII
DEFINITE INTEGRALS
Various types of definite integrals.
In this chapter we shall investigate various definite integrals which contain either Bessel functions or functions of a similar character under the integral sign, and which have finite limits. The methods by which the integrals are
evaluated are, for the most part, of an obvious character; the only novel feature is the fairly systematic use of a method by which a double integral is regarded
as a surface integral over a portion of a sphere referred to one or other of two systems of polar coordinates. The most interesting integrals are those
discussed in
integrals, for
122 12*21, which are due to Kapteyn and Bateman. These no very obvious reason, seem to be of a much more recondite
;
become apparent from the recent work by Hardy described in 12*22. and important types of integrals, in which the upper The numerous limit of integration is infinite, are deferred to Chapter xm.
The reader may here be reminded of the very important integral, due to Sonine and Gegenbauer, which has already been established in 11*41, namely
f" <@v
J
(C S
*>
,, Sm ***
.
= *r(2y + m)
2*- l .m\r(v)
<@v+m (Z)
Jv+m (z)
'
Z"
z*
12*11.
The formula
(1)
iWh (z) - 2
is
r(y+1)
J {z sin
|
6) sin"+'
cos*"* 1 Odd,
which
valid
when both
R (/*)
and
R (v)
exceed
1,
The formula was stated in a slightly different form by Sonine*, Rutgers and Schafheitlin+, and it may be proved quite simply by expanding the inte* Math. Ann. xvi. (1880), p. 36; see also Gegenbauer, Wiener Sitzungsberichtc, mcxxviii.
(1884), p. 979.
(2),
f Nietno Arehiefvoor Witkunde, (2) vi. (1905), p. 370. J Die Theorie der Bettel'tehen Funktionen (Leipzig, 1908), p. 31. Schafheitlin seems to have been unaware of previous researches on what he describes as a new integral.
V
374
'
[CHAP. XII
+1
6 cos2 " +1
6d6
= 2
,^
2' i+,
.
'
,_ +2w
rWr m!r
i
(/*
+ + l)T(v+l)j
'
m^
tt
sin 2
^ *
2
cos'-"* 1
0d0
= 2
1=0
() m (A 2V + " +2rw+1
m!r(/t +
+ w + 2)
obvious.
effect of the factor sin'* +1
is
in the integrand
T (/t + m + 1)
in the denominators.
sin
6 as the factor,
R (v) > 1
(2)
and
fi is
factors
J*7,
(5 sin 6)
M0 - gH^fy
In particular, by taking
(3)
= , we
have
(-) f
formula* which
f
* is
M (2 sin 6)
sin 0d0 =
from (1)
H^
(z).
A
(4)
easily obtained
is
M (* sin 0)
p^i'V^ff J,
(*),
when R(v)> R (/*) > 1. This may be proved by expanding / (z cos 0) and integrating term-by-term, and finally making use of Lommel's expansion
given in
5*21.
The
Fp.
and
FM +
functional equation, obtained from (1) by substituting functions to be determined, + 1 in place of the Bessel functions, has been examined by Sonine, Math. Ann.
,
529552.
have been given by Beltrami,
xin. (1880), p. 331, and Rayleigh, Phil.
Some
Istituto
p. 92.
Lombardo Rendiconti,
[Scientific Papers,
I.
Mag.
(5)
xn. (1881),
(1899), p. 528.]
is
(1)
obtained
two
368373
(2)
vn. (1907),
8890.
12*12.
integral.
An
the device (explained in 333) of integrating over a portion of the surface of a unit sphere with various axes of polar coordinates.
If
(I,
m, n) are the direction cosines of the line joining the centre of the
*
Due
to Rutgers,
(2)
12*12]
DEFINITE INTEGRALS
dm whose
longitude and co-latitude are
<f>
375
and
J, (z sin 0) sin* +1
f
cos2* +1
0d0
cos +1
sin 2" <ctyd0
Jo
= (i*y* +r+1
%*> * sin2* +1
Jo
1
.'0
= (hzY + * +l
w m * n ' +1
2
2
<fc
>Jm>0,n>0
JJl>0,m>0
= (i*>t+, +1 ff JO
'
008 '
sin*+f,,+1
cos2*
<f>
sin2" +1 </><tyd0
2r<^+ y +t)
and the truth of Sonine's formula
is
Jo
An
integral involving
is
same device*
fin
{z sin2 0)
cos2 "+*0d0,
Jo
in which, to secure convergence, If
R(v) >
cos2
\.
cos
<f>
we
write
= 1 - sin2 20 cos2 f
equal to
<f>,
we
is
r( y
f^a) /oT^
2*+r(v+i)r(*)J
sin4y+lffco84y+1 ^ sin2
^^^
sin
J#
sm 2<w<w
(iff
(*)"
[[
-n )*"
2
"n*v da>
'
2+*r(*+i)ra)
f
(*
J-(zsinO ) s
^+l6cos
"
2 ''
'
<
>coa,i ''
dd0d<f>
Fr^+T) io*" ^
so that finally,
(1)
sin
^ sin
' +1
* cos2
w
-
by
1211
2
(1),
2
2 0) sin "+>
f*Jw (z sin
(2)
0)
Jw (z cos
cf.
cos2"+' 0<*0
^M^lffi
'
* This integral
has been evaluated by a different method by Rutgers, Nieuw Archief voor Wisateo 12-22.
kunde,
'
376
[CHAP. XII
Some integrals which resemble this, but which are much more difficult to evaluate, have been the subject of researches by Bateman, Kapteyu and Rutgers; see 12 2.
-
reader
As a simple example of an integral which may may prove that, when R(v)>
&,
(a*
^
left in
the form
ir(v + ^)r(i) Jo Jo
V
vm.
V
(1894), p. 136.
12*13.
The formula
(1)
ly^san 0)J
is
^4^Mg+f21
is
which
and, in
valid
- 1,
fact,
1211 from
it
making
Z -*-0.
of z
formula
and
method
is left
to the reader.
We
exceed
, we
* r(
+ *)
k"
a*wfivu*Y(hzy
Jo
.
e dd
oJo
eizl+iZn cos *
'oii?n>0,n>0
\\
m^ n2+i 8 Q
[
i>>
yf,
da
J OJ J >0,J>0
JO.'
-Jir
e -r/f JO m>0 /
J.
isin9
(zn+ZQ
*>
*1
Qi sin
Jo JJ >0
=
I
Jo Jo Jo
*
''
^ S ^n ^ cosi" ^ sm2
1
d<l> d0.
3536.
12-13]
DEFINITE INTEGRALS
377
the exponential function involved here is a periodic analytic function of 2ir, and so, by Cauchy's theorem, the limits of integration with respect to yfr may be taken to be a and 2-rr + a, where a is denned by the
Now
yfr
with period
equations
vr cos
= z,
ts sin
= Z,
and
i/r -f
or
= V( + *").
a
-\fr,
i i
If
we adopt
a for
Jo Jo Jo
and this integral may also be obtained from its preceding form by replacing z by r and Z by zero. On retracing the steps of the analysis with these
substitutions
we reduce the
triple integral to
io Jo Jo
= r(y+|)r(j)
r(j/+l)
giw?
m^ w2 +1 dw
JJ m >0,n>0
F(v +
1)
JJl>Q,m>0
i!ii)_li} f'f
V^
CO9
'sin2* +2 " +2
COS2*
<^m
2 " +1
<<&*> c0
f' e
iwcos<
sin^+2,+2
^^
Jo
) = ^r (/i+i)r(, + i)^^r
initial
and
final
Sonine's
own proof
to the
domains in which
analytic continuation.
In Sonine's formula, replace Z by *J(Z* + C2 - 2Zfcos0), multiply by sin </(* + ^ -2 cos 4>)* v and integrate. It follows from 11-41 (16) that
,
"
(2)
**{?
J^tv^ + g+r-gggcoBfl)
(z^
~*T{v + \)T{\)) 9
+ Z' + ^-2Z^co8<t>)i^"^
R(v)>-$.
9tt<p
'
provided that
JR( M )>-1,
This result
(1903), pp. 442,
is
also
due to Sonine,
ibid. p. 45.
443
13
W.B.F.
^
[CHAP. XII
378
12*14.
An
namely
integral which
fw cog
I
.
first
of Sonine's integrals,
(z cos
cos
s(r)
j o
sin
Jv_i (z sin
sin ty)
has been evaluated by Gegenbauer*; we shall adopt our normal procedure of using the method of integration over a unit sphere.
It is thus seen that
l" e iz cos cos*
(z
gin
f*
sin
e* s
^ C/
( cog
0) sin +i 0de
^tz sin
r
(
rm
>^V
r'
(cos ' s
J
e te <*+*"to+)
^
1
cos
^ sin2
"
sin2,"_1 ^
= ^ffit?,7 1
1
f
1
ff
0/ (n) m2- d
= (^smjr-i
V") i Vi)
el>(?COSl/(+W8in(<r)C/
J.'>o
ft"'
n2l,_ ld6>
= ^/VrT/ix
A
?in
Airy-*
f 2ir
W*
(f) Jo Jo
/"*"
efaincoB(*- (J. ( s i n
f 2w
/
r (")*()
2ir.
** sin * cos *
CS
is
Isin
cos
(<f>
sin
d<bd0,
Jo Jo
<f>
with
period
If
we
retrace the steps of the analysis, using the last integral instead of its
is
equal to
ff
J-
ia
>o
* (I JJwi>0
=(
^Sm ff4
I (") A (i)
r
da>
r(v)
^^^.
I ()
,'o
for
Gegenbauer's function,
cos
<f>)
X
*
t This
Wiener Sitzungsberichte, lxxv; (2), (1877), p. 221 and lxxxv. (2), (1882), pp. 491502. was proved by Gegenbauer, Wiener Sitzungsberichte, lxx. (2), (1874), p. 433; en. (2a),
(1893), p. 942.
'
12*14, 12'2]
DEFINITE INTEGRALS
2
<f>
379
sum
first
Qr
(cos 6)
C/ (cos yjr)
! 'sin 2 Jo
""1
<f>d<f>.
We
*
|
eteooscoS *
j
= r!
{z si
n
)
Jo
(
sin'+i^d^
p(
^ ,^ 2"^
S1
(
"*
C/(cOSVr) f
008 '
0/(008 0)8^0^,
and hence, by
(1)
332,
'e^costfcos *
Jo
j, (^ s n #
i
gi
^
=
Odd
(y )
v sin*-*
If
we equate
I
real
and imaginary
parts,
(2)
Jo
_
and
(3)
f "sin (z cos Jo
f(-)*
)*
sin""1
(r even)
(r odd)
(0
cos
<^>)
J^ (z sin
sin
yfr)
{0,
(_)*<r-i>
/^V sin- f C
-t
"
(cos
^) /+,.(*).
(r odd)
12*2.
In Bateman's expansion of
formula*
2
.'0
write
<I>
<f>
<)}
cos2" +1
<f>
we deduce
(1)
that,
both exceed
cos <<&
1,
21
Jo
/M
<)
sin
<f>
= 2
n=0
( )n /M+ +2 l+
,
(z),
* Journal
fUr Math.
149175 [Werke,
vi. (1891),
pp. 184202].
380
that
is
[CHAP. XII
(2)
( J
V, (t) J(z-t)dt = 2
=0
Jo
^ (0 /
(,
y = J o {^-i (0 + J, +
(01
^ (* fi.
1)
dt
-2
M+ (#),
J0
(VM (0/,<>-of=
I
This formula
is
due
to
independently by Kapteyn f,
It will be observed that
special cases
integral values of
combined with
that
(4)
|
Jo
dt
= sin z,
fi
cosz,
Jo
when
and
and when
fi
By
with v
that, if
R (/*)
J? (v) are
Jo
'
\fi
v)
It seems unnecessary to give the somewhat complicated inductions by which Kapteyn deduced (3) from the special case in which fi = v = l, or to describe the disquisition by Rutgers J on the subject of the formulae generally.
12*21.
A simpler formula
(1)
*
Jo
Pcos (z - 1) J (0 dt
= zJ
(z).
To prove
this,
we put the
it is
easily
verified that
d?u
and therefore
u = zJ
where
(z) 4-
A cos z + B sin
z,
and
B are
constants of integration.
London Math. Soc. (2) in. (1905), p. 120. Some similar integrals occurring in the theory examined by the same writer, ibid. (2) iv. (190G), p. 484. t Proc. Section of Sci., K. Akad. van Wet. te Amsterdam, vn. (1905), p. 499; Nieuw Archie/
* Proc.
voor Wiskunde,
(2) vn. (1907), pp. 2025 ; M6m. de la Soc. R. des Sci. de Lige, J Nieuw Archiefvoor Witktinde, (2) vn. (1907), pp. 385 405. M4m. de la Soc. R. des Sci. de LUge, (3) vi. (1906), no. 5.
12-21]
DEFINITE INTEGRALS
is
381
Now, when z
and so
small,
u-z+
(z>),
A = B = 0,
is
established.
by
differentiation that
(2)
P sin {z - 1) J
.
(t)
dt
zJx (z),
Jo
partial integration,
sin
and,
by a
(3)
(z-t). Ji (t) dt =
am z zJ
(z),
Jo
The formula
(4)
Jo
is
Psin (z - 1)
^
*
\
dt
= -l
/*=<>
which
valid
when
(fi)
> 0, is
of a
more elaborate
it.
and the
result
is
required to prove
v
We
write
= V, (z - t) J* (t) dl,
Jo
^ + v-\yo"(z-t) +
=
{' Jl
Jo(z-t)}J(t)dt + j;(z)
Jo
^r-Mt)dt+j;(z) zt
t
= \*J(z-t)lP-dt + j;{z)
Jo
by 122.
By
(cf.
.
7*33),
.
we deduce
,
that
= A cos z + B sin z + /*
v
f*
I
,/,({)
sin (z - t)
*t
dt,
Jo
(** +8 ),
and, since
^ T(fjb + 2) +
when
when z
is small, it
follows that,
R (fi) > 0,
0.
A=B =
Hence we obtain the required
result.
By
(5)
we
find that
- >^~
dt
=- I
(-)" 6W J"^^ (4
382
12*22.
[CHAP. XII
As a typical example of a very powerful method of evaluating now give a proof of the formula (cf. 12*12)
(!)
finite integrals*,
we shall
l*W&t)J.Wt>&0^+*Mt-$$g^y ^,
is
which
valid
when
is
R (/*)>-
and
R (v) > -
this chapter, because
The method
it
involves the use of infinite integrals combined with an application of Lerch's theorem t
on null-functions.
iV
Let
(r) t
By changing from
$ 13*2 (5)
polar coordinates
(r,
|,
and using
we
see that,
whenever t > / ()
|
then
f"
exv(-r*t).f1 (r)dr =
exp
{-^J^zx^x^^dx
f*v(-y*t)Jn(zy*)y* + 1 dy
j\xV (-r*t).Mr)dr,
f
t
and hence, by an obvious modification of Lerch's theorem, /2 (r) and this establishes the truth of the formula.
;
(r) is
identically equal to
12'3.
Y (*).
(1894), pp.
n () has been obtained by Chessin, American Journal, xvi. 186187, from the formula
Y
V
I
+ +n + 5+... m2
;
Jo
, *
dt;
if
we
T (), we obtain
T. W -(r+*W.r.W-y
i
'""r-d
1)
'
-/
I
{z sit)
must express
my
own developments
thanks to Professor Hardy for communicating the method to me before of it. The method was used by Ramanujan to evaluate
many
may
use
it
examined
earlier in
this chapter.
The form
such that
when
>
0,
exp(-J-2t)./(r)dr=0
o /:
t,
then/(r)
is
identically zero.
CHAPTEK
13*1.
XIII
INFINITE INTEGRALS
Various types of infinite integrals.
subject of this chapter
is
The
sign. The methods of evaluating such integrals are not they consist, for the most part, of the following devices
in
powers of
its
argument and
inte-
grating term-by-term.
(II) Replacing the Bessel function
by
by one of the generalisations of Bessel's changing the order of the integrations, and then carrying out the integrations; this procedure has been carried out systematically by Sonine* in his weighty memoir.
integral,
(IV) When two Bessel functions of the same order occur as a product under the integral sign, they may be replaced by the integral of a single Bessel function by Gegenbauer's formula (cf. 12*1), and the order of the integrations is then changed j\
two functions of different orders but of the same argument may be replaced by the integral of a single Bessel function by Neumann's formula ( 5*43), and
is
(V)
When
then changed.
may be
replaced by the
involving
;
Gamma
functions,
then changed
numerous that it is not possible to give more than a selection of the most important integrals, whose values will be worked out by the most suitable methods care has been taken to evaluate several examples by each method. In spite of the incompleteness of this chapter, its length must be contrasted unfavourably with the length of the chapter on finite integrals.
are so
;
* Math. Ann. xvi. (1880), pp. 33 60. t This procedure has been carried out by Gegenbauer in a Wiener Sitzungsberichtc.
number
384
13'2.
[CHAP. XIII
It
(i)
lyMJM dt ~7(^-V
R (a) > 0, and, in
simplest method
order to secure convergence at the upper limit of in-
where
root is
numbers R(a ib) are positive. That value of the square taken which makes a + V(aa + &) >
|
\
I-
The
is
coefficient
by
tegrations
without
difficulty.
It
is
thus
found that
Jo a
8
ib cos &
+fe2 ),
= l/V(a
and the formula
is
proved.
Now
re- at Jv {bt)tJo
dt.
first investigated in all its generality by Hankelf, in a memoir published posthumously at about the same time as the appearance of two papers by Gegenbauer \. These writers proved that, if R (fi + v) > 0, to secure convergence at the origin, and the previous conditions concerning a and b are satisfied, to secure convergence at infinity, then the integral is
equal to
M~2~
f
'
+ v+l
2
fV +
.
1
'
_^\
a>)'
To
|
is
< a
|
If
in powers of 6
term,
we
I
find that
e-**J {bt)tr-*dt=
"~
\r/ lio.iv
(-)*" (& bY +3m
!
^ + ' +am_1 *~ at *
a" + " +aw
_ ^
* Journal JUr
r (/x + v + 2m)
m .o m r (v + m +
191192.
1)
Math.
t Math. Ann. via. (1875), pp. 467468. 433443; ibid, lxxii. (2), (1876), pp. 343344. J Wiener Sitzungsberichte, lxx. (2), (1875), pp. in 3-32; in the la the former, the special case M =i+1 was investigated by the integral given integral for J,{bt). Poisson's substituting result by general the obtained Gegenbauer latter,
13'2]
INFINITE INTEGRALS
final series
385
\
The
< a
|,
and
so the process of
Hence
re-otJAtyP-'dt
(Wayr(fi + v)
a"I>+l)
The
result has, as yet,
fp
+v
2
'
/*
+ +!
!/
'"
+1
'
6'\
a*)'
I
R (a) >
and
< a
\
and
so,
by the principle of
6.
may
re-rtJrWV-'dt
Jo
a*l>+l)
V + a)
l
M
\
'2
+1 '" + 1
i.
'
'
*)
(Wrfr+)
//* + *
2
'
i-/ +
2
. +
ft
'
aa + 6V*
by
Lamb J may
be regarded as
typical.
By combining two
(4)
r
Beasel functions,
1
it is
r -*r (60^"
=cot
*
"
"V + &*)*<
.ff
r(+i)
aFl
l"T"
"a-
v+1,
'
*+W
,, a2 + ft2J,
-cosec^^^^^,^^^^^, g
provided
Proc.
/? (>*)
>
(v)
London Math.
may be
Bromwich, Theory of Infinite Series, 176. t Wiener Sitzungsberichte, c. (2), (1891), pp. 745766; Gegenbauer also expressed series, whose general terms involve toroidal functions and Bessel functions, as integrals with Bessel
functions under the integral sign.
ZProc. London Math. Soc. xxxiv. (1902), pp. 276284; (2) vn. (1909), pp. 122141. See London Math. Soc. xxxv. (1903), pp. 428443 and Basset, Proc. Camb. Phil. Soc. v. (1886), pp. 425433.
also Macdonald, Proc.
386
obtained by choosing
functions.
[CHAP. XIII
/* and v so that the hypergeometric functions reduce to elementary Thus, by taking p equal to v + 1 or v + 2, we obtain the results
(5)
Jo
(6)
r.~,. Wr *_.2Bfe+ii, +
(a 2
+ o2 )*
fV^,(tor^,
R(v)> $, R(v)> 1
te
^ ^)
r +
Jir
provided that
respectively.
These formulae were obtained by Gegenbauer, Wiener Sitzungsberichte, lxx. (2), (1875), 443; they were also noticed by Sonine, Math. Ann. xvi. (1880), p. 45; and Hardy, pp. 433 Trans. Camb. Phil. Soc. xxi. (1912), p. 12 while Beltrami, Atti della R. Accad. delle Sci. di Torino, xvi. (18801881), p. 203, and Bologna Memorie, (4) n. (1880), pp. 461505, has obtained various special formulae by taking /*=1 and v to be any integer.
J".
- /-,(*) dt 7-
y(d?+W)- ay
vb
(8) (8)
[Note.
e-<j at) dt e Jv {bt)dt 6V(a2 + ft2) Jo It was observed by Pincherle, Bologna Memorie,
#
w( a *+ b2 )-<*r
(4) viii. (1887), pp.
125
143,
that these integrals are derivable from the generalised form of Bessel's integrals ( 6*2) by Laplace's transformation (cf. 9 15). This aspect of the subject has been studied by Macdonald, Proc. London Math. Soc. xxxv. (1903), pp. 428 443, and Cailler, Mem. de la
Soc. de Physique de Geneve, xxxiv. (1902
satisfied
The
differential equations
by
(5)
and
(6),
qua functions of a, have been examined by Kapteyu, Archives (1901), pp. 103116.]
The
integral
-^jo
J
(1863), p. 46, as
sinn, itt
J^bt)
tdt
a limit of a series of Legendre functions (cf. 14 64). The integral does not seem to be capable of being evaluated in finite terms, though it is easy to obtain a series for it by using the expansion
COSechT< = 2 2
5 -(2n +
l),rt
is large) will
be obtained in 13*51.
same general type are given by Weber, Journal fur Math. lxxv. 92102; and more recently the formula
r Jy(bt)rdt = (2b)T(
Jo
et-l
is valid
+$
Jn
r,
=i(2-* +
6y + i'
which
when
R () >
and 1(b)
1
<
13*21.
It
was noticed by Hankel that the hypergeometric functions which occur in the integrals just discussed are of the special type associated with Legendre functions; subsequently Gegenbauer expressed the integrals in terms of toroidal functions (which are known to be expressible as Legendre functions), and a little later Hobson* gave the formulae in some detail.
* Proc.
(1893), pp.
4975.
13'21]
INFINITE INTEGRALS
obtain the fundamental formulae* of this type,
387
To
we
shall
change the
notation by writing
a
where a
is
cosh a,
= i sinh a,
we thus
(1)
fV<cosh x (t s i n h a ) #. dt Jo
= T (fi + v +
1)
provided that
(fi
v)
>
1.
The
Set.
Math.
formula.
f%-<^,g,(*8inha)P'<ftJo
sin
^^ +
{ft
T f> - v +
1)
v) it
Q/ (cosh a),
provided that
R (p + 1) > R (v)
j
The
Legendre function f
(3)
when
[%-<*
Jo
jy (j s n ) p. dt = T (fi + v + 1) Pf (cos ),
i
and hence we
(4)
find that
T^
.
Jo
[V ""gy,
(tain V
= )*(& '
~x
sin u7r
7 , Sin(/A+1>)7T
r (^ v + 1) -^ 7T
Some special cases of this formula have been given by Hobsou, Heaviside, Electromagnetic Theory, in. (London, 1912), p. 85.
and by
An
(5) (5)
[**'*** I () IA)
Jo
cfr
= &-j(cosh)
V^tt)
'
Vt
has been studied by Steinthal}. This formula is connected with formulae of the previous type by Whipple's transformation of Legendre functions, which
* Since,
generality is lost.
by a change of variable, the integrals are expressible in terms of the ratio of b to a, no The various expressions for Legendre functions as hypergeometric series which
97 204. are required in this analysis are given by Barnes, Quarterly Journal, xxxix. (1908), pp. The reader will remember that it is customary to give a different detiuition for the Legendre
cf.
(1896),
471
15'5, 15*6.
X Quarterly Journal, xvm. (1882), pp. 337340. Proc. London Math. Soc. (2) xvi. (1917), pp. 301314.
388
[CHAP. XIII
a.
The more
e- tt
T ^*I
.
(t)V- i
Jo
cosvtt
Q*
1-
(cosh a)
In these formulae,
R (/* + v) >
v in
(6),
and
22 (cosh a)
>
1.
On
replacing v by
we
find that
Jo
sinn*-* a
and
when
and
R (cosh a) > 1.
If we take cosh
(8)
a 0, we
o
deduce that
p KAt)^dt^^r(^y(t^y
ft=l,
(7)
When
(9)
becomes
Jo
sin
and hence,
if
v=0,
a
/,
If
j,_ arcsinh
s/(a 2 - 1)
- a 2)
arccosa
we
replace
a by + i'6, we
/'"
find that
io<
/a j,
in-
+ * arc sinh o
and
so,
when / (b)
|
<
1,
(io)
/".<*).*>*-
j^.
1889), p. 32.
The former
of these
is
[Note. Various writers have studied the Lipschitz-Hankel integrals from the aspect of potential theory to take the simplest case, if (p, <f>, z) are cylindrical coordinates, we have
;
o /;
1
2 V(p +*8)'
It is suggested that, since e~# (zt) is a potential function, the integral on the left is a potential function finite at all points of real space except the origin and that ont the plane
3=0
it is equal to 1/p, and so it is inferred that it must be the potential of a unit charge at the origin. But such an argument does not seem to preclude the possibility of the integral being a potential function with a complicated essential singularity at the origin, and so
this reasoning
must be regarded as
On
the axis of
z,
the integral
13-22]
INFINITE INTEGRALS
389
the reader
For various researches on potential theory with the aid of the integrals of this section, may consult Hafen, Math. Ann. lxix. (1910), pp. 517537. For some develop-
deduce from the results of the preceding sections combined with 11-41 (16) that, if all four of the numbers R(a ib ic) are positive and R (/* + 2v) > 0, while tsr is written in place of V(&2 + c9 26c cos <f>), then
It is easy to
(1)
dt
<**
["
, wr(/i+2>>
7ra^+4
T(2i/+l)J
^ /+ 2, +
'
i
'
T
'
_^\
aV
sin8
# ^
to an elementary function if
fi
=1
or 2
/;^<^,<co* =
/i
^^p2^)a.
The
case
= 2 may
These formulae, or special cases of them, have been examined by the following writers: Beltrami, Bologna Memorie, (4) n. (1880), pp. 461 505 ; Atti delta R. Aecad. delle Sci. di
201205; Sommerfeld, Konigsberg Dissertation, 1891; Gegenbauer, Monatshefte fur Math, und Phys. v. (1894), p. 55 and Macdonald, Proc. London Math. Soc. xxvi. (1895), pp. 257260.
Torino,
xvi. (1880
1881),
pp.
By taking
/x=
1, v = l
t
in (1),
we
*
find that
(" - a J ll?dt = e t Jo
so that the integral on the
(1896), p.
left,
[\/( a 2 + 2-2cos0)-a}(l+cos0)cty,
(5) xlii.
2ir J o
expressible as
is
an
elliptic integral.
An
(3)
integral which
may be
(bt)
j o
cos at
(ct)
=
jo
v{a2
+ (6 + c)2 _ 46csin2
;
This was discovered by Kirchhoff * as early as 1853 the reader should have no difticulty in deducing it from 1321 (10) combined with 11-41 (16); it is valid if all the numbers
R(cb id)
are positive.
* Journal
390
[CHAP. XIH
v e- at f- J^{bt)Jv {ct)dt
sin 2 "
<f>d<f>
)'
(a
+i'
which
is
valid
when
R (aibic)>0
and R(fi)> \,
due
to Gegenbauer,
Wiener
Sitzungsberichte, lxxxviii. (2), (1884), p. 995. It is most easily proved integrals of Poisson's type for the Bessel functions. In the memoir cited
is
functions
(cf.
1323).
13*23.
pp. 397
formula due to Gegenbauer, Monatshefte fur Math, und Phys. IV. (1893), 401, is obtained by combining the results of 13*2 with the integral
formula of 5*43
it is
thus possible to
result obtained
means of integrals of trigonometrical functions*. The general by Gegenbauer is deduced by taking the formula
/M (bt)
(bt)
=-
tJo
J^+ v
("2bt
cos
(f>)
cos
(/a
v) <f>d<f>,
;
multiplying
if
R (a)
2at
by e~'lat VkJrv and integrating from > 1(b) and R (p + v) > - h then
it
|
to ao
it is
thus found
that,,
r e~
Jo
dt=-
"""Jo
0*-+"cos
(fx
-v) <f>.d<f>dt
= -[
tJo
Jo
The
difficulties
hence
(1)
_ r (fi + V + f) 6+
7T*
f**
COS*-*-''
<f>
COS (fl-v)(f>
1
Jo
/i
(a
+ 6coB
) M
"
r+ *
*'
This
result, in
= v = 0,
1881),
p. 204.
As
(2)
/*
=1
and v equal to
t
and to 1.
It is
found that
/." e
~ iatJ
>
M J*M *~%j&+W
(2 2 +6')*-2("2 +&2 )^
(3)
f" c-~J*(b)3t
* See also
an
earlier note
by Gegenbauer,
ibid. pp.
379
380.
13-23, 13*24]
INFINITE INTEGRALS
elliptic integrals
391
is &/>/(*
K and E
+&2
).
Beltrami's
is
/''-^W-aift,
wtW
>
Replacing 6 by
we deduce from
(2)
that
(5)
/V"-/l(6l)/.<*)*- a^-*)
where R(a)>\R(b)\, arid the modulus k t of the and (4) may be modified in a similar manner.
It was stated by Gegenbauer that the integrals in means of elliptic integrals, but he did not give the
(2), (3)
and
results in detail;
deducible from the results of this section were given by Meissel, [Jahrbuch liber die Fortschritte der Math. 1890, pp. 521 522.]
13*24.
The formula
J
Jo
in which
v.
(t)dt
T(^)
2"-* +i
p-*+i
r (v - $n + i )
Weber*
for integral values of
;
obtained by
was extended to general values of v by SoniDef pletely general result was also proved by Schafheitlin \.
The
result
The formula
is
of a
Jo t"-^ 1
rm
is
a Wo
2v r
(u
lim
fV-H + l
v+1
>
'
V'
at once obtained.
direct
Bessel function,
method of evaluating the integral is to substitute Poisson's integral for the and then change the order of the integrations this is the method used by
;
is
/*
and
and
partial integrations.
If
v
we suppose that
set
p. 230.
The
special case in
which
= was
by Stokes as
a Smith's Prize question, Jan. 29, 1867. f Math. Ann. xvi. (1880), p. 39.
v. (1905), p. 347.]
X Math. Ann. xxx. (1887), pp. 157161. Cf. Bromwich, Theory of Infinite Series, 172.
392
[CHAP.
TTTTT
R (fi) <
vergent)
r(o+)jr
R (v) > \,
=
1
we then have
/"(
L Fo^
By
+>
/"*"
2-r
(,
+4)
ra) j + .
<"*~ J. <)*''*
/"*"
/
cos-'1
sin2 "
0d0
Jo
"^-^r^-i/i + i)'
the theory of analytic continuation, this result
is
valid
when p and
v are
R (fi) < R {v + ).
real
When
R(fi)
>
0, we deform the contour ir.to the positive half of the and we at once obtain the Weber-Schafheitlin formula.
The integral*
/("+>
H,(-<)cfc
difference in the
may be
analysis
and
so,
by
Euler's formula (adapted for contour integrals), the factor cos \ fitr has to be replaced by sin \pnr.
It is thus found that
f
(o
+)
H
,
(-Qcft
provided- that
R (fi) < R (v + f
1 /"
and
(fi)
^ 0.
into the positive half of
When
R (/*) >
the contour
may be deformed
(t)dt
'
provided that
If
(3)
we take
2"-+ i
result,
H
1
iTT.
This
Hi(2Q<ft
* 2
_2_
W*
was
for both small
Diffraction.
* Generalisations obtained
/'J.
~ir^**'
1 \ + l&ry _
cos(2j?+^7t)
27r 8 ' 8
^'2
'
und
a>
1^
(2t)dt
and
large values of x.
The
last integral is of
( 10*7)
in
many
(7) v. (1910),
13'3]
[Note.
(4)
INFINITE INTEGRALS
By
differentiating (1)
393
j / (01ogft--y-log2;
been investigated by Lerch, MonaUhefte
filr
I.
105112.
for functions of the second kind, corresponding to (1), is
The formula
f
provided that
#(i>)|
v it)
dt _
</?(/*- v)<.
when
i>=0.]
13*3.
its
generalisations.
The
(1)
integral formula
|" J
(at)
exp (-
^)
which
will
be
discussed in 14r2. This integral differs from those considered earlier in the chapter by containing the square of the variable in the exponential function.
It is supposed that
|
argp
complex number.
It is equally easy to prove Hankel'sf
(2)
Jo
0*-1 dt
by a
direct method.
To
it
must now be
supposed that J
R((i + v)>0.
To obtain the
result,
we observe
723)
Jo o
is
I^(\a\t).\ex V (-^)\.\t^\dt
J, (at) in powers of
an
integer.
t Math. Ann. vni. (1875), p. 469. See also Gegenbaaer, Wiener Sitzungsberichte, lxzii.
(1876), p. 346.
J This restriction
(0+)
Cf.
/Jo
and integrating term-by-term.
Weber also evaluated
(2)
by expanding
(2),
may
be disregarded
if
we
by the contour
Series, 176.
394
It
f<*>
[CHAP.
Xm
'
\m ( 1 f]\v+2m
ran
Jo
m= orn\r(v
=o
+ m + l)J
)
"
rx r
2/>" +'l + 2m
r (i/ + m + 1
and this
If
is
right in (2),
(3)
transformation
("jriatJexyi-pH^.t^dt
Vr(,ti)
is
ex
P("v)- fl
{i'-to+H'+iiij,)fi
is
an even
In particular, we have
(4)
t> dt
= f^ex?
is
(-
^)
This integral
38;
susceptible to
if
In order that the hypergeometric function on the right in (2) may be Rummer's second transformation (4*42), we take fi=l and, we replace v by 2i>, we then find that
;
(5)
replace v by
v, it is
*
(6)
-"
when
(7)
|
exp
if
("
tan
) I/"
V7r
(*?)
-* 0, (a
7
* Kw ($)
sec
w]
fi
(i)
< '
and,
we make p
00
Z
2v
(at)dt
tan
Jo
u>
vir
,
when
(8)
R (v)
<
%,
by using
7'23
and, in particular,
0.
Jo
rY (t)dt =
13*31]
INFINITE INTEGRALS
(5)
395
III.
Formulae
and
(6)
(London, 1912),
p. 271.
left of (3) is
suggested by Basset,
vm.
128
with the help of Laplace's transformation by Macdonald, Proc. London Math. Soc. xxxv. (1903), pp. 428443; see also Curzon, Proc. London Math. Soc. (2) xm. (1914), pp. 417 440 and Hardy, Trans. Camb. Phil. Soc. xxi. (1912), pp. 10, 27, for formulae obtained by making p2 a pure imaginary.
;
For some applications of the integrals of this section to the Theory of Conduction of Heat, see Rayleigh, Phil. Mag. (6) xxn. (1911), pp. 381396 [Scientific Papers, vi. (1920),
pp. 5164].
13*31.
integral.
The
cussed
to
modify
it
and
if
arg|>
<
\ir,
we thus deduce
J o
that
(&**)
P-
dt
"
rVSW)
/."
il
exp ( "
^
(
" +-' **
^
/*
to unity
when
so that
({ab/p*)"
a2 +b 2 \
[*
fab cos
d>\
?J
If
integral sign,
we
find that
(1)
j\xV (-p*r)J
is
(at)Mbt)tdt
ex V (-^?)/(^).
|
|
This formula
valid if
R (v) > 1
|tt.
Like the result of 13*3, this equation is due to Weber, Journal fUr Math. lxix. (1868), Weber gave a different proof of it, as also did Hankel, Math. Ann. vin. (1875), 470. The proof given here is due to Gegenbauer, Wiener Sitzungsberichte, lxxii. pp. 469
p.
228;
347. Other investigations are due to Sonine, Math. Ann. xvi. (1880), p. 40; Sommerfeld, Konigsberg Dissertation, 1891 Macdonald, Proc. London Math. Soc. xxxv. (1903), p. 438; and Cailler, Mem. de la Soc. Phys. de Geneve, xxxiv. (1902 1905), p. 331. Some physical applications are due to Carslaw, Proc. London Math. Soc. (2), vm. (1910),
(2), (1876), p.
;
pp.
365374.
396
13*32.
[CHAP.
XHI
When the Bessel functions in integrals of the type just considered are not of the same order, it is usually impossible to express the result in any
simple form. The only method of dealing with the most general integral
Jo
is
J"M (at)
dt
and
seems unnecessary to give the result here. v /x, Macdonald* has shewn that the integral
equal to
An
when a = b
if
A.
+ 1* + iA
J.(at)JAat)e^(-pH2 )t^dt^
J
o
^ p^+vr(fi+1)r(v+1)
,
ffi
8
3
+ v+l
2
'
fi
+ 2 \ + n+v
'
'
P+1
>
+ 1 >P + P + 1 ~p)>
>
a*\
now be shewn
f-
that,
when
m K)
to 4/(3ir).
JMMt) at ^
t+>
Jo
xxx. (1882),
This result was obtained by Struve, Mem. de V Acad. Imp. des Set. de St Petersbourg, (7) p. 91, in the special case p = v = l ; the expression on the right is then equal
(v)
In evaluating the integral it is first convenient to suppose that both exceed . It then follows from 33 (7) that
v (t)
R (ft) and
r JMJ +
t
f >
j) coa
_ Q cos2v 2
^ gin
$ sin
^ d0d4>dt
London Math. Soc. xxxv. (1903), p. 440. 9991000. t Wiener Sitzungsberichte, lxxxviii. (1884), pp.
13*32, 13-33]
INFINITE INTEGRALS
fact that tr 2 sin (t sin 6) sin (t sin <) does not
1/tf-
397
exceed numeri-
In view of the
absolutely,
and
sin
sin
<j>,
Since *
00
j*
sin
(t
sin 0) sin
t*
(t
sin
<)
_ ~ I^tt sin
\\ir sin 0,
<j>,
(0
{0
^ >
<f>) <f>)
we
\tr
cos8*"8
cos8
""8
<f>
sin8
Jo Jo
sin <f>d0d<f>
+ $ir\
J
cos8*
-8
cos8"
-8
<-6
sin
sin8
<j>d<j>d0.
we have
cos*4
j J
1)
cos8
"-8
<&
sin
"8
<
= - cos2""
r(j*
1
<f>
cos8*" 8
J
sin"
d0
+
J
cos8*-1
</> .
cos8*"8
<f>
sin8 <-W<
+ y -i)r(p 2IV+i> + *)
J.
The other
/
so
we have
'
Jo
**+'
whence the result stated is evident. The extension over the range of values of fi and v for which merely R (/* + v) > is obtained by the theory of analytic
continuation.
It
may be shewn
Jo
in
(2 )
r HM (QH,(Q Jf
r+>
using 10*45
r(M +*)r(j)
the case
/*
= 1.
i
By
jo
we
find that,
when
(/i)
and
R (v) exceed J,
**+"
(2M -l)(2v-l)
/"
/"**
f|* {l-oos
(<
sin 0)} {1
<*
"2'*+"-*-r( M +i)r(i/+i) Jo Jo
Jo x cos2* -2
cos2
"" 2
<
Now,
if
a and
j8
are positive,
it
398
[CHAP. XIII
that
round a contour consisting of the real axis and a large semicircle above
r
}
d{
Hence the
proving
(1),
triple integral
and consequently
will
(2) is established in
if
the same
way
as
(1).
The reader
3)
this
/;^^=
may
integrals
2*+" 0* + v - 1) r 0* + ) r
(v
+ )
which
and
p and
v for
R (v) > \
and
R(n + v)>l.
The
J^
^ +|>+2
<*'
^++ i
may
13*4.
The
Jo
in which a
and 6 are supposed to be positive to secure convergence at the upper limit, was investigated by Weber, Journal fur Math. lxxv. (1873),
pp. 75
80,
namely,
(ii)
(i)
x=n=
0,
i/
l,
X = -,
/*
= 0, v=$.
The integral was evaluated, for all values of X, p. and v for which it is convergent, by Soninet, Math. Ann. xvi. (1880), pp. 51 52; but he did not examine the integral in very great detail, nor did he lay any stress on the discontinuities which occur when a and 6 become equal. Some years later the integral was investigated very thoroughly by Schafbut his preliminary analysis rests to a somewhat undue extent on the theory of heitlin
J,
The special case in which X=0 was discussed in 1895 by Gubler who used a very elegant transformation of contour integrals unfortunately, however, it seems impossible The analysis in the 0. to adapt Gubler's analysis to the more general case in which X
;
* Some related integrals have been evaluated by Siemon, Programm, Luisemchule, Berlin, 1890 [Jahrbuch ilber die Fortschritte der Math. 1890, p. 341] Wiener Sitzungsberichte, t See also 13 43 in connexion with the researches of Gegenbaaer,
lxxxviii. (2), (1884), pp. 990991. X Math. Ann. xxx. (1887), pp.
(1887), pp.
xlviii.
161178. The question of priority is discussed by Sonine, 582583, and by Schafheitlin, Math. Ann. xxxi. (1888), p. 156. in die Theorie (1897), pp. 3748. See also Graf and Gubler, Einleitung
148.
13*4]
INFINITE INTEGRALS
is
399
13*2.
The first investigation which we shall give The conditions for convergence are*
(R(p.
+ v + l)>R(\)>-l,
(ab)
(a
[R(p+v + l)>R(X)>0,
it
= b)
We
<
a.
The
constants
o, fi,
defined
by the equations
+ 1,
|\
[V
=7-a-
ry=V+l,
=7 1.
down
to the
end of 13*41.
known
Jo
that
r-J>(t)M>>t)
dt=
lim
^W.(H^
t
;
C+0J0
on the right
is
we
replace b
integral
R(z)>0&nd\I(z)\<
now
fy-^-'^zr^'
=
Jo
e
_M J-fi(at)Jy-i(zt) dt
PV
U=o
'
m\T(y + m)
e
. 2 I
So
-ot
w!Ti V(y + m) Jo
(crf )
^p+sm-1
dt>
provided thatf
P
'
m =o wi!r( 7 + m) Jo
is
it is
is
the case
when
z\<c.
Hence, when z
| |
<
c,
J/
*^0
mfo
w!T(7 + m)
(a2
x ,FX (0
* It follows
+ m,
- - m
-+1
^p)
from the asymptotic expansions of the Bessel functions that the conditions
R(p + v + l)>R(\)>-l
are sufficient to secure
/*
-v
is
an odd
integer.
400
[CHAP.
XHI
r(a-/8 + i)r(j)
i_o_ m
,
i.
__*__)
c)
Now
sFiia
+ m^-p-m;
| j
J; x)
(1
do not exceed in absolute value the alternate terms in the expansion of */%)~ A ~ 2m where A is the greater of 2a and 2/S 1 1; and, similarly, the moduli of the terms in the expansion of
,
1
(a
+m+h
(1
fi
>
x)
V)-^-
/V*.
(-.y(|g)T+2m-i
QaY-t r(2a+2m)
(a'
w =o ml
T (7 + in)
+ c )-**
2
l\T(l
+
,
ir(-|)|(i-^)-^--n
|r(|-i9-)r(a + m)|J
where
|
= (^/(a + c But this last series is absolutely convergent when < V(a* + c ) c, and it represents an analytic function of z in this domain.
9 2
a?
).
/,
y
C to
^(m-i(*o A
(-)* (j*)** 8- 1
_ | "
(ja)*-*
T (2a + 2to)
2 a m ,o m!r(7 + m) (a + c )*+T(a-/3 + l)
provided that
satisfies
R(z)>0,
\I(z)\<c,
|*|< v (as +
/
c3
)-c
Now
and take
take
6<v (a + C )-(7,
/
< c < G,
so that also
6<
V(a"
+ d*)-c
13-4]
INFINITE INTEGRALS
in the last integral formula
if
401
Then
done,
we may take z = b, and when this has been we use fonctions majorantes just as before, we find that the resulting
terms
less
m -o
iJ\
+
,
ir (-|)i(i-yx)-^-^-n
where
X =&l(a* + &).
is
|ra-y8-m)r(o + m)|
J'
with respect to
c -*
Hence, by the test of Weierstrass, the original series converges uniformly %c^C, and therefore the limit of the series when c when
the same as the value of the series
when
= 0.
We
c--oJo
&
and therefore
Ja -p(at)Jy -i(bt) d
Jo
"
(a
+ +
(1)
io
to say
^^
*-2r ^a^r^ra-^)-
^' 7 '^'
that
is
provided that < b < a, and that the integral obtained by Sonine and Schafheitlin.
If
that,
is
convergent. This
is
the result
we interchange a and b, and also fi and v, throughout the work, we when < a < b and the integral is convergent, then
["
JO
find
Ja .p(at)Jy -
(3)
pT^
(bt)
at
27--0 &2-y+i
q-gr(q) r (7 - a) T (q - fi +
1)
402
[CHAP. XHI
and
(3) are not
same
function.
when a = 6; and
it will
phenomenon
The
in
some
detail.
13*41.
critical case
we
have,
Jo
assuming that
(//,
to secure convergence.
Now
consider
P=**
***
where z
is
(z) positive.
the integrand in ascending powers of a and integrate term-by- term, this procedure being justified by the fact that
When
convergent.
We
/
Jo
dt
+ 7 + 2m) - +7 + ) m-OJ m! r (a - /3 + m + 1) )a-^+y+2m-i m 2m) r (a p (2a + + 7 + 2m) (_) (| a _ ~~ l *""+"" m)V(a-/3 + y + m)' mir(a-fi + m + l)T(y + m =o
.
ft
^a-Hm-1
p (g (a
F (7+ m) T
Now
and so
left is
an analytic function of
c, is
when
R (z) > 0,
its value,
when
tions
But, by Barnes' theory*, the series on the right and may be represented by the integral
its
analytic continua-
J_ (
and
xi
27rJ_ aci
(ia)--*-*-*
41 "- 1
+ r(a-/tf
^
|
analytic
tr.
It is supposed that the contour consists of the imaginary axis with loops to ensure that the poles of T (- s) lie on the right of the contour, while the poles of T (2a + 2s) and of T (a - /9 + 7 + 2s) lie on the left of the contour.
evaluate the integral by modifying the contour so as to enclose the poles on the left of the contour and evaluating the residues
When
\z
< 2a we may
* Proc.
London Math.
Soc.
(2) v.
(1907), pp.
59118. See
13*41]
at them.
INFINITE INTEGRALS
The sum
403
in ascending powers of z
hence,
when
R (z) >
/"
Jo
Ja-
fi
(at)Jy^{at)
_1 ~2 wl
1
(_)m(^ a
rn\
Y -a-g-m-i
T(l
(-) (^a)- w
Now
c
22 (7
a ft) > 0,
c
and
so,
-*
0,
we deduce
1
n (1)
.
fJ
Ja-,(at)Jy.
F=^
(at)
aC
"2r(l-/3)r(7-a)r( 7 -^)
provided that
i2 (a)
> 0,
R (7 - a - ft) > 0.
7;
is
1),
there
is
therefore no discontinuity
The
(2)
result
may be
^^dt
qq)*- 1 r (\) r (\n + \v - $\ + ^)
provided that
If
/a
is
We
place of
shall
ft
make a change
i>
by writing
if
and
in
__]_[-* "2wt.L.
"
(|o ) to+M
i
,+
*- A r(o
(-)**
-1
_1
(la^-"
1
"1 w *
r(-X-i)r(a + im)
404
[CHAP. XIII
and hence
w
unless
aK
2^r(J5 +
\ = 0. This should be compared with the more obtained from 13*4, namely that, when b < a,
f"
(4)
general formulae
Jo
2* a*-r-K r(a-p)
U+l) a*-^r(a-p)r(p T (p + 1\ + 1)
> a,
(aQ Ja-p-i (bt )
dt
* Fl
~P
'
a>)
'
and,
when
(5)
Jo
/+P
a+JT(a-iX)
= 2>6^-^r (a
Since X
4 + i)r(ix- P
when X
(| a
is
'
)
p / y -(
tt
-^
j.
i'
n - j. +1 iX; tt+y +
,
as \
1;
p)-
when a
^ 0,
the functions on the right in (4) and (5) do not tend to limits
b.
On
becomes
[**
).+-i
jr (2a + 2s)p r (-
s)
=a
is
( )*/(2a).
(6)
Jo
J. +p (a*) J.^-x
(bt) dt
-I
(- W(2 aX
U
according as 6
<
a,
= a, 6 > a.
Since
\
=a
is the
mean of its
limits
when
and
= a + 0.
The
( 7)
fy-coM^- ^j:-/'
is also easily
which
te
Amsterdam,
116.
iv. (1902),
pp.
102103; Archives
103
13-42]
13*42.
INFINITE INTEGRALS
Special cases of the discontinuous integral.
special cases of interest are obtained
ft,
405
Numerous
by giving
special values
to the constants X,
To
when two values only are given, the first is the value for b ^ a, the second for b > a and the values are correct for all values of the constants which make the integrals convergent. The following are the most important special cases * J^at)j^bt) a (b/ay/fj,, ai (1) [R(ri>0] t Jo (Ha/W/*the third for 6
;
a, and
>a
Jp. (at)
sin bt
sin{/*arcsin(6/a)},
dt
(2)
F Jo
JM (at) cos bt
Jo
=
[fi {b
a* sin \iltt
[R(fj,)>-1]
2
+ V(6 2 - a )}"
{fi
i> _1 cos
dt
(3)
=
I
a cos
11
[R(ri>0]
[fi{b+^(b*~a2)}'
sin
,,
\fjj
v( 2
(4)
Jo
or 0,
[B( H.)>-2]
aM COS
|/47T
cos
,
(5)
|
{/j,
V(a2 -62)
-(
1
<x>
or 0,
a**
[(m)>-i]
sin ai7t
/
-'o
V(6
-a
(6,
r Jo (at) sin
7 (a) cos
0,
fa
dt
=
-j
/V(6
-a
2
).
(?)
bt.dt
fiM =
-: ,
-n
fiir
Jo
la
Math. lxxv. (1873),
they are associated with the problem of
are
known
The
Numerous other special cases are given by Nielsen, Ann. di Mat. (3) xiv. (1908), pp. 82 90. and (5) diverge for certain values of n when a = b. t The former was known to Stokes many years earlier, and was, in fact, set by him as a Smith's prize examination question in Feb. 1853. [Math, and Phys. Papers, v. (1905), p. 319.]
integrals in (4)
406
[CHAP.
XHI
(8)
(at)
JM_, (bt) dt =
1/(26),
[R (fi) >
0]
(0;
and
if
we put
fi
= 1, we
f
(ibid., p. 80),
(9)
Jo (at)
(bt)dt
1/(26),
The
article
result of putting
/*
= in
(8) is
known
Some
other special formulae have been found useful in the theory of Fourier series by
Leipziger Berichte, lxiii. (1911), pp. 369
(5)
W. H. Young,
Math,
387.
v
und
> 0,
we
have, by 5*51
(5),
<n
J\ + n {x) = 2v
J^(t)~
.'o
= 1,
and so
(10)
I
.W |<1,
is
|^ +
(*)j<l/V2,
an interesting generalisation of Hansen's inequality which was discovered by Lommel, Miinchener Abh. xv. (1886), pp. 548 549.
find it interesting to
jV (aO{l-/oW)Y =
In the special case in which the Bessel functions are of the same order, Gegenbauerf found that by his method Weber's integral could be evaluated
in a simple manner.
If
have
f"
(\oby
"T
+ 1) r (J) J o Jo t (a* + fc*-2a& cos <*>)*" (ab)" T (v - \\ + ) [" f* sin " <f>d<f> 2 -2 2* T (v + \) F (I) r (\ + ) J (a +6 2a& cos 0)"-*A +i Jo(a* o
(v
"
f"
Jv
[t
V(q
+ 6 - 2ab cos
2 2"
<fr)|
Sm .,.,, Wdt
.
(2),
(1884), p. 991.
13-43]
INFINITE INTEGRALS
407
is
by
13"22.
b"
/b\ n C w
v-^r( + i)rft)raui)
J* {(i
.?.
U)
J.
tv- (*-*
Now
-*)-*(V(,)}
= (n +
C^
(5),
^
we
see that
(n
{(1
- *p+ Cs(t)} - - ( + 1)
-*)*-*
l)!"
.'
(1
CW*)***
(1
-l
=J
*)'-*-*+*
-^ s|
1
(1
-'^ G
(n + 2p-2v-l)(n
+ 2n-l)
r1
M
^
;
dz
2v
so that
+n+ l
>J
ii-*y-kC*n-i(s)dz,
0+i (cos r Jo
<f>)
(n
Hence
it
..
follows that
f Jo
and
this agrees
The method given here is substantially the same as Gegenbauer's but he used slightly more complicated analysis in order to avoid the necessity of appealing to the theory of analytic continuation to establish the result over
the more extended range jK (2v
1)
1.
By expanding
(I)
<f>,
Cj(a\J
wrfr-fo+j)
*M
(2v + 41 - A.
'
2v
3- \ +i
;
v+1
4a 262
>@TFf)-
which is valid whether a > b or a < b. This result was given by Gegenbauer, and with this form of the result the discontinuity is masked.
to
examine the
critical case
obtained
408
13*44.
[CHAP. XIII
The
integral
I
J^(at)Jv
(bt)dt
to Gubler*. It
is
Jo
will
first
now be
investigated by the
method due
convenient
[
.'o
J.(at)Jv (bt)
t
dt
even though this integral cannot be evaluated in a simple manner by Gubler's (v) > 0, R (X) > \, R (/* - X) > - 1 and, methods. It is first supposed that
as usual,
a and
b are positive,
and a >
b.
From
(2), it is
evident
equal to
We
and the
Fig. 29.
line R(z),
only at z
i;
and then,
and
under
consideration,
R\hbt(z-l/z)}^0;
and the repeated integral converges absolutely, since
K
Jo
is
t
I
J-
convergent.
The order
of the integrations
may
therefore be changed,
and
we have
r J(at)JAb t)^_ dt
Jo
If
f<
>
f"
]
Jy. (at)
~2frij. a>
exp
H-
z-*-1 dtdz.
;)}
we
write
b(z-l/s)
*
a(e-l/Q,
3748.
Math. Ann.
13-44]
INFINITE INTEGRALS
is
409
j
taken
x
for
which
^ 1, we
have,
by
132,
dt
_ j_ ~ 2*i j _.
/>+>
Qay r{p-\ + i)
[\a (i
+ 1/0}*-"-'
V(ji
+ l)
rv + i)
x
-ji^
f/x,
.'-.({T+i/cr^
4,1
- A. + 1, /* + X;
/it
+ 1; i^r) d*>
by Kuramer's transformation*.
Next consider the path described by , when z describes its contour. Since the value of with the greater modulus is chosen, the path is the curve on the right of the circle in Fig. 29; and the curve is irreducible because different branches of z, qua function of , are taken on the different parts of it. The
curve meets the unit circle only at e iu> where
,
a> is
a sin m.
Now
functions of
both the original integral and the final contour integral are analytic \ when (X) > 1, so long as a b. Hence we may takef X =0,
provided that
R R (fi) > 1
_ br + a ~r(aT+6)'
t (6t
6
ttr
+ a) +6
'
and the r contour is that shewn in Fig. 30; it starts from - b/a, encircles the origin clockwise, and returns to b/a) where the contour crosses the positive half of the real axis, we have arg t = 0.
_.
dz
Since
z(l
+ ?)
we
[ ^ J,, (at)
J o
(0+
'
( J
t**"-*-1 (6t
'
+ a)-it"+*+l
l
>
(ar
&)*<"*- dr
b/a
fi+)
Ima**
, 1
W *(-H-i)
formula
(1
\
] _x
+-, , a2 J
b2
Wc+^+d
(1
+ W )*C+*-i)
,
,
dfO.
* Journal fur
(1908), pp.
Math. 115119.
(57).
t If
X^fcO,
14
410
[chap,
xm
If we expand in ascending powers of b*/a? and substitute the values of the Euler-Pochhammer integrals, then Gubler's result
J^
I
(at)
Jv (bt) dt
yrUM + iv + i)
is
Lfi+ v
+1
n+l
manifest.
Fig. 30.
13*45.
The
integral
*'<&*<& *.
\ j |
which converges if R (a) > I (b) and R (v + 1 X) > R (fi) is expressible in terms of hypergeometric functions, like the Weber-Schafheitlin integral, but unlike that integral it has no discontinuity when a = b.
,
\
To
I
evaluate
j
it,
expand
(bt) in
powers of
it is
b,
< a
j
convergent
By
using 13 21 (8)
-
found that
Jo
MV ' + n+l)J b"Y(\v-\X + \it + \)T(\v-\\-\. + M+l nv A+l r> + i> V \4-/4+l V \ fJb+1 V+l ^^
-ow!l (v
\)
'
a*)
and, in particular,
(2)
/;
(1)
K^MH^U^^L^lR,
|
|
provided that
R (v + 1) > R (/*)
and
R (a) >
I (b)
/j,
Formula
=v=
and \
is
and
1.
* Electromagnetic
13-46, 13*46]
13*46.
INFINITE INTEGRALS
411
To obtain the
Mat)J
t
(ot) dtt
replace 6 by VX& 2 + c2 26c cos <), where b and c are positive, multiply sin 2 " <f>/(b2 + c2 - 26c cos <f>) iv and integrate. It is thus found that
f
l
by
,/M (a)
J, (bt)Jv (ct)
^
if
^-r
2
(\bc)"
(,
f-
+ i)r(i)Jo
J ^ ^"^~ sin io
f
(at)
( rt)
****
absolutely
where
-Er
is
convergent
\ -r v + 1 = /*,
by the formulae
( a>tsr )
J, o
1-x"^-- 1
.:.-i,.,fw..
.a
It follows that
/
(i)
Jo
*6
4W-TTrp7n
A
is
V
2
sinS
^^
62
0,
arc cos
+ c -a
^r Zoc
2
>
t
numbers
according as a2
is less
(6-c) ,(6
provided that both
+ c) exceed |.
2
In particular
(2)
/J
%
,
Multiply by a" +1 and differentiate under the integral sign with respect to and we then obtain the interesting result that, if R (v) > \,
f
(3)
J
o
(at)
Jv (bt) J
(ct)
=
dt
2" -1
(a6c)
>
when
a, 6, c
if a, 6, c
This formula is due to Sonine, Math. Ann. xvi. (1880), p. 46 other aspects of it have been investigated by Dougall, Proc. Edinburgh Math. Soc. xxxvu. (1919), pp. 3347.
;
412
[CHAP. XIII
It has been observed by Macdonald* that the integral on the left in (1) is always expressible in terms of Legendre functions. The expression may be derived from the integral on the right in the following manner:
When
we have
(a2
.'
a, b, c
= sin \A
d<f>
sin 6
- 62 - c + 2 be cos 0)*-"2
1
sin2 "
<f>
= (26c)*-"-
(cos Jo
<f>
- cos Ay-"'r
sin2"
(j>d<j>
^^
Jo
=
and
4*-'
(6c)*
sin*-* jj.
(/a)
T{V
rfJ+%~
(v)
therefore, if iJ
and
exceed
triangle,
we have
(4)
J"
J.
(at)
M P ^(
cos
)-
If,
however, a2
>
+ cf,
a?
and we write
- 6 - c = 26c cosh ^,
2 2
we have
Jo
2 [ (a
- 6 2 - c2 +
26c cos
<f>)*-"
= (26c)*-"- 1
J
(cosh
= (26ccosh^-"- 1
l,
Ii^^i>
2
^),
when a >
2
(b
+ cf, we
have
-1 COS
I>7T
.
C* t
sinh*
*^ ~i-M
Q
,(cosh<#).
In
/n\ (6)
like
j^
f rr
Jo
j-i <
0* COS
Z>7T
(X2
6,
- l) ** *
(brf(hey^ An{
c
;+9)w
may
0,., (A )
^* + *
irs
where 26c
X=a
-f b-
+c
2
;
and in
this formula a,
be complex,
R (a ib ic)
are positive
;
due
to Macdonald.
London Math.
142149.
13-46]
[Note.
see 5 71.]
-
INFINITE INTEGRALS
The apparent discrepancy between these formulae and the formulae
is
413
of Mac-
donald's paper
Qn m
formula
Other formulae involving three Bessel functions may be obtained by taking 11*6 (1), replacing z by x, multiplying by
2Jp (xcos$)/xk
and integrating.
It
is
0)
J ( X C0S
^ C0S ^) J i X S * n
(f>
S^n
^) Jp ( X C0S
cosp 6
^) ^A^l
cos*1
cos*1
<3>
sin"
< sin"<f>
u==
oL
nir(/* + n
l)
+lv-ip + \\ + n + l) X u+ v + p pX u v p + n + 1, ^ ^P .^,(
^
+, 1
cos2 6\
2 J*,
( n,
p.
+v+n+1
i/
*>
+1
sin 2
<j>)
x-^-w,
when
and cos 6
Some
is
/a
+ + n + l;
have been given by Gegenbauer in a letter to Kapteyn, van Wet. te Amsterdam, iv. (1902), pp. 584 588.
Some
If cti,
it is
a2
. . .
extensions of formula (3) have been given recently by Nicholson*. am are positive numbers arranged in descending order of magnitude
if
ctj
> 2 am
m
II
R(v)>1,
ft* dt
then
/
oo
(8)
J J
n=l
Jv {an t)-^^0;
is
by induction, by substituting
Gegenbauer's formula of
1
11*41
R(v)> |]
for
Jv (a m^ t) Jv (am t), and then changing the order of the integrations. When a a ... am are such that they can be the lengths of the sides of a polygon, the integral is intractable unless ra = 3 (the case already considered), or m = 4.
lt
2
,
329.
Some
in 13-48.
w
THEORY OF BESSEL FUNCTIONS
Ox, a.2 ,
414
[CHAP. XIII
When
16 A 8
so that
IT (a1
+ aa + a + ai -2a n
s
),
is
simplest
first
to obtain
for v
= 0;
+ a4
2
>
fla+Gts.
.
according as
2 2
A ^
Oj
as a 8 a 4
<f>,
and
replace
(a 2 t)
{a 3 t)
by
ra
J
a rr JJ/'^^-r^T^W.
t in
dt
Jb^
Jv ((M) J (<M) J*
i.
^**
> * *
^ #
Gtj
or a2
+a
We
write
- (g - a y __ (a! + a y - (a - a ) w - (a a) (a, + a ) (a 2 a
sr
2
;
2 ,
2
so that the
upper limit
for
a;
is 1
or
A/^(a
a.i
a3 a i ):
called 1/k.
We now carry out the process of analytic continuation (unless a, -f a4 =a 2 a3 when the integrals diverge at the upper limit if v 0), and we get
-!-
f"
nJ
n=l
-..\
(a n t)tdt
[ [{(!
-1 or
+ a*) - '}
2
1/fc
{**
- (i - *) [^ - ( - 3>
2
1
2
}
{(a,
+a
3)
-st
}]-*
Wsr
-
(9)
^.
V{(1
- # ) (1 - k2 a?)\
2
Hence
/_1_ ^.
fo
.'0
/
V
nJ
n=l
(a n t) tdt
7T
l7rV(G-iaa4)
v #
W(<
first
where If denotes the complete elliptic integral of the whose modulus is less than unity is to be taken.
* For other values of v f
it is
kind,
We
still
13-47]
INFINITE INTEGRALS
415
when
R (v) >
is to
so that it is equal to
^Wi^Wm-h
d hence*
(1
_^x>,-* S * )] "' +C
,
no)
r \j (atv*
*- -
a2
""sr
r ^>
13*47.
Several discontinuous integrals, of a more general character than the WeberSchafheitlin type, have been investigated by Sonine f and Gegenbauer J; modifications of these integrals are of importance in physical problems.
some
The
first
example which we
shall take is
f0,
(a<b)
ls))
=
To
if
|!^>p-v_iW(a! _
(o>J)
secure convergence, a and b are taken to be positive and R(v) >R (/*)> 1 a = 6, then we take R (v) > R (fi + 1) > 0. The number z is an unrestricted complex number, and the integral reduces to a case of the Weber-Schafheitlin
integral
when
0,
is zero.
The
that, if
we
c>
then
Jo
MV
1
J (<
+ *-)*"
J- (bt) (' +
'
Too /YM-ooi
- ssJ.
b-
L,
f
""""' exp
f"
- t* + z-\~\ dudt
)J
,
L* l"
az*~]
The
c+coi
r(a?b*)u
An
corrected.
R (v)
between and is obtained by analytic continuation. t Math. Ann. xvi. (1880), p. 38 et teq. X Wiener Sitzungsberichte, lxxxviii. (1884), pp. 990 1003. phys. de Geneve, xxxiv. This formula is also investigated by Cailler, Mem. de la Soc. de
(19021905), pp. 348349. The convergence is absolute only when R (y)>R (m + 1)>0; for values of this condition, the formula is to be established by analytic continuation.
||
not covered by
416
[CHAP. XIII
When a < b the contour involved in the last integral maybe deformed into an
indefinitely great semicircle
axis,
is
zero; but,
6*2 (8),
A related
(2)
integral
/V,(
may be
that in evaluating (/*) > 1 convenient to suppose that jarg.2 <^7r, though we may subsequently extend the range of values of z to |arg^|<^7r by analytic
the integral
continuation.
From
2J
left
of (2)
is
equal to
- a ( u + -^^-)
dudt
by
6*22 (8);
and
Now make
arg z-*-
If
we put
= iy,
where y >
0,
we
find that
)rrre-i<--
V(tf
+ ) p-'
;
^^
[y
v(a2+ 6a)|]
and it is supposed that the path of integration avoids an indentation above the singular point, and that given to V(<2 y2 ) which makes the expression positive when
1
<
= y by
t>y.
If
axis
we had put z - iy, we should have had the indentation below the and the sign of i would have been changed throughout (3).
real
In particular
(4)
Jo<pt)
Jo
(?=59
is
V(a8
+6)
S
Lamb,
Proc.
London Math.
pp. 122141.
t
With certain
limitations,
a and b
may
be complex.
13-47]
INFINITE INTEGRALS
last
417
The
formula (with the lower sign*) has been used in physical investigations by
Somnierfeld, Ann. der Physik ttnd Chemie, (4) xxviii. (1909), pp.
682683;
see also
p. 72.
we
divide by
&**
provided that a
>
and
this
established independently
Similarly, from (2)
we have
if
>
and
i2 (ji)
>
i/
1.
2i>,
In (5) replace
follows that
by
3
a by 2
sin
to 6
= 7r.
.
It
m {)
The
r ^MV(<
Jo
+ ^)
~W+W~
when
is
'o
this is valid
22 (v
|)
1.
case of interest
(8)
when
2p
= 2/a + 3,
f;^^^=^H,<
;
so that
(9)
j;3g(*--r**-J&$*.<w;
and these are valid if R (v) > \. The last formula was established in a different manner (when i>=l) by Struvef and from it we deduce the important theorem that J, when v > $ and x>0, H(#) is positive. Struve's integral is of considerable value in the Theory of Diffraction.
Some
by
multiplying by b
to
b.
^m+i f Jo
(bt)
^ + ^
t*
dt
a,
whichever
is
the smaller.
My thanks are
me
418
If b
[CHAP.
XIH
a,
when
>
we
put u
a sin
provided that
this is
to
If we replace a by u in (1) and then take a ^ b and integrate with respect u from a to oo after dividing by u v ~ we find that, when z is restricted to be
l
,
positive,
(w2
Too
foo
+6
2 )''
9l-l
Tjji
= r 7w y ^-x
1 (v) z
**
^"^'-'1 ^-M-i(^)exp{-<2
Jo
(^+62 )}rft;^
Jo
/
aia
z*
jo
by
13 3
(4),
(11)
Jo
provided that a
positive
M+1 = 6 2> -! r (y) *m ( *)> + ^ )iy+1 * and i (v + 2) > R (/a) > - 1 the restriction
(ff
;
that z
is
removed.
Formula
(12)
J M ()
Jo
(
^+
^)iv
at-
gv
where i2 (v + 2) > .R (/*) > and 6 > a, has been generalised in two ways by Gegenbauer*, by the usual methods of substituting Neumann's integral and
Gegenbauer's integral
(cf.
13-1) for
The
first
method gives
2*- i r(/*)
6"
(az)Ji(az)
zK+v
provided that b
*
> 2a and
R (v + X + f ) > R (/*)
> 0.
pp. 10021003.
(2), (1884),
'
13*48]
If
2
-or
INFINITE INTEGRALS
419
<f>),
(14)
Jo
P J*W
(?Tfy
at
_2*- T(ji)
1
(az)
(cz)
if 6
> a + c and
i2 (2i/
it
+ f ) > i2 ( > 0.
> 2a,
By
(15)
induction
follows that, if b
/o
J.M
v+ ,r.
method
~*
a,
is
M = -5^-' 7 * L
and
R(nv + ln + %)>R(n)>0.
If the induction of the second
first
method
once,
we
find
still
further generalisations.
The
(16)
when
z--0 is
nv - x
iy,,.{bt)u[JAt)-\r-
*- 2
k)
?[f^Tj]
this has
dam,
Sci.,
te
Amster-
two-dimen-
sional displacements)
as follows
" A man starts from a point and walks a distance a in a straight line he then turns through any angle whatever and walks a distance a in a second straight line. He repeats this process n times.
;
between r and
"I require the probability that after these n stretches he r+Sr from his starting point, 0."
is
at a distance
generalised form of the problem, in which the stretches may be taken the help of to be unequal, say (h, 2 , -.an, has been solved by Kluyverf with subsequently the discontinuous integrals which were discussed in 13*42; and
The
RayleighJ gave the full details of the analysis of the problem (which had been examined somewhat briefly by Kluyver), and then obtained the solution of the corresponding problem for flights in three dimensions.
If sm
is
. . . ,
am (m =
1, 2,
. . . ,
n - 1), and
if
6m
is
the
* Nature, lxxh. (1905), pp. 294, 342 (see also p. 318); Drapers' Company Research Memoirs, Biometric Series, in. (1906). Wet. te Amsterdam, vm. (1906), pp. 341350. t Proc. Section of Sci., K. Akad. van 321347. [Scientific Papers, vi. (1920), pp. 604626.] xxxvn. (1919), pp. % Phil. Mag. (6)
420
[CHAP. XIII
all
angle between sm and a m+1 then, in the two-dimensional problem, the angle m between ir and it are equally probable.
values of
Now
let
Pn
{f",
a lt
an ) denote the probability that after n stretches less than r, so that the probability between r and r + Br is
2, ...,
dP n (r;
It is then evident that
a 1} a2 ...,an ) dr
,
Pn (r;
a,,
a2)
. .
a)
=
J
... /
where lf 2 ...,0n -z assume all values between assume only such values as make*
,
it and
while
n_ 1 is to
for
lt
...
0n_z
Now
1342)
rf^(r)j;(.)*=n'
and
so,
<'>
iy tuple
integral, the
may be taken
to be
ir, tr).
We
n_1
and
t,
and,
remembering that
n_lt
we get
r\
J
vJ
(rt)Jo
(s n
= 2irrf
Jo
J, (rt)
(sn -x f)
(an t) dt
by
11-41 (16).
We
next
8
make
2
the substitution
n_2
,
S _i
and perform the integration with respect to cess we deduce ultimately that
-Pn (^
J
By
a1}
Og,
. .
an )
=r
r
Jo
w-l
and
In
problem
it is
no longer the case that all values of m are equally likely. which m is regarded as a co-latitude) are
varies from
to
tr.
~2
m d0m and m
,
factor
to the
remembered that tm
is a
#m-i
13-49]
If
less
INFINITE INTEGRALS
P(r
r,
;
421
is
ay, 02,
than
we
,
a \p) denotes the probability that the final distance deduce, as before, that
...
,
rtteL-pf '('...
~ 2) r ($))
.'c.'o
I" VVP
JoJnt-l
x
n\in*->em .dede^...d0,de
l,
where the integration with respect to 6n- extends over the values of which make sn < r. The discontinuous factor which we now introduce is
n-i
^J.^^-SS^^Io,
and then, since by 11*41 (16),
1
(sn
>r)
^p wr+e^de^ - r
infer that
1
(*p
- *) r <*) -^-^r-
W)b~
>
we
P.(r; ,a if
...,a m
|p)-r{r<fc^
a are all equal to a, and w is large, we may When the displacements a^a^ approximate to the value of the integral by Laplace's* process. The important part of the integrand is the part for which t is small, and, for such values of t,
*
so that (13-3)
Hat)*'-1
aH\
Pn (r; a, a,
....
a \p)
f^/'^O^^Oexp (- ^) dt
w,WT ^ 'TV^)&'^
1
'(ip
This process of approximation has the cases p = 2, p = 3, while Pearson has published various arithmetical tables
connected with the problem.
13*49.
''^
The
integral
J{a( + )lfr,lft(C+)
(*
rff
+ *)*
and
(? +
i2
(ji
*)"
is
convergent
if
a and
6 are positive
+ v) > 1 when
;
a = b the
last
R (/a + v) > 0.
The special case of the integral in which /*=0, v=$ has been investigated by Gallop, Quarterly Journal, xxi. (1886), pp. 232234; and the case in which a=b has been investigated by Hardy, Proc. London Math. Soc. (2) vn. (1909), pp. 469. The integral is obviously to be associated with the discontinuous integrals of Weber and Schafheitlin.
* La thiorie analytique de$ ProbabiliUs (Paris, 1812), chapter in. The process recognised as a somewhat disguised form of the method of steepest descents.
may
be
422
[CHAP. XIII
To evaluate the integral in the general Hardy is effective; suppose that a^b, and
-ft
method discovered by
R(v)> ,
(/*)> so
tna
Write
may be substituted for the second the integral's which will be used are absolutely conin place of t + f, and let z - = Z, so that the integral to be
t;
Poisson's integral
all
evaluated becomes
fi_co
J{a(Z+t)}Jp (bt)
(Z+ty
v
(46)'
=
'
f
f
00
f'J^[a(Z + t)\
r ( ,+W(4j ,L
(kby
L
r
z^ty
(
co
^ btc08 ^ sm'^ d^ dt
cos
r( y +i)r(f)J
I
J o
~*^ cos
~V
r*
(at)
{6 ( *
"^
sin2v
]
*****
sin2v
= r(y + 4)ra) J
2 (4&y
cos
* d * dt
***
- (2>ro.7i)i>+i) J.
by a
special case of 13'4 (2).
is
(a*
^
=\
t
This integral
expressible in a simple
a case
considered by Gallop, or
when a = b, the
two results
We
(1)
f"
sma(z + t)
Jo {H) dt
= wj
f
(bz)>
(6
^ a)
/ON
sina(^
+ )
r ,,,.,.
cosw^.dM
.,
r^+^r^ + i)
by integrating
dt
it
r(^ + y )r(j)
The reader
z+t
(bt)
round the contour formed by the real axis and an indefinitely great semicircle above has to be supposed that there is an indentation at z when z is real.
it
The
integral
To
evaluate
it,
we observe that
+t
z+t'
13-5]
INFINITE INTEGRALS
423
and
so the integral
may be
P
J -oo
f- sin a
(z
+ t)} J
sin
dt+ f JO
sin
a (z +
f
.'O
~
1)
(bt) dt
f.'
. ( + 1)
?
-oo
r oo
_2
sino(*-M)
Z
cos
+t
rf<
= 2 cos as
JO
ra
roo
sin ai Jo (&0
^+
/"
oo
w (* +
^o (&0 ftdw
'
.'
- oo
2s
'o .'o
/oo
cos
m (z +
Jo (&0 dtdu
ra
/>
/
= 2 cos az
Jo
sin at
{bt) dt
+ 2z
sin
ttz
sin ut
(bt)
dtdu.
Jo Jo
(4)
\t\sma(z + l
t)
Jo (&0
,,^-n eft
2cosaz
-77^
rrt
2*
sinw^ -
j--
du
2 COS a^ =
V(a
+ Zz
b-)
Jo
but,
when a <
6,
!
(o)
J_
'-J (bt)dt
= 0.
large
number
by considering
(az)dz,
supposed that
<f>(z) is
an algebraic
and that a
is positive.
We
take the
first
type
when
upper half-plane; the contour is (z) has no with its centre at the origin, axis real above the semicircle taken to be a large at the origin) which joins (indented axis real together with that part of the
singularities except poles in the
We
when
<f>
(z)
type by inserting loops starting from and ending at the indentation, one loop passing round each branch point, so that the integrand has no singularity inside the contour.
plane; the contour
(cf.
131) which
is effective
is
in evaluating
Bessel function one of the integrals discussed in 65, and change the order of
424
;
[CHAP. XIII
where 8
an
converges absolutely
when the
produces no theoretical
13*51.
difficulties.
Before Hankel investigated the more abstruse integrals which will be discussed in Chapter xiv, he evaluated a large class of definite integrals* by considering
1
f
zo^H^jaz)
first
is
and
\R(v)\<R(p)<2m + ^.
The first inequality secures the convergence of the integral when the radius of the indentation tends to zero; and (as a consequence of Jordan's lemma) the second inequality ensures that the integral round the large semicircle
tends to zero as the radius tends to infinity.
r.
It
{Hv u
(ax)
- e* JT,w
(axe**)}
ZiriJo
(x'-^)m^
X_
J_
f<"+>
^HV(az)dz
(^_ r2^m+i
, "
l
j_
47riJ
p tiwwes.
(_ r2)+i
(5) that
2m, J
_j__ (A 2.m!\rfW
r ,-, B
(tfr
(ar)] Kar'
It follows from
362
(1)
["[(1
J o
Y. (ax)]
-^
xf>~*dx
This result can be expressed in a neater form by writing r = ik, so that (k) > 0. It is thus found thatf
(2)
J"
[cos
(p
- v) tt Yv (ax)]
.
^
\tn
\M-i
-n(ra)
* Hankel's
investigation of the integral with
Bessel' schen
'.(*
work was published posthumously, Math. Ann. vm. (1875), pp. 458 461. A partial v=n, p=2n + 2, m=2n was given bj Neumann, Theorie der
Functionen (Leipzig, 1867), p. 58. t The evaluation of integrals of this character which contain only one of the two Bessel
is effected
functions
in 13*6.
13'5l]
INFINITE INTEGRALS
notice the following special cases of this formula
J (or)}
425
{cos
J^
^-^^ = l-^i-jZ
A' _ m (ak)
.
2 ~(^+i^ + i~~ The former is valid when 2m - f < 72 < and the latter when - 1 < R (v < 2m + 1 For an extension of (4) to the case when m is not an integer, see $ 13*6 (2).
(4)
(i/)
The
...
special formula
/""
(5)
Jo
~^^f
xJ (ax)dx
(a*),
xvm.
(1881), p. 194,
dynamics, n. (Cambridge, 1889), p. 19; while Nicholson, Quarterly Journal, xlii. (1911), p. 220, has obtained another special formula
} o ( ax ) (6)
dx
(ak)
/:
x* + i:*
k
integrals.
'
Some
though
integrals
it is
most convenient
resembling those just given may be established here, to prove them without using Cauchy's theorem.
(ax) dx
x*
- ,
+ k*
~Wo
1 [i*
Jo
aF+T*
fCJo
= ~{I
by
10*4 (11),
,.
(ak)--L
(ak)\
>
R (k) are
IT
,
Jo(ax)dx
;xi
More
generally, if i? (v)
> , we have
Iv (ak) - L (ak) =
and
since,
T ( v + i)Y(l) ll"
W
*""
by a
_
~r A?
ft*
(2fc)
6~
oo
< 2 and a
is
j_.
~^
,
= ^r
Jo
[(1 H-
*-)
^ (a*) +
(1
- *"*) H (a*)]
426
[CHAP. XIII
and
(8)
[cos \vtt
0,
(ax)
sin \vir
H (ax)]
where a >
R (k) >
and
-\<R(v)< 2.
is
The change
somewhat
similar integral
f
xv
(ax)
dx
'
Jo
x?
k*
which converges
If
we choose
Jo
by
616
v
(1),
rx K
{ax)dx
_ Y(v + l)
+#
r(i)
Jo io
#)*>
8 )-**
7r2A;,
'~ 1
4 COS
{H_ (ak)
V7T
- F_, (a&)},
when we use
1041
(3).
Hence, when
R(v)> ^,
and
therefore,
when
f" Jo
.R(i>)<,
Kv
(10) v '
_
0)
vtt
{H (ak) '
{
v (ak)}. v Ji
The
integral
f ^
J -^
v
(ax) dx
R (v) >
R (k)
2
Jo^+^"^"T(i + i)r(i)Jo./o
=
and so
(
r <**(**) dx _
gay
+ k*
sin
ViWd*
* Wiener Sitzungsberichte, lxxii. Gegenbauer's result is incorrect because (2), (1876), p. 349. he omitted to insert the term - X, v (ak) and consequently the results which he deduced from his formula are also incorrect. A similar error was made by Basset, Proc. Camb. Phil. Soc. vi. (1889),
;
p.
11.
pp. 422
The
424.
correct
result
Vierteljahnschrift,
xlvii.
(1902),
13-52]
INFINITE INTEGRALS
\
427
may now be
R (v) > f
which
by analytic continuation.
be evaluated in the form of a series by this method
is
An
integral
may
psinha^
J
sinh irx
;
which
is
it is
supposed that
\R(a)\
-1.
By
taking
round the contour used in this section, we find that the definite integral
times the
sum
of the residues of
sinh
at the points
i,
tt2
It follows that
(12)
r*^^J (bx)a?*
J o
sinh irx
'
dx=it
na.Kv {nb).
=i
The
An integral
und
13*52.
integral.
27TIJ
This differs from Hankel's integral in containing the (complex) number ft The conditions for convergence (with the second
135) are*
a>0,
The contour
Fig. 31,
\R(v)\<R(p)<2R(H) + i.
is chosen with a loop to exclude the point ik, as shewn in and then there are no poles inside the contour; and the integral round
Hence
;"
I
\H
to
(ax)
(p-*W H
w (axe ni W
dr
re-1
tik+) Z9-iH
(z*
v
-JLf
~~
w(az)dz
27riJo
+ k*)*+1
\irj_
Now
2m iviJo
(*
+#r +1
As
= lg(Jp -()"
* we
r^p-^r^ + i)-
in 13-51,
take R(k)>0.
428
[CHAP. XIII
H
x
(1)
(az) in ascending
J^
w-eKp-r-a^w
sinv7r.
r f,,
V2
r(/*+ +
"/ '
2
m=0
1) L
J+ ^+
W
and therefore
00
'
(!)
Jo
[cos (/>
x '~ dx
l
1
TT^rP-aM-a
2 sin vrr
F (/* 4-
1)
"^
.
+
.
TV
(ja*)*-
r (j p - \v)
fp-yp-y
a'JrVI
Fig. 31.
whether the integral of this type which contains a single Bessel function cannot be evaluated; it seems that the only effective method of evaluating it is the method which will be explained in 136.
It is natural to enquire
13*53.
Integrals resembling those of 13*51, except that they contain a product of Bessel functions instead of a single Bessel function, have been investigated
to tjie integral
*rij*
in which a
K{bz)
(z*-r*T*
'
^ b > 0,
m
|
is
a positive integer, r
R O)
4-
(/jl)
C
13-53]
INFINITE INTEGRALS
is
429
[When >
first
kind,
It (p)
may
be replaced by It(fi) in
this inequality.]
When a = b, the presence of a non-oscillatory term* in the asymptotic expansion of the integrand shews that we must replace 2m + 4 by 2m + 3 in the inequality in order to make the integral, when taken round a large semicircle above the real axis, tend to zero as the radius tends to infinity.
The contour to be taken is that of manner of that section, we find that
[r
13*52
and
if
we proceed
in the
(1)
ds
^ (M
Hvil)
(ax)
" *" r
-
(f-lr^
id \ m
strata?)
[^"-o)ff. oh.
Numerous
It
by Hankel.
integral
if
^
(2)
(az)
<L\l [^()^^
() +
^(^^K^](^^|-,
The
= 2, m =
;
and
^
tt
.
is
a Bessel function
mention
9
it is
(oar)]
we take
jt
=v
and
>
1,
we
see that
i
(4>)
P J (ax) J (bx)
6.
-^L. -
** J* ( br)
(l)
ar)>
according as a $
The existence of the discontinuity in the expression for this integral was pointed out by Hankel.
If
we modify formula
f"
(3)
we
see that, if a
v) if
>b >
and
R (k) > 0,
=
then
(5) J
^rp
-^m
(&*) [cos
H/* -
A.
(ax)
+ 8in$(fjk-v)Tr. Yv (ax)] dx
'
I(bk)Kv (ak).
* Since
HM
<2>
(az)
is large.
430
[CHAP. XIII
J^,
m=
+ sin
2
and
i J (p
^=
ll
we have
(6)
Jo
,,
rcosiCo+ a ^ [cos(p+
,_
F + fi n v) it. Y v (ax)] dx
v
In this result replace p by p + v, a by V( a' + c + c2 2accos 0)*", and integrate with respect to 6 from to 7J-; we find from Gegenbauer's formula, 11*41 (16),
0)
\ J
j.
[cos \ (p
+ p) 7r J v (ax) + sin | (p + p) it Y
and we
(ax)]
dx
<
^-I^bfyl^cfyKriak)*?-*.
This process
may be
find that, if
a>b+1c,
( fi )
r JO
then ~ *P I
/ L +K
,t
to)
M=i
/*
+ sin \ [p + p. + (N- 1) v) ir Y
.
(ax)]
dx
= -!. (bk)
Again, by considering
UI
n=\
(c n k)
(ak) k-*.
.
SjJi^[n-W] *"<>*
round the contour previously used, where both
b
and
p.
(9)
I" x^~k*
[n
Jflibx)]
[C0S
'
* (/
+ Xfl " v) W
.
'
Jv {ax)
2
+
provided that a
If p
sin \ (p
|
>S
(6)
+ 2/i - v) 7T F (cm-)] dx = - [II 7M (&&)] # (a&) A?" and {p + X (/*)} > -R (*>)
I
|.
is an even integer, the integral on the left involves functions kind only; a result involving the integrals of products of functions of the first kind of this type was given by Gegenbauer, who overlooked the necessity for this restriction (cf. 1351).
+ 2/i v
of the
first
An
is
obtained by considering
fy.
^IM^+ni
b
HV(az)
>
0,
\R(v)\<R(j>)<2m + 4, + R(ji).
It follows that
_J_
Ar r,^ j.i^<-r+! *,
"
(or)
13-54]
INFINITE INTEGRALS
431
and, in particular,
5660.
An
1351
(5) is
due to
when a and k
are positive,
=\
This repeated integral
j" /*
-r^-p J>
{ V(p*
+ ** -
may be
when p
p sin
is large.
Now make
a change
by writing
<f>
p cos
<f>
= k + r cos 6,
X
('
.
= r sin 0, [* Jo(ar)rdr
and we have
J*( ar ) rd0dr
2jfer
J or +
O oB0+2Jf~J o v'(^
+ 4^)'
be .established.
To
H ^{az)dz
v
(z*
+ bk y +i
i
Pig. 32.
It is supposed that
is
\R(v)\<R(p)<4>R(fi)
+ ^-.
'2 ,
24.
432
[CHAP. XIII
arg k
<
\tt,
the points
eiArV2,
e^ks/2.
By
(2)
j
(\p- ^vf
2/x) 7T
(ax) + sin
(\p- \v-
2/x) tr
(ax)]
-4 - w k V 2 )""* (
(akW 2Y + m T(p + \ v + m )
/
^ +4jk,^
4
+1
COS Qg
p+y-4/n+2m "*
i/
2sini/7r.r(/t+l)Lm=oi!r(i/+m+l)r(i 3+ii/-/A+i?)
(akl^2)- v+2m T(ip-jv + jro) p - - 4/z + 2m 1 _ v COS 4 ~mt w!r(-i/+m + l)r(i o-ii/-/A + iw) ""J
/
'
5-41, it is
seen
is
4 = v = p. + \, while 4 = = /4 + .
i>
2 = i/= /* + Up 2 = v p. + \
The corresponding
considered in 13*6.
13*55.
Sonine's integrals.
number
by
66,
The most
is
eUz+k)
2tti
circles
8,
\z\
R,
m is an integer and k is not a negative real number. = z & tends to zero as 8-^0, provided that R (p) > R (v) round The integral = R tends to zero as R -* oo provided that round \z\ integral and the
It is supposed that
|
j \ |
R(p)<
By Cauchy's theorem we have
2iriJ(^ +
m + $.
^M+lV
w!
13*55]
INFINITE INTEGRALS
433
e *
~a\ 1
2ttiJ
(#e'"') v
'
- T**H m
(xe- l " )\
dx
=~
i. Mm+i
"
In particular, taking ra
1
= 0, we
c
get
[ J
(2)
kr !!^ (k) = 1
fao gi(X+k)
.'
7r
?
A;
(*) {sin (p
v)ir
If
we consider the
integral
1
r
p-iiz+h)
cr--
we
find that
1
rco
(3)
p-i(x+k) -p-i
-
*>)
tt
7r] cr.
0,
we
get
2 T
(4)
/(&)
r/7x
fsin(a?+&) v ' X -r K
,
r J
(x)dx,
/e\
T7-
//\
2 f* cos
(a?
+ k)
r/
The
last
two
results are
due
to Sonine*.
More
(6)
generally, taking p
=v+
and
we get
iYv (k)},
/ ,,1(2:+*:) ~if
ir J o
* 1 x+k
J, (x)
dx=
^
f
2i cos vir
cos(x
^
+
(6)
k)
^
1
(2),
1
.
sm (. + *) A.
,
.,
,vj.
and
(7) of 13*42,
efo
(*)
= --
irJox + k
P ^?
+-
sin
(fr
cos
6?)
dd,
tt.'o
356(2), 9-11
*
I
Yn (k) = - -
TJO
O + A) Jo (*) dx + T
'
I
.'
sin (k cos
i.
(1890), pp.
434
13*6.
[CHAP. XIII
We
shall
now evaluate
Bessel function, under the integral sign, the definite integral of reversing the order of the integrations.
and
As a
first
(aw)
dx,
in
which
it is
R(v)>0,
and a
is
R(2fi
+ 2)>R(p + v)>0;
may
converge.
The
i
equal tof
af- i
{
f(-s)
\axy^a
r <** +
y
_A_r
1
(//,
_ _Ti_~:
s)
x (jar*. jfr+*--
&.
1)
When
,-v
^
}
we
find that
af-
(ax)
dx
o^-\
'
!o
(a?
+ krr +1
'
_ q"fr+-^r(ip + i y )r(/i+i-ip-iF) l2 (p + v.p + v +I 2*+i r(p + l)T(p+l) 2 \ 2 a^-pr(^ + p-M-l) *-p. u+2-""^/
,'
'
'*'
4;
a '*^
The hypergeometric
if
in certain circumstances
the former
if
=v+
2 or p
= v + 2fi +
when
2,
the latter
i/.
By
is
valid
-R(p)<R(p)<2R(ji) + $.
In particular, taking p = v + 2,
(2)
we
find that
Jo
(x*+kr +l
>ro.+i)^"
}'
p. 50;
it is
when
-KR(v)<2R(lt) + $.
* This method is due to Lerch, Rozpravy, Math. (1896), p. 233]; he shewed that
v. (1896),
f
/
2at
fit
1 y
/*""
T(-s)x2'd8
]
ooi
r(
+ i)
'
made
of the. method.
The change
may
"
13-6]
INFINITE INTEGRALS
435
A formula of some interest is obtained by making p = l, p> = \, the hypergeometric functions then reducing to squares or products of Bessel
functions
It is thus
;
is
found by making p
=l
v, p,
=v
\.
5 '41)
(3)
Jo
7*T*F
;
= hv {hak) Ki
"
{l<lk)'
provided that
R (p) > 1
and that
provided that
R (p) > \
stf~x
Next consider
Jv (ax) dx
be supposed that
Jo f
in which a
>
and arg k
| J
< \ir.
It is first to
R(p)>0,
The
integral is equal to
J2(4^
+ 4)>i2(p + v)>0.
Iff"
87n'J_ wi
r<-)
^-'(i^^j.^
r(i;+* + i)r(M +
i)
1)
Lm=o ! r(v + m +
1)
T (i/ + i v - 1* + |m)
x cos (p
(-)m (a/fc/V2)^ +4
+ i v fi + \ m) ir
^
This expansion
laid
is
m! T (2/* - }p + \v +
on
p.,
and p are
+ JgL >(,>)>-(*).
first series
Now
functions,
namely
p-2 v fi + ^
By
5*41
or
4 = = /i + .
i/
(5)
4*rr ~ (2*r r ( + i) J
-
'
w w
*
(a *)
'
(6)
Jo
(^ + 4*T + *
M2^=il^HT) J ^
72 (v)
^^ ^>
(
The former of these is valid when in both, a > and arg k\<\rr.
|
> , the
latter
when
R (p)
> , and,
436
Finally, as
[CHAP. XTII
13'55,
we
shall consider
x of'1
Jv (ax) dx
+ ky+i
'
Jo
(x
in which a
>
ir.
R(v)>0,
The
j
:
integral
/ root
I
is
equal to
1
'
T(-s)T(p-{-v +
27rij_ocf
irk*-*-1
.
2s)r(fjL
+ l-p-v-2s)
V2
'
r(v++i)r(/* + i)
T
(-)"
r(v +
_ y
M=0
The
first series
reduces to
(ak)
when
/*
is
then
expressible
/(_.
by Lommel's functions
f
00
(cf.
107).
r __
x v Jv (ax)dx
irk"
,.
....
provided that
number of the
may be
__
r (- s)V (/a+i>+
2s
+ l)(^y* + " +M
2Tri] -aoiVi
in
r (p, + v + 2,5 +
which the poles of V ( s) are on the right of the contour while those of 1) are on the left this expression is easily derived from 5*41
;
6'5.
xp
~l
J,,,
(ax)
2
Jv (ax) dx
two functions of the type
3
/:
by these methods.
final
(.r
+F)x + i
The
result is a combination of
is
and the
a 2 >P.
13*61]
INFINITE INTEGRALS
437
by an
integral, is
Jv (b/x)
p-i
(1)
Jn {ax)
Jv (b/x) dx
+1
V"
'
+ 2^+P+ir(
This
is valid
ft+ i)r(^+^+ip+i)-
6 are positive
^ +1
'
-T- +1 a~ - +1 ~2 +
"
'
1
'
~W)
l6";
1;
when a and
and
had been given previously by Cailler, Mem. de la Soc. de Phys. de Geneve, xxxiv. (19021905), p. 352; Bateman, Trans. Camb. Phil. Soc. xxi. (1912), pp. 185, 186; and Hardy, who discussed the case of functions of orders + , (see 6*23).
An
lim
is
due da
4^r^ {1+21o^^-^
\
(,/+1)
-^( 2 --
,,)
i]'
\ f
&X SIT!
V7T
d%
+ 4^"l"It is easily
f
/
y)
{ (2) - * (' +
1)
proved that
x f
jo
d# i# exp^^)^^)^^)-^^^.
sin
i/tt)
_ j_
r- rt+.<
r(-)r(* + 2),ft)-
r(- s )r(2 + 2) rfs + s+i)r(2-i/+s)(2^),M
!
*+<
s(s
+
1)
!
i>r(i/
_1
and
(2n
+ l)!(2p)-*
1)
4 =i w.(n+
this series is
T(v + n +
T(2 -
*/
+ w)'
an integral function of
1/p.
j
To
at
small and
and
when p is argp \<i'ir, we observe that the last integrand has double poles 1, and simple poles at f 2, - f 3, ...
|
* This integral was brought to my notice by Mr G. G. Darwin, who encountered problem of Diffusion of Salts in a circular cylinder of liquid.
it
in a
[CHAP. XIII
ds_
438
J_
f*+**
s (s
V(-s)T(2s + 2)
SiriJi^i
+ 1) T (v + s + 1) T (2 - v + *) (2p)M
{f(2)-yfr(v +
^-^^T)
and
so (2) lira
-*-+o
_ (l-2v)^nv7r
~2ttv{1-v)
dx
+
(X>
4^=l0 {1+2logi8 -^
(,,
+ 1) ~^ (2 " I,)}
]
2
+ 2 2H(n-2)r(i/+l-^)r( 2 -''-).(-2)! t 3
In the special case v
(3)
(-)"(2j>)T(in)
= 0, we
f
find that
lim
iiioM) +
exp ( _
p^) /t
,
.
w w ]
j,
2
f ~ _i 2^ + Mfo7r
13'7.
(2m
l)!'
1141
is
obtain-
able by taking the cylinder function to be of the first kind and substituting the result of 62 (8) for the function under the integral sign.
"ski. L.i
and
if
sm *'
exp i(
l
we
find that
z*\
dtd *'
\
of the integrations,
/"+'*
f,
Z* +
(zZ\ dt
but the former restriction may obviously be replaced by R(v)>- 1, and the latter may be removed on account of the symmetry in z and Z. It is also permissible to proceed to the
This result
is
proved when
R (v) > - \
and
\z\
<
\Z\;
limit
by making \z\-*\Z\.
13-7, 13-71]
INFINITE INTEGRALS
6*21 (4)
439
By
and
(5),
we
find in the
provided that
(i>)
> 1 and
<
\.
pp. 152
was obtained by Macdonald, Proc. London Math. Soc. xxxn. (1900), differential equations, and he deduced Gegenbauer's integral by reversing the steps of the analysis which we have given. The formulae (2) and (3) were given by Macdonald, though they are also to be found in a modified form in Sonine's memoir, Math. Ann. xvi. (1880), p. 61.
The formula
(1)
A further modification
of the integrals
on the right in
(2)
and
(3)
integrals, see
13*71.
The expression of
(Z)
(z)
as an integral.
We
product
shall
v (Z)
next obtain a formula, due to Macdonald*, which represents the v (z) as an integral involving a single function of the type
v,
namely
(1)
ww-trM-*-*?]*'*
is
This formula
when
\ir\
\ArgZ\<7r,
arg
it
<
7T
but
it is
convenient to prove
when
X,
x,
and
is
to extend it
obviously to be associated with 11*41 (16), is of some importance in dealing with the zeros of functions of the type v {z). It is possible to prove the formula
without the rather elaborate transformations used in proving 1141(16); the following proof, which differs from Macdonald's, is on the lines of 26.
By
6-22(7) we have
(X)
(x)
=\
**
t
*
e
J
\ J 00
oo
~|
J
e-ivT-Xoo&MT+U)-xco*h(T-U)
dlJdT
oo J -oo
v,
If
in the integral
/
* Proc.
e -Xco&(T+U)-xcoh(T-U) flJJ
-oe
Lon'Um Math.
169171.
440
[CHAP. XIII
dv
~^~>
r
it
becomes
exp
a\ v +
If
+ a? + 2Xocco8h2T\l ~
*f
and so we have
K. (X) K, (.) = I,
[ -~-M"*
e
exp
(|
^}]
AT.
and, on performing the integration with respect to T, we at once obtain and x are positive. Macdonald's theorem when the variables
13"72.
IIL
(z)Kv (z) = 2f
Jo
,00
K^tfz cosh
i)
cosh
(p-v)tdt
(/*
=2
.'o
K^
we use
0-s
(2z cosh
t)
cosh
+ v) t dt,
fi
To obtain
these formulae
622
(5)
IL
{z)K v (z) = -\
T*
(cosh
H-coshM) cosn
00 JJ 00
^ cos h vudtdu.
The repeated
integral
double integral.
is absolutely convergent, and it may be regarded as a In the double integral make the transformation
t
+ u = 2T,
t-u = 2U,
and
it is
apparent that
v
h (z)
(z)
= \\
"J
00 J 00
But
2 cosh
/i(T+U) cosh v (T - U)
(ft
srcosh
v)
T cmh\iL v)
U+coshjjtr
v)
v)
T cosh
v)
(/a
+
(/a
v)
U
U.
U + sinh
(/j,
T sinh
+ v)
we interchange
we
K (z) Kv (z) = \\
If
~ e %z cosh
T C09h u cosh
(/a
Tcosh (/a
-v)UdTdU.
(1),
we
U we
obtain the
first
form of
(1).
and
if
we
T we
220223.
13-72, 13-73]
INFINITE INTEGRALS
441
The formula
(2)
T Jo
cos 9) cos
is
(ti-v)0 d0,
Neu-
we take
fi
v,
we
find that
I (z)K v (z) = ~
7T
.
(**
o
(2zcos0)co$v0d0.
then replace
/x
More
generally, if
;
we
find that, if
R (v
we take m) <
I
ft
= m and
then
'
and
by m and
v,
1,
(4)
Im (s)
K
,
(z)
-
K
} -
dO.
If
we combine
r
,
(3)
with
2
6'16 (1)
we
find that
cos (u cos 0) cos
i>0
+ |)
1
;
fl*
(4s)"
provided that
5>
and in particular
<
/(s)/f (*)=|
()
V(tt ,
+ 4tf
13*6,
formula
(3).
(z)
+ Yv
(z).
The
which represents
J* (z) + F
(z) is
difficult
to establish
is
rigorously.
It is first necessary to
positive
(= x), and it is also necessary to appeal some such principle, in the course
( 6*2 1)
H (x) =
7TI
.'
ex sinh w
-" w
dw.
_c
] j
which the integrand tends to zero as w -* oc on the contour, it is clear that when an exponential factor exp { \w-} is inserted, the resulting integral converges uniformly with regard to \, and so it is a
From
the
manner
in
continuous function of
\.
Hence f
lim
HJ$ ( X ) =
*
exp {- Xw2
8inhtt,
w dw.
Phil.
Mag.
234; Quarterly Journal, xlii. (1911), p. 221. 2266 ; Trans. Camb. Phil. Soc. xxi. (1912),
is
48.
In this integral (as distinguished from those which follow) the sign lim
commuta15
w.
B. F.
442
[CHAP. XIII
By Cauchy's theorem, the contour may be deformed into the line / (w) = \tr, so long as X has an assigned positive value writing t + \iri for w, we get
:
H (x) =
lim X-^+0
a\v-ai
r-
W*
rx
exp {- X
J
(t
-x
\-rri)')
eixcoaht
- vt
dt
in Hardy's notation.
In
like
manner
jyW (#)
e-'jcch -,
<j tt
\tnf\fulfilled
|
.
dt x I
J CD
ei
- vu
du
(in
integral
-X
^ (t)
We
^'
1'
(a?)
#< 2 (x)
>
=G
T X
[*
./
e <*(coBh*-corii )-(*+)
(^^
exp [ X (t
^rriy
X(u \irif].
then
r " il,hr- 8,,r (dZW).
Make
the substitution
^tt2 [J* (x)
+ u = 2T, t u 2U and
2
+ F
(x)}
=gT r x
J
.'
8te "
co
inh
exp {- 2XT 2
- 2X ( U + Jui*)
8
}.
it
may be
replaced by
U is
performed
+ F
(x)}
=G
L-'o J
r I* e -x
+
2iesinh Tsinh
'-lvT
d UdT
''
13-73]
INFINITE INTEGRALS
consider the integral
443
We
first
[" exp {- 2\
J CO
dU,
T is positive. When T is positive, the tf-path of integration may be deformed into the contour I (U) = %ir; if we then write U =v + \iri, where v is real, the integral
in
which
becomes
[
J -cc
exp {- 2\ (v +
7Ti') ]
= 2 exp (2\tt )
2
'
<fo
2 exp (2XTT2 )
e-torinhroodidv
2
-'o
- 2 exp (2\tt )
To approximate
inequalities
2
r
Jo
{1
- exp (-
when X
is
small,
we use the
+ 8tt
X 2 v2
so that, for
and
1,
f"
J -en
=2
exp (2\7T2 )
|"[ "
e-2*nhreo8h
jv
_ 2 \ + 8tt \ ) 6
2 2
If
we
J CO
f*
in a similar
manner, we find
it
equal to
2K
where
(2x sinh T)
T)\
<6 ^
X
1,
provided that
T is
positive.
On
as
lim s-^+oJs
I
we
find that
\Tf*\j?{x)+Yvnx))
lim
- (2\ + 8tt \ )
2
2
exp (2\7ra )
j^^^sinh T)\
(o/x
1
l
1 e- 2 " r dr
J/*-oJ
ll
+
A
lim
lim
I"
++06-*+0J6 L
*/*
444
[CHAP. XIII
T is
small,
convergent
X (
Jo
\~K
(cy*
ll
(2xsinhT)\
Jh-o
-*
e^ T dT
making \
f
is
+ Y* (x)} =
"
2" r
) rfT.
Jo
when x > 0,
J, 2 (a?)
If
+ F
2,
(a?)
I*
we
z,
replace x
by
of
provided that
result
E (z) > 0.
r 2
we have the
(1)
J* (z) +
0)
t)
cosh 2vtdt,
provided that
R (z) > 0.
is
v (z )
I^- Y
OV
d
(z)
-^4^=-OV
7T
rK,(2zsm\it)e-^dL
.!
To prove
this formula,
x),
we
first
suppose that z
is
we
replace
by
and then
OV
OV
I yt, $H
2i
5-5-.
<
2>
(x) {x)
dH"{1)(x) dy
H (x) dH'*W a dv
v
G\
e-y(t+u)
^tdu)
= - 4? G
IT
I
f~<X> J
QD
=*
^G rr
-T--
^^
Cr
For the
full details of
a,
202.
( (
13-73]
INFINITE INTEGRALS
positive,
445
Now,
T being
8
.'
we have
I" 00
U + \ irif) e
(t>
2i** inh
T>inh
u dU
=2
and, since
i>e-2*
sinh
"
+ 7n) exp {- 2\
function of
(i>
Tcosh
js
an
0(jd
v, it
last integral is
2iri
(A),
00
(\) is a function of
is
where the constant implied in the symbol to x its integral with respect to T from
In
like
T such
that
convergent.
manner,
(""
J
(2
00
Hence
it
follows that
"
/" fa)
W dY ^ - F, W ^ -^ = - - fT
fa)
9l>
-^sinhTcosh*-2,T rf
^r
is
dv
TTJo J -ao
The extension
to the case in
real part is
made
as in (1).
is
=tf [,r,, )
and hence that
?W. r
; (, )
^].,-,,[,, (t
X
-*
i
,
4|i-r. W
^],
w ^(^Lf^,)^..
(8 )
v v
'
cv
Cv
nz* J
(o
'(zicoshZT-v^K^vBrnh^e-^dT.
may
JO
(5)
J^z)
* ain( *~ v)
446
13'74.
[CHAP. XIII
Since
K (^)
is
a decreasing* function of
, it is
Jf{x)+Y*{x)
is
v,
when x
when x
is positive.
is
approximately equal to
2/(irx),
is large,
we
x{J*{x)+Y*{x)}
and prove that
it is
v>\, and
that
it is
an
increasing function oi x
It is clear that
when v<
\.
^[x[J/(x) + Y*(x)}-]
= f [K
.
(2x sinh T)
+ 2x sinh TKQ
'
2vTdT
8^
7T
K
+
2
T cosh 2vT
,
dT,
parts.
Hence
=~
integrand
"
f
t' Jo
Now X tanh XT
is
0,
>
1 or
< 2v <
last
this
Next we prove
that,
when x ^ v ^ 0,
(a?
-i*y[JS(x) +,*{*)}
is
an increasing function of
If
x.
we omit the
2
i'
-i /7r2
we get
2 (x*
v2) sinh
'
(2x sinh
t)}
and
is
to establish the
theorem stated
it is sufficient
positive.
* This is obvious
13-74]
INFINITE INTEGRALS
447
We
2j>
2
twice integrate by parts the last portion of the second term in the
integral thus
sinh
Jo
'
(2x sinh
t)
cosh 2vt dt
v sinh
r
\
sinh 2vt
{sinh
'
-v
J
-y-
(sinh
tK
'
(2x sinh
t)}
sinh 2vtdt
= =
L
d
-j-
Jo dt
v
I
tK
'
(2% sinh
(2a;
t)}
sinh 2vtdt
2a;
sinh
cosh iT
sinh
)
t)
sinh 2i/d
.'o
= x sinh cosh tK
,
(2% sinh
(2a;
cosh 2vt
Jo
)]
+ =
Jo dt
)
j-
[sinh
cosh iT
sinh
cosh
2itf eft
* J5T
'
(2# sinh
t)]
cosh
2i/i eft;
equation
= 0.
- 2x* sinh
t)
'
K (2x sinh
T
)
cosh 2v
2
+x
= a;
and
coefficient of
j
r
I
jfiT
(2# sinh
sinh
t)
- 2 sinh
cosh
2i>
rf ~=-
(sinh't
-j
rt
)~1
cosh
sinh
2irt
eft
jfiT
(2a?
[tanh 2 1 cosh
2i>t
+ 2v sinh
positive
;
sech
2i/] oft,
Jo
this is positive
is
is positive,
and the
result is established.
are
2/tt, it
(a?
i^)*
{Jy* (x)
+ rr
(x)}
results that
when x^.v^\,
An elementary proof of the last inequality (with various related inequalities) was deduced by Schafheitlin, Berliner Sitzung*beric1de, v. (1906), p. 86, from the formula
(cf.
5-14)
(4r-l)JVr(0* = i( a 2 + 62
where <$
(x)
_^Qf^ + ^;
(A )
+ 2(l-^)^2 (a;) + ^; 2
(. t .-)],
448
[CHAP. XIII
is
dYv (v)
dJv {v)
left
To obtain
this result,
we observe
may
be
=1
irv
-s[
But,
for
K (x)
is
t, 2v sinh (^tjv) is a decreasing function of v, a positive decreasing function of its argument, we see that
- V< K f
is
a decreasing function of
v,
and therefore
[MpY
> lim
d Yv {v)
dv
dJv ( v) v 1AP)
.
dv
=
stated.
lim
8'42
;
and
13*75.
It
is
Jv
(z)
+ F2 (z) from
(2* sinh
t)
cosh 2vt dt
.'0
for
we
cosh*
m=0
(2m)!
T "
(2^)!
where
and,
i2p
<
cosiC(i>7r)|
when
v is real
and
is
so large that
p + %>v,
Mp lies
between
and
(2p)l
'
13*75, 13-8]
INFINITE INTEGRALS
449
We
7r
m=0
(2w)!
Jo
that
is
to say,
by
1321
(8),
(1)
JS(*) +
proved when
;
it may be extended over the wider range and p exceeds v \, the remainder positive, and z is arg z < 7r and, if v is real less than, the (p + l)th numerically after p terms is of the same sign as, and
this
j
is
|
term.
13*8.
Ramanujan s
integrals.
Some
Ramanujan* from an
_^cos
j
ff.e
av
2M+ - 2r(/A+
f)T(v-ft
which
is
valid if
R (p + v) > 1.
e ittd
The
r.
J
..ro. +B r(.-
,
(2cos^)^+- 2 eitY
r 0* +
0;
- x)
y . >t ,
>
( jf
.
<7r ),
<[>r)j
where
t is
any
real
number.
in ascending powers of
By expanding
formula,
it is
x and
y,
this
seen that
a)
/1%^^-W
7r
;
if
t,
the integral
is zero.
In particular
(2)
J oo
I" Jt(x)J.s{x)dS
= J^{Zx).
pp.
In view of the researches of March, Ann. der Physik und Chemie, (4) xxxvn. (1912), 2950 and Rybczyriski, Ann. der Physik und Chemie, (4) xli. (1913), pp. 191208, it seems quite likely that, in spite of the erroneous character of the analysis of these writers J, these integrals evaluated by Ramanujan may prove to be of the highest importance in the
theory of the transmission of Electric Waves.
* Quarterly Journal, xlviii. (1920), pp. | Cf. Modern Analysis, 9 7, 11-1.
294310.
a, (1915),
Royal Socccxv.
pp.
123124.
CHAPTER XIV
MULTIPLE INTEGRALS
14*1.
integrals.
The
difference
this chapter
and the
last is
more
than one of mere degree produced by the insertion of an additional integral sign. In Chapter xiii we were concerned with the discussion of integrals of perfectly definite functions of the variable and of a number of auxiliary parameters; in
sign are to a greater or less extent arbitrary.
will
the integrals which are now to be discussed the functions under the integral Thus, in the first problem which
satisfy the
now be considered involve Bessel functions only seems desirable to investigate them somewhat fully because many of the formulae of Chapter xiii may easily be derived from them, and were, in fact, discovered by Weber as special cases of the results of
incidentally;
The
but
it
this section.
Weber's researches* are based upon a result discovered by Fourierf to the equation of Conduction of Heat
du
dt~dx*
is
= if"
< is
I"
(x + 2X^t,y+2Y</t,
its
+ 2ZJt)
x exp {- (X 2 +
where
F + Z )} dXdYdZ,
2 2
an arbitrary function of
first proved that, if
<I>
three variables.
Weber
equation
{x, y, z) is restricted to
be a solution of the
?)
<
*
3? + ? + !?**
<>
32 <i>
3 2 4>
Journal fur Math. lxix. (1868), pp. 222 237. La TMorie Analytique de la Chaleur (Paris, 1822), 372. The simpler equation with only one term on the right had previously been solved by Laplace, Journal de VEcole poly technique,
t
vni. (1809>, pp.
235244.
14*1, 14*2]
MULTIPLE INTEGRALS
451
then
(2)
u
<I>
provided that
has continuous
and second
differential coefficients,
and
way*
triple integral as
a single integral
[cf.
we change
to polar coordinates
<f>,
by writing
2YJt
= rsintfsin
+ r sin
<f>,
2Z^/t
= rcos0,
=-
Now
fir
3>Oc
(4tt01 Jo Jo J-*
+ rsin0cos<f>,
sin
<,
z + rcosd)
x exp
consider the function of
r, vr (r),
j-\
r2 sin
d<f>dddr.
iff(r)=j
<f>,
It is a continuous function of
coefficients
positive value
with continuous first and second differential and the result of applying the
;
operator
1
d\
drj
,.
r2
to vf (r) is
dr\
We proceed
differential
shew that the last integral is zero. If we make use of the equation (1), which <S> satisfies, we find that
to
the difficulty! caused by the apparent singularity of the last integrand on the polar axis, we consider the integral taken over the surface of a pole sphere with the exception of a small cap of angular radius 8 at each
To avoid
caps can be
made
bounded at the poles, the integrals over the by taking S sufficiently small. small arbitrarily
on the
left is
If we perform
*
A sufficient condition is
12*41.
should be bounded when the variables assume all real values, Cf the corresponding two-dimensional investigation,
.
Modern Analysis,
t This
difficulty
452
to
<f>,
[CHAP. XIV
The
first
-?/".[sin ^]r^'
and
this can
is
can be
is zero.
made
arbitrarily small
by taking
8 sufficiently small,
and therefore
it
Consequently
(8)
l&a + *T.(r)-0,
,
so that
ts (r)
A =
sin kr
+ B cos kr
,
where
all
and
B
r,
are constants
values of
r.
and
B
we
since rst (r) and its derivate are continuous for must have the same constant values for all values
;
of
If
we make
r-*-0,
see that
B = 0,
Hence *
u
t&w
r,
!o
exp (~
sin
kr rdr
-
= exp ( - kH) * ( *
*>'
'
and
this establishes
Weber's
result.
O (x, y. z) is a solution
and
if
f(r)
is
an arbitrary continuous
function of
then
(4)
f"
r r ^(X
>
Y Z)fW{(X-xY+(Y-yy + (Z-zy}]dXdYdZ
>
=
The reader
will
-y
-ii_'
r)
sm &r
rdr.
have no
difficulty in
make
This integral
is
exp ( -
-\ cos At
dr = v/(irt) exp ( -
fc2 1 )
with respect to
k.
MULTIPLE INTEGRALS
One such
set of conditions
is
453
that 3> should be
to infinity,
and that
(r
- oo
),
q > 1.
simpler formula established at about the same time by Weber* is that, a function of the polar coordinates (r, 0) which has continuous first and second origin is u differential coefficients at all points such that 0<r <a, whose value at the and which is a solution of the equation
A somewhat
then
k (r,
is left
0) c0
= 2n-tto ^o (^")
/ -JT
to the reader.
14*3.
The formula
(1)
("info P
J
J j
f" -'
F{R,&).Jo[u\/{B? + r*-2Rrcos(4>-<l>)}lR(d<PdR)
=27rF(r,4>)
was given by Neumann in his treatise f published in 1862. In this formula, F(R, 4>) is an arbitrary functioD of the two variables (R, 4>), and the integration over the plane of the polar coordinates (R, <)
is
a double integration.
In the special case in which the arbitrary function is independent of <I>, we inreplace the double integral by a repeated integral, and then perform the to reduces formula tegration with respect to < the
;
(2)
Jo
The extension
(3)
namely
["udu
was effected by Hankel. In this result it is apparently necessary that though a modified form of the theorem ( 145 1452) is valid for
values of v
;
when v \
(2)
(3) is
The formulae
and the proof of
and
much more
(3) is
Math. Ann. i. (1869), pp. 811. Temperaturzustand eines homogenen t Allgemeine Ldsung des Problemes ilber den statiovaren wird (Halle, 1862), pp. 147 begrenzt KugelJUlchen nichtconcentrUchen KSrpers, weleher von zwei 409410. 151. Cf. Gegenbauer, Wiener Sitzungsberichte, xcv. (2), (1887), pp.
% Cf.
'
454
[CHAP. XIV
any additional
complications.
Following Hankel, many writers* describe the integrals (2) and (3) as "Fourier integrals" or "Fourier-Bessel integrals."
On
account of
;
its
proving (1)
and at
greater simplicity, we shall give a proof of (3) before this stage it is convenient to give a brief account of the
who have
stated,
He
lim
rRF(R)dR f Jv (uR)J
lim
X-oo Jo
(ur).udu
sir
IT
and then applied the second mean-value theorem to the integrand just as Substantially the same proof was given by SheppardJ who laid stress on the important fact that the value of the integral depends only on that part of the JJ-range of integration which is in the immediate neighbourhood of r, so that the value of the integral is independent of the values which F(R) assumes when R is not nearly equal to r.
in the evaluation of Dirichlet's integrals.
A different
mode
of proof, based on the theory of discontinuous integrals, who integrated the formula ( 13*42)
P
\y,
<r \ (R>r)
&
after multiplication
r" +1
by F(R)RdR, from
to oo
so as to get
J2+
[J
J"
J,+1 (ur)
Jv (uR)F(R) RdRdu =
J*
F(R) dR;
is
and then, by differentiating both sides with respect to r, formula (3) obtained but the whole of this procedure is difficult to justify.
;
at once
A proof of
||,
but, according to
it is
The extension
* See e.g. Orr's
is
effected
by replacing the
paper cited later in this section. statement of a mode of deducing (3) from (1) when r is an integer was made by Weber Math. Ann. vi. (1873), p. 149, but this was probably later than Hankel's researches, since it is dated 1872, while Hankel's memoir is dated 1869.
t
t Quarterly Journal, xxin. (1889), pp. 223244. Math. Ann. xvi. (1880), p. 47. Proe. Camb. Phil. Soc. v. (1886), pp. 425433. See Gray and Mathews, A Treatise on Bessel Functions (London, 1895), pp. 8082.
||
1 Math. Ann.
14*3]
MULTIPLE INTEGRALS
455
be discussed in
14*5
14*52.
An attempt has been made by Orrf to replace the Bessel functions by any cylinder functions, the w-path of integration being a contour which avoids the origin but some of the integrals used by him appear to be divergent, so it is
;
difficult to
The same
criticism
It will
be
is
shewn
( 14*5)
if,
the integral sign are not necessarily of the same type, the repeated integral
not, of necessity, convergent.
be stated that, if r be a point of discontinuity of F(R), the expressions on the right in (2) and (3) must be replaced by J
It should
${F(r-0)+F(r + 0)\,
just as in Fourier's theorem.
For the more recent researches by Neumann, the reader should consult
(Leipzig, 1881).
his treatise
Neumann's formula
(1)
was obtained by Mehler as a limiting case of a fact, it was apparently with this
but
it
(cf.
14*64).
more direct method of proof is given in a difficult memoir by Du Bois Reymond|| on the general theory of integrals resembling Fourier's integral. The proof which we shall give subsequently ( 14*6 et seq.) is based on these researches.
from a result obtained by
Subsequently ErmakofflT pointed out that the formula is also derivable Du Bois Reymond which is the direct extension to
for
r. r. /-. f* {x
a
'
Y) cs
(a
(x
x)
+ ^ ( f_ y)]
dxdY> dadP'
>
= u cos to,
(3
= u sin at,
to.
and
Ann. n. (1873), pp. 146161. t Proe. Boyat Irish Acad, xxvii. a, (1909), pp. 205248. i The value of the integral at a point of discontinuity has been examined with some care by
Cailler, Archives des Sci. {Soe. Helvttique), (4) xiv. (1902), pp.
* Math.
347350.
v. (1872),
Math.
||
Ann. v. (1872), pp. 135137. Math. Ann. iv. (1871), pp. 362390.
f Math.
Ann.
pp. 639640.
456
If
(r, <)
[CHAP. XIV
and (R, <f>) be the polar coordinates corresponding to the Cartesian (x, y) and (X, Y) respectively, the formal result is fairly obvious when we replace V(X, Y) by F(R, <>) but the investigation by this method is not without difficulties, since it seems to be by no means easy to prove that the repeated integral taken over an infinite rectangle in the (a, /3) plane may be replaced by a repeated integral taken over the area of an indefinitely great
coordinates
;
circle.
F(R,
<I>)
is
F (r,
<f>)
which occurs on the right in (1) must be replaced by the limit of the mean value of F(R, 4>) on a circle of radius & with centre at (r, when 8 -*-0. <f>) This was, in effect, proved by Neumann in his treatise of 1881, and the proof
will
be given in
146
14*63.
A formula which
is
(3),
namely
484
cf.
12 2.
14'4.
The
Hankel
generalisation of
(cf.
Neumann's
was
effected
by
143)
may be formally
stated as follows
Let
F (R)
be
an arbitrary function of
subject to the
condition that
\"F{R)s/R.dR
exists
and
is
absolutely convergent;
and
let the
\.
X
Then
(1)
{"udu
an
interval in which
F(R)
has
The proof which we shall now give is substantially Hankel's proof, and it same general character as the proof of Fourier's theorem it will be set out in the same manner as the proof of Fourier's theorem given in Modern Analysis, Chapter IX. It is first convenient to prove a number of lemmas.
is
of the
* It
it is sufficient
t>
would be more
14*4, 14*41]
MULTIPLE INTEGRALS
457
14*41.
lemma of Riemann-Lebesgue*
in the theory
Letf
F(R)*/R.dR
a
exist,
and
{if
it is
an improper integral)
let it
be
absolutely convergent ;
and
f
J a
let
v~^
\.
Then, as
X -
oo
it is
is
b is finite; in the
second part
is finite is is
removed
F(R) *JR
is
bounded
also
removed.
j
(I)
F{R)\/R be K.
|
xJt x2
...
<r n _,
{x
=a,
xn =
b),
where e is an arbitrarily small positive number and and lower bounds of F(R) \/R in the rath interval.
Write
interval,
|
Um and Lm are
the upper
F(R) *JR = F{R m_ ) V#m-i + m (R), so that, when R lies in the mth w m (R) \^Um L m Now, when v ^ \, both of the functions of x,
1
.
x*Jv {x),
are
i*tiJv (t)dt,
Jo
.
Let
bounded when x > 0, even though the integral is not convergent as x * oo A and B be the upper bounds of the moduli of these functions. It is then
clear that
*
F{R)Jv (\R)RdR
2 F{R m ^) V-R*,-,
=1
-
[*m
Xm-l
(\R) V-K
dR
a>
+ 2
^ 2
|
.'
FjR^ ^R^W -^
i
w-l
j**r
m {R)Jv {\R)^R.dR\
|
Xm-1
Jv {x)\Jx.dx + 2
/*
0>
I
AdR
m (i)
2.Bwg
*
* Cf.
\*
Ae
VX-'
% 9-41.
;
Modern Analysis,
t The upper limit of the integral may be infinite preserves the analogy with 14-3 (3).
irrelevant factor
"
458
[CHAP. XIV
By
is o
sufficiently large (n
last expression
can be
made
less
remaining fixed after e has been chosen) the than 2Ae/^X, and so the original integral
r\F(R)\ s/R.dR<e,
J e
F(R) Jv (XR) Rd R
U
\
Ja
f J a
F (R) J
we get
(XR)
RdR\ + ^- r\F(R)\</R.dR;
\
\XJ C
(I),
J F(R)J (XR)RdR
v
\
2BnK
2Ae
J a
The choice of n now depends- on e through the choice of c as well as by the mode of subdivision of the range of integration (a, c) but the choice of n is still independent of X, and so we can infer that the integral (with upper limit
;
(Ill)
If
F(R) V-B
is
it is
unbounded
in a
number p
St \F(R)\^R.dR <
8
J8
By
(I)
and
intervals,
we get
J a
A,
I
V^
\
where
is
now
still
F (R)
e,
choices of both
14*42.
The
We
is
shall
and
F(R)*JR .dR
exists
and
Jo
absolutely convergent, then
Jo
K
lim
rF(R)
exists.
A-ooJo
[JO
Modern Analysis,
9-41.
14-42, 14*43]
MULTIPLE INTEGRALS
\,
fi
459
e,
ex
number
F
where
If
it is
F(R)\JR.dR<
A
we
is
write
F (R) J
(uR) J, (ur) uR
<f>
(R, u),
clear that*
["
j
f%(22,
u)duldRr<i>
Tj P
<j>
(R, u) dRi\ du
r|
&
T
u) dui
dR~ r | f v (r,
%>
^1 d U
^J
Since this result
?-\F(R)\<JR.dRdi
'0 J
is
e,
we
infer that
P [ <f>(R,u)dudR Jo Jo
r<b(R, u)dRdu,
is
Jo Jo
left
assumed
X -*
oo
it is
evident from
K-m-aoJ
=
and
lim
*F(R) {
(uR)
X-oo.'o
{Jo
14*43.
integral.
Next we
the
part of the path in the immediate vicinity of r, provided merely that has an absolutely convergent integral.
*
F(R) \/R
are X
The and
460
[CHAP. XIV
not a point of the
prove that,
if
is
interval*
then
t
udul
Ja
F (R) J
= 0.
we
find that, if
JQ
We invert
f
lim
X-^oo
F(R)
a
\ \
(uR)
(ur)
udul
)
RdR
\
[JO
TftlTf
i
j*
-
/,+.
(XB) dR
(Xr)
lim
XrJv+1
J*
</
(XB) <iR.
fEgV^figgn
are
ea;
it
Riemann-
Lebesgue lemma
and
so
("uduf
J0
Ja
F (R) J
(uR) J, (ur)
b.
RdR =
provided that r
is
a^r$
14*44.
It will
The boundedness of
J a
JO
now be shewn
that, as
t f Ja J
remains bounded, provided that a and b have any (bounded) positive values. It is permissible for a and b to be functions ofXof which one (or both) may tend
to
r as X -*
oo
Let us first consider the integral obtained by taking the dominant terms of the asymptotic expansions, namely
2
f
I
cos
.'0
7T\/r.J a
uR \vtr
\ir) cos
- n
7T
h ("sin
X (R r)
cos (X (R
sjrj a |_ CHb-r) si n f
R-r
x dx
f
I
A(6+r) cos
i>tt)
VrL.'A(-r)
Jx(*+r)
dx + cos
vrr
log
a + rj
+
b
r*l
.
14*44]
MULTIPLE INTEGRALS
first
461
convergent
The
integral
is
bounded because
/
J -00
&
dx
is
and the
second integral
the integral the limit of
is
bounded because
x
is
dx
is
convergent ; and so
as
now under
1
:-
consideration
r/" A (*-) sin
I
bounded, and
b
its limit,
\ -* oo
is
a;
^r wr
[J K(a-r)
dX + COS 1/7T
log
+ r~
,
a + rj
J (uR)J(ur)uR*dudR
Ja J
= -A- f TTtJrJa
Jo
f " [Ittw/,,
(w#)
Jv {ur) J(Rr)
\vir
Now,
R of an
domain of values of
function of r
R and r, and
is
when r
The
limit
shewn
to
be bounded, and
,
it
converges to a
whenever
rHb-r) sin#
J A(a-r)
does
so.
The second
integral
roo
may be
sin
42
-
inr
J rb
r}
\/r
a J A
d [_uK
(uR hnr
\ir) cos
\ir)
-i
cos{uR-Ivtt-
where
^ (\),
< 2
(X) and
.
\ and
i2
to zero as
\ -*
oo
462 Hence,
for all
[CHAP. XIV
the integral
bounded as \ -*
oo
and
it
.K(a-r)
does
so.
14*45.
Proof of HankeVs
all
is
integral theorem.
Now
Since
that
of Hankel's theorem
F(R)
an internal
F(R)\/R; and
therefore
we may
write
F(R)s/R = Xi(R)-X*(R)>
where X\ ( R ) an(^ X* ( R ) are nionotonic
(positive) increasing functions.
we choose a positive number After choosing a positive number fluctuation in the interval (r 8, r -f 8) 8 so small that F(R) has limited total
e arbitrarily,
and
also
Xl (r
'
ej
'
X2
(r- 0)-
<
ef
we apply the second mean-value theorem, we find that number intermediate in value between and 8 such that
If
r+S
f
X
-'0
there exists a
f Xi
WJ
+
"
(w
= Yj (r + 0)
(Xi ( r
J
'
(uR)
(ur)
r\
/
JR dudR
.
Jr
fr+S
+ 8 ) ~ Xi
Afi
+ )1 J
a;
,
r+( J
,
J" ( uR ) J
(ur)u*JR. dudR.
sin
Since
as
Jo
,
<tx
7r,
1444
that the
first
X - oo
term on the
-*- oo
on the right does not exceed Ce of the modulus of the repeated integral
Hence,
if
while 8 remains fixed. And the second term in absolute value, where G is the upper bound
(cf.
1444).
lim
A-*.
it
f
J r
Jo
follows that
+
lim
exists
P
X
*-*.*> J r
JO
X\ (r
Xi( R ) J( uR )
J*( ur ) uRhdudR
and
is
equal to
+ 0)/\/r.
14-45]
MULTIPLE INTEGRALS
treat
463
We
8,
and
the interval (r
r)
and we
I
infer that, if
lim
A-*oo
J r8 J
rr+8 fK
then
exists
lim
A-^eo J r -S J
and
is
equal to
C F(r+0) + C,F(r-0).
1
We
integrals*,
evaluate
and
By
the
theory of generalised
Jo Jr
lim
i>-0
f" Jo
exp (p^u*)
roo
r+ *
dRdu
J r
rr+S
= lim
p-^oJr
,.
exp (- p*u 2 )
.0
fi
p+
R*
/flr\ Di ,_
by 1331(1).
of integration,
as
^) -*- 0.
Hence
C^
p _fc0 2jt>V7rJ
lim s
exp x
-I
|
4p 2
I j
dR
= lim
and similarly
C'2
ri//>
I
-}
exp ( #2 ) <&e
p-*o "J"* Jo
1
lim j
r
I
exp ( xz) dx
= .
We
lim
exists
/'
A-- J r-S Jo
and
is
equal to
t\F(r+0) + F(r-0)}.
C,
* Hardy, Quarterly Journal, xxxv. and C2 se 1452.
,
(1904), pp. 22
66.
For a
different
method
of calculating
464
But,
if this
[CHAP. XIV
by
1442,
v
Jo
also exists
(ur)
RdR
and
is
equal to
stated in 144.
The use of generalised integrals in the proof of the theorem seems to be due to Sommerfeld, in his Konigsberg Dissertation, 1891. For some applications of such methods combined with the general results of this chapter to the probleme des moments of Stieltjes, 8188. see a recent paper by Hardy, Messenger, xlvii. (1918), pp.
14*46.
The proof given by Hankel of his formula seems inadequately. The first is in the discussion of
lim f
|
which he replaces by
lim
f
In order to approximate to this integral, he substitutes the first terms of the asymptotic expansions of the Bessel functions without considering whether the integrals arising from objection to the proof), the second and following terms are negligible (which seems a fatal and without considering the consequences of \R vanishing at the lower limit of the path of
integration.
The second
point,
which
is
lim
*-.>
J r+t J o
this is zero if tends to a positive limit
Jv (uR)Jv (ur)uRdudR
must be bounded
-*-0 as X-^oo
after proving
and
is
if
{-=0, he takes
if
and
14*5.
to
We
shall
now
l"
(uR)
K (ur) RdR,
v.
in which the order v of the unrestricted cylinder function &,(z) is any real* number. The lower limits of the integrals will be specified subsequently,
since
it is
For definiteness we
suppose that
<r
<&, (z)
{cos a
where
*
<r
The subsequent
simplified
and no
p^Q.
14-46-14-51]
MULTIPLE INTEGRALS
of the
465
( 14*41),
The analogue
Riemann-Lebesgue lemma
( v
namely
that
f F(R) e
Ja
(XR)
RdR = o (l/y/\),
rb
provided that
.
\
<>.
F(R)*JR.dR
precisely the
exists
and is absolutely convergent, may obviously be proved by methods of 14*41, provided that a b ^ oc and
,
fa^O
[a
if
O^v^l,
>\.
>0
if v
The theorem
theorem
is
The modified
V (
u r) u s/R
du dR
J a J t is
bounded as
A,-hoo
functions of
X which have
while t remains fixed; as in 14*44, a and b may be The number t is positive, finite limits as X -*- oc
.
though
it is permissible for it to
8ina?
dx
/:
oo
same
/:/:
it
<&9 ( R) <(c
(r)
u sjR du dR,
.
would be found that the convergence of this integral necessitates the convergence of the
/A(6-r) 1-cos.r
dx;
is
,
integral
and
so, if
X (6-r)-*-oo as X-*-<c
divergent*.]
14-51.
*5
|.
F(R)*JR.dR
exist
and
be
an
and
Jo
letO^v^^. Then
(1)
-,
(* udu
r F(R)
/
<&.
RdR
+ vir)
f
00
n/
>
o sin (a
R r
iv
"
viv\ fjp
* This point
466
[CHAP. XIV
an
interval in which
F(R)
has
As
in 14*42,
Jo
r udu Jo r F(R) 9?
RdR
F (R) f
J
=
A.
lim l"
-. oo
.'
uRdu dR,
exists.
Jo
1
&,(itRy&y (ur)udu
-i*
X
2
ur
r t% v+1 (ur)
^ (uR)
R -r
2
2<r 2 sin
vnr)
R
R
v
TTsini/TT
rv
(R -r )'
1 2
Hence we
K
F (R) \
uRdudR
2
n\
as
X. --
2<r
sina sin (a +
yTr)
/-
R*-r*>
v/
-*-
^ p^ p
oo
.
oo
and so the
last
Now
number
if if
e,
r<R^r +
r- 8 ^R<r.
Now
take
("
Jo
F(R)
Jo
namely
+ 8),
(r
+ 8,<x>).
and we find that
in 14*45,
(ur)
uRdudR
2a*sinasin(a
7T Sin VTT
+ *-7r)jf'-
-|l
<@y
Cr+S r\ fr+S
^>
+ F(r + 0)<s/r.
J r
(uR)<@y (ur)uRldudR
(uR) <@v (ur) uRHudR
Jo
X
+ F(r - 0)
r
.
( ( J r-tJ
<&,
+ V,
14-51]
MULTIPLE INTEGRALS
\r)\
467
is
where
is
arbitrarily small
when
e is arbitrarily small.
The making
so,
by
after
-* oo
we
X
infer that
Jo
is
r udu Jo f
IT
a sin (a
Sin VJT
vir)
R - r"'
2v
rco
fr+6
+ F(r + 0) Vr
lim
-*0.'0
<@v Jr
r tfv
(uR)
(ur)
uR*dRdu
+ F(r-0)<Jr.\im
I I*
To prove that the limits exist and to evaluate them simultaneously, take F(R) = R when r<R<r+B and F(R) = for all other values of R.
We
r'+i lira
t-0J0 Jr
rr+S
lim / / S-*oJo Jr
;
^{uRY^^u^uR'^dRdu,
and similarly
Jr-S
<@ v
$ v (ur)uR*dRdu (uR) r
r
+ 8. We
then have
dRdu
^h-i (ub)
{6" +l
- r+ ^, +1 (iir)} #r (ur) du
l
l
Jo
=
when
lim
P -*.+oJo
is
abso-
< p < 1 v.
Now
/"
{b' +1 Ji, +
(ub)
/^ %
>
468
[CHAP. XIV
and these are all absolutely convergent. They may be evaluated as cases of Weber's discontinuous integral of 13 "4, and hence we find that
{6-+ 1
|
K+
2
(ah)
- f+ ^ +1 (ur)\
<& (wr)
_ <r
r* sin (a
err
2%>
"
sin a sin (a
sin (p7r
The
when p -* 0,
r*\
ft
.
is
reducible to
62"
vir)
7T Sin
J/7T
'log 1-t-2
L
\
/,
+2 log r
"^
/, , i*^(l, -; l-v;
V
yfr
r*\ t-J
W
r
(1)
+ ty ( v)
-*
\,
some algebra
and the
when
+ 0,
is
simply ^^r".
In like manner
it
may be shewn
r
(
J
that
= \<x*r\
and
[
so
we have proved
that
v
udu
j
(ur)
RdR
Trsini/TT
provided that
%v%\, F(R) is
<@ (z)
er
and
14*52.
It is evident
from
14-51 that, if
F(R) *JRdR
exists
and
is
absolutely
Ja
convergent, where a
(1)
> 0,
then
r<@v+1 (Xr) * (Xi2)]
lim
^^
0) + F(r-0)}, provided that r lies inside an interval in which F(R) has limited total fluctuation and F(R) is defined to be zero when ^R < a, if the order of the
$<r*{F(r
and
\.
14*52]
MULTIPLE INTEGRALS
now
469
We shall
Let
It is
unrestricted order.
<*>
(R, r
X).
<D (22,
X)
- <,,_> (R,
X)
(Xr)
and
so,
we have
(2)
[3>
(R, r ; X)
4^ (R,r;
result,
X)]
JRF (22)
dR = 0.
lim
is
[4>, (R,
r ; X)
- <D n (22,
= 0,
where n
any positive
integer.
lies
2
between
$,
(1)
<J> n (22,
X) 22F(22)
v,
<*22
- |<r
[F (r
+ 0) + F (r - 0)},
and
values of
we deduce from
(3) that
(4)
lim P<l>(22,r*
is
This result
practically
indicated in 14*46.
To obtain the
<>
result in
{<$v (z)
\tf
Y (,) = Y
=
r
v
v
- Jv (22) Yv (z).
Then
u r^, (l
^)^^- <r)^^
left is also
equal to
[22iU (uR) <@ (ur) - r<@v+1 (ur) f (uft )] = w22 [ Yv (22) J, +1 (u22) - Jv (22) 7, +1 (22)] [ Yv (r) Jv (ur) - Jv (r) Yv (ur)] -ur\Yv (r) Jv+1 (ur) - Jv (r) Yv+l (ur)] [ Yv (R) Jv (uR) - Jv (R) Y, (u ft)] = wFv (22) Yv (r) [RJ, +1 (uR) Jv (ur) - rJ v+1 (ur) Jv (uR)] + $u{Jv (22) Yv (r) - Jv (r) Yv (22)} [22J, +1 (uR) Yv (ur) - R v+l (uR) Jv (ur) - rYv+1 (ur)J(uR) + rJ+l (ur) Yy (uR)] [RD v+l (uR)D v (ur) - rD v+l (ur)D v (uR) (R)\ (r) Y -\u{Jv (22) Yv (r) + Jv v - RD v+l (uR) D v (ur) + rD v+1 (ur) D v (uR)] - uJv (22) Jv (r) [RYV+1 (uR) Yv (ur) - rYv+1 (ur) Yv (uR)], \Dv (z) = Jv (z)+Yv (z), where \D v (z) = Jv (z)-Y,(z).
*
Math. Ann.
vi. (1873),
pp. 146161.
470
[CHAP. XIV
r/(R)RdR
J
a
;
exists
and
is
absolutely convergent
and consider
Carry out the integration with respect to u, and replace the integrated part by the sum of the four terms written above, divided by R* r2
.
Since
is
J,(B)F.W-J.(r)r.(B)
r,
bounded near
r, it
total fluctuation in
any bounded
interval
containing
of terms tend
to
follows that the integrals corresponding to the second group zero as \->- <x> by the generalised Riemann-Lebesgue lemma.
happen to
cancel.
When we
lim
use (1),
A f
we
p / (R)
say
<& (ur) 9$ v
(uR) uR
dudR
=
that
is to
${J*(r)
+ Yv>(r)}.{f(r + 0)+f(r-0)},
"
(6)
f
uduT f(R)<@
(ur)9$ v (uR)RdR
-!{/, (r) +
in
in
F (r)}
(5),
{/(r
+ 0) +f(r - 0)},
an interval
in the
which the cylinder functions are defined by which f(R) has limited total fluctuation.
and r
lies inside
Apart from
by Weber
integral theorem.
will
We
shall
now
theorem which
be the subject of
discussion in the sections immediately following. It is convenient to enunciate the theorem with Du Bois Reymond's* generalisation, obtained by replacing
the Bessel function by any function which satisfies certain general conditions.
The
generalised theorem
is
as follows
the
(I)
Let "^(Z, Y) be
is
pair of real
("
I -00
exists
*
oo
and
(t),
is absolutely convergent.
390. Neumann's formula (cf. 14-3) is obtained by writing iv. (1871), pp. 383 and the conditions (I) (III) are substantially those given in Neumann's treatise
Math. Ann.
g{t)sJ
published in 1881.
14*6, 14-61]
(II)
MULTIPLE INTEGRALS
is
let
471
denoted by
expressed in terms of polar coordinates, let it be F(R, 4>) have the property that (for all values
of $> between
functions o/<I>.
(III)
F(R, and
4>),
let this
total fluctuation in
0, 4>) be
integrable
If Q(R,
let
<I>)
(
in
0,
R),
fl(R,
3>)
denote the total fluctuation of F(R, <J>) in the interval tend to zero uniformly with respect to 4> as R-*-0,
it, ir),
a number of
sectors the
sum of whose
!
angles
small.
Since
F( R,
<I>)
- F(+
0, <J>)
^ Ci(R,
<t>),
this condition
necessitates that
F (R;
that
<I>)
-*
F (+ 0,
g(R)
is
(IV)
Let
rR
g (R) \/R
be a continuous function of the positive variable R, such bounded both when R-*-0 and when R ~ oo
.
Let
g(t)tdt
= O(R), and
let
G(t)
dt
be convergent.
Then
is convergent,
JO
(dX d Y)
and
is
equal to
r
2*-.[i*W(+o,<i>)]
Jo
ffW?. t
dt
where ffiKF (+
0, 3>)
meansf
2tt,
'
shall prove a
number
of
Lemmas,
14'61.
we have the theorem that if T is an unbounded domain^, surrmmding the origin, of which the origin is not an interior point or a boundary point, then, as A. -*- oo
Corresponding to the result of
JIT
*
is
The
object of the exception is to ensure that the reasoning is applicable to the case (which
which & (X, Y) is zero outside a region bounded by one more analytic curves and is, say, a positive constant inside the region, the origin being on the boundary of the region. f The discovery that the repeated integral is equal to an expression involving the mean value of F[R, <) when the origin is a point of discontinuity of F(R, $) was made by Neumann, Ueber die naeh Kreis-, Kugtl- und Cylinder-functionenforUchreitenden Enticickelungen (Leipzig, 1881), pp. 130131. J For instance T might be the whole of the plane outside a circle of radius 5 with centre at the
of considerable physical importance) in
or
origin.
472
[CHAP. XIV
theorem of
type used in
a theorem of a much weaker character than the view of hypothesis (II) of 14*6. The reason of this is the fact that (,/A) for certain values of R, and this seems to make arguments of the
14 41 inapplicable.
first
of
<t>,
that
T is
as the difference f of
tonic functions Xi (-^> )> %2 (-^ )> F(R, 3>) in the interval (0, R).
wnos e sum
is
If
and
for
that, for
and
lf
rKR
= Xl (,<*>)
C
*,.,dt
J kR
+ X (*i.*) 0(*)y l
Ax
f
J
kR ,.,dt
<
G(t)
kRt
Since
choose
G (t)
t
dt
is
convergent,
if e is
so large that
dt
ff((
>
<,
not
less
Also
Xx (R, <D)
<x(*>.*)-i-P(+o.*) + *|^(+o
and similarly
*)!,
IttCJR,*)!**
whence
it
(< >,<t>)
+ $F(+0,<t>) + $\F(+0,)\,
follows that
Jf
F(R,<&)G(\R)
(dRd)
R
{*<>, *)
{n(oo,<t>)
<2eT
J -ir
+ *(ao,<I^^(+0,*)l}d*
+
\F(+0,<?>)}d<!>;
= 2e r
and, since
e sufficiently small,
we may
*
bounded, this can be made arbitrarily small by taking it is independent of the outer boundary of T. Hence proceed to the limit when the outer boundary tends to infinity.
3>) is
F(+ 0,
and
This
is
the case
some
of the conditions
when g (R)sJ {R) then G (R)=RJt (R). imposed on J*' (/?, *f>) are superfluous,
;
It is
14-62]
MULTIPLE INTEGRALS
the modulus of
473
and
this is the
theorem to be proved.
14*62.
integral.
We shall next prove that the existence and absolute convergence of the integral
r r
are
sufficient conditions that
<t<
t)
(dXdT)
fttdttf
JO
J
CD
(X, f
lira
Y) g {u V(X 2
.
+ F )} (dXdY)
2
.
Y)\ g{u>J{X*+Y*j\udu{dXdY),
./
exists.
e,
there exists
such that
,
J
where
J. is
JV(B *)|2((Md*)<^,
>
<jr
|
(w)
j
Vw*
We
j
then have
k
.
f r [ F(R, <i>)g (uR) udu R (dRd<P) \J -wJo JO - P ^ r F (R, <P) g (uR) R (dRd<P) udu JO J-irio
=
i
r r rF R -vJ JO
(
fi
>
^ ^r
o^)
*
f
(<^*&)
-f
Jo
J -it J/3
R (dRd)udu
!
< ji
J -irJ
f*
["
fi
/"*!
(dRd<P)
Jo
+
<
Since this
is
.
aT i' r -n
J
J
F(R,
4>)
R* (dRd^)
I
u*
fi
e,
we
infer that
Jo
=
the integral on the
exist.
left
lim
A-*>
["
oo ./
-WO
P F(R,)g(uR)udu.R(dRd<t>),
Jo
is
assumed
16
to
w.
b. f.
474
[CHAP. XIV
then
on the right
exists,
-lim
14*63.
K-*.aoJ
f -v
f>(ie,4>)(?(X^)^^.
J
-ft
We
e
are
now
due
to
Neumann
and
stated in 14*6. first take an arbitrarily small positive number then choose the sectors in which the convergence of (R, 4>) to zero is
We
uniform, in such a
way that
the
sum of their angles exceeds 2ir e. We then 4>) < e in these sectors whenever R ^ 8 and we
;
n (R, <D) +
to be
F(R,
<S>)
and
fG(u)
where
^ f ^ 8.
f*
Now
Hence
f Jo
G(\R)^\ =
dR
/"**
I
Jt
\
I
G(u)-\<2G.
J a*
dii
where and is
tj
| |
is less
uniform,
less
than
2BG in
Hence
I
it follows
that
"
l'F(R, ) Q
Hence,
I
/" /* *<* )
./
-WO
(Mi)
^
-tt Jti
= 2eC {2tt + }.
- ftriW( + 0, )
f (.) JO
2?)
* U
<2eC(2ir +
that
Urn"
is
+ o(l),
to say
'
f K-^colJ-wJo
f JF(#,
4>)
(\R) (
^?> -
2,r
JWF<+ 0, 0)
Jo
f G () **
14*63, 14-64]
MULTIPLE INTEGBALS
left is
475
Now the
small,
expression on the
independent of c
and so since
e is arbitrarily
we
That
is
to say,
Km
exists
A-xJ-irJO
f f *(**) 0(XS)<-^
-"f
and
is
equal to
2-rJlF(+O
Applying the result of been proved.
*)J"0(iO ^' Jo
= J (w), we
= uJ
x
x
have
tg(t)dt
(w),
Jo
so that
G (u) = uJ
Jo
(u),
and
* r gw -H- /.-(.)} . -c u
Jo
Hence we have
(1)
puduj
9
I"
<I>,
sin
<I>).
If
(2)
we change the
origin,
we deduce that
y+
sin 4>),
and
finally,
(3)
ruduT
Jo
J-wJ-aa
F(R,<P)J
<f>)]
RdRd
when
(R, 3>)
= 27rJ*LF'(r,0),
of the values of
where 0lF(r,
<f>)
F(R,
<E>)
(r, ^).
14*64.
Neumann's
t
5 ?!L2 f ' f" /(S, <l>) P (cos 7 ) sin @d4>de f(0 ) = =0 **?r Jo J -n
by a limiting process
cos
;
in this formula
= cos
cos
(<J>
$).
The formula
is
obtained f by constructing a solution of Laplace's equation, which has an assigned value f(0, <f>) on the
135137 ;
cf.
476
[CHAP. XIV
by Mehler
is
is
5'71
made
indefinitely large,
R = k%,
coordinates
(,
= k0,
3>), (k, 6,
<j>)
noted by F(R,
<l>),
and
Pn (cos7)
ra-
the function of position /(, <E>) is then debecomes approximately equal to J (nvr/ic),
where
2
R?
r2
<f>).
We are
*>,<)
If
= Km 1
-*.>
2n
=0
lf *Tr
r
J
F(R,^)J
(n^/ K
JO
-n
)^d^. K*
now we
write
w//c ^= u,
we get
F{r,$)=-^udu^ j"
which
is
F(R,<P)J (uvi)RdRd<P,
Neumann's
result.
But
many
Thus, although we
00
know
that
/"" /""
S
is
/o\ n 2 4-1
n=0 \*/
(-)
-IT-^I *T Jo
f(,)Pn(cosy)sm@dd
.'
-a-
which assumes the value f(0, <f>) on the that we may put p = k in the series necessitates a discussion of the convergence of the series on the surface of the sphere and the transition from the surface of a sphere to a plane, by making k -* oo , with the corresponding transition from a series to an integral, is one
surface of the sphere, the theorem
;
r< k\
CHAPTER XV
THE ZEROS OF BESSEL FUNCTIONS
15*1.
quite general theorems mainly concerned with the fact that Bessel functions
have an infinity of zeros, and with the relative situations of the zeros of different functions. Next, we shall examine the reality of the zeros of Bessel functions (and cylinder functions) whose order is real, and discuss the intervals in which the real zeros lie, either by elementary methods or by the use of PoissonSchafheitlin integrals.
necessarily real,
Next, we shall consider the zeros of Jv (z) when v is not this function as a Weierstrassian
product.
We
of assigned order, and finally consider the rates of growth of the zeros with the increase of the order, and the situation of the zeros of cylinder functions of
unrestrictedly large order. full discussion of the applications of the results contained in this chapter to problems of Mathematical Physics is beyond the scope of this book, though references to such applications. will be made in the
it is
v,
of the functions
The
zeros of functions
whose order
is
In
and by Rayleigh*; and more recently Hermitef has examined the zeros of Jn+l (a?). The zeros of this function have also been the subject of papers by RudskiJ who used the methods of Sturm; but it has been pointed out by Porter and by Schafheitlin that some of Rudski's results are not correct, and, in particular, his theorem that the smallest positive zero of Jn ^(x) lies between J (n + 1) tr and \ (n + 2) tt is untrue. Such a theorem is incompatible with the inequality given in 153 (5) and the formulae of 15*81, 15*83.
>
Sohwerd, Die Beugungsertcheinungen (Mannheim, 1835); cf. Yerdet, Lecons d'Optique i. (Paris, 1869), p. 266; Bayleigh, Proe. London Math. Soc. iv. (1873), pp. 95103. t Arehiv der Math, und Phyt. (8) i. (1901), pp. 2021. % M6m. de la Soc. R. det Sci. de Liege, (2) xvm. (1895), no. 3. See also Prace MatematycznoFizyczne, in. (1892), pp. 6981. [Jahrbueh liber die Fortschritte der Math. 1892, pp. 107108.] 8 Porter, American Journal of Math. xx. (1898), p. 198; Schafheitlin, Journal fur Math. cxxn.
Phytique,
(1900), p. 804.
478
15*2.
[CHAP.
XV
of J,
(z).
(z)
has an infinity
integral
and a formal proof of this result by an analysis of Parseval's due to BesselJ. It was subsequently observed by Lommel that Bessel's arguments are immediately applicable to Poisson's integral for J, (z), provided that \ < v ^ \. A straightforward application of Rolle's theorem to x v Jy (x) is then adequate to prove Lommel's theorem that J,(z) has an
of real zeros
is
;
infinity
of real
zeros,
The Bessel-Lommel
and x
lies
when - \ <
< \,
between mir and (m + 1) ir, then J (x) is positive for even values of m, (0, 2, 4, ...), and is negative for odd values of m, (1, 3, 5, ...). Since Jv (x) is a continuous function of x when x > 0, it is obvious that Jv (x) has an odd number of zeros in each of the intervals (\ir, it), (7r, 2ir),
(fir, Sir),....
Some more
be given in 15-32
ir
1536.
then,
let
x = (wi +
\6)
where
<
^ 1;
by
we have
'
and
so
sgn
{x)
= sgn
cog i 7rM
{(2w
+ g), _ ,),->
Now
where
(-)' vr
f
.'
|
8r
^^^^--^ du,
^m
{(2m+ey-u*}i-'
du.
COS^im
\m
If
now we
{(2m
it is clear
that
tV
= Cf^^sin^irU.dU,
Jo
and, 8ince||
*
v^
fr(U)
is
Comm. Acad.
La TMorie
Sci. Imp. Petrop. vi. (17323) [1738], p. 116. Analytique de la Chaleur (Paris, 1822), 308.
BeueVsehen Functionen (Leipzig, 1868), pp. 6567. This is the point at which the condition r ^J is required ; the condition v > - & ensures the convergence of the integral. f The reader will prove this without any difficulty by regarding r as a continuous variable and then differentiating /, (U) with respect to r.
Studien iiber die
||
'
15-2-15-22]
It follows that
479
and
so
sgn
{*'
-J-
(,
= sgn(-l)"-,
since vm'
That
when J <
sgn
v
/
<
\,
(mir
+ \6v) = j_
and from
The
theorem follows in
Nlerlandaises^vu. (1872), pp. 351 358, with the help of a method which resembles the Bessel-Lommel method. Baehr's result for Ji (x) is that the function is positive when x
lies in
(x),
the intervals
(0,
and that
it is
negative
when x
lies in
the intervals
gated in this
(Qir,
6n-), ....
way by
C. N. Moore,
Annals of Math.
156162.
The
It
results just stated are of a less exact nature than the results obtained with the aid
of slightly
more
refined analysis
by Schafheitlin
( 15*33
15-35).
was noted by Whewell, Trans. Camb. Phil. Soc. ix. (1856), p. 156, that zero between 2 and 2 ^2, and that the function Hq (*) has some real zeros.
15*21.
(x)
has a
It is easy to prove that 9%, (z) has no repeated zeros, with the possible exception of the origin *. For, if $v (z) and ^y(z) vanished simultaneously, it would
V <WV (z) = 0, vanish would at the common zero of that all the differential coefficients oi9o v 9^K (z) and $*/ (z), and then, by Taylor's theorem, "^P, (z) would be identically zero.
follow,
by repeated
15*22.
It will
that if j lt j >2 ... are the positive zeros of arranged in ascending order of magnitude, then, if v > 1,
,
now be shewn
Jy (x),
<>,3 <
This result
is
{tf"
+*
J^ (x)} - ar J
(x)
first
at least one zero of x~" J, +1 (x), and the second shews that between each consecutive pair of zeros of x*+1 J, +1 (x) there is at least one zero
J (a,)
is
now
obvious.
by Sturm, Journal de Math.
i.
(1836), p. 109.
480
If
[CHAP.
XV
is
Jv +i(x)
any
Jv (x).
real
The
^v+1
(x).
when
four
mathematicians published proofs almost simultaneously; the proof which has just been given is due to GegenbauerJ and Porter ; the other proofs, which are of a slightly more elaborate character, were given by Hobson|| and
van VlecklF.
It has
l) r
/ +1 (X),
at
any positive zero of Jv (x) the functions Jv+1 (x) and Jv+2 (x) have the same sign but at successive zeros of Jv (x) the function J +1 (x) alternates in sign, and so there are an odd number of zeros of Jv+i (x) between each consecutive pair of positive zeros ofJv (x) interchanging the functions /+2 (x) and Jv (x) throughout this argument, we obtain Porter's theorem that the positive
; \
zeros of
Jv+Q (x)
(x).
is
A result of a slightly more general character than the theorem of 15'22 due to A. C. Dixon**, namely that, when v> 1, and A, B, 0, D are constants such that AD j* BC, then the positive zeros of A Jv (x) + BxJv'{x) are interlaced with those of CJ, (x) + DxJJ (x), and that no function of this type can have a repeated zero other than x = 0.
The latter part of the theorem deducible from 5'11 (11),
f*
is
Xjy
(X)
f
Jo
when x
is
positive
+ BxJv
(x).
in
which
t
Cf.
a, /3 and v are real, and x is positive. Gray and Mathews, A Treatise on Bessel Functions (London, J Monatshefte fUr Math, und Phys. vm. (1897), pp. 383384. Bulletin American Math. Soc.vr. (1898), pp. 274275. Proc. London Math. Soc. xxviii. (1897), pp. 872373.
II
American Journal of Math. xix. (1897), pp. 75 85. ** Messenger, xxxn. (1903), p. 7 ; see also Bryan, Proc. Camb. Phil. Soc. vi. (1889), pp. 248264.
IT
15-23, 15-24]
481
if
we
v
observe that,
*'<)--
x{AJv (x)+BxJ,'(x)y
2l/;*<o*
and so
is
(x) is monotonic. The positive zeros of <f> (x) are therefore interlaced <f> with the positive poles, and from this result the former part of the theorem
obvious.
If the function
(x)
+/3Yv (x),
we have
af@:(x)
2vft (a sin inr
-\ @?i$)tdt = \x
Jo
%<
+ /3 cos vv)
d<@v (x)
d
'
ir
sin vrr
dx
dx
1 < v<
/9
and
A<@v (x) + Bx r @
'
(x)
Again, since
<@ {x),
x<@v\x)
\* d<@ (x) dx
the theorem
is
d{x<@'(x)} dx
= - \ {(a? +
i^)
<@* (x)
+ &<@:* (*)},
lies
VV. whether v
between
and
1 or not.
The
theorem in which
4 = 1, 5 = 0, C = v, D = -l.
15*24.
v.
we
If
(*)
W(x)
= yJ (x) + 8Y (x),
irx
then
^
and
<
Now it is known that, at consecutive positive zeros of #r (x), <$,' (a?) has opposite
signs,
therefore,
from the
is
to say & 9 (x) has an odd number of zeros between each consecutive pair of positive zeros of 9^ (x); similarly <@ {x) hasan odd number of zeros between each consecutive pair of positive zeros of 9% 9 (x) and so the zeros must be
;
interlaced.
If we take one of the cylinder functions to be a function of the first kind, we deduce that all real cylinder functions have an infinity of positive zeros.
* Olbricht,
cussion of
Nova Acta Caes.-Leop.-Acad. (Halle), 1888, this result with some instructive diagrams.
pp.
dis-
482
15*25.
[CHAP.
XV
An
zeros
is that,
(z)
has no
1,
if possible,
that a
is
a zero of
is
{z)
not
real.
It follows
for
Jv (z)
that a
not a pure
+ ? +
1)
would be a
Let
or
so that a
is
also a zero of
J (z), because
Since v
(z) is
a real function of
z.
>
1, it
follows from
51 1
(8) that
tMt)JAa
j
t)dt
= ^-^[
Max)_^_ ^(c^-gj
= 0.
and
so,
since a2
2
tf
positive,
exist,
may be used A Jv (z) + BzJv (z) has all its when (A/B) + v<0.
l
to
and
B
it
> 1,
the
series for
-y- {z
A <B Jv
(z)}
which
is satisfied if
and
/So
'
(?)
such that
/3
^/8
2
.
15*26.
It is not possible to prove by the methods of 15*25 that Yv (z) has no complex zeros in the region|| in which arg2 < ir. But it has been proved by SchafheitlinU that Y (z) has no zeros with a positive real part, other than
|
|
more
generally,
<$v
(z).
'"ri
Yy {-z)2i cosi'T7(-), and hence, by 3-63 (1), Yv (z) cannot vanish unless v is half of an odd integer. This type of reasoning is due to Macdonald, Proc. London Math. Soc. xxx. (1899), pp. 165179. In this paper Schafheitlin also sub**I Archiv der Math, und Phys. (3) i. (1901), pp. 133137. When arg*=ir,
jects the
Y,(z) = e T
complex zeros of Yi
(z)
to a similar treatment.
15*25-15'27]
483
For
let /9
is
so that ft
also a zero of
Y (z).
Y (z), and let j3 be the conjugate complex, Then, by 511 (8) and 351 (1),
l't7.(/3t)Y,(frt)dt
=
and
/s^ft2 L
Fo(/3ar)
-is
I,(A * )
"r-J-^09--^ u g'
so, if /9
= pe*", we
have
and the expression on the left is positive while the expression on the right negative when to is an acute angle.
15*27.
is
the zeros of
Jv (z).
The proof which was given by Fourier that the zeros of J (z) are all real was made more rigorous and extensive by Hurwitz*, who proved (i) that when
v
>
1,
the zeros of
a positive integer or
v lies
has 4s + 2 complex which 2 are purely imaginary, (iii) that, if s is a positive integer and between 2s and (2s + 1), Jy (z) has 4s complex zeros, of which none
To
we use the
notation of 97.
We
positive zeros,
zeros, (iii)
take the function g2m,(0 which has, in the respective cases (i) (ii) in 2s 1 positive zeros, 1 negative zero and 2s complex
2s complex zeros.
/
We
now prove
that, if
2 f (f) = =o n (y + n 1
!
yn
ii
-r
i)
f ()
has at least as
many complex
zeros as g<m,*(Z)'
After Hurwitz,
we
'
write
<Pm {?,
V)
i-q,
1' "
where
<f>m (>
rj
= it}. The
(:
v) are easily
shewn
to
be
1
;
a real function, K
' \
it
follows that if
is
a complex zero of
<M
Again,
it is
</>
*?)
<>.
not
difficult to
( 9*7)
(ft
/)
that
w+ i (fc
*)
<*>,
for
246266; cf. also Segar, Messenger, xxa. (1893), pp. 171181, a discussion of the Bessel coefficients. The analysis of this section differs in some respects from that of Hurwitz ; see Watson, Proc. London Math. Soc. (2) xix. (1921), pp. 266272.
* Matft. Ann. xxxin. (1889), pp.
484
Hence,
[CHAP.
XV
is
the curve
negative
m when
<j>
(, 17)
<f>
=
,
lies in
rj) is
m (ff, rj)
m+1 (f
lies
<j>
wholly inside
,
m (f
rj)
0.
zeros of
g^, (f)
lie in
bounded regions of
Now,
since,
by 965, 97,
Jv
r(!/+2m +
l)
can be made arbitrarily small in any bounded domain of the f-plane, by taking m sufficiently large, it follows from Lagrange's expansion * that the number of
number of zeros ofg2mv () and so / () has 2s complex zeros. None of these zeros is real, for if one of them were real it would be a limit point of two conjugate complex zeros of gtmiV (), and so it would count as a double zero of/(); and/() has no double zeros.
zeros off, () in
in that area
is
when
m is sufficiently large
Again, from the series for/ (f ) it is seen that, when v lies between - (2s + 1) and (2s + 2),/ () must have one negative zero, and it cannot have more than one negative zero, for then gm> v (f) could be made to change sign more than once as X varied from to 00 [since g*m,,{C) can be made to differ from /() by an arbitrarily small number], and this is impossible.
zeros.
If we replace f by \z*, so that negative values of correspond to purely imaginary values of z, we obtain the results stated in the case of Jv (z).
For a discussion of zeros of Besscl functions in association with zeros of polynomials based on rather different ideas, the reader should consult Lindner, Sitz. der Berliner Math. 5. It may be mentioned that Hurwitz has extended his results to Ges. xi. (1911), pp. 3
2531.
15*28.
It has
Bourget's hypothesis.
that,
when
wi.
v is
positive integer
zeros,
have no common
other
seems that this theorem was not proved before 1929, except in some asm = l and 2 (cf. 15'22).
The formula
Jv+m (z)
* Cf.
Modern Analysis,
95.
15*28, 15*3]
485
shews
that, since
Jv (z)
of J(z) and
/+, (z)
zeros, the
common
zeros
Rm-i,y +1 {z) = 0,
i.e.
number when
been proved by Siegel* that J,(z) is not an algebraic number and z is an algebraic number other than zero; hence follow theorems which include Bourget's hypothesis as a special case.
v is a rational
Jp (z) and Jv+m (z) have no common zerosf; for such zeros are algebraic numbers and it is known that no algebraic number % can satisfy the equation
When
v is half of
an odd integer,
it is
cot (z
- \vir - tt) -
pfo
y)
when
v is half of
ir
%
an odd
integer.
is irrational
is
may be
rational
Jv (z)
when v is rational an
;
yields
inspection of the polynomial R m -Jt v +i(z) now immediately an elementary proof of Bourget's hypothesis in the cases tn 3 and m = 4.
15*3.
amount of
interesting information
by a
discussion of the
Jy (x) and related functions, when v is positive, differential equation satisfied by Jv (x) together with
we
shall
now
zeros.
The reader will find a more systematic investigation of these theorems in various papers by Schafheitlin, notably Journal fur Math. cxxn. (1900), pp. 299 321 Archiv der Math, una" Phyt. (3) I. (1901), pp. 133137; Berliner Sitzungsberichte, in. (1904),
pp.
8385.
. . .
will
be
...
called j , j,', j The smallest positive zeros of " will similarly be called yv , yv\ y v
Y (x),
F' (x),
Y " (x),
v
We first prove
(1)
It is obvious
*
that
j*>v,
jj>v.
70.
numerous applications of SiegePs fundamental theorem. t This was noticed by Porter, American Journal of Math.
5153. I Gf. Hobson, Squaring the Circle (Cambridge, 1913), pp. 44, Some related results are due to Watson, Proc. London Math. Soc. (2) xvi. (1917), pp.
171.
165
486
[CHAP.
XV
equation
it is
<
and
with
(x)
is
positive,
xJv
'
(x) is positive
and
increasing,
and so
(x) increases
x.
Therefore, so long as
functions so that
J,,
< x < v, both Jv (x) and xJJ and jj cannot* be less than v. vJv"{v) =
-J '(v)<0,
v
and
so
J "(x)
v
zero.
(2)
j,"<p.
Next, since
T J
,
.,
+2
(x)
2(i/ v
+ l)L
>
|1
v(v
-^
+ 2)\
J
T Jv (*)
,.
|1
L ~
2(,/
+1)0/ +2)1
-A
(),
the expression on the right is positive so long as x < v + 2. Now, if ^V were less than V{" (" + 2 )} the expression on the right would be negative when x is equal to jj (which, from a graph, is obviously less than jy ), and this is not the case.
(
Therefore
(3)
jJ><J{v{v
15*22 it follows that
2)).
Now, from
JJ
so that /(>)
<j v
and
Jv+*{j*)
If
now we put
x=j
'
in the
formula
it is
obvious that
(4)
jv'<j\2v(v
l)}.
Similarly,
(/
by putting x =jv
2) |l
in the formula
+ 3) Jv (x) + 2 {v +
that
2 {V
+ 3)
|
</,+, ()
+ ( + 1) J* () = 0,
we deduce
jr
and therefore
(5)
* Cf.
15-3]
487
In
j. (*)
+ /
1)
=-
v jl
- ^=i j Jv (x) - (v +
J/ (*)
J\v( v
-l)}<j w"<W-l).
Some
complicated formulae
+1
(x)
- Jv (x) R
4 +3
(x),
R
i.e.
s,
v+ i(j)>0,
3jv*
2
- 16 (v + 2) {v + 4) ja +
16 (v
1) (v
2) (y
+ 4) (v +
5)
> 0.
Since j is certainly less than f (v + 2) (v + 4), by results already proved, j v must be less than the smaller positive root of the equation
3a; 1
16 (v
2) (p
+ 4)^ + 16 (v +
1) (p
+ 2)(i/ + 4) (p + 5) = 0,
5)}.
j,<V{M" +
!)("
+
,
3,
,+a (#)
(x)}
j;>V^(^ + 2)),
i>
and,
when
>
4,
(9)
j/WK' +
x*
S)},
'
these inequalities being derived from the consideration that jv the positive roots of the equation
lies
between
- 3 (p +
2 2) x*
+ 2i>
(i>
1)
(i/
3) (v
+ 4) = 0.
The
discussion of
y v requires
slightly
We
use
J*(x)+Y*(x)
is
a decreasing function of
this is obvious
from
13 73.
Hence
it
follows
that
in this interval
positive, since
(x) is negative,
is
and
it
JJ (jy )
obviously negative.
Hence
(10)
jJ<y
<j,-
488
15*31.
[CHAP.
XV
It has already
(x) cos
and so there are an infinite or 9%, (x), has an infinite number stationary. Such values of x number of positive values of x for which it is /i which exceed the order v (supposed positive) will be called lf 2 fi3 ., where
//.
,
. .
/x 1
</i2
</tt3
<
....
We
The
shall
first
is
that
A (x) defined as
-2x*<@; 2 (x)/(x*-v*Y,
and
so
2)
....
Since
(/*),
is
now
evident.
more interesting
(7-21)
^r(*)(^ycoB(* + a-W-iir)+o(^).
This indicates the possibility of proving inequalities consistent with
i^,(/*)|
= 0(l/vV)
when
/in is large.
It can in fact be
(I)
fa,
shewn that
1/
|
<@v (x) when x takes the values fa, The values assumed by (aj2 9 )* c members are less than *J(2/tt). sequence whose an increasing ...form
\
fa,
(II)
A*r+2
form a
(i)
The values assumed by x^\9^,(x)\ when x takes the values fa, fa+i, decreasing sequence whose members are greater than \f(2jir)
provided that
v> V3,
(ii) fa'
13)}/(4i*
- 3).
A (x) <@S (x) + 2B (x) <@, (x) <&,' (x) + G (x) <$;* (x) = @ (*),
where
A (x), B(x),
=
We
have
' (x)
[A' (x)
-2(x*-v*)B (a-)/* ^V
8
}
(*)
+ 2 {R (x) + A(x)-B (x)/x - (a3 - v*) G ()/*} + {C (x) + 2B (x) - 2G(x)/x] <&'* (x)
= D(xY$J*(x),
where
London Math.
170171.
15*31, 15*32]
489
provided that
A (x)
is
B (x)
[B' (x)
v*).
(3a*
and so
@ (a?)
is
is
= 2/7r.
see that
Since
1, 2, ...,
v9)* *^
(/*)
2
we
\
values
v )*
9%v (/*)
less
than V(2/?r).
(II)
If
A (x) = x,
then
4
2
<0,
provided that 4n?
> 3 and x
a
(4v -3)ar*-8i/2
(^ + l)a* +
i/*(4/
- 1) = 0.
In this case (a?) is a decreasing function and we can apply arguments, similar to those used in theorem (I), to deduce the truth of theorem (II).
15*32.
Schafheitliris investigation of the zeros
o/J
(x).
By means of the integrals which have been given in 6*12, it has been shewn by Schafheitlin* that the only positive zeros of J (x) lie in the intervals (m7T + f tr, mir +^ir) and the only positive zeros of Y (x) lie in the intervals (mir + \ir, mir + 7r), where m = 0, 1, 2, ....
We
(x),
with slight
modifications,
and then we
shall
of the type
J
(where v
lies
(x) cos
and $), by the methods used by Schafheitlin. and %ir of a. Schafheitlin's investigations were confined to the values
between
From an
612 (7),
6
'
Jo(
it is
'"ttJo sinflVcostf
obvious that,
+ \ir,
,
and
so
sgn Jo ()
+ it).
3144.
490
[CHAP.
XV
J (x)
(m 4- 1) ir 0,
J o
and then
/.(.)
sin
cos
The
last
integrand
is
<
< 20
or
20 <
is
<
\tt.
sin
is
ycos
is
an increasing function* of
an acute angle.
between and 20 there corresponds a value which sin {\0 0) has the same numerical value, but has the positive sign, and the cofactor of sin {\0 <f>) is greater for the second set of values of than for the first set. The integral under consideration is consequently positive, and so J (x) cannot have a zero in any of the intervals (mir + fair, mir + ir). Therefore the only positive zeros of Jo (x) are in the intervals (mir + ir, mir + %ir).
to each value of
for
Hence
15*33.
Theorems of Sckafheitliris
type,
when
< v ** \.
We
where
shall
now extend
and
^P (x)
Jv (x) cos a Y
(x) sin a,
<a<
ir
\ < v ^ \.
the crude result that the only positive zeros of
^f (x)
in the intervals
(mir
+ f 7r + \vir a,
mir
+ ir a)
*A) ~~r{v
for,
when
we have
and
so, for
sin-^ + i)r{i)i mir a < x < mir + ir 4- \vir a, sgn [sin (x + a v0 + \0)~] = sgn ( l)w
x,
aff>
Consequently the only zeros of 9y (x) lie in the specified intervals, and there are an odd number of zeros in each interval, with the possible exception of the first if a > f ir+ \vir.
such values of
v
(x) is
not zero.
in the intervals
(mir+ f 7r + \vir
* Its logarithmic derivate is
a,
mir
+ \ir + \vir
+ J tan 0.
(2x
- sin
cos 0) cosec 2
15*33, 15*34]
491
ir,
where
m = 0,
except that,
(""
if
is sufficiently
near to
there
may be
We
shall
a).
Write
x = (m + l)ir-a-(l-2v)4>,
an angle between
where
is
and
tt.
With
this value of x,
(-)*2"* f* cos-*flsin
{(1
- 2) (\0- $)}
^^.^
To each
and
value of
between
and
2<f>
\ir for
which
sin{(l 2v)(\0-4>)}
Again
is
an increasing function of
2a?
provided that
> max
(2i/
+ 1) sin
cos
+ (v - 1) ^-^
^ J.
and
when x > %
since v
Hence,
if
x > and
mir + \ir
we have
and
this proves the
= sgn (- 1 ) m+1
when
15*34
Theorems of Schafheitlin's
function
*$w (x)
\<v<.
^a<
ir,
as before, in which
now supposed
that
\<v<\.
zeros of
9!> v
W e shall first prove the crude result that the only positive
lie in
(x)
the intervals
(mir
o,
mir \ir
+ \vir a)
. . . .
*+'*'
(,,
tfl
^.
for
when
*
a< x < (m + \) ir a,
(x)
15-24,
we
see that
there cannot be
t If x<i and m=0, the reasoning fails when T + vir-a<l. t If a>(if- \)t, the interval for which m=0 is, of course, to be omitted.
492
[CHAP.
,
XV
we have
+ a v0 + $ 0)] = sgn ( l) m
obvious.
in the intervals
(mir
tt +
+ %vir a),
near to
ir,
m = 0,
is sufficiently
there
may be
a),
a,
^7r 4- \vir
a).
We
f(0)
that,
is-
15 33
only now,
if
6 =/($),
it is sufficient to
prove
when
then
f(2<f>-ir)<f(24>+ylr).
To
4l
T^r^rl - **
2i>
+i
sin (20
+ +) sin (20 - ^
I J
)
[cosec 2 (20
an increasing function of
[cosec2 (20
and
therefore,
a fortiori,
by an increasing function
if it is
is
because 40
an acute angle
if
and so
%_T(
is
alwayt positive
positive
when ^=0,
and
i.e.
when
v
<f
+ &, lie
in
(i7r
\tr + \inr a,
mar
+ a v0 + \0)
now be
explained.
We shall now
which exceed
(2i/
prove that,
if v
> J,
those zeros
+ S)/ir lie in the intervals (mir a + %vir + \tt, mir a + \vtt + f tt)
(2v
where
15-35]
493
made lengthy and obscure by the use of arguments equivalent mean-value theorem when the explicit use of that
theorem
is
obviously desirable.
As
so that
'
= Jv (x) cos a
(x) sin a,
(, "r(.+j)r(i)l (
sin^tf
cot
dtf
r (v
f )T
(|) J o
33
cos*+i
dtf -
Now
as
cot2v+1
e- 2*^49 increases as
2
increases from
2
to
.
.
increases from
It will
to
77-,
where
2 is
= arc tan-
be observed that
when x
is
large
compared
with
v.
Now
suppose that #
rrnr
lies
between
mir a + \ir
(v
a + \ir (v |) and
|) 4- f 7r,
^ so that
X + a
{V
0!
W7T.
2y-l
7T
2i/+3'2
< ^1<
2,
2y
2i/
+ 2 ir + 3'2' v +6
provided that
arc tan
+h
x
<
4i>
We
so that
is
2.
Then, by the second mean- value theorem, there exists a number and X) such that
{
O,
between
+%)
!o
sin2
"^
J,d0\
atf
cos'+itf
^+i
cos(x
+ a-v0o -l0
cogi>+}
o)
|.
* Journal
(1900), pp.
299321.
494
[CHAP.
XV
is
stationary
such values of
the
fraction is equal to
/cos* * 0.
Hence nenCe
co S (x
+ a-v0
-W
(0, 0^),
"fl>
cob*+0,
and therefore
SDT1 Sgn
(
cos ( m7r
+ft)
_ cos (g + + a
v0
-*fl>) |
is
cos-**
and
^7r,
cos"+*0o
_ Sm /_ ! vm gM }- s 1>
a;
+ \0)
when
between
we
under consideration,
= sgn (-
l) m
a + vir \ir,
a + \vir + \tr,
mir
a+
(mir
mir
a + \vtt + J ir),
and v
is
and
when a =
an
integer.
The reader will observe that this theorem gives no information concerning ^v {x) when v is large it will be apparent in 15*8 that there are a large number of zeros less than (2i + l)(2i/ + S)/tt, and that interesting information can be obtained concerning them by using Debye's
the smaller zeros of
;
integrals.
15*36.
Bdcher's theorem
on the zeros
of^
(x).
A result of
discovered by Bocher from a consideration of the integral formula 11*41 (16). The theorem in question is that 9^ (x) has an infinite number of positive zeros,
2;
where j
is
the
To
= 0, z=j
(Z)
in 11-41(16),
and then
f ^o () d<f> - tt^
.'o
(>)
= 0.
to-7r, i.e. as w
increases from
Hence %\{^t) cannot be one-signed as increases from Z to Z+j and so ^(nr) must vanish
<f>
;
value J of
in the interval
),
(Zj Z+j
,
).
Since
is
is
number
lie
(greater than j
Bocher's theorem
now
evident.
+ 5)jr
for
v. (1899),
pp. 385388.
Cf.
Modern Analysis,
3-63.
15-36, 15*4]
[Note.
495
By
as = cv
du
\
v
I
CV
,
where
u, v are
two solutions of ^?
oar
+ ^-s+=0 cy2
curve
2
and
By
taking
v=J
must vanish
at least twice
{J(a?+y )} and the curve to be x2 +tf2 =*jo2 Weber* deduced that u on any circle of radius^.
^n
is
(r) cos n6 satisfies the requisite conditions drawn with centre on the axis of x and
subtending an angle
circle.
than
ir/n at
the origin,
$n (r)
Hence the
positive zeros of
,
wn {r)
+ cosec 5-
These results are of interest on account of the extreme simplicity of the methods used
to prove them.]
15*4.
On
the
We shall
Jv (z)
as a Weierstrassian product,
and then develop expressions involving quotients of Bessel functions in the form of partial fractions but as a preliminary it is convenient to prove the following theorem, which gives some indication as to the situation of those zeros of Jv (z) which are of large modulus. In this investigation it is not supposed that v is restricted to be a real number, though it is convenient to suppose
;
that v
is
When
is
some
now be
proved.
Let
G be
iB
where
\iril (y), iB
B is a (large) positive number. We shall shew that when in is a sufficiently large integer the number of zeros of z~" J {z) inside G is precisely equal f to m. Since z~* J (z) is an integral function of z, the number of its zeros inside
v v
Cis
dw
1 [ liri) c
JrlM dw
J(w)
(1869), p. 10.
a real negative number (and for certain complex values of v) there may be pairs of zeros on the imaginary axis in such circumstances the contour C has to be indented, and each pair of zeros is to be reckoned as a single zero.
When
v is
496
[CHAP.
XV
We now
on
all
in turn.
It is first to
be observed that
the sides of C,
#( (w)
+1?2 ,(W)],
is large.
(w) and
r)
2<
(w) are
(l/w)
when
||
Now,
B -* oo
%/,(>)
fiB+lmHo) /""
idw=\il{v).
iB-lniI(v)
this is equal to
tB+l+Jnr+Jir
fiB+mTT+Ji/ir+iir
Q
\
.
/^Y)
s [ + **(>) + ^+-5.i
7>v
2v
+ l.
log
iB + mir + ^vTr +
^ + ^-
lir
/(l/)
n/ 1/m +Q(i/J)J
~|
as
B oo.
-iw + iflfcO + i,
same
is
value,
and
now considered
m + \v + .
and to do
Lastly
this
we have
we
first
when w
|
is
large, is equal to
riB+mv+iw+ir
Now
and so
tB+Hr+J>w+J
1_ riB+mw+ivw+i*
2iri J
jr+i ( W )
T ^kW
dw
-u iB+mv+iv*+\w
~-i(2i/ + l)+0(l/m).
Hence the
limit of the integral
is
m+
(l/m).
* Allowance is
15-41]
If
is
497
sufficiently large,
less
we can
;
(1/w)
numerically
integer, it is
must be an
That is to say, the number of zeros of z~*Jv {z) between the imaginary axis and the line on which
exactly
ra.
Note. The approximate formulae quoted for the functions of the third kind shew that the large zeros cannot have a large imaginary part and so all the zeros of / () lie inside a strip whose sides are parallel to the real axis and at distances from it which are
;
bounded when
15*41.
It
is
is
bounded.
infinite product.
express
Jv {z)
as a product
of 'simple
factors'
of
Weierstrassian type, each factor vanishing at one of the zeros of Jv (z). In order to express J,(z) in this form, it is convenient first to express the
logarithmic derivate of z~ v J {z) as a series of rational fractions by MittagLeffler's
theorem*.
are
|
The zeros of z~y Jv (z) R(h,n) >0 and R(j r,i) <
|
jv,s>
wheref
Wtl
,
^(>
>2)
jy<3
...
being
all
unequal (15*21).
iB,
jv<2
whose
where A and B are positive, and we suppose that j>m are the zeros of highest rank which are inside the rectangle.
vertices are
We now
consider
z
2irtJ*J)W(W 2
/,+,
(w)
Z)' J (w)
dw,
is
where z is any point inside the rectangle, other than a zero of J (w), and v not a negative integer.
z,
>,i, >.a
>
The
residue at z
is
+ Jv,n
since
by
3*2.
It follows that
=
* Acta Soc. Seient. Fennicae, xi. (1880), pp.
2-rri
]j)W(w z)
.
If
jy>n )=0
for
any value
of n,
we choose jvn
498
[CHAP.
XV
We
Jv+1 (w)/J,(w)
an odd function of w,
to consider the
right-hand half of D.
We
we take
take
A = Mir + M (\v \) where M a positive integer and then M to be at least so large that M = m, which possible by 154, and
4ir,
is
is
functions
iy r
,i(w),
*7 f2
(w), defined in
be
J in absolute value.
Then
bounded whenever
g
(W\mr~J)
and when the expression does not lie within bounded and w is not arbitrarily near a zero of J, (w) so that, from the asymptotic expansion of 7*21, J, +1 (w)/Jr (w) is bounded on
;
these limits,
I (w)
is
That
is
to say
Jv+l (w)jJ
(w)
is
the rectangle
D as B and
M tend to
Hence
2itijdw(w z) J p {w)
and therefore f
1
K '
if
_ JjtiW,,,,^
jv,n)
Jy(z)
integrate,
n=l
l*->.n
find that
=1 \*
When we
we
This
is
This formula was assumed by Euler, Acta Acad. Petrop. v. pars 1, when v=0, aud subsequently by various writers for other values of v
cf.
15*5, 15'51.
due in substance to Graf and Gubler, Eirdeitung in die 130, and it was given T/ueorie der BetseVschen Funktionen, I. (Bern, 1898), pp. 123 explicitly by Kapteyn, Monatshefte fUr Math, und Phyt. xiv. (1903), pp. 281282.
The
is
Because
TTa
**
t If we
AiB, -A'iB, we
'
15*42]
499
power
series;
the
such a series have tieen expressed as determinants by Kapteyn, Proc. Section of Sci., K. Acad, van Wet. te Amsterdam, vm. (1905), pp. 547 549, 640 642 Archives Neerlandaises, (2) xi. (1906), pp. 149168. Some associated formulae have just been published by Forsyth, Messenger, L. (1921), pp. 129 149.
15*42.
An expansion which, in some respects, resembles the partial fraction formula obtained in 15*41 is as follows
in which
x and
while z
to take
and v are
it is
convenient
R (z) > 0.
The expansion was discovered in the case i>=0, as a special form of an expansion occurring in the theory of integral equations, by Kneser, Math. Ann. lxiii. (1907), pp. 511 517. Proofs of this and of related expansions for integral values of v were published
later
by Somnierfeld, Jahresbericht der Deutschen Math. Vereinigung, xxi. (1913), pp. 309 method of proof has been criticised adversely by Carslaw, Proc.
Soc. (2)
London Math.
It
xm.
(1914), p. 239.
be noticed that the expansion has some connexion with the expansions which will be discussed in Chapter xvni.
may
Fourier-Bessel
H Q) - H (Xw)
zt-w*
is
(w)
Jv (xw)
J,(w)
dw W
'
left side
of the rectangle
The
integral round the indentation tends to zero with the radius of the
;
when x and
\R. (Xw)
when
-*-oo
Hence the
limit of the
and so the
sum
Now
the residue at z
FTE7 {J
"
{Xz) Yy {z) ~ Jv
{z)
F" (Xz)]
'
500
[CHAP.
XV
( j,
M ) (*
-j\ n )}
Vi
v>
v \Jv,n/\ z
~~
\J'(A
\_J
(X
\Y'(i
v,n)
4t Jv (j*,
nX)J
(j v< n x)
Jv (xw)/Jv (w) by the contour integral, see Carslaw, Proc. London Math. Soc. (2) xvi. (1917), pp. 84 93; Carslaw has also constructed some similar series which contain Legendre functions as well as Bessel functions, and these series represent the Green's functions appropriate to certain physical problems. See also Beltrami, Lombardo Rendiconti,
For a generalisation of this expansion, obtained by replacing
96 V {xw)l<$v {w) in
336
15*5.
(2
*Jz).
An
of
was
devised by Euler*, and applied by him to determine the three smallest zeros
(2Jz).
magnitude be
a^ot,,
...,
then by
15*41,
j (2v*)= n(i-).
w
by Euler;
if it
-log/ dz
(2\/*)=i
oo
w=1 an
oo
- 2
n=l
m^O^n
then absolutely convergent.
;
provided that
00
<a
t ;
and the
last series is
Put
l/am+1
then
--J
Replace
(2</z)
=J
(2Jz)
2 a^z.
(2>s/z)
on each side by
p+p
2s
"~
1*
2s 3 s
.
'
"
* Acta Acad. Petrop. v. pars 1, (1781) [1784], pp. 170 et teq. Math, xxxiii. (1846), pp. 363 365 should also be consulted.
From
and unequal.
15-5]
501
multiply out the product on the right, and equate coefficients of the various
= <T \ = <r <r = 0-3 O-2 + ^O-j, -dbf = < + o" gkr sdsv = a - a* + i"s - ^rO"2 + sh "i
1
lt
2
x
,
-1^
<*4
<>
whence
o-i
o-6
= ^$r,
Since
<
0^
<
02
<
a3
<
. . .
it is
evident that
,
and so
<rm+1
<*!
By
m
1
^ml^m+l
1000 000
1-414 213 1-442 250 1-445 314
1 1
2-000 000
1
-500
000
1-454 545
1
4 5 6
-447 368
1-446 089
By
writing
l/On^d-',
-2
< am
<x'
m+1
<
<r'
m /a.z
Euler deduced that a 2 = 7*6658, and hence that o3 =1863, by carrying the
process a stage further.
These results should be compared with the values
ax
pp. 153155.
= 1-445796,
a2
= 76178,
a3 = 18-72,
derived from the Tables of Willson and Peirce, Bulletin American Jiath.,Soc. in. (1898),
The value of a, is given by Poissont as 1 446796491 (misprinted as 1 46796491); according to Freeman J this result was calculated by Largeteau for Poisson by solving the quartic obtained by equating to zero the first five terms of the series for J (2 Jz) the magnitude of the sixth term is quite sufficient to account for the error.
;
* This
system
is
for
an algebraic equation,
M4m.
t Proe. Camb. Phil. Soc. ra. (1880), pp. 375 377. Thiorie Analytique de la Chaleur, p. 310, footnote.
Of.
La
502
15*51.
[CHAP.
XV
to calcu-
v (z).
15*41)
and writing
2 -^
=i J"
i
v,n
we
a-
find, after
Rayleigh, that
'"=
2a
(i'
l)'
a<
2>
24
(i*
'
28
(l/
3)(v
+ 4)'
7i/+19
29
(i/
+ iy(v +
2)<(v
+
/i
S)(v
4)(j/
5)'
The
(z)
and
(z) are
deduced to be 2*404826
and 3831706.
Immediately afterwards Cayley f noticed that o-( r can be calculated rapidly when r is a power of 2 by a process which he attributed to Encke J, but which is more usually known as Graeffe's method of solving an equation.
>
The method consists in calculating <r( r when r is a power of 2 by starting with the given equation and forming from it a sequence of equations each of which has for its roots the squares of the roots of its predecessor; and aJ?) then rapidly tends to a ratio of equality with l//2r,i.
)
ov< 8 > to
be
+ l)8
(i-
+ 2) (i>+3) 2 (v + 4)2
(i/
+ 5) (v+6) (v + 7) (v + S)'
>
It
was observed by Graf and Gubler|| that the value of ov<r can
o-
easily be checked
by the
formula
J
W=22*;
i?r/(2r)!,
where
Br is
number
this formula is
equation
J *w-(s)* 8ini
Extensions of some of these results to the zeros of zJv (z)+hJv (z), where h have been made by Lamb, Proc. London Math. Soc. xv. (1884), p. 273.
'
is
a constant,
and 1, has recently been The smallest zero of Jv {z\ for various values of v between tabulated by Airey, Phil. Mag. (6) xli. (1921), pp. 200205, with the aid of the RayleighCayley formulae.
London Math. Soc. v. (1874), pp. 119124. [Scientific Papers, i. (1899), pp. 190195.] t Proc. London Math. Soc. v. (1874), pp. 123124. [Collected Papers, ix. (1896), pp. 1920.] t Journal fur Math. xxn. (1841), pp. 193248. Die Aufibsung der hoheren numerischen Oleichungen (Zurich, 1837). Einleitung in die Theorie der Bessel'schen Funktionen, i. (Bern, 1898), pp. 130 131.
I!
* Proc.
15-51, 15-52]
[Note.
503
The procedure
sum
its largest root seems to be due to Waring, Meditationea Analyticae (Cambridge, 1776), p. 311 other writers who were acquainted with such a method before Graeffe are Euler (cf. 15*5); Dandelin* Mem. de VAcad. R. des Sci. de BruxeUes, in. (1826), p. 48; Lobatschevsky* Algebra, or Calculus
;
257.]
15*52.
The most
method
is
of calculating the
is,
in substance,
due to Stokes f,
some
Stokes' method will be sufficiently illustrated by his own example J whose zeros are the roots of the equation
(&),
7*3.
It will
(x,
,
P (x,
0)
and
0) are
zw P( *'
m 0) ~ X
ax 0)
-27(8^ +
1
1.9
1.9.2.5.49
^!(8^
1.9.25
nt *' Q(
For
Q (x,
0)
is
negative
The
function cot (x
\ir)
it is
is
x = mr \ir,
and so
\v)
that there
one zero
\tt, wit
+ \ir),
denote the (r
to-1
l)th terms of
P (x,
m-l
0)
and
Q (x,
0)
we may
P (x, 0) =
where 6 and
* I
1
ur
+ 0um
Q (x,
# and
0)
= 2
vr
-I-
vm ,
m which
lie
between
and
1.
350
( 6-4)
and
(x), for
verify,
by
{1-P2-Q2}/P2,
where P,
positive.
stand for
is
ultimately
504
[CHAP.
XV
cot(a?-^7r)
^ 2
r=0
V,
Vm
,
ur
0um
in
which
it is
values, are
If a? were supposed complex, there would be a number of contours in the a?-plane each of which enclosed one of the points nir \ir and on which cot (a? \ir) exceeded the modulus of the quotient on the right.
|
By
panded
Biirmann's theorem * the modified equation would have one root inside
mr tt, and
this root
can be ex-
powers of
mr \tt.
We
thus obtain an expansion for the root of the equation in the form
*
'
r=0 (nir
- Sir)
;
and
it
is
and X r (0, 0i) are independent of n but depend on readily perceived that the first in of the coefficients are actually
independent of
and
lt
so that,
when
r
r
x
<
m, we
r.
may
write
f
Now
the
{0
)=f
as
is a bounded function of and and 1 vary between and 1 and it is clear that the upper bound of the modulus of the function in question is 0(n~2m_1 ) as n~*- oo Hence, when and X are given their actual values which they have at the zero under con-
sum
sideration, the
sum
is still
(w~ am_1 ).
That
to nir
is
to say, it has
^7r),
and
its
value
been proved that there exists one zero (nearly equal may be written
J
,
= (mr-%Tr)+ 2
w+1
is
(n--*)-
zero
\ir
first
2
,=o (nir
^iryr+1
.
^~
If
where
yjr
as x
-*
then
tan
y oo 8a?
* Cf.
33
512a?
'
3417
16384a?6
% 7'31.
Modern Analysis,
15*53]
505
sothat
^~8^-384^ + 5l20^"-*
,
25
1073
and
25
1073
5120a?
-( x-{mr
This series
3779
is
all
the zeros of
{pe),
to at l eas*
nve
1.
15*63.
any cylinder
function,
- \vrr - i^ + ) = p^
It seems unnecessary to prove the existence of such zeros (with large positive real parts) or the fact that they may be calculated as though the (z, v) and Q (z, v) were convergent, because the proof differs from series for
The expression for the large zeros of a cylinder function of any given order was calculated after the manner of Stokes by McMahon*; but the subsequent memoirs of Kalahnef and Marshall J have made the investigation more simple
and have carried the approximation a stage further with no greater expenditure of work in the calculation.
Following Marshall we define two functions of
equations
z,
called
M and
ty,
by the
M cos
ylr
= P(z,v),
M sin
yfr
= -Q(z,v),
oo
if cos (z
J
AnndU
of Math.
ix. (1895),
pp.
2325
225232.
t ZeitschriftfUr Math, und Phy*. lit. (19Q7), pp. 5586. t Annals of Math. (2) xi. (1910), pp. 153160. Cf. Nioholson, Phil. Mag. (6) xix. (1910), pp. 228249.
W.
B. F.
17
506
[CHAP.
XV
{j$} -z-hw-l'*-*,
and use
and,
when we
363 (3), we
find that
cty
2/(irz)
dz ~JS(z) + Y/(z)'
so that,
by
7*51,
When
ljz 9,
is
expanded
as far as the
term involving
we
<ty
dz
_ fi-1 _
2 *
3 2
(p
2V
2 10 ^
-s-
(n
- 1) (5/r>
1535/* 2
+ 54703/a - 375733)
'"'*
2 15 zs
in this equation
integration, that
/x
It follows,
on
/*-! _r r~
23 s
(/i-DQ*-25)
3.2 7 23
+
and
so the equation to
(/i
- 1) (5/x3 - 1535/*2 +
7.2
is
15
375733)
'"'
be solved
-M7T - \VTT
7r
^TT
+ a ~ - -^
g-^V
is
"
If
/3
= (n, + |v -
^)
a,
7
Zrvft
^-1
2
s
( /i
-l)(7 /A -31)
3 2
.
Qt-l)(83/t a - 982/*
15 2 10/38
.
+ 3779)
ft
Yv (z) sin o
are given
by the
asymptotic expansion
(n
+ \v \)ir
This equation
-l 8 {(n + \v - \) ir - a}
4i/
2
(4p-1)(28i/ 2
-31)
ir
384
{(n
+ %v- \)
aj
= \)
1797, but
no clue
is
was given by Gauss in his notebook with the date Oct. 16, it. [Cf. Math. Ann:
15-54, 15*6]
[Note.
507
analysis in a
The fact that J* (z) + F2 (z) has a simple asymptotic expansion shortens the manner which was not noticed by Marshall he used the equations
;
latter
l+
z*
J^
J/3'
series for
M.]
to cylinder functions.
The method
and of
when the
cylinder function
a Bessel function of
CLZ
the
first
or second kind.
for the large zeros of 9$J (z) is
fi
^
where
fi 1
+3
7/x,
+ 82fi - 9
8
= (n + ^v + \)7r
*
zeros ot
*-* is
AThe
where k
zeros of
is
fi
+7
7/a2
+ 154/a + 95
384ft
s
8&
constant, and of
Jv'(z)Yv'(kz)-Yv'(z)Jy'(kz)
have been treated in a similar manner by McMahon. Kalahnet has constructed tables of the zeros of the former function when k has the values 1*2, 1*5, 2*0 and v is 0, , 1, $, 2, # while it has been proved by Carslaw, Conduction of Heat (London, 1922), p. 128, that these zeros are all real when v and k are real. The zeros of Jv (z) Yv (kz) - YJ (z) Jv (kz) have been examined by Sasaki, Tdhoku Math. Journal, v. (1914), pp. 45 47.
'
15*6.
when
its
order
is varied.
The equation
has an infinite number of roots, the values of which depend on v since / (z) is an analytic function of both z and v, so long as z 0, it follows that each
;
an analytic function of
first
v.
when the
function of the
v
kind
is
replaced
is
(z) cos
y(,s)sina,
where a
any
is
If j denotes any particular zero of Jv (z), the rate of change of j, as v varies, given by the ordinary formula of partial differentiation
*'o->| +
P^
= 0.
* Annals of Math. ix. (1895), pp. 2529. t Zeitschrift fUr Math, und Phys. liv. (1907), pp. 55
86.
508
Since
zero,
[CHAP.
XV
(j)
= Q,
it
= Jr +i(j)j*0,
0,
so long as j is not
when R(v)>
"
(2)
of
Jv (x)
;
in-
Equation
R. des Set.
deduction from
was stated without proof by Schlafli, Math. Ann. x. (1876), p. 137 and the was established in a different manner by Gegenbauer*, M4m. de la Soo. de Liege, (3) n. (1900), no. 3, in the case of the smallest zero of Jv (x).
(2)
it
We
where p
= Jv (z) cos a
as follows
is definable
Yv (z) sin a,
is
constant
(i.e.
independent of p).
To prove
this
c is
l
{Jv (c)
,
r ;
is
constant, so that
dc[d
and therefore
\Tv (z)Y]
A*
(Y,(2
= 0,
W'JJ
ircdp
vx
[_
'
dp
'
dp
] z=0
Hence, by
(3)
~ = 2c
dp
p.
"
Jo
f
K (2c sinh
t)
r**dt.
c is
is positive,
this formula
shews that
an increasing
less
if c is
a zero which
is greater
p,
than
then
c is
an increasing function of
has been
when
p tends to
the equation
pit) = 0.
vitiated
in the latter part of Gegenbauer's memoir is by his use of Rudski's erroneous results ( 15*1). t Berliner Sitzungsberichte, v. (1906), pp. 82 93 ; Jahretberieht der Deutachen Math. Vereini*
pp.272279.
15*6]
509
complex
number and
a (complex) number, with a positive real part, then c is an r analytic function of v and so, as v varies, the zeros of @v (z) vary continuously, and they can only come into existence or disappear when c fails to be an
c is
;
analytic function of
v, i.e.
when
= 0.
(z) It follows that the positive zeros of p (z) are derived from those of except that one positive zero varies, as variation v continuous a process of by disappears whenever v passes through one of the specified negative values.
If
we now choose a
v,
so that*
(a/rr)
0$a<7r, we
variation of
no zeros disappear during the process ofzeros which are so large that the formula of case the and so in
is available,
.
l,
the formula
a ^ tt mr + *vtt * 2
gives the nth positive zero,
in order of
If,
8(rwr
4^-1 ^ + $ Pir-iir a)
;
...
when the
magnitude.
(a/Tr)
however, v has varied so that it finally lies between (a/7r) k and 1, where & is a positive integer, k zeros have disappeared, and so
&)th
positive zero.
584
it
is due to Macdonald, Proc. London Math. Soc. xxix. (1898), was applied by him to the discussion of the zeros of Bessel functions of the
1.
If we draw the curve ^* (y) = 0, it evidently consists of a number of branches starting from points on the negative half of the #-axis and moving upwards towards the right, both x and y increasing without limit on each
branch.
If
we take any
(<>, y<>)
it
and a
#-axis, it is evident
^x (y) ^
.
exceed v
equal to
meets each of the lines in the same number of number of zeros of $* (y ), qua function of v, which the number of positive zeros of ^, (y) qua function of y
which are
less
than y
who took
Pig.
This is a generalisation of a theorem due to Macdonald -f> and the cylinder function to be a function of the first kind.
33
Jx (y) = 0,
the length
A much
larger
diagram of the same character has been constructed by Gasser j, who has also constructed the corresponding diagram for F(y) = 0. The diagram for
<$x (y)
is
of the
as that for
Jx (y) 0,
(2)
except that
* This does not lead to any real loss of generality. t See a letter from Macdonald to Carslaw, Proc. London Math. Soc.
xni. (1914),
p.
239.
510
[CHAP.
XV
of
number
on which 2x
y
is
an odd integer.
L
Fig. 33.
[Note.
of 9BJ
(4)
(2),
find it interesting to
(3) that, if
d is a
zero
dd
and hence,
if
2d ^ = ~-^
C*
(c'
cosh 2t -
r 2)
dt,
d> v
|
v.
The sign of dd/dv has also been discussed (by more elementary methods) by Schafheitlin,
Jahresbericht der Deutschen Math. Vereinigung, xvi. (1907), pp. 272
279;
used by Schafheitlin
is
extremely complicated.]
15*61.
The mode of increase of the zeros of Jv (x) when v is increased has been examined by Rayleigh* with the aid of arguments depending on properties of transverse vibrations of a membrane in the form of a circular sector. If the membrane is bounded by the lines 6 and 6 tr\v (where v > ^), and by
* Phil.
Mag.
pp.
5358
[Scientific
Papen, yi.
Cf. Phil.
Mag.
pp.
54454B
444446].
15*61, 15*7]
511
fixed,
the circle r
= a,
placement in a
and if the straight edges of the membrane are normal vibration is proportional to
the dis-
The effect of introducing constraints in the form of clamps which gradually diminish the effective angle of the sector is to increase v and to shorten the periods of vibration, so that p is an increasing function of v, and therefore
(since a
and
an increasing function of
v.
v.
That
is
to
{x)
and
/'
using arguments of this character, Rayleigh has given proofs of a number of theorems which are proved elsewhere in this chapter by analytical
By
methods.
15-7.
The zeros of
(z).
number (zero v (z), where v is a given positive and z lies in the domain in which axgz < ir, have been studied qualitatively by Macdonald*.
The
zeros of the function
included),
| |
From
6*22, it is
obvious
&rgz \^\tt. This may be proved at once v (z) has no zeros for which from a consideration of the integral given in 1371 ; for, if z = reia were such a zero (r > 0, \ir < a < tt), then z = re~ia would be another zero but the
|
K that K
that
(z)
it
K
rr
^ = If" exp {- v ^
f
^cos2a)
^ Jo >0,
j* -)
/r*\dv
which
is
contrary to hypothesis.
If a
is
equal to +
|
\tr,
we have
and so
(z)
of z lying
and
or
may be shewn
is
number
\,
unless v
the number
In the
first place,
ir,
unless v
* Proc.
t This
is
London Math. Soe. xxx. (1899), pp. 165179. not the number given by Macdonald.
K
512
is
[CHAP.
XV
an integer
for
(re 1ri )
= j*"* Kv (r) + = 0,
sin vir
.
iri
Iv (r),
we must have
(r)
(r)
+ irlv (r) = 0.
is
Since the Wronskian of the pair of functions on the left of the equations (rr/r) cos vir, they cannot vanish simultaneously unless cos vir = 0.
Now
(z) as z describes
a contour
consideration
T and 7, their equations being z = R and z = 8, number of zeros of v (z) in the pair of quadrants under equal to the number of zeros of zv Kv (z) inside the contour,
j
and
change in phase of z v
(z) as z traverses
the contour.
Now
is
(z)^ -
arg {*"#(*)}
JjRexpirt
+\ug[rK
L
(s)}
Jexp(-ir)
two terms* tend to 2tt(v |) and respectively, because when \z\is large or small on the contour,
As
i2-*-oo
first
(z)
~ s*e-V(!*-),
zv K> {*)
~ 2--
V (v)
respectively f.
The
last
/-o\
o 2 Tarc tan L
4.
TrcoBV7r.Iv (r) v
.
. :
1R
,
.
'
(r)
+ 7ramv7r.Iv (r)j s
Now
{r) is
is
a positive
is
increasing function, and so the last denominator has one zero if sini^r negative, and no zero if sin vir is positive.
If therefore
we take
less
when
when
r-*-30
is
being numerically
sign of cos
vir.
the value assigned to the inverse function than two right angles and having the same sign as the
Hence the
total
(z) is
(z) in
v
*
I
This
I
is
valid
when
arg z
<
t.
t The second of these approximate formulae requires modification when r=0. X The two zeros of 2 (z) are not very far from the points - 1-29 0-44/.
15*8]
513
is
number
\.
integer,
When v $
zeros for which
is
an
(z) is
a polynomial in z multiplied by a
function with no zeros in the finite part of the plane, and so the
number
of
R (z) <
is
exactly v
\.
ir
we
write z
^e***,
we have
K
and so
{z)
has a sequence of zeros lying near the negative part of the imaginary axis. The zeros of large modulus which belong to this sequence are given approximately by the roots of the equation
v
K {z)
tan (f
it
\vir
may be verified that they are ultimately on the right or left of the imaginary
i>7r
is less
i.e.
The
1,
i.e.
when
There
15*8.
is
= ir.
The previous investigations, based mainly on integrals of Poisson's type, have resulted in the determination of properties of zeros of Bessel functions, when the order v is not unduly large. This is, of course, consistent with the fact that Hankel's asymptotic expansions, discussed in Chapter vn, are significant only when v* is fairly small in comparison with the argument of
the Bessel function.
The
fact that
of large order suggests that these integrals may form an effective means of discussing the zeros of Bessel functions of large order; and this, in fact, proves
to be the case *.
are valid for functions of any positive order, though they gain in importance
We shall adopt
H (v sec ) =
v
e-"dw,
contour in the plane of
where
on
it.
We shall
x and
v sec
j8
indifferently
when x > v.
514
If
[CHAP.
XV
w
u and
e-^du,
J oo
J
e~ VT dv
Hence,
if
we regard
/3
as variable,
raa
and
e
define
- "7
+ ni-ip
arg
J _oo_j(5
dw
# = 0, and to vary continuously with /3, it remain a positive acute angle for all values of /3 between and ^v: and, moreover, by 8*32, it cannot exceed J ir, since dv/du < \/3to
will
^ will be defined by
= i/(tan/8-) + x-ir.
It is then evident that
H <(vBec/3) = ffl#*,
w
where Jtt
is
Jv () =
If
and
cos
9,
F (a?)
= Jtt sin .
Yv (x) sin a,
it is clear that the only zeros of 9i> v (x), greater than v, are derived from the values of "9 which make "9 + a equal to an odd number of right angles.
It is easy to
shew that
SP increases with x,
when
v remains constant.
For
we have
,
Hence,
of
2 - ^, /(?ng) ^ > - arc tan ^ >0 *-arctan^ Y 0. ^ + ^ ^ as x increases, W increases steadily, and so, to each of
(fl;)
the values
W for which
y\T
= (m +
is
ir
- a,
(x).
x.
is
Next we
to prove
it
;
also
an increasing function of
This
is
theorem of a much
1374
required
we thence have
*J(x
2
-v*)
dx
J, (x)
(a?)
>
it is
2
clear that
(l/x)]
[x
^ and
in
so
<
J y
( V )
:
.
>
<X < ? 7r
1
left is
15*8]
515
To form an estimate
x when
v is large,
v
;
we
write
Jv (v)=-Y(v)ta.ny
hence, from 8*42,
we have
lim
yv =
tt,
and,
when
v is large,
r<|)
V3(
so that,
210
rm.^)* r Q).
when
is large,
is
an increasing function of
v.
following Table gives the sexagesimal measure (to the nearest half minute) of the angle whose circular measure is yv it exhibits the closeness of 7 to its limit, even when v is quite small:
The
10
12
00
yy
30 28 39' 29 23 29 38' 29 45' 29 51' 29 54' 29 56' 29 57$' 29" 58' 2958'
The
yv
is is
values of
the case
J* {v) dv
and
this inequality has already
Yv (v)
~^r >
been established in
1374.
has a zero for which
We
where
are
now
fi)
v (tan
- /3) + % mntr - o,
;
is
and therefore at
first,
the positive zeros of <@y (v sec ), with the possible exception of the
v (tan & )
(w
i)
it
a and
it
mtr
yv 'a,
and the difference of these expressions (when v is not small) slightly exceeds -fair.
From
and
so, if
H ^(X)e
v
ia
increases from
a-^Trto^ + oasZ
the number
to any value exceeding v, increases from of zeros of <WV {X) in the interval (0, x) is the
+
+
a)lir
+ .
vs
1374
*^| rf*
decreases as 8
increased.
This result
the shews that the interval between consecutive zeros of <$y (x) decreases when the rank of
zeros is increased.
Porter;
cf.
15-82.
516
15*81.
[CHAP.
XV
Jv (x) and Y (x). It has been seen ( 15'3) that Jv (x) and Yv {x) have no
The smallest zeros of
v is positive,
(0, v),
when
and it
is fairly
obtained in
842
in 8*43
- J, +
.)|
=-
$$%$
0)
lies
As
1
increases from
(l/*Jv),
which v (tan
8
between f tr
- 0) = y tan
(v ~ *).
Hence,
if
we
tan(f-f7r)=-Q(,i)/P(,),
the value of so obtained
is
which
is
(i>~ J ).
The value
tan(-i7r)
of which the smallest root is
smallest zero of
Yv (x) is
v
The formula
xxxiv. (1917),
of 8*42 for
p. 193.
Airey's formula
Jv (x)
Jv (v), Jv
'
iy\
Jv" (v),
....
J- V (x) was
given by Airey.
This zero
v.
may
lie
any-
where between
Jv (x),
15-81, 15*82]
It does not
517
section.
seem to be possible to make further progress by the methods used in this now make a digression to explain the methods of Sturm (which have been applied to BessePs equation by various mathematicians), and we shall then give an investigation which leads to the fascinating result that the two expressions which, in this
We
shall
section,
-1
(i
),
Jv (x)
and
Tv (x)
in
the errors
v is large.
[Note. An elementary result concerning the smallest zero of Jv (x) has been obtained from the formula of 5-43 by Gegenbauer, Wiener Sitmngsberiehte, cxi. (2a), (1902), p. 571; if /* = v + e where e < 1, then the smallest zero of J iv + (x) is less than twice the smallest zero of Jv (x), because, for the latter value of x the integrand cannot be one-signed.]
<
15*82.
Sturm*
for
The
manner are of some interest, though they are not of a particularly deep character, and most of them have already been proved in this chapter by
other methods.
is
is that,
normal form
in which the invariant / is positive, then the greater the value of /, the
more
increases.
an application of this result, we may take a theorem due to Sturm Ann. sci. de Pltcole norm. sup. in. (1866), p. 72, that, if v2 - i be positive and c be any zero of <gf (x) which exceeds J(v* - ), then the zero of ^f (x)
of
(ibid. pp.
As an example
174
which
is
c+
This result follows at once from the consideration of the facts that the function
3&9&V (#)
is
annihilated
by the operator
dx*
( 4-3)
1,2
and
that,
when x > c,
+ \ A
-$=
x* /
>
c2_2_i.l
^ .
pp.
A slightly more abstruse result is due to Porter, American Journal of Math. XX. (1898), 196198, to the effect that, if v 2 > \ and if the zeros of <@ (x), greater than ^(v 2 - ),
, ,
magnitude are clf c2 c3 ... then cn+1 cn decreases as n increases. This has already been proved in 15*8 by another method.
in ascending order of
Other theorems of like nature are due to Bdcher, Bulletin American Math. Soe. in. 205 213; vn. (1901), pp. 333 340; and to Oasser, Bern Mittheilungen, 1904, pp. 92135.
(1897), pp.
* Journal de Math. i. (1836), pp. 106 186; an account of recent researches on differential equations by Sturm's methods is given in a lecture by Bdcher, Proe. Int. Congress of Math. i. (Cambridge, 1912), pp. 163195..
518
15*83.
[CHAP.
XV
of large order.
proceed to establish a number of results concerning cylinder functions of large order which are based on the following theorem of Sturm's type
:
We
d2 u
such
that,
zt + i^-o,
x
d2 u2
^? + /..-o
Mi'
when x = a,
Wj (a)
= u2 (a),
(a)
=m
'
(a),
and
let
I and I2
I
a^x^.b, and
also let
(x)
and
lies
x
of u
on the
left
(x).
maximum
point of
is
on the
maximum
u 2 (x)
j
max
positive,
Ul
Uj (x)
< max
that, so long as
(x)
d2 u2 ~ d2 u^ U2 ~dxv dxJ
=(T ~
,
1
*'
UlU*
'
and
so,
when we
integrate,
[du
Since the expression
du
~\
1
now under
du 2
we have
n \ (1)
Hence we have
du
x
-
^dx-'^dx-^
djujuj)
dx
and therefore
W%0, LMiJa
that
is
to say,
(2)
To
it is
of u, (x)
and u 2
(x), if
is
positive immediately
may
The theorem
is
Math.
i.
15*83]
519
time Ma(#)
is
still
first
and
it
value
a.
first
The
is
therefore proved.
Again, if w/(a)
before it vanishes,
and at
this point,
we have from
(1)
/*,).
Therefore the
Finally
first
maximum
of u^ (x)
we have
(ft*)
= max u^ (x),
manner postulated more oscillatory* than
completely proved.
Uy,
When
i/ 2
two functions,
(x)
and
it,
(x),
(x) is
is less oscillatory
than
v^ (x).
We
ing
Jv (x)
now apply the theorem just proved and F (x) when v is large and x is
1/
Our procedure
will
more
oscillatory
In the
x=
ve;
(3)
[^ +
function which
is is
"*
(**
= ^(ve9)
for
(4)
[^+2^]u-0,
> 20 when 6 > 0:
solution of (4) is
<x /2*i/0*>
since e* - 1
The general
.-**,(?*);
and the constants implied in this cylinder function have to be adjusted so that u and its differential coefficient are equal to <@ (ye) and its differential
coefficient at
*
6 = 0.
The reason for the use of these terms is obvious from a consideration of the special case in which 2} and I2 are positive constants. f These results supersede the inequalities obtained by Watson, Proe, London Math. Soc. (2) xvi. (1917), pp. 166169.
520
[CHAP.
XV
which
than 90 {x),
when x ^ v,
is
now endeavour to construct a function which is (slightly) more oscilthan latory ^?r (x), in order that we may have 9$ v (x) trapped between two functions which are more easily investigated than <$ (x).
The formula
for
We
the
we should
where
from
yfr
(0) is
a function of
to be determined.
It
might be anticipated
8'43
but
is
it
that the suitable form for ty (0) would be tan3 fi, where sec /3 = e e ; appears that this function leads to a differential equation whose solution
0,
its degree of oscillation depends on the relative values of v and and we are not able to obtain any information thereby.
such that
The
is
by
v{t(WW}^iW()l
known
to
be
4*31)
2
i/
2 (e *
1).
It is consequently
(0)
which
is
*'(*)- <(*-!*
by determining whether^
1 for this
*(0)-0,
yfr
value of
2
(0),
2
+"'(0)
_ 3 (^(0))
4.Vf'(0)J
5_ | +'(0)1
2^'(0)
36J^(0)j
T/r(0)
> "
"
When we
replace e* by sec $,
f'(0)
we
find that
= tan/3,
= tan-,
+->-* a
and hence we have
(5)
*"-<*>=-
J;*^
3(tan/3-)^
sin s
cos
'
/3V(l+itan2 /9)
sin3 /3
3
Now
is
^3(tan-/3)
cos
V(l + tan2 )_
it
is
of tan2 /9.
*
equation in
The multiple of the cylinder function cf. 4*31 (17). its normal form
;
satisfies
a differential
15*83]
521
is
Hence, since (5) is true when #= 0, it is true when a certain angle between arc tan V{V24 3} and \ir
O
**
/9
where
/S
The sexagesimal
measure of
is
59 39' 24"-27.
we
is
slightly
more
oscillatory
than
J (v sec /8),
;
so long as*
# $ #>
We
greater than v
which ma^es
Then
20 \,a/i/,
give
Since,
a;
i/
v (tan
2
!/ 1
/3
- ) = J
(v~ J ).
lies
= + ^A,
this form,
we
see that the value of the zero of *@v (x) which is next greater than v
is expressible
in the form
v
+ $\M +
0(v~i).
When
/"
functions of orders
that
\,
= T926529 +
1-855757
(ir*),
is large, is
+ v*x
+
is
0(jt*).
In
like
i/*
x 0-931577
0(ir*).
The
first
maximum
of
/ (a;)
may be
The
which makes
r (|)
vanish J, then the
i.e.
(W
J. (v) J% (f)
first
maximum
of t/"(#)
at the point
at the point
i/
i/J
x 0-808618
(i/~i).
maximum of the function Yv (x) cannot be treated in this manner because its The first maximum is on the right of its first zero this follows at once from 15*3, because F (x) increases from - oo to O as x increases from to the first zero. For an investigation of the maximum value of Jv (x) qua function of v the reader
first
;
cols.
j3
435
438.
*/3 3 is
we have
to consider values of
for
which
bounded;
small values of
/3.
t The permissibility of this follows from the second part of Sturm's theorem just given. X This value of is approximately 0*685548.
CHAPTER XVI
NEUMANN
16*1.
SERIES
The
definition
The
the investigation of
various types of expansions of analytic functions of complex variables in series whose general terms contain one or more Bessel functions or related functions.
These expansions are to some extent analogous to the well known expansions by the theorems of Taylor and Laurent. The expansions analogous to Fourier's expansion of a function of a real variable are of a much more recondite character, and they will be discussed in Chapters xvni and XIX.
of an analytic function
Any
2
n=0
is
a n Jv+n (z)
called a
Neumann
is
general series
although in fact Neumann considered* only the which v is an integer the investigation of the more due to Gegenbauerf.
series,
;
'
To
mann
distinguish these series from the types discussed in 16*14, the description Neukind has been suggested by Nielsen, Math. Ann. lv. (1902), p. 493.
'
have already been discussed in Chapter v. It will be sufficient to quote here the following formulae
:
=o
ni
J.(z
+ t) =
m= -oo
"
where
r2
= Z + z* - 2Zz cos
1
<f>.
expanding an arbitrary function then we shall investigate the singularities of the analytic function defined by a Neumann series with given coefficients and finally we shall discuss the expansions of various particular functions.
shall first discuss the possibility of
We
into a
Neumann
series;
the reader
For a very general discussion of generalisations of all kinds of series of Bessel functions, may consult memoirs by Nielsen, Journal fiir Math, cxxxn. (1907), pp. 138
35.
(2),
(1877), pp.
125127.
16-1, 16*11]
NEUMANN
SEBIES
523
Various expansions of types which resemble Neumann's (other than those given in this Webb, Phil. Trans, of the! Royal Soc. cciv. (1905), p. 487 and
Nielsen, Atti delta R. Accad. dei Lincei, (5) xv. (1906), pp.
490497.
16*11.
coefficients.
Let f(z) be a function of z which is analytic inside and on a circle of R with centre at the origin. If C denotes the contour formed by this circle and if z is any point inside it, it follows from Cauchy's theorem that
radius
Now, by
9-1,
-i
t
M=0
U0.(*)./.W;
on the contour. It follows at once
and
that
(1)
/(#)- I an Jn (z),
=o
where
(2)
an =
^fc f(t)O
n (t)dt;
and
this is
Neumann's expansion.
/(*)- 5
b n z,
we
see that
-^,
^k**
bj
ton {t)dt
w=0
m 1)! m
\
!
(n>l)
and
so
fOo
=b
<3>
=M t*2-(^i-^ 6_.
Neumann
(>1)
sum
=0
The
given
is
construction of the
Neumann
series
is
consequently
338340.
35.
. (1872), pp.
524
16*12.
[CHAP. XVI
Neumann's* analogue of Laurent's theorem. Let f(z) be a function of z which is analytic and one- valued in the ringshaped region defined by the inequalities
r^\z\^R.
Let
C and
circles
= R,
\z\
= r,
between the
circles,
we have
,
f()= JK '
//(*)&
2-TriJc
t-z
/()*
2ttiJ c
z-t
f(t)Jn (t)dt.
c
^^ ^M^f/(t)On(t)dt+X^On (z)j
o
Consequently f(z)
(1)
is
where
(2)
^j
f(t)O n (t)dt,
c
an'=^ff(t)Jn (t)dt.
is
/<,)-
*=0
i . ik*+ n=l *
we
have, as in 1611,
w
(**)
I a
=n 2
2 n-*w-
*-
ft*-,
(n
^ 1).
ttn
1 6*13.
By
A n>v (t)
defined in 9*2,
Gegenbauerf has
if
= R,
and
G denotes
the
'
1m Jo
and so
(1)
*"/(*)=
5
n=0
an J,+n (z),
* Theoric der BeueViehen Functionen (Leipzig, 1867), pp. 36 39. t Wiener Sitzungsberickte, lxxiv. (2), (1877), pp. 124130. See Wiener Denk$chriften, (1884), pp. 298 816 for some special cases of the expansion.
xlviii.
16*12-16'14]
NEUMANN
SERIES
525
where
(2)
a.-^.Jo /(*)-A.r(0*
is
/(*) =
S Kz\
=o
then
(3)
On = (v
+ n)t2'+-
m=0
r{v + n - m)
W!
Neumann's expansion
of 16*12
may be
of
of 9*5, namely
n .^ (t)JIIL+in (z)Jp+ ^n (z)
)
-= 2 B
Z
\
n =0
\
which
is valid
|
when z <
|
t
|
we can
is
analytic
when
(1)
< r,
5
=o
an Jli+in (z)J+in(z)
given by the formula
is valid
when z <
| \
r,
and the
an =
coefficients are
(2)
^jc f(t)B
n=0
ni
^{t)dt,
z
j
C being
= r.
with
This expansion
this,
is
due to
Gegenbauer*
(3)
an expansion
closely connected
namely that
f(z)= I an'Jn*(z),
where
(4)
a^ =
tf{f)Cln {t)dt,
and Q it) is Neumann's second polynomial ( 9 "4), is valid provided that f(z) this expansion was obtained by Neumann f. is an even analytic function
;
Gegenbauer's formula has been investigated more recently by Nielsen, Nouv. Aim. de Math. (4) II. (1902), pp. 407410.
is
=0
ni
* Wiener Sitzungsberichte, lxxv. (2), (1877), pp. f Math. Ann. in. (1871), p. 599.
218222.
526
[CHAP. XVI
i
n=0
b n z+n
-2
=0
where
(6)
16*2.
Pincherle's theorem
and
its
generalisations.
let
Let
5 an Jy+niz)
n=0
be any
this series
and
its analytic
Let also
The
function defined
by f(z)N and
be called
The Neumann
series converges
and
this
domain
is identical
lim
<i,
Neumann
series has
power series, and the circles of convergence of a associated power series are identical.
Neumann
series
and of the
due
much
further,
can be proved that f(z) has no singularities which are not also
.
singularities oif{z)N
To prove
this theorem,
we
write
2
*
i.
r> + , + t)r(tr * w
7779. 151180;
Nyt
t Bologna Memorie,
pp.
Math. Ann.
lv. (1902),
493496.
16*2, 16*3]
NEUMANN
circle of
SERIES
527
convergence*,
cos
^
{W(l
-)}
0(^)^3^.
it
continuation
follows that, if
<f>
(z) is analytic
any value of z, so also is f(z), provided that the path of integration is suitably chosen and so all the singularities off(z) must be singularities of $ (z).
Now
cp
(z)
may be
an
2 S
and a theorem due
to
T(v + n+l)
Hadamardf
bn zn
,
states that, if
on zn
F
then
all
(z)
= 2
n=0
F,{z)= 2
=0
F (z)=
3
I
w=0
b n cn zn ,
the singularities of
t
/3 is
some singularity of F
(z)
and y
some
singularity of
2 (z).
n=0 r(v
is
Tfr+n + l) + n + %)
(f>
;
at the point
;
of f(z)x
and
this
z=l,it follows that all the singularities of (z) are singularities and therefore all the singularities of f(z) are singularities cff(z)N is the theorem which was to be proved.
The reader should have no difficulty in enunciating and proving similar theorems \ connected with the other types of expansions which are dealt with
in this chapter.
16'3.
Various special
Neumann
series.
The number of Neumann series, in which the coefficients are of simple forms, whose sums represent functions with important analytical properties is not large we shall now give investigations of some such series which are of
;
special interest.
By
r - Pn (cos20),
2 2
to be truncated
if
for
which
R (v + n + %)
is
argument
is
t Acta Math. xxn. (1899), pp. 5564 Hadamard, La Serie de Taylor (Paris, 1901), p. 69. t For such theorems concerning the expansion of 16-14, see Nielsen, Math. Ann. m. (1899),
p.
230
et seq.
528
[CHAP. XVI
I Jm J ^
where the contour
If
^p{^(<-l/Q}^ - JL m= ^lj ^ _ +
(0+)
2ti
C08 2 ^
1}
lies
circle
1
= 1.
is
write (t ljt) = w, so that the contour in the w-plane (large) closed curve surrounding the origin, we find that
now we
5 JU.()P.(co.M)-^,j*
and
so
V{(*'+l)(w'+sin0)}'
we
(1)
n=0
*7r J
^"""du,
The
interesting expansion
^vW-*o
2.4...(2n)
{2"+I>
I)} + n + l)
who proved
(1),
This expansion is easily derived as but the following direct proof is not without
By Neumann's
formula
5 '43)
we have J
J"
and,
if
^-7ri
V{<(l-0}'
we expand J (z VO
2t "
r / u\ J* <* VO -
Ti
* 2
o
(2i/
+ 2w + l)r(2i/ + w + l) m!r(2* + l)
& (- m,
2(2i/
2v
+m+
2v
+1
we
find
on integration that
9/1 Jv\h z) ^
=0
_
.
+ 2m+l)
*
1
"
, v r 9H-sm+i\*/i
vm.
(1887), pp.
elliptic
functions of modulus
by Nielsen,
J If
B + i) < 0,
(j
we use loop
16-31]
NEUMANN
r
7T
.
SERIES
529
where a> =
If ml
i
>
(-rn,2v
+ m + l;
2,
*)<**
tr
= i=J=.
by
TT.WilJo
Vam
(1
_ tr d
"
[<r
'~!,^- tri]
dm
w partial
integrations.
is
It follows that
and
when
|A|<1.
If
we
write
, t
(1
,
- *)
,
we
find that
am hm =
,
f\"-* * JO
1
{1
ht)~i dt
=I
It is
T Jo
M H"i (1
now
- "^*
_1
! 3 ...(2n
1)
2.4... (2n)
Jo
v+ft _ i /i (
\_|
'
1.3...(2w-l)r(i/
2.4...(2w)
~tt' and
T(i/
+ rt + ^)r(^) + n + l)
'
16*31.
The Neumann
series
summed
by Lommel.
The effects of transforming Neumann series by means of recurrence formulae have been studied systematically by Lommel*; and he has succeeded by this means in obtaining the sums of various series of the type
which an
is
a polynomial in
n.
m (z) *= S
(v
+ 2w + l)/2ni (v + 2n +
1)
Jv+2n+1 (z),
.
=o
(v
S8>m (z)
+ 2n + 2)/
2)H
_1
(i>
+ 2w + 2) / +!m+2 (*),
where fm (v)
is
49.
530
[CHAP. XVI
we have
(1)
- 64v m
z
,
(z)
= I
=o
fnn, (v
=/2m(V
provided that
+ l)Jv (Z) +
2^ m
(z),
/2m
(i/
fm (v) satisfies the equation of mixed differences* + 2n + 3)+/m('+2n + l)=2(i/+2n + 2)/2w^ + 2n + 2).
1
(i;
is
We
it is
(2)
- M*, m
z
it
(z)
fm^
{v)
J^ (z) + 2^, m_
Hence
,
follows that
S4v m {z) = W*m ( v + !) Jy ( z ) ~ K/a-i (") J*+M + m and so 4v m (z) = " 2 f^ 2"/^ (v + 1) </ (*)
1
,
Z2 S#v
m-i (z),
=o
Therefore, since
#,,_, (z)
= 2
=o
Jr,+2 +1
(*)
=s
*>
/"*/,
(0 *,
we have
- j
and
similarly,
w |j
r(1 _
for
n+
t
j-)
(**)
>
&y m(z)
/r+l( *)
t.-r(^- + *)
( ** )
}'
* Recent applications of Neumann series to the solution of equations of mixed differences are due to Bateman, Proc. Int. Congress of Math. 1. (Cambridge, 1912), pp. 291294.
16*32]
NEUMANN
SERIES
531
The
were also investigated by Lommel, but the results are not so interesting. As examples of his expansions the reader may notice that
These results were given by Lommel, though his formulae contain numerical
errors.
16*32.
The Neumann
of the series
series
summed by Kapteyn.
00
The sum
iz 2
Jo
* the sums of the alternate terms of the series have been expressed by Kapteyn
in the form of integrals
from which this integral may be deduced, and may be deduced from this integral.
We
The
\z
methods.
series
may be
n=0
\z
r (0+ > J
f<
0+)
2 <- +
M= o
I*-"
it
-J
+ a
('
(u
Hereto
~ (iy /
Now,
let
f tUh
'
exp
I**
" *)
ia
(s)} **
to
be the
circles \u\
= l,\t\ = A>1.
Then,
if
m = | ( 1/t),
we have
3 (?
ST
van Wet.
f
are
+ m/ = 53
te
1)
* {*" ( " D] da - * [J
; ;
(a)
"
^ (a)/
'>-
* A'jeuip
Archief voor Wiskunde, (2) vn. (1907), pp. 2025 Proc. Section of ScL, K. Akad. Amsterdam, vn. (1905), pp. 494500 Kapteyn has subsequently summed other series,
is
962969. The interchange of summation and integration any points on the contours.
permissible so long as
tu
> 1,
where
t,
532
[CHAP. XVI
Therefore, on integration,
I = Ce-^ + 1
where
&
f
.'
V"
o
-"'
{Jo (v)
- J, (v)/t) dv,
see that
C is
independent of
a.
By
taking a
= 0, we
C=l/t.
Hence we have
+1 =
-.
*exp fo (*
-a+
*) (*
- ^)
{tJ (v)
J
- J, (*>)} cfoj
(v) dv,
dt
4 Jo
= 0.
Consequently
I
that
is
nJ.
()
J. (a)
-|
T ^"V"/
(
J
/o (U)
*'
to say
(1)
n=l
If
aJ -li^!lj
*
u
(a-v)dv.
we
select the
this equation,
we
(2)
-5[
4
J
Mz ~
\
(
v)
zand
+ ^^^1 z+v
)
/. (
*.
which
is
Ta !,
2n
Jm ( *> J (a) =
4 \-T=-
TUT].
-*
4J 4j
*-
*+o
1V
*>
i
'
*-
when we
integrate by parts.
it
Hence
(3)
follows that
n=l
+ f \^P +V Jo[^
'
^^\j.{t-v)dv, ZV
)
which
is
16-4]
NEUMANN
SERIES
533
difficulty in
S
n=0
{v
Jv
'0
V) -f~ ZV JM-v)dv
(z
v)
.,
16*4.
series.
have been studied from the standpoint of the theory of by H. A. Webb*. His theory has been developed by Kapteynf and subsequently by Bateman|. The theory is not so important as it appears to be at first sight, because, as the reader will presently realise, it has to deal with functions which must not only behave in a prescribed manner as the variable tends to + oo but must also satisfy an
functions of real variables
,
Neumann
In
fact,
theory seem to be included in the functions to which the complex theory is applicable, and simple functions have been constructed to which the real
variable theory
is
inapplicable.
is
The
result
based
is
that ( 13*42)
Jo
<Wa W <J*n+i W
00
| 1/(4w
+ 2)
the'
m=n
),
so that, if
type
2
n=0
a*n+i
JM+1 0),
we have
and
if
term-by-term integration
id
permissible,
/, o
//<= a=r2
+ 2) Jm+1 (*)
f"
We
(1)
/() =
shall
(4
J -^^f{t) dt
and we
conditions
(I)
The integral
Jo
exists
and
is absolutely convergent.
* Messenger, xxxin. (1904), p. 55.
t Messenger, xxxv. (1906), pp. 122125. X Messenger, xxxyi. (1907), pp. 3137.
V
534
(II)
[CHAP. XVI
coefficient
for
all
positive values
(III)
satisfies the
equation
t)
(2)
Jo
^ *&> {f(v +
v
+ f(y - 1)} dv
when
x.
We
now proceed
to
sum
(4n
the series
S=
and we
that
first
+ 2)^+1
(^r% /(0<fc,
L
is
evident
2
w=0
for positive
Jm (01^1
and
2 Jm +i (#)
\
is
I
convergent.
effect
integral,
we may
(unbounded) values of t, Hence, since f(t) possesses the interchange, and then,
8=
X/(0
(4w
/<>
l?o
=
2.lo
/(0 Jo|"T=ir +
- j J,
(t
-j^lMx-iddvdt
+ v)\
,
1 f* f
- v)
J,
(t
I fI
Jo (*
- V)
v)
f*
V
^
1)
{f(t
+ V) +f(t - V)}
dtdv
(x
Jo
\ Jo
f(v - 1)
^P dtdv.
t
We now
we have*
Jo
7 V (x - v)f(v ;
^P dtdv
*
= =
I" f Jo (u oJ o
- t)f(x - u)^P-dtdu
t
J (u)f(x
u) du
Jo
=f(x)
The
first
J^{u)f'(x u)du.
Jo
by writing
transformation
is effected
v=x + 1 u.
16*4]
NEUMANN
J
" f .0
SERIES
535
Hence
(3)
I
=0
(4i
+ 2) Jm+1 (x)
dt -^^-f{t) c
Now
write
Ji(0
Z
so that
'
{/(+*)+/<*-)}* 3
v,
m
an absolutely
in such an interval
F(v)
is
a continuous function of
convergent integral.
If then
as (0, X),
f Jo
J (x-v)F(v)dv = 0,
it
follows
by induction from
\F(x)\^fjF(v)\dv,
* hat
\
F{xy
^W^V
|
where
is
F(x)
in the interval
and n
is
any positive
integer.
If
we make n -
x>
it is
clear that
F(x) =
0,
and so the
necessity of equation
(2) is established.
The
from
sufficiency of
is
evident
(3).
been pointed out by Kapteyn that the function sin (x cosec a) is which equation (2) is not satisfied and Bateman has consequently endeavoured to determine general criteria for functions which satisfy equation (2); but no simple criteria have, as yet, been discovered.
It has
for
one
[Note.
If f{x) is not
separately
and then
it is
f{x)=7, a^Jn
n=0
{x),
where
a =
2 /
f(x)
Ji(\ x dx
\)
->
a= n
J
_J(x) Jn (*)
r^j
>
> 0),
The
536
16*41.
[CHAP. XVI
of reduced functions.
series
f(z)=X
cn z
n
,
f(z) B =
2
=0
cn
.n\zn
supposed convergent
for sufficiently
may be
represented
by the integral
jo
The
function f(z)i,
maybe termed
If the
Neumann
series
f(z)= 2 an Jn (z),
tt=0
= 2
n=0
an
1
\
-'
e~*
Jn (tz) dt
an
*
2,
|V(l+^)-l) w
Now
put
z
-t
Bf/G^).-i^Hence,
function of
if
+?
.(
the
Neumann
is
series for
f(z)
is
2 an Jn (z),
tt=0
=o
2 an n
i-r / u-r/'
provided that this function
is
More
that
(1)
/(*)-S
an J,+n (z),
then
(2 >
295368.
16*41, 16*5]
LOMMEL'S FUNCTIONS
if
537
In like manner,
(3)
then
[Note.
If
ef"smbz=S, an Jn
{z),
then
I ,=
nil
result,
26z ( 1
+ g2
(l-2az-z*?+4b*z*-
This
which
is
16*5.
variables.
which are of considerable importance in the theory of Diffraction and which are defined by simple series of Neumann's type, have been discussed exhaustively by Lommel * in his great memoirs on Diffraction at a Circular Aperture and Diffraction at a Straight Edge.
Two
The
n,
Un
(to,
z)
and
Un (W,z) = 2 (-W7)
Jn +*m(z),
J-n- (*>
(2)
Jn^n{z)
==C08
(w
{2
z2
mr\
2w-^-)^
(4)
tfn+i (w, z)
~ -^
The last equation may be derived from the preceding equation by replacing n by n + 1.
There
is
no
difficulty in
we
write
U.(w,z)= 2
i=0
k. b.
(-)**(
7) \zj
/,+(*).
Akad. der Wist. (Miinchen), xv. (1886), pp. 229328, and the definition of Vn {w, z) in it differs from that adopted subsequently by the factor ( - l) m Much of Lommel's work is reproduced by J. Walker, The Analytical Theory of Light (Cambridge, 1904). The occurrence of Lommel's functions in a different physical problem has been noticed by Pocklington, Nature, lxxi. (1905), pp. 607608.
529
664.
The
first
memoir
W.
B. F.
18
538
[CHAP. XVI
z,
factor
only
The corresponding generalisation of (2) gives a series which converges when v is an integer. And consequently it is convenient to define Vp (w, z) for unrestricted values of v by means of the natural generalisation of
namely
F (w,
z)
(3),
(6)
= cos
+ (f + 1^
y ) + u-+*
(J
>
z )-
It is evident that
(7)
(w, z)
+ Uv+i (w, *) 7
(z),
(8)
V, (w, z)
As
special formulae,
we deduce from
U. (z, z) -
(9)
V (z, ) = \
1
(z)
+ cos z},
(10)
U.iz, z) = (8),
(z,z) = I sin z;
(-)^4,W}
"% (-)
(12)
Um+1 (z, z) = - V
>1
(z, z)
Wl J^
+l (z)\
provided that n
in (11),
and
>0
(13)
The
2 m=0 CQS
m0.Jm (z),
2
in=0
sin (to
+ ) 6 Jm +
.
(),
which are closely associated with Lommel's functions, have been studied by Kapteyn, Proc. Section of Set., K. Akad. van Wet. te Amsterdam, vn. (1905), pp. 375 376, and by Hargreavcs, Phil. Mag. (6) xxxvi. (1918), pp. 191199, respectively.
16*51.
The
variables.
It is evident
(1)
by differentiating
v
16'5(1)
^U (w,z) = --U
(w
1
>
z),
and hence
^
and consequently
d2
JJ V
(w, z)
z* = -
U. +2 (w,
z)
--
16*51, 16*52]
It
is
lommel's FUNCTIONS
that
539
now evident
U (w, z)
is
is
where
is
A and B are
independent of
z, it is
clear from
v
165 (6)
that
F_ K+2 (w, z)
Therefore
{6)
dz->-;dz-
w>-\w)
J- +z(z)
These equations are due to Lommel, Miinchener Abh. xv. (1886), pp. 561
563.
16*52.
We
(1)
(w, z),
namely
^U (w,z) = -^U
y
+1
(w,z).
OW
~ U.
(w, z)
= 2
m -Q
(-)" (v
=4 S (-y*(w/sy^*{j +m* =0
and
so
(2)
L Uv
(w> z)
= u~* {w> z)
ww)
u +i (w
"
>
z)
we deduce
that
^Mttf.sH-^F^M;,*),
2
(4)
AV
(w, z)
z).
If
(5)
now we take
w = cz,
where
we deduce
that
2^U (cz,z) = cU y
(cz,z)-(l/c)U+1 (cz,z),
(6)
Hence we get
4
d*
d?
'
= C* Uv^
2U"
cz> *)
<
>
z)
540
[CHAP. XVI
z),
follows that
Uv {cz, z),
are particular
(7)
^+(c+iy^cj . (z)+c^Mz).
v
2
The
(8)
is
of
some
K
interest
we have
{w, 0)
= 2
*;/
1X
and so 7,, (w, 0) and F_ +2 (/, 0) are expressible of one variable by the equations
in term's of
Lommel's functions
HO)
Of these
The
V. y+2 (w ,0)
results, (l}r (8)
Ji w
l{
is
^
^
when n
should be noticed
(11)
Um {w, 0) = (-)
cos
\w - 2
\ZZ
(12)
(13)
= cos(u/ + |rwr).
Hence
(14)
follows that.
(w,0)
o)
= l, Vn+1 {w,0) = 0,
= (-) m=Q
(
(i5)
v.m (w,
'
\&Vn>)\
"7
(16)
V^
{w, 0)
= (-) 2
/J*"L
16*53.
The formulae
(1)
tf,,
(w, z)
= -^\
Z
.0
r1
/_ x {zt)
cos
[\w (1
- *')}
t"dt,
(2)
U^ (w, z) = *
~
> v
J^
{zt)
sin
{\w (1
- i*)}
*<fc,
Jo
which are valid when R (v) > 0, may be verified immediately by expanding the integrands in powers of w and then using the result of 12*11 (1) in
16*53]
lommel's functions
For other values of
v,
541
they
may be
^( w^) = - 27
_
I
(4)
^M- g^^^
wp
t
when
_^_J
J^i-MQ.cnllwil-ty.i-tYdt,
i
r(0+)
^(-*0.rin{i(l--)}.(-^d,
in
increases from
tt
to ir as
R (v) > 0,
(5)
z.
f
J
J^ (zt) exp
By modifying
we can obtain
integral representations of
(w, z)
and
Let us consider
(1
JO
if
/_! ()
exp {+ \iw
- *>)}
*>
<&
and at the upper
The
limit
when
R (v) >
when R(v)<,
w and
To
coincides
ambiguity in sign being determined by the ambiguity in sign in the integral; such a modification in the contour is permissible by Jordan's lemma.
tt, this
arg= +
When we
we
find that
z,
as in 13*3,
w"
/,_, (zt)
J o
exp \iw (1
{
- 1*}} V dt
.
^. mir ;+ m)
(
'
2 * +2W
J,
"exp(?
}M)A
2"
.to
Vw)
.
'
that
is
to say
()
r <fc = exp
*?
^ + ^)
w > 0,
When we
z
this formula,
we
see that, if
> 0, and
(7)
then
^J
+ ^ -^),
;
(8)
^^(*)ntt(l-lO}.ri-rin(|+^-^).
542
[CHAP. XVI
that
16*5 (6)
(9)
V t_(w,z) =
-
z
.'
J^(st)cM{tw{l-t)}.fdt,
1
(10)
Fi_
(w, z)
=-
\w (1 - 1*)} V dt.
.
now
R (v) > 0.
- f))
Changing the
(11)
notation,
we
see that
fao
Vv (w, *) - -
w^
z v 1 -
{\w
(1
J*
(12)
V
w
(to,
z)
= - ^_
ji
- *>)}
^
<ft,
provided that
and
-2
are positive
and
R (v) > \.
The
(13)
Um (z,z) ^
ff+i(*,*)
p
f
.'o
j^
/*-*
{zt) cos
(1 __
<8)}
<9n
- <)}
<s)
|
.
2n~ 1
(14)
=p =
.
(ft) C08
n, (1 _
^
eft.
f
!
Again, from
(15)
(6),
we
see that
and, in particular,
(16) v '
Jo
last results should
fV.w^WI ^).
8
sin \ 2 /
wcos\2iW
;
The
The formulae of this section (with the exception of the contour integrals) are found in one or other of Looomel's two memoirs.
16*54.
all to
be
It is evident from
165 (13)
(z*/w, z) are
16'54, 16-55]
LOMMEL'S FUNCTIONS
543
To
^ [cos (**) U
zt)
ztj
= sin (}*) T- w * tf
.
(^ *) wtU^ (
,
.
ztj]
(wt/z) 1 -".
integration
we
find that
~
=r
x
,,)
^g,
),
and, similarly,
(2)
f\
(zt)
cos (\wt*)
*- dt
lf+1
(^ v
,oy
it
follows that
f1
==; z
)}
t"
dt
tc '-
Cr
-*(^
)'
w (1 - *)] V dt
.
differ
The reader will find some additional formulae concerning Lommel's functions in a paper by Schafbeitlin, Berliner Sitzungsberichte, vm. (1909), pp. 6267.
16*55.
Some
functions of the type TJV (w, z) with functions of the same type in which the second variable is zero, provided that v is equal to \ or
When we
write v
-\
in 16-53 (5),
z)
we
get
i
TJk (w, z)
iU% (w,
= (~) j + (^A j
exp [
exp {
(\w-zt-
\wt*)} dt
($w+zt- $wt*)}
dt.
'
544
[CHAP. XVI
Now
(W
and we
find that
4-
z)2
'
(w
zf
2w
2w
(w, z)
iU%
(w, z)
n = eT* (-) f
,4
exp { <n (1
- f )} d
1
(-
~ P)l d &
in the respective integrals, awci V"> a/8 are * oe interpreted by the conventions*
.
V "
v v
/S>
= w z
V(2w)
'
Hence we have
(w, z)
i U$ (w, *)
_eW-J
+ e fe ()
When we take o-f2 and Sf
in the formula
(1)
)}
<*
J
2
as
new
results
combined
(w, z)
iU (w, z) - te**{Ui(2*.
+ **{17(2S,O)0 I (2M)},
and, as a corollary,
(2)
O)}.
These formulae are due to Lommel, Miinchener Abh. xv. (1886), pp. 601 605; they are reproduced by Walker, The Analytical Theory of Light (Cambridge, 1904), pp. 401402.
16*56.
Fresnel's integrals.
It is easy to see
from
1653
(1)
and
(2) that,
0-h.i <.
0)
= 2y_^
(y)
I-* sin ft w (1 -
I-)) <ft,
so that
(
X)
ovJrv
<2 "
\
_1
f
2 " -1
cos
J7 +1
(w, 0) sin | /,
(2)
_1 v 2" i (v)Jo
I
\*
sin (A wt2 )
= TJV (w,
0) sin
$w - Uy+l (w,
0) cos \w.
16*56] If
LOMMEL'S FUNCTIONS
we
take v
545
\w = z \m# we
t
see that
(3)
^oo.(i^A-ij;Q*oo.i.yV^()A
- [ Uk (2z, 0) cos z + U% (2z, 0) sin z]/</2 = I + [F} (2s, 0) sin + F, (2*, 0) cos z]/</2,
and
(4)
^^(I^A-l^^rinlA-l/V^)*
= [ 0"j (2s,
0) sin
= i ~ [ ^* (2*,
We
integrals*
(
J0
Jo
fsin(A7r*2 ) dt.
(2)
The ascending series, originally given by Knockenhauer, Ann. der Physik und Chemie, xu. (1837), p. 104, are readily derived from the ^-series, namely
<>
,, ( ,,)=
(^f _ ^. + _^__....}
1
>
while the asymptotic expansions, due to Cauchy, Comptes Rendu*, xv. (1842), pp. 554, 573, are derived with equal ease from the K-series, namely
(8)
W,--(^-LJO
+ ...}.
Tables of Fresnel's integrals were constructed by Gilbert, Mem. couronnees de I' Acad. pp 152, and Lindstedt, Ann. der Physik und Chemie, xvn. (1882), p. 720; and by Lommel in his second memoir.
(V ()ft-2
r
n=0
S Jy+m+1 (z)
Z n+l
=
(10)
*
J^ n (z)
l)'
n~o(vi-l)(
+ 'd)...(v+2n +
3).,.( y
Jz
2 ('JAW" =o
("
l)(
+ 2-l) jr
+n (*).
i.
zn
[Oeuvreg,
Mim.
(1866), p. 176.]
R (v) > - 1.
546
[CHAP. XVI
also
by
differentiation.
Other formulae
due
to
Lommel
(11)
are
JO
f'j-i(t)di
2*COsi* [
X (-)J n + i(i*)]
t
\_n=0
(12)
JV (*)cft-2in|#^2(--)^
t
+l (i#)]
verified
by
differentiation.
The
cos
(
(
at
Jo
b\ t = t/lP'
,
dt
f*
/
Bin [at
\
Jo
r) * tjl
b\
tdt tdt
when a and
b\
tdt
b are positive
cos
j
r + *-irF
,
b\
dt
r*
]
/ {
ain
, at+
t)Tt
be expressible in terms of Lommel's functions of two and unity respectively. This discovery is important because the majority of the integrals representing such functions contain Bessel functions under the integral sign.
to
t,
it is
seen that
<>
( 2)
i>(-+Di!*-*J>Mr!.
f>(- + )
by
*|L- j;* (- +
l
)*-/xtt+?
{2</(ab)},
-.
) i
and
since,
6*13 (3),
l-fjsin (at
it is sufficient to
=J
< a.
We now
*
write
c
= V(6/a),
2</(ab),
= $(l-c*)/c,
the lower sign
is
When
taken
it is
supposed
that the integrals have their principal values. t Me$enger, xxxvm. (1909), pp. 129 132.
16*57]
LOMMEL'S FUNCTIONS
t
547
ceu
b\
dt
{<+l/(<">
.
rfa
5=+17J5^j(
where
it,
is
The only
is
at
id,
and so
2-iriJ
T (0*
+ t*)V(t*-1)
2tV(^+l)'
it
As the radius of the large semicircle tends to infinity, the integral round tends to zero by Jordan's lemma, and hence
/"
cos(arr).TdT
ii
(^+t)V(t-1)
sin (arr)
a
rdr
Thus we have
f* 08
Jo
/
.
&\
<fo
tre~
~ 6)
a+b
[* sin
(a;
cos
0
<ft
cos
tf>
<f>
d<f>
'
yi
c&
+ <F"
4*/*
2^" Jo
+ cosa
But
and
so
cos
we
(3)
cos(a*
Jo
it is
-) =-^
r<-6
g
>
wl^J^i*).
2 _ r TTt^J^z),
.
Similarly
f\ (4)
found that*
*
.
Jo
h \ tdt F Bn / (* + -)_,
and
(5)
/ b\ dt Pjr cos^ -^=j7rsin(a + 6)-7r 2 (-^(JV^*), + -J r
fl
(6)
Pj sin^ +
o
-J I
i =-j7rcos(a+6)-7r 2 (-)mi
csw /am
(a;).
The
548
[CHAP. XVI
(7)
( Wi
*)
+F
(,)
-Pj
e
00.(^+5)^,
in
(8)
0;< w ,)+f.( W
,*)---pJ o
(^; +
2)
rr7
>
w < .
LommeVs functions.
An
(w, z)
by
integrals
is
the appropriate
series.
We
thus get
, w,i..^f (0+) .__/'. - <y <*< (-r (*)-"- J _ ^ ex P U v+ *n+l \ u) t t \
"
u. <,
*>
1 5 gi, s
o
When
which 1 and get
J
the contour
is
=\ w
\
Now
\iw are
_
vrri)
iw
iz*
and
so
ir
i
x
(0+)
(WO"
(4
*\dt
Making a
/o\ (2)
slight
m F^*)-Ssi..
X
(
+)
^/w >"
/<
*V*
+ iw
In general it is impossible to modify the contour in (2) into the negative half of the real axis taken twice, in consequence of the essential singularity of the integrand at the origin. The exception occurs when z = 0, because then
the essential singularity disappears, and
/q\
vt
m
=
{0+ Ht/v>ye* dt
and hence
(4)
(w, 0)
w
-
sinvTr f^exp^w-ie-iwtr
Jo
+w
du,
is
Ja
+ argw|< J-t.
16-58, 16*59]
LOMMEL'S FUNCTIONS
on the right
in (4) is called Gilbert's integral*.
549
Formula
(4)
16*53 (11)
by a transformaas,
From
is
when
and
w are
positive,
Vv (w, 0)
*
and
C uv-\
Jo
c e
-kuw du _
-r(i- v ).(fror
a positive decreasing function of
Vv (w,
From
Vv (w, 0)
have
and
Uv (w,
we
p-l
/_ \m
where p is any positive integer. We choose and then, by 16*58 (4), we have
(~) p y^+tp (w, 0)
p to be so large
that
R (v + 2p) >
7T
oo
/ Jq
1+U
exp ia
*)
-{/;
= 0(w-'-'),
when
\w\
is
M F+2p-lg-JW^M
arg w
<
tt.
Hence
(i)
for the values of
^
w
sin vtt
7T
<
~i, r(i-,->:).(t
under consideration.
are both positive, ( ) p
When v+ 1p and w
as,
Vv+2p (w,
and
is
lt
Jo
( )p r(l-i/-2p).(|w;) +^'
,'
same sign
as,
and
is
may be
V
|
|
{,s)~ 2 (-nzjivy+^J-^iz)
TO=0
;
when w
and z are fixed but it is not easy to obtain a simple expression which gives the magnitude and sign of the remainder.
is
large while v
* Mint. courorme de VAcad. R. des Sci. de Bruxelles, xxxi. t MUnchener Abh. xv. (1886), pp. 582585.
(1863), pp. 1
52.
550
[CHAP. XVI
Uv (w,
z)
are
(3)
U U
{w,
0)~cos(w-i/7r)+ 2
r=
'
(4)
(w,
z)~cos($w + y*/w-$inr)+ I
(-^(^^-^J^^^).
These results were given by Lommel*, but he did not investigate them in any detail.
is
or 1 and c
is fixed,
while
large
and
positive, has
is
K
as
t
{ex, jc)
-,
-^ j
,
cos
J c* (1
- * )}
2
N ow, if c> 1
tions that
(5)
increases from 1 to oo
and hence
/
it
(cx,x)~{\ -~_^co8{x +
""" 2 \* c2
has a
it
maximum at
1/c;
-t2 ) + (xt+ \vir \tr\ qua function of t, and hence, by the principle of stationary phase ( 8*2),
follows that
(6)
V, (ex, x)
~ -^ cos
\%x (c
+ - ) + \vtr \
is at one end of the range of and so the expression on the right in (6) must be halved. We consequently have
Finally,
integration,
(7)
Vv (x,
x)
~ \ cos (x + \mr).
This equation, like (5) and (6), has been established on the hypothesis v> J; the three equations may now be proved for all real values of v by using the recurrence formula 16'5 (8).
that
MUnchener Abh.
572573.
pp. 259269.
CHAPTER XVII
KAPTEYN SERIES
17*1.
Definition
of Kapteyn
oo
series.
Any
2
in
nJ*+{(
+ n)*},
a Kapteyn
first
which *
and the
series.
Such
name
they were
z,
systematically
in
investigated,
by Kapteyn f
an
important memoir published in 1893. In this memoir Kapteyn examined the question of the possibility of expanding an arbitrary analytic function into such a series, and generally he endeavoured to put the theory of such series
into a position similar to that
Neumann
series.
Although the properties of Kapteyn series are, in general, of a more recondite character than properties of Neumann series, yet Kapteyn series are of more practical importance; they first made their appearance in the
was discovered by Lagrange and rediscovered half a century later by Bessel; and related series are of general occurrence in a class of problems concerning elliptic motion under the inverse square law, of which Kepler's problem may be taken as typical. More recently,
in the
modern theory of Electromagnetic Radiation. The astronomical problems, in which all the variables concerned are real, are of a much more simple analytical character than the problems investigated by Kapteyn; and in order to develop the theory of Kapteyn series in a simple manner, it seems advisable to begin with a description of the series which
occur in connexion with elliptic motion.
Bessel.
The
motion in an
focus, attracting the particle according to the inverse square law, is as follows
The semi-major
are denoted
axis,
axis,
ellipse
by
a, b,
and
taken as coordinate
most
part, be
r is zero.
t Ann.
tei.
(1893), pp. 91
120.
J Hist, de VAcad. R. de$ Sci. de Berlin, xxv. (1769) [1770], pp. 204233.
pp. 113138.]
Berliner
il
552
[CHAP. XVII
ellipse to the
centre of force.
taken as origin of polar coordinates, the radius vector to the particle being r, and the true anomaly, namely the angle between the radius vector and the axis of x, being w. The eccentric
centre of force
The
anomaly, namely the eccentric angle of the particle on the ellipse, is denoted by E. The time which has elapsed from an instant when the particle was at
the positive end of the major axis
is
called
t.
is
t
complete revolutions.
The geometrical
(1)
a(1 ~
1
+ e cos w
T = a (1 - coos E\
'
tan \w
(3)
sinw
;
= -^1
are deducible
and an integrated form of the equations of motion (the anaSecond Law) supplies the equation
(4)
M = E-e8mE.
is that of expressing the various coordinates r, w, E, which determine the position of the particle f, in terms of the time t, that is, effectively, in terms of M. It is of course supposed that the variables are real and, since the motion is elliptic (or parabolic, as a limiting case), < et. 1.
Kepler's problem
The solution of the problem which was effected by Lagrange was of an approximate character, because he calculated only the first few terms in the
expansions of
E and
r.
E as
solution given by Bessel depends on the fact that (4) a continuous increasing function of such that the effect of
increasing
M by
to increase
total fluctuation is
a function
of
in Fourier series,
The
Dynamics of a
Particle. See, e.g. Plummer, Dynamical Astronomy (Cambridge, 1918), Ch. f Kepler himself was concerned with the expression of E in terms of M.
m.
17-2]
KAPTEYN SERIES
e sin
553
so, for all real
In particular
values of E,
it is
E is
e sin
E = 2 ^ n sin wi/,
n=
l
where
An =
j
E sin nMdM
E cos nl/>
H
= _
f*
cos
L 2
M7T
Jo
W7T/o
^*
n7rj
dM
=
2
mr
cos
'
nM.dE
= - Jn (ne).
n
Hence
(5)
it
follows that
and
problem con-
The series on the right is a Kapteyn series which converges rapidly when e < 1, and it is still convergent when e = 1 cf. 8*4, 8*42.
cerning the eccentric anomaly.
;
The
radius vector
is
GO
=B + 2
/*"
Bn cos nM,
e cos) dM
E)*dE
where
1 B = TT
(*
Jo
=when n 0,
it J o
= l + e*,
while,
Bn = IT
L
2 f*
i
(1
e cos E) cos nM dM
7T
Jo
_ r(i-.j)..if[Jo
rf(^g) dJf
<3
"o
2c f*
mr
sin
J
*A'(~X
so that
(6)
O,
= 1 + ie - 2
8
ft
/'(n<?)
cos
nM.
n-l
554
[CHAP. XVII
w - M is
of the true anomaly is derived from the consideration that an odd periodic function of M, and so
00
where
Cn = -
(w M) sin nMdM
_>
r-
it
Jo
,[_ 8(,-JQc,Un L
nrr
.(*
\d>M
Jo
.
nnJa
^
/
=
and
2 V(l
-e
is
cos (nff
1
W7r
- nc sin E ) ecos.E'
,
This expression
Bn
The
not such a simple transcendent as the coefficients A n most effective method of evaluating it is due to Bessel*, who
2
-e ) 1 COS i?
V(l
where
+ 2fco8E y
/i
1
+...,
V(l
On making
the substitution,
we
71
=i
17*21.
A large class of expressions associated with the radius vector, true anomaly and eccentric anomaly, are expansible in series of much the same type as those just discussed. Such series have been investigated in a systematic manner by Herzf, and we shall now state a few of the more important of them; they are all obtainable by Fourier's rule, and it seems unnecessary to write out in detail the analysis, which the reader will easily construct for himself.
First,
we have
r cos
w = x ae =
a(l-(?)-r
so that
(1)
/i\
rcosw =
a
.
?. a -f i 3 e +
w=1
%g nM,
and next
/aN (2)
Berliner
rsinw
a
n =i
Abh. 1824 [1826], p. 42. Nneh. era. (1884), col. 17 28. Various expansions bad also been given by Plana, Mem. delta R. Acead. delle Sei. di Torino, (2) x. (1849), pp. 249332. In connexion with their convergence, see Cauchy, Compte$ Rendu*, xviu. (1844), pp. 625643. [Oeuvres, (1) vin. (1893), pp. 168188.]
A$tr.
17-21, 17*22]
KAPTEYN SERIES
565
while
(3)
cos
if
E ae = - \e +
2 *=i
2
i
Next,
(4)
m is
cos
mE = m
{/_ (ne)
t7" w+OT
(ne)} cos
nM,
(5)
sin
mE m 2
of a/r
is
" 1 a n
{/"_
(ne)
The expansion
(6)
= 1+2 2 Jn (ne)coanM.
1
The expansions
(7)
w=e+
6*
(8)
sinw^V^-e*) 2 2/n'(ne)sinnJlf.
of cos w/r*, sin wjr* are of a simple form, 2n/n'(ne) cos
The expansions
(9)
namely
-cosw 2
i
nM,
(10)
[Note. It is pointed out by Plummer, Dynamical Astronomy (Cambridge, 1918), p. 39, that these are readily derived from the Cartesian equations of motion in the form
a3 cos to
r*
"**
3~> +
cPy
a3 sin
y.
-Ot
combined with
(1)
and
(2).]
17*22. Sums of special Kcupteyn series. The reader will observe that, in the case of the expansions of even functions of M, the results simplify when we take the particle to be at one of the ends
all
equal to
or to
ir,
equal to a (1
e) or to
a (1
+ e). From
the results
Herz
n=i
(2)
n=i
n
(ne),
5- = 2 I* 1 ~ * n=l
Jn (ne),
jr.' (ne),
(3)
(T
i^ -
(3) that,
^L = J
(-)-> Jn i^-S +
1 e
n=i
* It is seen
from
when
insertion of
col. 69.
a constant
term.
m is equal to 1, the expansion (4) has to be modified by the These two formulae were given by Jaoobi, Attr. Naeh. xxvni. (1849),
(1891), p. 149.]
[Gen.
556
[CHAP.
XVn
by
< 4)
L
(
\m r (Ism
oo
{ne)
>
(_)-i r r d? d* m
(5)
~i
= 2
- )"-
n J (we).
Since
dM
if
j-j-j
l-ecos# dE*
e
on the
left in (4) v /
and
(5) v / can
sufficient labour.
it is
we regard
and
M as
sin 8
easily
de\isin
E (1 - e cos E))
^_ ~~
cosE
dE
87
Z? cos w " ^
4- c e
E (1 e cos 2?)
#(1 -ecosE? 1 91
sin
sin x "
E " =9e
_
(1
esini?
-ecos^)3
i
=
so that,
_a_
9ilf 1
-ecos^'
by
17'21 (6)
w= i
as the lower limit,
and
therefore, if
we
1
integrate with e =
1
(6) v
'
= - - = - 2 sm.(l-ecos2?) sinif
-.
=rp:
=-
%
=i
sin
nM
['
.
Jo
find that
Jn (nx) doc.
If
,_>
we
M, we
cosi
;a t? /i ~TZZrW\* a sini(l-ecos#)
~~
cosJlf
~;~a jit sin'ilf
"*"
'
n (l-ecos#)
OB
ft
=2 2
=i
na cos n3f
J (nx) doc.
Jo
The
last
and
(7)
Thus, as cases of
(8)
(4)
and
(2ne)
(5),
Schott proved
(ibid. p.
110) that
JiW2j^
last of these
= ^lL^4j J^JV^Sii*)**.-^^.
obtained by taking if equal to
The
may be
and w
in (7).
17 23]
17*23.
KAPTEYN SEBIES
Meissel's expansions of Kapteyn's type.
series,
557
Two
(1)
extremely interesting
9
namely
*?
Z JmVne)
n ~i
n*
+?
l"+f
+ (!*
+ ?)(* + )
<S\
V
'
e'f
f
"(l
f)(8
+ p)
ft
have been stated by Meissel * who deduced various consequences from them it is to be supposed at present f that < e $ 1, and f; is real.
cos2yiif _
2 wti n +
7r
cosh
(ir
1M)
1
a 2|"
'
2sinh7r
(which
to
ir.
is
valid
when
(2ne)
M^
it),
replace
.Vtf+F~Wo
=
2
__ 1
__!(**
/"**
[
r}^
1)
-TrJ.^t
tt
"H
fl)
_ IV
*J
e,
wJo
J
is
|sinh7r|
Now
o
and hence
it is
TO =i
(2w)!
Jo
sinh7r|
by a formula due
evident.
to
mtir(m + i + if)r(i + i-f) Cauchy||; and the truth of Meissel's first formula
follows in like
,c
'
is
now
series
|
w=l
* Astr.
cos (2n
= (2n-l)*+f ~
-1)M
sinh (|tt
- M)% 4fcosh^
'
Nach. cxxx. (1892), col. 363 368. t The extension to complex variables is made in 17*31. % See Legendre, Exercices de Gale. Int. n. (Paris, 1817), p. 166. It is easy to see that the term independent of e vanishes.
||
Mm.sur
les
Cf.
Modern Analysis,
p. 263.
568
[CHAP.
XVn
Now, since the series obtained from (1) and (2) by differentiations with respect to * are uniformly convergent throughout any bounded domain
of real values of
,
we may
differentiate
We
-i
Jm (2ne)
,ro
'
the latter
is
The
to
1, 2, 3, 4, 5
3 are
!?
i'
2.'
~8
'1
*-** + *
2 32
5 s
72"
The
m 1, 2, 3 are
e
2'
2~18' 2~
8i
+ 460'
m=4,
5.
Conversely,
it is
2m
is
*/ (2ne),
2m 1,
is
2 6J_1 K2n-l)},
where o and bn are even polynomials in 1/n and l/(2n 1) respectively, of degree 2m.
17*3.
It
Simple Kapteyn
series with
complex variables.
was stated in
171
that, in general,
series,
Neumann
Kapteyn series are of a more and we shall now explain one of between the two types of series.
is,
In the case of
Neumann
series it
the Bessel functions in the form of a power series in the variable, and then to
rearrange the resulting double series as a power series whose domain of con-
vergence
is
Neumann
series.
in the respective polynomials are not
and c**-1
t-^
2.(m!)
and
<-
r"
2.1.8...(2m-l)
17-3]
KAPTEYN SERIES
of
559
;
Kapteyn
for the
series
2,a n
domain
which
*exp^(l-*')
,.
l-W(l-*)
pii^i^anl'
while the double series obtained by expanding each Bessel function in powers of z is absolutely convergent only throughout the domain in which
[#|.expV(l-|| 1 )
< Um
;
and the
the
first
latter
domain is smaller than the former thus, when the limit is 1, domain is the interior of the curve shewn in Fig. 24 of 8*7, in
;
which the longest diameter joins the points i 1, while the shortest joins the points itx 0*6627434 while the second domain* is only the interior of the circle jr - 0C627434.
j |
series, if
we
a portion of the domain of their validity; and the proof for the remainder of the domain either has to take the form of an appeal to the theory- of analytic continuation or else it has to be effected by a completely different method.
As an example
we
shall give
, = 1 + 2
lies in
1 Jn (nz),
provided that z
<1.
l
venient to describe
+ vXl-**)
This domain occurs so frequently in the following analysis that it is conit as the domain K; it is the interior of the curve shewn in Fig. 24 of 87.
Formula
(1)
is,
of course, suggested
by formula
(2) of 17'22.
To
we
write
+ 2 2 Jn (nz) = 8(z),
= 1/(1 - z).
and then
Since
it
J^^f^^pmj^
(1898), pp.
* For an investigation of the magnitude of this domain, see Paiieoz, Journal it Math. ziv.
(1849), pp.
z.
tup, (8)
560
[CHAP. XVII
we
we can
find a circle
that on
(2)
is true,
the inequality
I
*P {*(- 1/Pi
then
/.v
V
'
W
/ox
= J_
2wt
l ( J(r+) 1
eft
*
To
where
all
investigate (2),
p, u, a,
we
recall
(/>
the analysis of 8*7. If z = pe**, t = eu+ie and m being positive), then (2) is satisfied for
,
values of
a p V(sinh u
+ sin
a)
-u<
and when u
is
left
has
its least
87)
g
1+VU-*
'
which is negative when z lies in the domain K. Hence, when z lies in the domain K, we can find a positive value of u such that the inequality (2) is
satisfied
when
1
= eu
Again,
if
we
in (3)
we
find that
(4i\
S(z>>=J-[
is
1/0}
1/*)}
<ft
'
2iri J (y+) 1
1
1 j
where 7
the circle
e~u
When we
'
and
so
2S(z)
is the
sum of
of the integrand at
its
poles which
lie
V and
is
7.
We next
proved
this,
we
only one pole inside the annvlus*, and, having notice that this pole is obviously t = \.
is
equal to
dt
2m'J (r+>y _ )
<*
log
[l-rexp{fr (*-!/*))]
Lf
+
dt
<nog[l-rexp{fr(t-l/t)} ]
{
2inJ (r+)
W
T
dt
dt dt
dlog[l-texp{-lz(t-l/t)})
dt
\$z(t-l/t)}-\
27riJ(r+ )
= J.
mJ
d\og[l-tr>ex V
dt
dt
(fjj.)
+-L.f
liri J (T+) i T+ )
dlos[-t^{-(t-i/t)}]
dt
dt
* The corresponding part of Kapteyn's investigation does not seem to be quite so convincing as the investigation given in the text.
17-31]
KAPTEYN SERIES
the
first
561
Now
for, if
we
write
then
U <1
|
may be
written
in the form
)
dU
dt
series involved is
ti J r +) U=o
<
j J
zeros of 1
- f exp [\z (t 1
1/t)}
in the annulus is
2iriJi r+
It follows that
>W
is
NK)>be 2/(1
28 (z)
+ exy{$z(t-l/t)}
t-exp{fr(t-Lft)\
at
t
and
It has therefore
domain K,
i.e.
been shewn that S (z) is equal the throughout the whole of the open domain in which the series
is
S ()
converges to the
of K, except at
= 1, but the
17'8.]
17*31.
We
(1) 7
shall
g v =i
M2n*)^ n* + ?
*
l
z*?
z*?
(1"
+
)
(l'+^M^ + l?)
z
+ (?) (* + ) (3 + + ...,
(2)
% J*-i\&ri-l) Z _
z>?
f!
(i+^)(3+r)(5'+r)
which are valid when z lies in the domain and is a complex variable which is unrestricted apart from the obvious condition that ft must not be an integer in (1) nor an odd integer in (2). These results are the obvious
extensions of Meissel's formulae of 17*23.
[Note. The expansions when is a pure imaginary have to be established by a limiting process by making approach the imaginary axis ; since the functions involved in and
(1)
even functions of
{,
no generality
is lost
by assuming that
R (f) is positive.]
562
[CHAP. XVII
is to say,
which we denote by the symbol 8 (z, <f>), and proceed to sum it by Kapteyn's method (explained in 17*3), on the hypotheses that <f> is a real variable and that z lies in the domain K. We define a complex variable yjr by the equation
<f)
yjr
z sin ^r.
<f>,
The
singularities of
yfr,
qua function of
<f>
are given
2
by cos^r =*
\jz,
that
is
None
of these values of
< is
and, as
<f>
in-
creases from
which can be reconciled with the real axis in the ^r-plane without passing over any singular points.
It follows that
if,
for brevity,
we
write
Usr*exp[is(t-l/t)},
then
S ( *'
l-tfa
dt
'
By
we have
and
so 2S(z,
<f>)
is
equal
to the
sum of
the residues
its^ poles
which
shall
lie
bounded by
T and y.
t
We
= e***.
By
equal to
o*+,y->
<
dt
,
nog(l-2ffcos<fr +78)
dt
(r+)
+ 2irij
= _!.( = 2,
* It
is
_lf
(r+>
dlogjPexpi-zjt-l/t)}]
dt
/nj<r+) L=o
\% Un co6(n + l)<^\~dt-r2 a
J
*
<f>
^
r
l+*/(l
-*)
so that
=-=**
!
|
<1.
17-31]
KAPTEYN SERIES
563
the integral of each term of the uniformly convergent series vanishing, just
as in 173.
Now
the residues of
1-ff'
1
r 1^COS^r
= 1 + 22 Jn (nz)coBn<f>,
function of
<f>
and the series on the right is a periodic which converges uniformly in the unbounded range of real
values of
<.
Hence, when
/
tr& dd> + 2 2
=i
to say,
,
d$ =
Jo
Jo
T r dd>. 1-^cos^r
That
is
w
(4)
f r*i^' Jo
"^-J+2i^ + f
=i
is
;
)
l
by Cauchy's theorem,
this
may be
in (4) is
Now, when the path of integration is the real axis, the integral on the left an integral function of z\ and this function may be expanded in the
form
m =o ml Jo
By changing
v
'
=i4n+f
-i
m=1
(2m)! J
K)
that (?)><>.
(fr-lf+p
=i(2m-l)!j*
lies in
raY
the domain
'
K and
...
By
ir),
(w, 27r),
+ 0, f w + 0,
. . .
for
-ty
we
infer that
Jo
sinh$7r./o
1
564
[CHAP. XVII
and that
Jo
cos"*" 1
6&d
"r
By substitution
all
(r +
8
}
ir
a
}
...
and writing 2^ for f in (5) we at once infer the and the mode of extending the results to explained. The required generalisations already been other values of has
in (5)
and
(6)
when
R () >
17*32.
The expansion of z*
into
a Kapteyn
series.
With the aid of Meissel's generalised formula it is easy to obtain the expansion of any integral power of z in the form of a Kapteyn series. It is convenient to consider even powers and odd powers separately.
In the case of an even power, zm we take the equation given by 17*31 (1) in the form
,
n\ {l)
27rtJ
*-ir(l+iOr(l-i)
l.
where the contour of integration is the circle f| = n+ $. Since both series converge uniformly on the circle, when z lies in the domain K, term-by-term
integrations are permissible.
Consider
of
2J,f|-. +
(+)
When $ w, there are no poles outside the contour, and so the contour may be deformed into an infinitely great circle, and the expression is seen to be equal to unity; but when m > n, the poles im are outside the circle and the
expression
is
sum
to
The
m+r
(m + w)! (m n
.
1)!'
is
therefore equal to
Next we evaluate
27rtj| _ n+i
(
dl;
*-*+*
17*32]
KAPTEYN SERIES
the origin
is
565
if
When m^ri,
But,
we take the
1.
the expression
is
seen to be equal to
j
when
m > n, there
= n + \,
and the
expression
is zero.
Hence we have
o)
If
w-<*-)-J H
we
replace
^fig-^
|
+ ...^.
we
n by n -
find that
2W
= S (m + n-l)\Jam (2m*)
t.
(m + n)\J2m (2rm)
1)!'
m?-\{m-n)\
**
w= +1 m^.(m--
and so
(3) V '
m^.(m-M)!
'
If
n=
1,
equation (3)
is
intervening analysis.
When we
given by
have to deal with an odd power, a2" -1 we take the equation 17*31 (2) in the form
,
J_f
2.{l' + ra H38
+f }...{(2n-l)* + f*}
a
X Zi
{i -aisL ATI I J |f|=2W
2
? + (2m-iy
dZ
+-r
H3 +r2 }-..{(2n-i)
2
8 a 2
+?}
Ji {i +r U3 +n--K2m-i) +r
a
^
}
m=i
i J^Ufc, _!),,_ 2
2
.=,+i
(" + "-
7 (m-^^m-ra-l)!
-K^- 1)^1
Hence
(6} (b)
(m + n-2)!^ _1 {(2m-l)^}
.
w -___ (m _
!
w)i
The formulae
and
(6)
may
jn
(7)
5 a*V- (**) -
is
w - 1)
(
/" {(n
+ 2m) z\
'
+ 2mr.m!
which
values
1, 2, 3,
K when
n has any
of the
This formula was discovered by Kapteyn*; the proof of it which has somewhat artificial, seems rather less so than Kapteyn's
Ann.
sci.
(3) x. (1893), p.
103.
566
17*33.
induction.
[CHAP. XVII
Kapteyn
series
We
shall
now give an
series,
alternative
method*
of zn as a
Kapteyn
1 ~~
JL=l + 2
z
m-l
2 Jm (mz),
complex variables in
1,
it is,
is real,
then
1 < z<
and,
if
z is
complex, then z
the domain K.
will
fact that
Kapteyn
W&8
series
2 am Jm (mz)
M=1
is
%
r=l
m WV
T^ +
=
}- 1
* CLm [Z
'
{mz)]
= (1-* ) 2
m=1
so that
it
a m Jm (mz),
(z)F(z) = (l-z')f(z);
F(z) can be determined in terms off(z) by quadratures.
Now, from
(1),
we have
I J^(%nz) =
and
so, if
^-
F(z)
==
Jfm
^^
then
Therefore, in the
(z^)F{z) = )sz\
domain K,
where
and
If
we make
z-*~0,
we
see that
A=B =
Consequently
(2)
*
0.
jP-2 I
m=i
Wataon, Messenger,
J"< 2w *)
rn*
xi/vi. (1917),
pp. 150157.
17-33]
KAPTEYN SERIES
like
567
In
t=o
The expansions
of sn
when n
is 1
Now
form
assume
that, for
zn
is
expansible in the
sn = n 2
bmn
Jm (mz),
(s) defined
by the equation
fnr
(z)
= (n + 2f X
m-i
By
we have
=< +2
>^S+*SS-< b+2
s
>'< i
-* >*"
,
= (n + 2)
On
integration
*n+a.
we deduce that
<f>
{z)
= zn+* + A
'
log z
+ R.
<f>(z)
It is obvious that
A' = B'
of zn+* is
*+*
= ( W + 2) 1
a
where
It follows at once
n+a
* bm
n-
by induction that
and so
fan' i
to say
2n
1 =i (2wi)^-
7=
:rn=n TT\ T (m -n + 1)
.?n a
That
is
=2
ntnm^rim-n + l)'
1
r(m + n)J(2iiu)
in the
to the
~ and this is equation (3) of 17'32. The expansion of zm is obtained same way from the expansion of z the analysis in this case is left
;
reader.
568
[CHAP. XVII
We
f 4)
<**)
is
= -
(K
+ 2OTr,, m
>
,
which
ex-
pansion
is
mZo(n + 2m)
is
n+1
.m\
% l/m ),
a
The expansion
there-
1,
17*34.
z)
in a Kapteyn series.
From
two preceding
sections,
we can
domain
z) when
will
z lies in the
K and
lies
be defined later in
this section.
Assuming that
1
1
>
|,
we have
.
t-z
Now,
^t
|
nZ i
?!!i w+1 *
m-
z
if
exp V(l - g 2 )
1+V(1-* )
2
= jr
'
is
expressible as
series if
2 wrc2
r (n + m) Fn+2W
1
-i-o(w
is
+ 2m)"+ .m!|<|^
convergent.
double series
provided that
Hence, when
t\>2
2;
exp V(l
f2
1+V(1-^)
rearrangement of the repeated series for \/(t z) is permissible, and, when we arrange it as a Kapteyn series, we obtain the formula
(1)
4nn.
sci.
(1893), pp.
113120.
17-34, 17*35]
KAPTEYN SERIES
569
where *
(2)
(3)
*.(*) -1/fc
< n (0
the last formula
;
2
9
w (Mn_2m+1
,
From
discovered
by Kapteyn
it
we have
- S*
by
w-m-1)! _ 1
<*
(rt-2m)2 .(w-m-l)
and
therefore,
91
(2),
2
$nt< n (t) = i
so that,
(4)
(*
^)
{|n*
n (n*)},
- 2 )
0 (nt)
+ sin mr + 1 cos
2
|n7r
when n =
Kapteyn's polynomial
<fb n (t) is
Neumann's
polynomial
It is
for,
O n (nt).
possible to extend the
now
domain of
by
8*7
combined with
and
when
and
t lie
in
(5)
n(*)<n(),
.
n(*)<n(i),
I
where
ft (z)
z exp V(l
-^
is
|MOI<*.(l'l)<(i)=i.
by
(4)
;
so that
(1) is valid
when
0(*)<0(1),. |*|^1.]
17 '35.
expansion of
l/(t
- z)
into
a Kapteyn
series.
Now
that explicit expressions have been obtained for the coefficients in the expansion
=<fl
(()
+2 |
n=l
(t)
Jn (nz\
(t)
it is
be defined as
n (1 - *2 )
the reader will find
it
+t cos2 \nn,
series
S* + 2 +T 1^S
-2) *
n =i
* Cf. Kapteyn,
W.
B. F.
19
570
[CHAP. XVII
polynomials, and
substitute suitable integrals for the Bessel coefficients and reduce the result to l/(tz) after the manner of 9'14.
Or
again, if
we
find that
{{T^f
and
then, dividing
e n2
W * (n2)
we
'
by
17'3
(1),
find that
(<
-l)(<-2)3
(P-\?{t-Z?
e
(?-\f(t-zf
=
whence the
nW
(nz)\
C+ft'+l)'***
(t) (t) is
-n JW.W
and hence
it follows
^-^3
^-^3
Yn (nt)},
J,
that
A n and
2?n are
independent of
but
it
A n Bn =0.
17*4.
series.
into
a Kapteyn
We
fl (z)
shall
following expansion-theorem
is analytic
f(z) = o + 2
2 OnJn (nz),
where
(2)
*.-%z JG.V)f(t) *,
path of integration
is
and
the
is the
curve on which fi
()
= a.
This result
obvious
when we
expansion
(<)
+2
nl
since D,
is
(t)
=a
and
D, (z)
This theorem
is
due to Kapteyn.
17*4, 17*5]
KAPTEYN SERIES
571
/(*)- 2 an zn
n=0
then
(3)
(4)
Oo=Oo,
a
'** (n
~ 2m>2
(w
-m"
1)!
a"-*
17*5.
Kapteyn
of
series in
which v
is
not zero.
The theory
Kapteyn
series of the
oo
type
i
% am Jr+m{(v + m)z\
in which v
is
.
made
to
pansion of zv
The
may be
possible to
prove that
(1)
the domain
to
of analytic continuation.
To
circle,
expand the
coefficient of z v+2r is
(-Y~
,'
m (v + 2m)" +3r
+2r
(r-TO)!
t 2"+2Tc> + 2r+l) to OT
I
(-)'-'" v {
When
known
r>
1,
is
to vanish identically
whenever v
v.
is
an integer.
It therefore vanishes
The expansion
by a comparison of the
coefficient of z 9
equation.
From this result, we can prove that, under the conditions specified
(2)
in
7*4,
j- = S
cJ
64*,
where
n\ W
* This
**,-,
'
< 0*659 by
Nielsen, Ann.
set.
pp. 4246.
572
[CHAP. XVII
And
the reader will easily prove that if/ (2) satisfies the conditions specified
in 17-4, then
(4)
*"/(*)
n=0
Sv
J*+n {(v
n)z],
where
(5)
a,
,
= ^U j/0)
a
2
(0 dt,
;
in
and hence
(fo
=1
'
X*
(v
"
2 m
where a
a lf
...
[2sote. Jacobi in one of his later papers, Aatr. Nach. xxvm. (1849), col. 257 270 [Ges. Math. Werke, vn. (1891), pp. 175 188] has criticised Carlini for stating that certain expansions are valid only when \z\ <0'663.... But Carlini had some excuse for his state-
permissible only in this domain, although the expansions are actually valid throughout the
17'6.
Kapteyn
series
.
series of this
v, and
with simple
coefficients.
The
results required in
by
z sin#.
throughout
n=i
so that
sin
""'
m
yy)
I JnHnz}
nti
z*
1.3
*g
,
n*+?
21 2 +
"
"2.4 ,(P
tm is
+ ^)(2 + t )
2 2
'S,Jn 2 (nz)/iv
t
a polynomial in z2 of degree m;
in a similar
sum
* Cf. Nielsen, Ann. sci. de VEcole norm. sup. (3) xvin. (1901), p. 60.
sci. de VEcole norm. sup. (3) xvm. (1901), pp. 3975. % Cf. Schott, Electromagnetic Radiation (Cambridge, 1912), Chapter vin.
17-6, 17-7]
KAPTEYN SERIES
573
l/^-sTobr^
(3)
j^M-^,.
general theory resembling that of 16 14
-
A
(4)
is
pansion
<*>
r( 2Ui)
io(. +
V
\
+ '-m!
^^ +w>^
which is easily derived from 175 (1) and is valid throughout K\ but it seems unnecessary to go into details which the reader should have no difficulty
in constructing, in the unlikely event of his requiring them.
17*7.
Kapteyn
series
If
y/an
1,
we have seen
the series
that the
Kapteyn
series
%an Jn (nz)
since,
represents an analytic
1,
is real, \Jn (nx)\< although when \z\ > the series does not converge at points which are not on the real axis.
But
when x
axis,
The behaviour
that
it
Kapteyn
series
may
be
summed
up-f-
by saying
it re-
K and
that
As an example,
let
Jn (fix) s=
It is evident that, if
<f>
yfr
x sin
yjr,
then
is
uniformly convergent.
Now, when x > 1, < decreases as -\jr increases from to arccos(l/#) and then increases to ir as yjr increases from arc cos(l/#) to it. If m be the integer such that the minimum value of < lies between 2mir and 2 (m -f 1) it, let the values of yjr corresponding to the values
0,
ym
2tt,
,
-4tt,
...,
2rmr, -2mir,
,
...,
2tt,
of
<f>
be
7, 7j, ...
8 m> &m-i,
$i>
and then
n
Jym
r^Jtr+r
J 8
n =l
made by
Professor Hardy.
574
[CHAP. XVII
lies in
the intervals
is
(7,.,
7r+ i) and
2
(8 r
8r+1 ) the
sum
of the
series
- %Tr<f> + %ir* + r (r + 1) tt + (r + 1)
2
ir<f>,
and, since
l(yfr
sjr,
j(yfr
d^r
= \Jr
4-
sin
-\Jr)
+ #
(-^
it
may be shewn
without
much
2
difficulty that
2
S = ^x* + %x + 2
r-0
{ (S r
r(S r ~7r ).
The reader
this
will see
may be summed by
method*.
The convergence of Kapteyn
series
17*8.
With the exception of the points + 1, the boundary of presents no means of Debye's asymptotic expansion the consideration of the convergence of the Kapteyn series 2an /r+M {(1/ + n)z]
features of special interest; because, by
is
v
and that of two similar
series f
*expV(l*Jn*, \/n
.s
)*
'
*Vt 1+V(1-* ))
2
with
written for
>Jn.
The
points
fail.
expansions
values of
v,
i ft
of
2 an J,+n {(p + n) x)
ensure both the convergence of 2 OnJv +n (v + n) and the continuity throughout the interval^ < x ^ 1.
is
00
it follows
that ^,anJv+n (y
+ n)
converges.
connexion the researches by Nielsen, Overtigt K. Danske Viderukabernes Selskabs, 127146, should be consulted. t If a^/^/n does not tend to zero the series cannot converge; and if it does tend to zero Sc^/^/re 5 is absolutely convergent, and so, if we replace each Bessel function by the first two terms of the asymptotic expansion with a remainder term, the series of remainder terms is absolutely
1901, pp.
convergent.
J Due allowance has to be made for the origin Bromwich, Theory of Infinite Series, 19.
if
<0.
17*8]
KAPTEYN SERIES
J^\{v + )x) /+ (v + n)
all
575
Again,since
is
values of
in the interval
j$#^ 1,
it
of convergence* that
% an Jv+n
is
;
{(p
+ n) x\
uniformly convergent (and therefore continuous) throughout the interval ^ x % 1 and this proves the theorem.
reversing the reasoning,
By it may be shewn that if "2a n v+n (v + n) converges, so does Sotn/n^, so that the convergence of Xcin/n* is both necessary
and sufficient for the theorem theorem of its kindf.
*
to be true; the
theorem
is
Bromwich, Theory of Infinite Series, 44. t This was pointed out by Professor Hardy.
(1917), pp.
Cf.
(2) xvi.
171174.
CHAPTER
XVIII
AND
DINI
SERIES OF FOURIER-BESSEL
18*1.
In his researches on the Theory of Conduction of Heat, Fourier* was led to consider the expansion of an arbitrary function f{x) of a real variable of x
in the
(1)
form
/(#)= 2
am J
(jm x),
where jlt jif js , ... denote the positive zeros of order of magnitude.
(z)
arranged in ascending
The
manner
arises
problem of a chain oscillating under gravity and in Euler's problem of the vibrations of a circular membrane with an initial arbitrary symmetrical displacement ( 1'3, 1*5).
In order to determine the
plied both sides of (l)by
coefficients a m in the expansion, Fourier multi-
xJ
and
].
xJ
J Jo
'
+n,
KtJi(jm),
m = n,
am
- jj
*/(*) J (jm t)
dt
...
If
we now change
denote the positive zeros of the function J (z), arranged in ascending order of magnitude, then
(3)
/()= 2
TO=1
amJ,(im*)>
where
(4)
am =
2
*
+1
\Jm)J
f tf(t) J v (jm t) dt
This more general result was stated by LommelJ; but, of course, neither = does the procedure which has been indicated establish the validity of the expansion; it merely indicates how
in the general case nor in the special case v
and
is
uniformly convergent.
* La ThtorieAnalytique de la Chaleur (Paris, 1822), 316319. t The omission of the suffix y, associated with jj , jj, js, ..., should cause no confusion, and considerably improves the appearance of the formulae. % Studien iiber die Bessel'schen Functionen (Leipzig, 1868), pp. 69 73.
it
18-1]
FOURIER-BESSEL SERIES
is
577
for the
somewhat deceptive;
is
subjected to appropriate
all
integral
tJ f Jo
isvconvergent,
i.e.
when
>
1.
Dini, however,
Several subsequent writers, while proving theorems for the latter range, asserted that the extension to the former range was merely a matter of detail; but it was not until after 1922 that
interest.
anyone took the trouble to supply the detail which is tedious and of no great In the exposition given here, it will be supposed that v> \.
The first attempt at a rigorous proof of the expansions (1) and (3) i& contained in some notes compiled by Hankel* in 169 and published postmore complete investigation was given by Schlaflif a year and an important paper by Harnack* contains an investigation of the expansion (3) by methods which differed
after the publication of Hankel's work;
humously.
A few
(5)
/(#)= 2
,
m=l
bm J v (X m x),
where \, \2 \3
of the function
...
z-{zJ v '(z)
+ HJ
(z)},
when
> \ and
is
The
(6)
coefficients in the
{(xw3
Jo
(\ m t)
dt.
The mode of determination of the numbers Xm subjects f{x) to what is known as a mixed boundary condition,' namely that /' (x) + Hf (x) should formally vanish at x 1.
'
The expansion
(5)
= 0)
in the
problem
when heat
is
radiated from
is the ratio of the the cylinder; in this problem the physical significance of external conductivity of the cylinder to the internal conductivity. * Math. Ann. vin. (1875), pp. 471 494. In the coarse of this paper, Hankel obtained the integral formula of 14*4 as a limiting case of (3).
t Math. Ann.
x. (1876), pp. 137142. % Leipziger Berichte, xxxix. (1887), pp. 191214; Math. Ann. xxxv. (1889), pp. Serie di Fourier (Pisa, 1880), pp. 190277.
4162.
578
It
[CHAP. XVIII
must be modified* by
when
H+v=
make
it
advisable to
name
The researches which have now been described depend ultimately on a set lemmas which are proved by Cauchy's theory of residues. The use of complex variables has, however, been abandoned, so far as possible, by Kneser f and HobsonJ, who have constructed the expansion by using the theory of
of
integral equations as a basis.
On
because
is
seems somewhat unnatural to use complex variables in proving theorems which are essentially theorems concerning functions of real variables. On the other hand, researches based on the theory of integral equations are liable to give rise to uneasy feelings of suspicion in the mind of the ultraorthodox mathematician.
The theory has recently been made distinctly more complete by the important memoir of W. H. Young, who has thrown new light on many parts of the subject by using modern knowledge of the theory of functions of real variables in conjunction with the calculus of residues. An earlier paper
by Filon|| which makes some parts of the analysis appreciably must also be mentioned here.
less synthetic
The question of the permissibility of term-by-term differentiation of the expansion which represents a function as a series of Bessel functions has been discussed by FordU, who has obtained important results with the help of
quite simple analysis
(cf.
184).
More recondite investigations are due to C. N. Moore**, who, after studying the summability of the expansion by Cesaro's means, has investigated the
uniformity of the convergence of the expansion in the neighbourhood of the
origin,
and
The reason why the uniformity of the convergence (or summability) of the expansion in the neighbourhood of the origin needs rather special consideration
is
that
it is
J v (Xmx) which
are valid
when
XmSc
is large;
x approaches
m,
for
The modification
was also noticed by Kirchhoff, Berliner Sitzungsberichte, 1883, pp. 519 524. t Archiv derMath.undPkys. (3)vn. (1903), pp. 123 133; Math. Ann: lxiii. (1907), pp. 477524% Proc. London Math. Soc. (2) vn. (1909), pp. 359388. Ibid. (2) xvm. (1920), pp. 163200. Ibid. (2) iv. (1906), pp. 396430. Cf. 19 21 19 24. IT Trans. American Math. Soc. iv. (1903), pp. 178184. ** Ibid. x. (1909), pp. 391435; xu. (1911), pp. 181206; xxi. (1920), pp. 107156.
||
18-11]
FOURIER-BESSEL SERIES
579
In the exposition which will be given in this chapter, the methods of the calculus of residues will be used to a far greater extent than has been usual in recent researches; this is a reversion to the practice of Hankel and
Schlafli and (in the special case of Fourier series) of Cauchy. The advantage of this procedure is that it results in a great simplification in the general appearance of the analysis throughout the whole theory. And, although it
seems impracticable to prove certain theorems (notably those* relating to fractional orders of summability) with the help of complex variables, the gain
in simplicity is so
marked that
it
possible if the methods of the theory of functions of real variables had been used more exclusively.
very
by using complex
variables, it
where
c x is
sufficient to
a constant, independent of z, when v is given and exceeds |,are prove all the requisite theorems concerning convergence at a point
(or summability at a point) and they are also sufficient to prove theorems concerning uniformity of summability throughout an interval of which the origin
may be an end point. Direct proofs of theorems concerning uniformity of convergence throughout such an interval require more elaborate inequalities, but in this work the use of such inequalities is evaded by deducing uniformity
of convergence from uniformity of summability by an application of Hardy's
convergence theoremf.
It
to the
may be stated
here that the theorems of this chapter correspond exactly theorems concerning Fourier series which are given in Modern Analysis.
tioned
In addition to the memoirs which have already been cited, the following may be menBeltrami, R. 1st. Lombardo Rendiconti, (2) xm. (1880), pp. 327337 Gegenbauer, Wiener Sitzungsberichte, lxxxviii. (2) (1884), pp. 975 1003 Alexander, Trans. Edinburgh
:
(1888),
pp.
(1889),
(6)
223260;
Volterra,
Ann. di Mat.
xxv. (1897),
p.
xiv. (1907), pp. 547549; Messenger, xxxm. (1904), pp. 7077, 178182; Rutgers, Nieuw Archie/, (2) vin. (1909), pp. 375380; Orr, Proc. R. Irish Acad. xxvn. A, (1910),
pp.
Inst.
1,
pp.
8385.
[Jahrbuch
The investigations by Alexander are mainly based on operational methods, while Orr dealt with expansions in which functions of the second kind are involved.
18*11.
In the special case of series of circular functions, it is necessary, as the make a distinction! between any trigonometrical series
00
Ia
*
+
m=l
(am cos
mx + bm sin mx),
9-1.
Such theorems have been investigated by Moore and Young, % Cf. Modern Analysis,
580
[CHAP. XVIII
and a Fourier
which the
am
It
is
If' =
I
7T J v
=-
If".
-.jf
J IT J
necessary to
which
will
make a similar distinction* between the types of series be dealt with in this chapter; any series of the type
00
m=l
in which the coefficients
2 am J v (jm x),
will
2 75 TT~\ v v+iKjm)
f (*) J "
^ dt
>
And
off(*\
if,
sum f(x)
for
any point x of
the interval
2
where X 1}
A-,,
bm
Jv (\ m x)
X3)
...
zJ v '(z) + HJ v {z),
will
be called
Dims
series
of Bessel functions.
by the formula^
l
+ xm J/
2
(xm)} bm
= 2K*
Jo
tf(t)
J v {\m t) dt,
And
if,
sum/(#)
for
interval (0,
Some writers have been inclined to regard Fourier-Bessel expansions as -* oo merely a special case of Dini expansions, obtainable by making but
make
this
(cf.
There are very few expansions of simple functions in which the assume a simple form.
One
*
The
is
due
to
(2)
xvm.
(1920), pp.
167168.
is
t It
convergent for
values of m.
when H + v ^ 0.
18-12, 18*2]
FOURIER-BESSEL SERIES
Another is x', which gives
2Jv{j x ? \ m=\Jm" v+i KJm)
rise to
581
the formal expansions
investigated in 1542.
(i)
x"=
r
i
/a\
_.
V
m=l
(*-m
S
'
It will
0$a;<l,
if
be seen subsequently that (1) is valid when + v>0. Cf. 1822, 18-35.
^x <
1,
and
(2)
when
H
i
\n2
r + 2 + Jv (X m <) dt =
(v
+ 2n) Jv (X m )-Xm Jv
'
(X m )
+ 2n
- 4 (+n)
+ *-> J" r (X m )
is
rf<
is easily established,
positive integer.
j?"
i>
any
may
in this
expressible in terms of
"i
known
functions and
/:
In order to calculate this when
function
('
K
Jv (\ m t)dt.
integer,
v is
an
McMahon*
(t)
dt=*Jx
(x)
which
is
18*2.
Schlafli.
which were described in 181 are based on the by Dirichletf in his researches on trigonometrical series of Fourier's type; this method of proceeding is obviously suggested by the
analysis used
fact that the trigonometrical series are special cases of the Fourier-Bessel
The
\.
f(x) = $a
+ %
(, cos
mx + b m sin mx),
where
am
=IT J
bm
7r J
\
it is sufficient
to prove that
1
/" ir
f(x) = lim
,.e.
{%
that
* Proc.
/(.)-
*,
g^
f* sin (n
10-74(3) in
American Assoc. 1900, pp. 42 43. The tabulation is most simply conjunction with Table I. (pp. 666697) ; see Table VIII. t Journal fUr Math. nr. (1829), pp. 157169.
by using
582
[CHAP. XVIII
n -*
oc
m=l "
v+1
\Jm)
and so
it is
sum
when n
is
is
large;
residues
and it is in this investigation that the use of the calculus more than desirable.
of
examination
is
2\ n*J (\ m x) J {\n t)
v
An
1833 shews that the difference of the two sums is readily amenable cussion, and so we are spared the necessity of repeating the whole
to the
18*21.
by
We shall now begin the attack on the problem of Fourier-Bessel expansions discussing properties of the function Tn x), defined by the equation
(t,
(\\
K**)
4 n [l,X)-
rn
/.
_ v ~
m=l
where
0<x^\,
and
is
+ \> 0.
due
to
will
be used
is
Hankel* and
Schlaflif,
though
function Tn (t, x) is obviously as fundamental in the theory of FourierBessel expansions as is the function
sin (n
sin (x
in the special theory of Fourier series.
+ ^) (x t) t)
'
Tn (t,
mth term
of the
sum
for
Tn (t,
x) as the residue at
of a function, of the
When
this
* Math. Ann. vm. (1875), pp. 471494. t Ibid. x. (1876), pp. 137142. J Proc. London Math. Soc. (2) xvrn. (1920), pp. 163200.
18-21]
FOURIER-BESSEL SERIES
583
Tn (t, x) as
jn and jn+1
it is
.
sides lies along the imaginary axis while the opposite side passes
between
The
moved
off to infinity
I(w)
-* oo
namely
2 2
2 {tJv (xw)
(3)
(4)
J v+1 (tw) - xJ (tw) J+1 (xw)}/{(t - x*)J v (to)}, ttw {/ (w) Yv (xw) J, (xw) Yv (w)} Jv (tw)jJy (w), it w [Jy (w)Yy(tw)-J v (tw) Y v (w)} J v (xw)jJ v (w).
The
first
of these was the integrand studied by Schlafli; the other two are
15 42.
is
xt and
<x+t<
all
2
(3)
when*
when
^x< t <
1.
We
at
tv=jm
(5)
Tn the case of
(2),
g (w) =
if
(tw) J +1 (xw)},
and then,
so that
w = jm +
where
is small,
we have
.-,
Jy (W)=0jy' (jm)
WJy>(w)
= 6*jm Jy'*(jm) +
It is easy to verify,
Jy' ( jm ) { J y" (jm) + Jy' (jm )} + .... using differential equation, that the Bessel's by
3
>
coeffi-
at jm
is
9'(jm)l{jm Jy*(jm)},
and
the residue at j m
is
Yv (jm x) - Jy (jm x) Yv (jm )} Jv (jm t)/Jy = - IT jmJy (jmX) Jy (jm t) Yy (jm)/Jy (jm) = 2 Jy (jm x) J, (jm t)/Jy (jm),
2
(jm )
by
3*63,
in the
*
is
is
dealt with
This
most
easily seen
in the
form
- HvP) (xw)
(2) are
139.
'
584
[CHAP. XVIII
at
next take the contour of integration to be a rectangle with vertices A n Bi, where B will be made to tend to ao and A n is chosen so
,
.
that jn < A n <jn+1 When it is desired to assign a definite value to shall take it to be equal to (n + ^v + \) it, which lies between n &ndjn+l
A n>
we when
is sufficiently
large ( 1553).
Now
it is
w = u + iv,
it
may be verified that when v is large, and either w>0, then the three integrands are respectively
(e-<*-'>l t'i),
(e-fe-s-t'M),
(e~ (t
-x) W),
and
so, for
An
B -* oo when
x and
We
_ j^
rA+i
ri
xt)
y
(7)
Tn (t,
x)
-I
An+
\
" w \J
(w)
(xw)
- J v (xw) Yv (,)}
J^lp
m
(0<t<x<
(8)
1)
Tn (t,
x)
=I
An+CCt
Appp
j
(0<x<t<l)
equation (6) it is easy to obtain an upper bound for Tn (t, x) j; for it is evident from the asymptotic expansion of 7'21 that, when v + \ is positive (or zero) and bounded, there exist positive constants cx and c2 such that
From
(9)
jv
m\ ^ exp l!f ^,
(
iJ,(^
c'
exp
F,
^ 0,
^
i
when
is
on the
line joining
A n ccitoA n + aoi
v.
and
provided that n
Hence
|<*-<>r.ft.)|
so that
;s? ^J_p{-<S-.->!.||*
**'
ttc 2
2
1
(10)
\Tn (t,x)\^
'
1*
- #*
1
(2
- x - t) *J(xt)
18*22]
FOURIER-BESSEL SERIES
585
f Jo
t"
+l
Tn (t,a;)(t*-a?)dt
fv+l rAn+aai
= "^7
7n
J
1*
J An-ooi
J" (xw
">
/ "+ 2
(^ ~ xJ
>- ( tw )
x V)J,-(W)
and hence
(11)
Ijo
f t"
+1
irc^A n (2
x t) *J{xt)
of the
[Note. Theorems obtained by a consideration of integrals involving Bessel functions first kind only can usually be made to cover the origin, in view of the fact that the constant Cx in equation (9) is independent of t in the interval 0<t < 1. Thus (11) may be
written
valid when 0^#^1, 0^i<l. This extension is not so easily effected when integrals involving functions of the second kind have been used because the simplest inequality corresponding to (9) is
\
ry (tw)\^ Cl
'
and
(8)
it is
to the integrand in
is large,
then
sin
_ *A n (t + x)l
sm$-jr(t + oc)j'
'
evident,
we
shall
which
difficulties
magnitude of the error in this approximation is not next evaluate some integrals involving Tn (t, x) by means of caused by the unknown error may be evaded.
18*22.
Integrals involving
Tn (t, x).
shall
now
t
lim t
n-ooJo
t'
+1
Tn (t, x)dt = a?
(0<x<
1)
(2)
lim rt +1 Tn (t,x)dt
n-^co J
= $x*.
(0<a-<l)
From
(3)
these
it is
obvious that
lim
f\''+
Tn (t,x)dt = lx.
be apparent that
w) dt -* x* + l
(0<a;<l)
it will
t'
+1
Jo
Tn
(t,
uniformly as n
-* oo
when x
lies in
the interval
1
^x^
where
A,
is
'
586
[CHAP. XVIII
We
boundedness of
Jo
in the interval in
r t' +1 Tn (t,x)dt
which
< t ^ 1.
was given by Young, Proc. London Math. Soc. (2) xvin. (1920), pp. 173 174, and the proof of it, which will now be given, is his. Formula (2) seems to be new, though it is contained implicitly in Hobson's memoir.
Of these
results, (1)
It is evident that
\\Tn (t,x)dt = Z
JO
3^%^r.
tn=\jm*>v+\\Jm)
When we
transform the
18'21,
1
sum on
manner of
we
f
00 *
2J, (xw)
dw
1
[
dw
tiai-vi
wJ v {w)
^irijA^aoi
wJ\w)
is
first
integral
/:
fW
(tx)dt=x---
f^
J
^(>^
Now
!/An-*>i
WJy (w)
C^A n \/x
c2
4ci
A n (1 x) six
it is also
is
evident;
t>
evident that
a* f
+1
Tn (t,x)dt-x + l
long as
Jo
oo so
^#^ 1
A.
(4
**+-* I
2J - (
y>
(0<tf<l)
Formula
(2) can
be proved in a somewhat similar manner (though the more elaborate) by using an integrand involving
It is easy to see that
u~*)
+l Jm*f v+i\Jm)
= hm
-^
J (w)
F (xw)
- Jy (xw) Y (to)}
V,
'
18*22]
FOURIER-BESSEL SERIES
587
Now take < x ^ 1 in the last integral and substitute for the Bessel functions the dominant terms of their respective asymptotic expansions, valid when w
|
is
large (7-21).
The
0(l/w*) when
0<#<1
is,
at mosti
J Jn-ooi
v?
An
\n)
is
Now
lim
the result of substituting these dominant terms A * +m s iD w 0-- x - i*") 7 ) sin (** ~ [
.
lim
b-*. 2iriJ
-*f-
008(2^-^-^^-^)
wcos(tt/-w tt)
We
shall
have to
it is
type; so
discuss, almost immediately, several integrals of this general convenient at this stage to prove a lemma concerning their
oo
integral
lim
is
f^**
j.
(* -*-* *)
(l/),
as -*. , if
If we put
-1<A<1; and the integral is bounded if'0<X<1. w=A n iv, where A n as usual, stands for (+$*+) w, the expression under
,
consideration
may be
2;Ucos(A-l)A.
L
tA cosh v /o 04*+^)
r "%'*
coshAg.cfo cosh
-sin(A-lM n
v
1
_|
-4
/""
2
^4 n *
/""
and the
first
part of the
Lemma is obvious.
(1 - A) =, we have* v (1 A) sinh Xv=sinh (fl ) ^ cosh v
Again,
if
^A < 1
and v
9j
["
dv
iOA
{">
dv
Lemma
is
proved.
Lemma
that
(1/n),
f
o
t"
+1
Tn (t, x)dt
= \x* +
when
*
The function sinh(t;-) has one maximum, at & sinh* (v - &)/cosh (v- f ) which is less than sinh (t> &)
and
its
V
588
Moreover,
if
[CHAP. XVIII
<x^
1,
we
from the
Lemma
f
+1
Tn (t, x) dt,
<x$
1.
P f^ Tn (t, x) dt
Jo
Jo
are bounded as
-*-
oo
when
Lastly
we
shall consider
+1
Tn (t,x)dt,
1
Jn
and we
when
< t
oo
.
and
<x%
1,
to integral is a bounded
that,
function
ofn,x and
i.e.
t,
as
n -*
It is easy to 1
x + 1 ^ 1,
shew by the methods which have just been used when t^x, then
when
Jo
f +1 Tn (t,x)dt
= I
l
%t +1
- im
B-*-<*>
/
A n -Bi
'>v\'")
sm w(lx). SJn (frW ^ 1/7T \ IT ) ~ w cos (w - *7r - tt) VZj J,* TTiat J An-Bi r^n+Bi cos ^^ + frw w \vtt \tc) / i'+i ~ \AJ ~ jb^Io 27m? _ w cos (w - \vir - \ir) 4n p+k fAn+Bi C0S(lV + tW XW hviT \ir) + hm ^ t J / dw.
1 \
"+*
[An+Bi
fl j
Lemma
and sO the
original integral is
i.e.
bounded when
l<#+ 1^1
first
and
when
is
< t^. x ^ 1.
bounded when 0<x^t^l,vre
shew that
integral
v+1
- J v+l (tw)
(w)\
and then apply the arguments just used in order to approximate to the integral on the right the details of the analysis are left to the reader.
;
be
ct
V +x Tn (t, x) dt <U,
I
JO
where
U is independent of n,x
and
when
A ^ x 4 1, A ^ t ^ 1.
18*23]
FOURIER-BESSEL SERIES
589
These results constitute the necessary preliminary theorems concerning and we are now in a position to discuss integrals, involving T (t, x), n which occur in the investigation of the Fourier-Bessel expansion associated with an arbitrary function f(x).
Tn (t, x),
18*23.
such that x
that, if (a, b) is any part of the closed interval (0, 1), not an internal point or an end point of (a, b), then the existence and the absolute convergence of
is
We
shall
now prove
f**f(t)dt
J
a
oo
-*-
tf(t)Tn (t,x)dt
= o(l),
wherej
0<#<1.
will observe that this
The reader
path of integration in
Jo
F tf(t)Tn {t,x)dt
,
which
is
of any significance, as n -* oo
is
the point x.
convenient to prove the theorem in three stages. It is first supposed is bounded and that the origin is not an end point of (a, b). In the second stage we remove the restriction of boundedness, and in the third
It
is
that t*/(0
stage
we remove the
Let
(I)
t-"f{t)
= F{t){*-<*),
the upper bound of \F{t)\ in (a, b) be K. Divide (a, b) into p equal parts by the points t1 t*, ...*,_, (t = a,t = b); and, after choosing an arbitrary p positive number e, take p to be so large that
let
, ,
and
e,
where
^
f J
(t
m_
m ).
-(*)
so that
a>
m (t) $
|
Um - Lm in (tm_
|
m ).
tf(t)Tn (t,x)dt =
*(*_,)
m=l
Jt m-i
f"
t"+*Tn (t,x)(t*-x?)dt
+ 1
* Cf.
Modern Analysis,
it ie,
9-41.
t If
x=l,
'
590
[CHAP. XVIII
Sc^Kp
\\f{t)Tn {t,x)dt
J a
TTC<?A n (2
x b) \/x
4c
2
fm
Jtm -i
2 7rc 2
(2 - - 6) V#m=l
a;
that
is
to say
/>
'/(*)r.('.)'|<
1wtf( g
e
_,. t) vg[^f+'J^>)
Now
when
(and therefore
we
An
so large that
A n > 2Kp/e.
That
is
to say,
by a
A n we may make
7TC2 (2
2
left less
than
8 Cl
X b) *Jx
-*-
which
is
arbitrarily small.
oo
and
When F (t)
fi,
be possible to
choose r intervals
such that
\F(t)\dt<e.
/u.
When
lies in
we use the
4c
2
inequality
if
If is the upper bound of \F(t)\ in the parts of (a, 6) outside the by applying (I) to each of these parts, we have
If
we take
e sufficiently
K) and
sufficiently large,
and
which had
be proved.
rb
(III)
If
&f{t)dt
exists
and
is
absolutely convergent,
we can choose
Jo so small that
p ]<*/(<)
and then, since we have
n
\dt
<
e,
4 Cl2
Trc 2
2
\i tf(t)Tn (t,x)dt
(2
x b)<s/x J
-a-2
eft.
18*24]
it
FOURIER-BESSEL SERIES
591
\\f(t)Tn (t,x)dt
j
7rc 2
(2
8d2 x
\r + l)Kp
b) >Jx
(v, b)
3e
An
when the
2
intervals
where
is
^ are
omitted.
Hence
Lebesgue
18*24.
it
small by taking
sufficiently large,
is
on the left can be made arbitrarily and so the analogue of the Riemann-
Lemma
completely proved.
now prove the following theorem*, by means of which the sum of the Fourier-Bessel expansion associated with a given function is determined
shall
We
1)-
and
let
and (if it
is
an improper integral)
.
let it
be absolutely convergent.
Let
"m =
**
/8
v+i
, ~
\jm)
f tf(t)J(jmt) dt,
< a < b < 1 and
where v
+ > 0.
Let x be any internal point of an interval (a, b) such that such thatf(t) has limited total fluctuation in (a, b).
00
Then
is
the series
its
2
m=l
a m J v (jm x)
convergent and.
sum
is
We
first
observe that,
m-\
J {/(*
2 am Jv (jm x) =
n-*oo
tf(t)Tn (t,x)dt,
J o
VT
1
(t,
x) dt
8* (*) s
V*
{IT' f(t)
{-/()
as
it
is sufficient
to prove that
Sn (x)-+0
*00
n-oc
convergence of
2
to the
0>mJv (jmX)
(2)
592
[CHAP. XVIII
We now
discuss
1
V+
.
[tr"f(t)
J iX
in detail,
Sn (x)
in precisely the
and the reader can then investigate the other integral involved same manner.
total fluctuation in {a,
b),
where
(t) t~"f (t) - or* f(x + 0) = X* (*) functions of positive increasing bounded are an d Xi (0 X2 (0
in (x,
b),
such that
Xi(*+0)-#(* + 0)0.
Hence, when an arbitrary positive number positive number 8 not exceeding b x, such that
e is
8.
+ 0)} Tn (t, x) dt
0)}
-/x+S p+i r
|
f ( t ) _ a;/ (a; +
<" +1
Tn (t, x) dt
X+S fa;+a
1
ra;+8
'
J x
X! (0
Tn (t, x)dt-
***
X
>
We now
from the analogue of the Riemann-Lebesgue lemma that the modulus of the first can be made less than e by taking n sufficiently large.
It follows
Next, from the second mean- value theorem and 8 such that between
it
number
r
J X
+S
t+>
Xl
(t)
Tn (t, x) dt =
xx (*
+ 8) J
('*'
*" +1
Tn <* *> dt
>
*+f
and, by 1822, the modulus of this does not exceed 2Ue; and similarly the modulus of the third integral does not exceed 2Ue. By treating the integral and *ina similar manner, we deduce that, by taking n between the limits
sufficiently large,
difference
between
0) +/(*-<>)}
+ 2)
00
e;
and
we have proved
its
that, in the
is
2
m-l
am J v (jm x)
is
convergent and
sum
and
this is the
Modern Analysis,
3-64.
18-25]
18*25.
FOURIER-BESSEL SERIES
593
Let f(t) satisfy the conditions enunciated in 1824, and also let/(<) be continuous (in addition to having limited total fluctuation) in the interval (a, b).
Then
uniformly to
This theorem is analogous to the usual theorem concerning uniformity of convergence of Fourier series *; the discussion of the uniformity of the convergence of the Fourier-Bessel expansion near x = 1 and near x = requires
rather more careful consideration, in the
first
1822 (1) is
untrue when x
to
not practicable
P^Tnit^dt,
Jo
when x and
The difficulties in the case of the neighbourhood of x = 1 are easy to overcome (cf. 18-26); but the difficulties in the case of the neighbourhood of the origin are of a graver character and the discussion of them is deferred
;
to 18-55.
choice of 8 which
(a
since continuity involves uniformity of continuity f, the was made in 18*24 is independent of x when x lies in
+ A, b -
A).
Next we
f xJ
provided that
v+1
{'""/CO
x, it
Riemann,
F
is
t+i{t-f(t)-x-''f(x)}dt
J x-\-6 X
a bounded function of x.
Now
11*1
l
l^r^^f^-^'f^)} dt U
I
\o
\t h f(t)\dt+\x-*f{x)\
[\+idt, Jo
+ A,
- A)
since f(x)
is
Modern Analysis,
9-44.
t Cf. Modern Analysis, 3 61. It is now convenient to place an additional (trivial) restriction on s, namely that it should be less than A, in order that the interval Ix- d, x + d) may lie inside the interval (a, b).
594
[CHAP. XVIII
and so
n
m=l
2 am J(jm x)-f(x)
,
and
theorem
stated.
18*26.
The uniformity of
the convergence
near
x=l.
the point x
is zero.
= 1,
sum
/(l-0) =
is
associated with/(j5)
is to be continuous in conditions stated* in 18*24, the combined with (a, 1) and that /(l) is zero, throughout (a + A, 1). are sufficient for the convergence to be uniform
We
shall
now prove
The
analysis
is
we take
V+i
/,
\t-"f(t)
-x-f(x)}
Tn (t,x)dt,
into two parts (0,
we then
8),
(x
8,
8,x + 8), (a* + 8, 1), if x ^ 1 8, or x ^ 1 8. And we then prove that the 1),. if
(x
8),
may
Again,
\f(x)\<
when
o\
$x$
1.
f{x)-x-f{x){\'^Tn {t,x) dt
Jo
when x
for
(U+l)e
(1-8,,
*
1).
The
interval (a, b)
is,
(a, 1).
- S t ), by
18*21.
18-26, 18*27]
FOURIER-BESSEL SERIES
595
f(x)-x-f{x)\\^Tn (t,x) dt
.'o
a; in (a + A, 1) by a choice of n which is independent of x\ and this establishes the uniformity of the convergence of
Jo
to the
\\f(t)Tn (t,x)dt
sum/(ar) in (a
+ A,
18-27.
tvhere (a, b) is
It is easy to prove that, if t*f(t) has limited total fluctuation in (a, b), any part (or the whole) of the interval (0, 1), then
From
this
when 0<#$1;
theory of Fourier
r*
Mo
where
c is
c,
a constant, independent of
when
lies in
the interval
(0, oo
).
Now
in (a, 6);
write **/(*)
= >M0 ~ ">M0>
(t)
are monotonic
|
'
*i (0 ** <L (M) dt
= ^(a)
= 0(\-*).
J a
<*
<2c{|+i(a)|
+ |*i(6)|}X-
is
A similar result
If
it is
holds for
yfr2 (t),
evident.
known merely
that
rh
f t*f(t)dt
J a
all
is
the
*/(0 Jv (M) dt
= o (1/V\).
(1889), p. 247.
* Quarterly Journal,
xxm.
596
This theorem
is
[CHAP. XVIII
in precisely
may be proved
shall write
We
when
t*f(t)
is
when the
into
bounded, with upper bound K, and leave the reader to confunction is unbounded, on the lines of 1823.
Divide
(a, b)
tx
t2 , ...
t -, 1 -(t
, <p
= b),
and
let
2
r=l
{t
m m _i) ( Um L m) <
where
Um
and
Lm
(t
m -i, tm ).
Next
and then
I
let
tlf(t)
F(t),
F(t) = F(tm^) +
oy
m (t),
tm
\\f{t)Jv (\t)dt\<K I
I
\Ja
m = l>
tU
(M)dt\+
I
J tm-i
m=l J['m-i
where
is
(t)\ in
the interval
(0, oo
).
1-8-23,
left is o
Hence, by (Ar J ),
The theorems of
made
(0^#<
1) in the forms
This
is
evident
when
it is
remembered that
Hence
the general
x^
1)
if x*f(x) has an integral which is absolutely convergent ; and, if this function has limited total fluctuation, the general term tends to zero as rapidly as l/jm
18*3.
to Dini's expansion.
We shall now consider a class of contour integrals by means of which we can obtain theorems concerning Dini's expansion, analogous to those which have been proved for Fourier-Bessel expansions, either in a direct manner or
by means of the corresponding theorems
00
for
Fourier-Bessel expansions.
is
2
>
bmJrCkmX),
where \i,X 2 \ ...are the positive zeros (arranged in ascending order of magnitude) of the function *Jw'(z)+HJ.(z),
* Proc.
London Math.
Soc. (2)
rem.
(1920), pp.
169171.
18 * 3 ]
DINI SERIES
597
where
+ i^O.
The
coefficients bm are to
bm
.
f Jo
so that
Before proceeding further, we shall explain a phenomenon, peculiar to certain Dini expansions, which has no analogue in the theory of Fourier-Bessel
expansions.
is
when
Further,
It
is
H+v= H+ v
if
origin.
only to be expected that these zeros should contribute to the terms of the series, and such a contribution in fact is made.
If
(1)
H+ v = 0, an initial term
2( J
/
+ 1)#"
[\+if(t)dt
Jo
H+ v
term
is
K '
(V + P)/.
(5l.)-V/r (^)io
zeros + i\
/,
/()y, (V)Ctt
'
These initial terms in the respective cases will be denoted by the common symbol <%(#), so that the series which will actually be considered is
where <%(#>
is
n x),
is
is
H+v = 0, H + v<0.
[Note. The fact that an initial term must be inserted when ff+v=0 was noticed by Dini, Serie di Fourier (Pisa, 1880), p. 268, but Dini gave its value incorrectly, the factor xv being omitted. Dini's formula was misquoted by Nielsen, Handbuch der Theorie der
Cylinderfunktionen (Leipzig, 1904), p. 354. For corrections of these errors, see Bridgeman, Mag. (6) xvi. (1908), pp. 947948; Chree, Phil. Mag. t (6) xvn. (1909), pp. 329331 and C. N. Moore, Trans. American Math. Soc. x. (1909), pp. 419420.]
Phil.
' '
598
[CHAP. XVIII
We now
...,
...,
(0 or
i\o).
The residue
of the function
at^
2Jy(jm x)Jy{jm t)
The residue
at
+ JJ (Xm ) + H JJ (\m)}
2 \ m2 J y (\fnOC) J p (Xmt)
(
*
The residue
at the origin
The
residues at
i\ when H+j/is
(xa
2X 2 /y (X
+ ir-) /,* (x ) - x 2 // (x
lies
;
Now
it is
let
Dn be
a number, which
not equal to any of the numbers jm numbers jm which does not exceed n
between X and X +1 so chosen that and let jjf be the greatest of the
,
Let
Sn (t,x; H) = 2
2J
m=l
^^~'^-S4
(x,t)
i>+i
\Jm)
where 4
(x, t) is
defined to be
0,
2X 2 /p (A ar)/> (X
(V + v ) L
2
CK)
~X
//" (,Xo)
according as
H+
i>
is positive,
zero or negative.
Then, evidently,
m-1
2 am Jy ( m x) - JT
(x)
- t bm J
m=l
H) dt.
when
We
shall
1,
< x<
Jo
are sufficient to ensure that, as n -* oo
,
18-31, 18*32]
DINI SERIES
599
This equation enables us to deduce the properties of Dini's series in respect of convergence* from the corresponding properties of the Fourier-Bessel
series.
18*31.
8n (t, x
H).
(t,x;ll)-
(t0)
1_
27rt
p
'
wi
f J -ooi
where the symbol P denotes Cauchy's principal value.' The integrand being an odd function of w, the second integral vanishes, and so we have
m K)
An
(2)
(t v nK
'
'
J)n - aDi
l>
,(w)}-
immediate consequence of
1821)
is
that
|iS;<-U;10|*. (2-X-t)y/(xt)'
cs is
where
independent of
n,
x and
t.
Also
and hence
!/. o
,
t"+*Sv. n (t,x;H)dt\^
/-j
t.
(2-x-t)Dn ^/x'
where
c4 is
independent of n, x and
18*32.
Sn (t,
x;
H)
We shall
now prove
the whole)
of the interval
and
absolute convergence of
J a I
"fif{t)dt
,
are sufficient
to
provided that
< x < 1. And, if b< 1, the theorem is valid when < x $ 1. The proof has to be divided into three stages just as in the corresponding theorem ( 1823) for Tn (t, x). We shall now give the proof of the first stage, when it is supposed that t*f(t) is bounded and a > 0. The proofs of the redifficulty.
*
x=l.
600 Let
[CHAP. XVIII
=
>
F(t),
and
let
^(0
be
if.
t2 ...
Divide
into
p equal
p
parts
by the points t u
tp^. (t
= a,tp = b),
number
t
e,
1
where
Let
so that
|
Um - Lm) (tm -
m ~!) <
Um and L m are
a
(<*_!, TO ).
F(t)
m (t) ^
j
Um L m
in
( OT_i, TO ).
Then
tf(t)Sn (t,x:H)dt
/,
= I Fit^f**
Hence, by 1831,
/"
"Sn (t,x;H)dt + I f
W
t+*a> m
(t)Sn (t,x;H)dt.
^7
so large that
3
[2Kpc<
and
if
we now take n
Dn ec >2Kpc
we have
*/()
s ft * ff )
;
*
|
<
(g
_*.V
Hence the
integral on
is
arbitrarily small.
tends to zero as
n -*- oo
the reader has removed the restrictions concerning boundedness and the magnitude of a by the method of 18'23, the theorem is completely
When
proved.
As a
corollary, it should
be observed that
Cb
a*
j
tf(t)Sn (t,x;H)dt
oo
when
^x^
1 if 6
<
1,
and when
^x
*S
-A
1,
where
is
an
18*33.
Dim 8
immediate consequence of the result of the preceding section the existence and absolute convergence of the integral
An
that
\*f(t)dt
/,o
with/(#) behaves are sufficient to ensure that the Dini expansion associated (or summability), as the Fourierin the same manner, as regards convergence Bessel expansion throughout the interval (0 < x < 1).
18-33]
DINI SERIES
it is
601
For
evident that
M
uniformly to zero
(x)
and
this
sum
(multiplied
by
>Jx)
tends
Now,
(
since the
it
1523),
follows
numbers \ m and jm which exceed v are interlaced has the same value that Dw may be chosen so that n
| |
after
a certain stage.
n
Therefore, since
x*
am Jv (jm x)~*0,
uniformly throughout
(0, 1),
x)}
uniformly throughout
(0, 1
A).
That
is
xi&
is
A)
and
its
sum
is zero.
when the
(uniformly)
series is 'summed' by Cesaro's means, or any summable and its sum is zero.
'
'
similar method,
it
(0, 1
A),
the series
m=l
series
X*M
is
(x)
+ 2 xtbnJ.Q^x)
m=l
the
convergent (or is
'sum.'
summable by
same
And if, further, the Fourier-Bessel series (multiplied by \/x) is uniformly convergent (or uniformly summable) throughout an interval (a, b), where
Q^a<b<
formly summable) throughout
(a, b).
1,
then also the Dini series (multiplied by \/x) is uniformly convergent (or uni-
(a, b)
where
0^a<6< 1,
* Cf.
Bromwich, Theory of
w.
b. p.
20
602
then the series
[CHAP. XVIII
m=l
converges to the
sum
* {/(* + <>)+/( -0)}
that
at all points
x such
a + A^x^b A, where
f(x)
is
is arbitrarily
6).
small ; and
continuous in (a,
18*34.
series at
= 1.
sum
of Dini's
We
series
shall
now complete
by considering the point x = 1 and we shall prove the theorem, due to Hobson*, that, if/(a) has limited total fluctuation in the interval (a, 1), the sum of the Dini expansion at x = 1 is /(l 0).
<J (r
A 4-
*?
Tn {t,x;H) = \
k
27rt'Ji>n _aoi
]D
2iri J Dn -9>i
w Jv
-
(w)
+ HJV (w)
'
where
</>
(w, x)
7T
[{w j; (w)
{w
is
Tn (t,x\ H)
(t,
when 0<
< x^
1,
the
These representations of
(t,
Tn
Tn
x) given
by
18*21 (7)
-
and
Tn
(t,
x H) analogous
;
to 18 21 (6) is
x; H) are strictly analogous to the representa1821 (8) the fact that there is no formula for the reason why Dini series were discussed in 18-33
series.]
Now
f Jo
when
v+1
Tn (t,l;H)dt
< t < 1.
We
have
/V T(tlH)dt- lim
JJ-^oc
'
Dn+m
[
(M)^, (*)<**
WJ W
(W) +
Tl J Dn-Bi
-JTl
HJ
(W)
B^oo
* Proc.
Dn - Bi \wJ+i(w)
\WVJ
London Math.
18-34]
DINI SERIES
8, it
603
follows from
1821 that
1
this is
(S, 1).
When
.
** t *S
h, it
is
(1/Dn).
And when t =
1, it
when n-*oo
It follows that
I bm J S (1) +=i
write
(\ m )
-/(l -
0)
(*,
H) dt.
Jo
Since t~"f(t) /(l 0) has limited total fluctuation in (a, 1) we may it in the form X\ (0 ~~ X* (0> where X\ (0 and X* (0 are bounded positive
t
decreasing functions of
such that
Xi(i-o)=x*(i-<>)=o.
number
Hence, given an arbitrary positive number B, not exceeding 1 a, such that
e,
we can choose
a positive
whenever
8 **
$ 1.
We
f
then have
1
V+
[t-"f(t)
-/(l - 0)}
Tn {t,l;H) dt
0)} T(*,l;
Jo
= P~V> {*-7(0-/(l
Jo
5)*
;
o(l) as w-*oo
2elim
in absolute value
(cf.
f +1 Tn (t,l; H)dt
is arbitrarily
18*24),
small.
It follows that
lim f
n-^at>J
+1
{f'f(t) -/(l
m=l
converges to the
This discrepancy between the behaviours of Dini series and of Fourier Bessel series ( 18*26) is somewhat remarkable.
604
18*35.
[CHAP. XVIII
extending
x= 1.
series
Because Dini
1, it
that /the condition that f(x) is continuous* in existence and absolute convergence of
(a, 1),
\\*f(t)dt,
Jo
may
be
in (a
+ A,
1).
is,
We
The reason
18*26) near
was the
fact that
v+ *T (t,x)dt \ t n
Jo
1),
as
was seen in
1822.
We
shall
+l
Jo
Tn (t,x;H)dt
1),
failure is
removed.
consideration of
1826 should then enable the reader to see without Dini expansion converges uniformly in (a + A, 1).
Jo
is
\ t*+*Tn
(t,x\H)dt
the
sum
of the residues of
v+l 7rt
at \ lt
Xjj, ...,
- {w YJ (w) + HYV (w)} J+1 (tw)] x J (xw)l{wJJ (w) + HJ V (/)}, or + iko if H + v < 0. at
x;
Hence
is
v+1
Jo
Tn
(t,
H) dt
the
sum
of the residues of
- (2/w) (H + v) J v (xw)l{ivJ v
and hence, when
'
(w)
+ HJV (w)),
< x ^ 1,
,
n
t>
+1
m r v T n (t,x; H)dt = x
*
H +r-\ v
z, +0*'
f
Jw (xw) dw r//\ Lu r
\ l
Without
restriction
18-35, 18-4]
DINI SERIES
is
605
of the order of
magnitude of
eip{-(l-)|J(t0)l]
and so the integral on the right converges uniformly when A ^ x $ 1. That is to say
to zero like l/(Dn Jx)
l t"^Tn {t,x;H)dt
1); and we have just seen that this is a sufthe uniformity of the convergence of the Dini series associated with/(<) to the sum f{x) in (a + A, 1) under the conditions postu-
The
differentiability
of Fourier-Bessel expansions.
In the earlier part of this chapter we obtained an expansion which, when written in full, assumes the form
We
is
(2)
shall
now study
it
/'(#)
This problem was examined by Fordf, and his investigation to Stokes' researches on the differentiability of Fourier series^.
Ford also investigated the
differentiability of Dini's expansion
analogous
when
H=-v,
but his
method
is
It is evident that
truth of (2)
if
we can succeed
in
proving that
( 3)
v+1
(jmtV x);
+ 1) Jr+i (*)}.
inside
fy{f'(t)-j/(t)\dt
exists
*
and
is
absolutely convergent.
The term
in
is infinite; this
wJv'(w) is more important than the term in Jv (w) except in the Uviit when shews clearly the reason for the difference in the behaviour of the Dini expansion
% Cf. Modern Analysis, 9-31.
from that of the Fourier-Bessel expansion (cf. 18-26). t Trans. American Math. Soc. vr. (1908), pp. 178184.
606
[CHAP. XVIII
by the formula
bm
=
{
*iV* Jo I V' (*) " "/(*)} J*+i ijm.pt) dt Wi (>,,) +j*m,*J'\+1 (i,r) JV - (" +
7v^)/o S r
*
r+V- <^<>*
I" jm,v
*/(*) Jv (jm,yt) dt\
jo
= j-
/
Q,
Jm,v
wu
provided that
t)\
= 0.
may be
t***f(t)~0 as t+0,
/(l-0) = 0,
f(t)
is
< t<
1.
18*5.
series.
combined with the expression of Tn (t, a) as a contour no easy matter to discuss by direct methods the question of the summability, by Cesaro's means, of the Fourier-Bessel expansion.
It
is,
of Riesz*
series,
will
It is evident that
l
(1)
(2)
ym.
(1910), p. 309.
18*5, 18*51]
FOURIER-BESSEL SERIES
607
where
(2)
Tn (t,x\R)= i
(i
- ip) -^ (}?*>/: [* m l)
and so it will be convenient to discuss the properties of Tn (t, x R) after the manner of 1822 before we make further progress with the main problem.
j
18*51.
Theorems concerning
is
Tn (t, x
R).
and
x,
and so we
defined by equation (2) of 18'5, it is a symmetric shall proceed to establish the properties of the
^ t^ x ^ 1, and we can then write down the < <r *% *% 1 by interchanging t and x in the
is
We
first
observe that
Tn (t, x R)
|
the
sum
of the residues of
(l
~) [J*{w) Y
(xw)-Jv {xw)
Y(w))
j^
obvious that,
when* t < x,
ZlJAn->i\
AJ
Jp (w)
Jp (tw)/Jp (w)
is
an odd function of vk
We
shall
now
i,
and when
is
when
on the ^x^ 1
and
^t%
1,
To obtain these inequalities, we shall use series of ascending powers of w when w\ is not large, and inequalities derived from the formulae of Chapter vn when w is not small.
|
When
.
$.
.r,
as
B - ao
There
no need
to
make an
iB to A n ill do not tend to zero indentation at the origin, because $ (w, x) is analytic at
the origin.
608
|CHAP.
XVm
We first
(1)
Jv (tw)/Jy (w). We
|
observe that*
J, (tw)
Jy{W)
<^exp{-<l-t)
I(w)\}
when
is
18*21 (9) when w is not small, and from the ascending series not large (i.e. less than jm).
|
when
is
We
is
equal to
convenient to
The
first
investigation h
quite simple.
and
33 that
(2)
^.-(^K^J.i^-^K^g-jjJ
I I
j
pixw
J.
\^v^\
Hence
(3)
(II)
|*< *)|<^exp{(l-*)|/()|}.
When
>
\ and
w is not large, it is easy to deduce J (w), Yv (w), Jv (xw) and Yv (xw) that
| |
from the
\^(w,x)\<ks \w\x-".
|
If
is
not small,
we use the
(5)
iW'WK^-
|ff,.()|<*gp.
W
|
whether
xw
I
is
large or not.
1
Hence,
|
(7)
<P (w, x)
},
is large,
If we now combine the results contained in formulae deduce that, whether ^%v^\ or v^^,
(8)
1
and (7) we
4> (w, *)
},
their values
numbers Jtj, k^, is , ... are positive and independent of w, x and may, however, depend on the value of v.
t;
x ^ 1.
18*51]
FOURIER-BESSEL SERIES
is
609
(1), it follows
when
that
(9)
< x ^ 1. Hence, by
\^{w,x)
I
O v \W)
^\<hr^(^ + x-')^\-{x-1^\I{^\\,
i
when
^ x ^ 1 and
and +
Q^.t^.1.
to the integral formula for
We now return
A n iv
deduce that, when
iv in
the
first
^t<x^
1,
ir.ft.|ji)i<^(^ +
^)j;^.*-*g+^.
(ot<*i),
t
We
(io)
(ii)
It is to
|
yj |<gffi^ tt .| a)
is
<o<.<i).
and
t,
be remembered that k6
|
independent of
so.
a:
so that
we may
make x t tend
One
of
to zero, if
we
desire to do
is
Tn (t, x R) when
x and
To obtain them, we
i
write
r-< t "' B
when
vertices
^ < a? ^ 1
+ iA n
,
A n iA n
easy to see that (9) is satisfied whether w be on the horizontal sides or on the vertical sides of this rectangle and the factor 1 (iv/A n ) does not
It
is
;
exceed
V2
in absolute value at
+ x~) V2
6A n we
,
is
infer that,
when O^t-^x^.
1,
and
similarly,
1,
(13)
Tn (t,x\R)\<^!^^.
enable us to discuss adequately
The
The reader will observe that the consideration of small values of x has increased the length of the analysis to an appreciable but not to an undue extent.
610
18*52.
[CHAP. XVIII
We
can now prove that the existence and the absolute convergence of
l
Jo
ftf{t)dt
are
sufficient to
at all points
x of the open
And
the
series is
*{/<* + Q)+/(*-0)}.
This theorem
series.
is
is
convergent is
summable
when
<x<
lira
1,
t>
+l
Tn (t,x\R)dt=
lim
V*
Tn (t,
R) dt
|
Hence
lim
it
follows that,
+i
when the
limits /(a:
+ 0)
exist,
then
X
n - oo J(o
t''
Tn (t,x\R)x-'f(x-0)dt+
now
lim (\'+ 1
n-*>oo J x
Tn (t,x\R)x-'f(x + 0)dt
R), defined as
We
are
sum Sn (x
n (t,
|
I 1 \ m=l (
JO
f
VT
1
and we
large.
.
t*+ l
Tn
(t,
sufficiently
The sum 8n (x R)
\
is
equal to
(t,
f jo
x\R)dt
+
positive
f
J X
v+l
{tr"f(t)
Now, on the hypothesis that the limits f(x 0) exist, if we choose an arbitrary number e, there exists a positive number J 8 such that - *-v f{x + 0) < e, (x^t^x + f *-7(0
I j
8),
ll
t- v f(f)
~ r*f{x - 0) < e,
|
(x
> t> x -
S).
We
now choose a
8,
o- (n), which is less than 8 for and divide the interval (0, 1) into six parts by
the points x
* Cf.
x a (n),
X It
is
x.
Modern Analysis,
9*4. t Cf. Modern Analysis, 8-43. convenient to take 8 less than x and 1 - x.
18*52]
FOURIER-BESSEL SERIES
(0,
611
In the intervals
(x
8),
(x
8,
1851 (10)
and (11); and in the intervals (x <r(n), x), of the form given in 1851 (12) and (13).
It is thus found that
x+
<r(n))
we use
inequalities
^^it^ sr
)} dt
>
V*dt
3A^
f*
tv+i dt
+ -, A-*V_ &*Jx}f
n
Zk6
(*'
+*
{*"/(*>
1
- *-/(* + 0)}
dt.
I
x +*
For any given value of e (and therefore of 8), the first and last terms in this expression can be made arbitrarily small by taking n sufficiently large, on
account of the convergence of
n
\t*f(t)\dl
/;
An
and,
if
U(n)
3A n*a(n))
V2
n,
J'
small as
this is
independent of
small
and
it
can be made as
e sufficiently
initially.
We
|
Sn (x R)
|
can therefore make the intermediate terms in the expression for as small as we please by taking e sufficiently small, and when this
|
first
and
last
That
is
made
arbitrarily small
by taking n
lim
Sn (x\R) = Q.
Hence
lim
n~oom = l\
lim
>
f
J
Tn (t, x\R)dt
+ x~"f(x + 0)
since the limits on the right exist.
lim
n-oo
f"H
f Jx
Tn (t,x\R)dt,
612
THEOKY OF
BESSE.L FUNCTIONS
is
[CHAP. XVIII
,
equal to
\xv
it
has
now been
2 am Jv (jm x)
summable (R) with sum l\f(x + 0) +f(x - 0)} provided that the f{x 0) exist and this is the theorem to be established.
is
;
limits
As a
continuous in
be able to prove without difficulty that, if f(t) is the summability (R) is uniform throughout the interval in which
is
a + A^x^.b-A, where A
18*53.
origin.
Cf. 18*25.
We shall now examine the uniformity of the summability (R) of the Fourier-Bessel expansion throughout an interval of which the origin is an
end-point.
tiplied
It will be supposed that the expansion is modified by being multhroughout by </x, and it will then be proved that, if t~"f(t) is con-
tinuous in the interval (0, 6), then the modified expansion is uniformly throughout (0, b A), where A is any positive number.
summable
Given
e,
8 (less than
A)
so that
whenever
x-8^t^x+8 and
> 0, provided
that
lies in (0, b
- A).
may be
taken to be independent of x.
We
now
write
Sn (x\R) =
|
(t,
x\R)dt
after the
manner
of 18'52.
We
express x*Sn (x
R)
as the
sum
* (
8),
A n 8*
'>rnrvw--/(.)|i* Jo
+ An (t-*y V2 U.- x+ (n) +s r sAj dt 4^5 [ dt + r
2kt e (x*+* + x)
["
f* -<-<*>
dt
3A n*
r*
I dt
J.. rW
A n LV 2
/Li.
!* + ,in)(t-xYJ
r\
xi
\{t- y f(t)-*-f(*)}\dt.
In this formula any of the limits of integration which are negative are supposed to be replaced by zero.
18*53, 18*54]
FOUBIER-BESSEL SERIES
upper bound
for
|
613
Now
this
oo
Now
it
(
1
18 22) that
-
F+
Tn (t,x)dt
so,
Jo
is
uniformly convergent in
(0, 1
A), and
x\R)dt
A).
(t,x\R)dt
R)dt,
Jo
t"
+1
Jo
Tn (t, x
i.e.
to x*f(x) in (0, b
- AY
TO = 1
2 am x*Jv (jm x)
is
(0, b
A)
with
sum
.'o
fV<o dt
and that
t~ v f(t)
is
exists
and
is
absolutely convergent,
continuous in (0,6).
18*54.
series.
We
series*
shall
2
m=0
a m xlJv (j m x)
(ii)
on the hypotheses
(i)
that
j\if(t)dt
exists
and
is
(iii)
is
summable
(R).
fm {x)
in place
that/,,, (x)
The
may
614
Consider
first
[CHAP. XVIII
I
m
(l-k)fm (x)
1
Jn/
(of Riesz' type) for
summing
the series.
l, it is
evident that
lim
exists
I (4lLZh\ fm(x) I
and
is
equal to lim
(l-Mfm
x).
it is
m=\
so that
2 fm(x) = o(n),
w--ac \
Jn
m=l
and therefore
lim
I (l-k)fm{x)=
li
m | (i_is) /m(a );
.
the limit on the right exists in consequence of the hypotheses made at the beginning of the section.
Again, since
Jn
whether
m be
o{n) or 0(n),
lim
it
follows that
2,(f-)/.W=o.
and
so
lim
and
Jo
exists
and
is
m=l
is
2 am xljv (jm x)
summable (C
1) with
By the same
is
reasoning, if /(a?)
continuous in
(a, 6),
uniform in (a + A, 6
is
A)
and, if a
(0, b
summability (C 1)
uniform in
A).
18-55, 18-56]
18*55.
origin.
FOTTBIER-BESSEL SERIES
615
We
of
has limited
can now prove, by using Hardy's convergence theorem*, that, if ^ fit) total fluctuation in (0, b), while f(t) is also subject to the conditions 18-53, then
1 am x*Jv (jm x)
JB=1
is
uniformly convergent in
(0, b
A)
with
sum
x*f(x).
equal to zero in
Let h {t) be an auxiliary function defined to be equal to f(t) in (0, b) and (6, 1); and let the Fourier-Bessel series associated with h(t) be
eo
w=l
co
Then, by
(0, 6
18*54,
m=l
A)
a mlJjm
is
a bounded function of
a;
and m. Hence, by
m=l
is
(0, 6
A),
with
sum
x*f(x).
Again
(am
by an analogue of the
Hence
2,
to the
sum
x*f(x) in
(0, 6
A)
as
18 56.
Summability of Dini
series.
Except when x = 1, the summability (G 1) of the Dini series associated with 18*53. f(t) may be inferred by combining the results of 18*33 and 18*51
all
The summability (01) may, however, be established independently f for < x $ 1 by replacing A n and the functions Jv (w) and
wY
F (w), which occur in 18*5, by y (w) and n and the functions wJJ (w) + to the analysis may be left the of details the respectively; HY (w) (w) + v reader, and he will find that when x = 1 the expression \ {f(x + 0) +f(x - 0>]
'
HJ
0).
8-5.
Modern Analysis,
616
[CHAP. XVIII
is
continuous in
may be
of convergence
The summability of Dini series (and of Fourier- Bessel series) by a modifimethod is of some physical importance. Thus, in Fourier's* problem of the Conduction of Heat in an infinite solid cylinder of radius unity,
cation of Abel's
if
is
symmetrical.
satisfying
Normal
the boundary-
condition are
J*
(*-m?\)
is
m=\
where the
coefficients bm are to
is
the Dini series associated with the initial temperature f(r). It is expressible as
00
is
evident
lim
2 6m J (Xm r)exp(-A-\TO
the Dini series
is
);
and
when
summable
(R).
18*6.
series
and Dini
series.
It has
been shewn by YoungJ that the existence and the absolute con-
vergence of
Jo are sufficient to ensure that if all the coefficients a m of the Dini series (or the Fourier- Bessel series) associated withf(t) are zero, then the function (t) must be a null-function.
*
xii.
La
(1906), pp.
Theorie Analytique de la Chaleur (Paris, 1822), 306320. Cf Rayleigh, Phil. Mag. {6) 106107 [Scientific Papers, v. (1912), pp. 338339] ; and Kirchhoff, Berliner
.
524.
H
> 0, and so there is no initial term to be inserted. t In this physical problem, t Proc. London Math. Soc. (2) xvni. (1920), pp. 174175.
18-6]
FOURIER-BESSEL SERIES
this
617
...,
To prove
= 0, 1, 2,
we may write
Um =
*) dt;
and the series on the right converges uniformly in (0, 1 - A) and oscillates boundedly in (1 A, 1). It is therefore permissible to multiply the expansion y **/(*) and integrate term-by-term.
It follows that
JO
f\+n>+*f(t)dt=
m=0
= 0.
Since
all
the integrals
n
t'
+w f(t)dt
Jo
(p 1,2,3,...)
are zero, it follows that t"f(t) is a null-function, by Lerch's theorem*, and the theorem stated is proved for Fourier-Bessel series. The theorem for Dini
can be proved in precisely the same way, and it is theoretically simpler because the Dini series associated with t" +*> does not foil to converge uniformly in (1 A, 1).
series
00
2 am Jv (jm x),
bm
(\ m x),
(where the coefficients am and bm are any constants) which resembles Riemann's theory of trigonometrical series f.
Such a theory
is,
which will be discussed in Chapter xix seems convenient to defer the examination of the series
ao oo
and
it
m-l
2 am J {jm x),
-
m-1
X bm Jv (km x)
by Riemann's methods
to 19 7, when the discussion of the series forms a simple corollary to the discussion of Schlomilch series.
* Lerch,
pp.
3743.
9632.
CHAPTER XIX
SCHLOMILCH SERIES
19*1.
Schlomilch' s expansion of
a function of a
real variable.
In Chapter
real variable
xvm
we
in the form
f(x) =
2 am J {jm x),
that, for large values of
where jm
is
J (z), so jm = (m + \v - i) 7T +
m,
(1/m).
in
That
is
to say, the
is
the series
argument of the Bessel function in a term of high rank approximately proportional to the rank of the term.
In this chapter we shall discuss the series in which the argument of the Bessel function in each term is exactly proportional to the rank of the term. By choosing a suitable variable, such a series may be taken to be
3D
2 OmJyimx).
It will appear subsequently that it is convenient to
(19-11;
cf.
is
simplified
add an by making a
initial
term
slight modi-
fication in the
they present themselves naturally in the investigation of a periodic transverse vibration of a twodimensional membrane, if the vibration is composed of an unlimited number
= 0)
Apart from applications the series present various features of purely mathematical interest; and, in particular, it is remarkable that a null-function can be represented by such a series in which the coefficients are not all zero
(19-41).
In some respects the series are more amenable to analysis than FourierBessel series, but the two types of series have many properties in common; and the reader will be right when he infers from a comparison of the arguments jm x and mx that the relevant range of values of x is (0, ir) for Schlomilch series, corresponding to the range (0, 1) for Fourier-Bessel series.
* Zeitschrift
special cases
r=0 and r = l.
pp.
t Phil Mag.
567571
2225].
19-1, 19-11]
SCHLOMILCH SERIES
a
series
619
19*11.
zero.
Schlomilch's expansion in
We now state and prove the expansion theorem discovered by Schlbmilch. The theorem is concerned with the expansion of an arbitrary function f(x) of the real variable x, and, with modern terminology, it is to the following
effect:
Letf{x) be an arbitrary function, with a derivate (x) which is continuous in the closed interval (0, v) and which has limited total fluctuation in this
interval.
(!)
/(#) = \a
+ 2 am J
(mx)
where
o
= 2/(0) + 2 [ n [**
/
/
[' f 7TJ0 JO
uf(u sin
<f>)
<j>)
d<bdu,
(2)
m = and
this
uf (u sin
{m >
0)
expansion
tt).
is valid,
and
interval (0,
Schlomilch's investigation
is
(3)
/(*) =
(4)
proceed to verify that the function g (x) defined by (4) actually is a solution of (3); we substitute the value given by (4) in the expression on the right of (3), and then we see that
We
-J o
g(xam6)de =
7r j o
/(0) + x sin 6 2x
f (x sin
<f>)
sin
<f>)
d<f>
\d6
-/() +
J
x
ft" ft*
f (x sin 6 sin
sin0 d<f>dd.
Now
replace 6
by a new
variable
sin
defined
by the equation
<f>
x = sin 6 sin
620
[CHAP. XIX
We
deduce that
sin sin
d<f>dd
TT.'o
g(xsm0)dd-f(O)= vr
2x
[*" [ 9 1,
x
/' (# sin
Jo
cos
<f>)
J o
sin
v d% d0
= 2^
n*
/"*
V.U
/ J^
f*
x
,
V0)
cos
= 2x
= #/
(x sin
X ) |- arc
sin
/cos 0\~|**
J J^
X dX
/'(a;sinx)cosxax
Jo
and
so,
when
it is
<jr
(x) is defined
by
(4),
g (x)
is
a solution of (3).
easy to verify from (4) that, when/'() is a continuous function with limited total fluctuation in the interval (0, ir), so also is g{x); and therefore, by Fourier's theorem, g (x) is expansible in the form
00
Now
g(x) = %a
+ 2 am cosmx,
m-l
where am =
2 [*
\
_
and
this series for
f (u sin
<) d<f>
cos
mwdw,
interval (0,
tt).
Hence term-by-term
we have
/<*)-!fV**)<W irJ
o
* "Jo
I
U<*o
I
= |a + 2 am J (mx),
m=l
and
be established.
same
as those
given by
equation
(2).
When the restriction concerning the limited total fluctuation of /' (x) the Fourier series associated with g(x) is no longer necessarily removed, is convergent, though the continuity oif(x) ensures that the Fourier series
19 * 2 ]
is
SCHLOMILCH SERIES
(0, tt);
621
a + X am J (mx)
uniformly summable (Cl) throughout (0, tt), with sum f(x); an application of Hardy's convergence theorem* then shews that the additional condition
is
am
is sufficient
(l/0n)
f(x) when x
to ensure the convergence of the Schlomilch series to the lies in the half-open interval in which < x % ir.
sum
For further theorems concerning the summability of Schlomilch the reader should consult a memoir by Chapman f.
series,
[Note. The integral equation connecting f(x) and (x) is one which was solved in 1823 g by Abel, Journal fur Math. i. (1826), p. 153. It has subsequently been investigated* by Beltrami, 1st. Lombardo Rendiconti, (2) xm. (1880), pp. 327, 402 Volterra, Ann. di Mat.
;
(2)
xxv. (1897),
p.
104
C. E. Smith, Trans.
American Math.
Soc.
vm.
(1907), pp. 92
106.
The equation
2x fhw rj
/
1
rr
/ (x sin 6 sin
<p)
sin
is
in
most simply established by the method of changing axes of polar coordinates, explained 3-33 this method was used by Gwyther, Messenger, xxxm. (1904), pp. 97107, but
;
in view of the arbitrary character of f(x) the analytical proof given in the text seems preferable. In connexion with the changes in the order of the integrations, ef. Modern Analysis,
4-51.
19'2.
The
definition
of Schlomilch
series.
Such series may be generalised by replacing the functions of order zero by functions of arbitrary order v\ and a further generalisation may be effected by taking the general term to contain not only the function (mx) but also a
function which bears to the Bessel function the same kind of relation as the sine does to the cosine. The latter generalisation is, of course, suggested by the theory of Fourier series, and we are thus led to expect the existence of
tt, it)
of the variable.
for insertion are
The
* Cf.
Modern Analysis, 8 5. t Quarterly Journal, xliii. (1911), p. 34. t Some interesting applications of Fourier's integral theorem to the integral equation have been made by Steam, Quarterly Journal, xvn. (1880), pp. 90104.
622
[CHAP. XIX
Bessel functions of the second kind and Struve's functions; and the types of series to be considered may be written in the forms*:
a
a
m Jv (mx) + bm Y, (mx)
I> + 1)
am J* (x) + bm H (ma?)
=!
(\rnxy
Series of the former type (with i/= 0) have been considered by Coatesf; but his proof of the possibility of expanding an arbitrary function f(x) into
such a series seems to be invalid except in the trivial case in which f(x) defined to be periodic (with period 27r) and to tend to zero as x -*- oo
Series of the latter type are of
is
much
They will be
called generalised
Schlomilch
series.
Two
The
first is the problem of expanding an arbitrary and the second is the problem of determining the
construction of analysis (resembling Riemann's analysis of trigonometrical series) with the object of determining whether a generalised Schlomilch
series, in
which the
coefficients are
not
all zero,
Math. Ann. Lil. (1899), pp. 582587 Nyt Tidsskrift, x. B (1899), pp. 7381 Oversigt 1901, pp. 1900, pp. 5560 K. Danske VidensJkabernes Selskabs, 1899, pp. 661665 127146 Ann. di Mat. (3) vi. (1901), pp. 301329.
;
;
Nielsen J has given the forms for the coefficients in the generalised Schlomilch expansion of an arbitrary function and he has investigated with
great detail the actual construction of Schlomilch series which represent null-functions, but his researches are of a distinctly different character from
those which will be given in this chapter.
The
investigation which
we
shall
now
an arbitrary function into a generalised Schlomilch series is based on the in his memoir on applications investigation given by Filon for the case v =
of the calculus of residues to the expansions of arbitrary functions in series of functions of given form. It seems to be of some importance to give such an
investigation
*
||
because there
is
set of
the second
t Quarterly Journal, xxi. (1886), pp. 189190. + See e.g. his Handbuch der Theorie der Cylinderfunktionen (Leipzig, 1904), p. 348.
London Math. Soc. (2) iv. (1906), pp. 396430. has to be assumed that - < v < . The results which will be proved in 1941 1962 of v. suggest that it is only to be expected that difficulties should arise for other values
Proc.
II
It
19*21]
functions J, (mx),
SCHLOMELCH SERIES
H (mx)
( ir,
so as to obtain a set
623
which is a normal orthogonal and consequently there is no method of obtaining the coefficients in a Schlomilch expansion in so simple a manner as that in which
nr);
( 18-1).
The investigation, which forms the latter part of the chapter, concerning the representation of null-functions by generalised Schlomilch series, is of
exactly the
same character as the exposition of Riemann's researches on Modern Analysis, 9*6 9632.
19*21.
The application of
the calculus
of residues
to
the generalised
Schlomilch expansion.
We shall now explain the method* by which it is possible to discover the values of the coefficients in the generalised Schlomilch expansion which represents an arbitrary function f(x), when the order v of the Bessel functions
lies
between
- \ and
\.
When
this has
been done, we
shall
prove directly that the Schlomilch series in which the coefficients have the specified values actually does converge to the sum /"(#).
This is analogous to the procedure which theorem in the expansion
:
is
GO
/(#) h+
2
m-l
(a m
coswe+/3Bl sinfiM?)
the values of the coefficients are discovered by multiplying the expansion by cos mx and by sin mx, and integrating, so that the values of c^ and /3 m are taken to be given by the equations
<%=-
/"
|
/(0
cos mtdt,
Pm =-l
If*
We then take the series in which the coefficients have these values,
2^1' f(t)dt+~
and prove that
it
namely
I
j
sum f(x).
Jy (mx) i H (mx)
(I
mx)'
'
J, (mx)l(%mxy,
H (mx)j(\ mx) v
We
shall write
(1)
*
Jv (z) + il3..(z) _
for the special case
is due to Filon; it was given by him, *=0, but the extension to values of v between
624
so that*
<f> v
[CHAP. XIX
complex
and uniform
variable z;
and evidently
J (mx)
v
i
t'H (mx)
i
'
<f>
+ mx).
We
now observe
that (
<f>
(mx)
is
the residue at z
(xz)
=m
of the function
w<f> v
&\WKZ
where
m = 0,
1, +2,
. .
and so we
(xz)
~dz,
S7TI J
SI
in
origin,
which the contour C is a circle, of radius + \, with its centre at the is an integer whioh will be made to tend to infinity. and
is assumed to be one- valued throughout the 2-plane, be analytic at infinity (cf. 19'24); its only singularity in the finite part of fche plane is an essential singularity at the origin.
and
By
tends to infinity,
provided that v
> |.
by calculating
residues, that
It is evident,
2 (-)m \F(m)
is
<f> v
(mx)
+ F(-m)
<f>
(- mx)}
'
sin trz
that
is
to say
(2)
2 (-) {F (m)
<j> v
(mx)
+ F(-m)
<f> v
(- mx)}
=The problem
2m
1
y
r
J
(0+)
ET/ F(z)
of expanding an arbitrary function /(a?) into a generalised Schlomilch series is consequently reduced to the determination of the form of
-s 2m
differ
F(z)
J
?"
sin
'
dz
vrrz
* The insertion of the factor (J z) v in the denominator makes # (z) amenable to Cauchy's theorem when the contour of integration completely surrounds the origin.
19*22]
19*22.
SCHLOMILCH SERIES
The construction of
the function F(z).
625
We now
2m J
and, in order to calculate
'
smirz
we
shall
is
(1)
F(z) =
i
w=l
?4#,
*
where yfrn (z) denotes the sum of those terms in the expansion of ir- 1 sinirz whose degree does not exceed n, and the coefficients n will be defined later.
The reader
M*)=-M*)=*-*iry,
With
^"sin^^/^sin^
- y
=i
It follows immediately that
(2)
i
*-<">
j
X n+1 (s
jt\
-'fV w *) &
=6 (0) i VfV
=i
*> 7rWcos * W7r
(/i
+ l)!
_ l
pqtg)
i-
+ l)
with /(a?) -/(0). For this purpose we have to assume temporarily that f(x) has differential coefficients of all orders at the origin, and then we define the
coefficients
pn by the equation
(3\ K)
= 1,2,3,...).
way that the sum
We next transform
of the series,
by which F(z)
form
* This type of series is fundamental in Filon's theory, and is not peculiar to Schldmilch expansions; thus, in his work on Fourier-Bessel series, sin *z is replaced by z~"J {z) and v fn (z) denotes the sum of the terms whose degree does not exceed n in the expansion of that function.
626
the coefficients
that they
[CHAP. XIX
may be convergent, we
\.
suppose that
<
<
We
_
then have
rgyrgn+v + i)
r
)
= - *ru-l)
r( *>
fm (0)
(1
-**+-*-**
/!
^r(i-,)Jo
^
_
it ['"
*/>] *
where
for dfdu.
Now,
we arrange the
series
in descending powers of
z, it is
^
w=1
- D)
'
and therefore
we need
to
sum
the series
y pn
and we are able to
effect
7T
W cos ntr
(4),
whence
we
find that
(6)
ii'"(n+T)r
have now obtained symbolic expressions for all the coefficients in the generalised Schlbmilch expansion of /(#), but it is necessary to transform these expressions into more useful forms, by finding the significance to be
attached to the symbolic operator
for the
We
-jjyzT\ ^ fcn
'
for
g6116'8,1 values
of *
and
value zero of z.
19*23]
SCHLOMILCH SERIES
627
19*23. The transformation of the symbolic operators in the generalised Schlomilch expansion.
We
/(<B) i \_tt(D-iz) j
r^e
^
The usual
interpretation of
^o
- f(tu) is
i:
e-^fi^dv,
;
where a
is
a constant of integration
sinh irD
.._,
.
and therefore
.
sinh-wi) T
u
.
,.
"1
.
7T
x(u ) = x( u 7r )>
;
where x (M )
is
an arbitrary function of u
and hence
fJa
it
follows that
_*Z
= r sinh^(D-f^)
[
7T
yV
J-0
'J w==0
- e-
that
is
I* 'e-^fitv) dv Ja J-0
to sayf
all
= 0,
1, 2,
....
F(z)
*
ir<f>,
(xz)
sin irz
numerous expressions involving symbolic operators of the types nnder l. (1839), pp. 22 32 and by Boole, Differential Equations (London, 1872), chapters xvi and xvn. t The expression for F(z) which is derived from this formula does not appear to have a singularity at the origin unless a is infinite or is a function of z ; but it seems unreasonable to be perturbed by this when we consider the nature of some of the analysis which has already been
interpretations of
The
628
at 0,
[CHAP. XIX
(1),
2, ... is
concerned,
we may omit
the second
and
where
<
2
>
ww^km
1922
(6)
F.
(1
/(tv) dv
dt.
Again, from
(3)
and equation
f
we have
j.gcoBinir
ti
(n
= r(i)/(0)
+ 1)!
r(i-).l
2r(*
(1 vx
py^dr "'" A
eft
^W^^W^*
<f> v
dt
The
v
first
= 0; when
term on the right in (3) is equal to T (v+ l)/(0), except when* = 0, the value of the term in question is zero.
We
(4)
f{x)
<f) v
(0)
(mse)
+F(-m)
in (4).
<f> v
(- mx)\
= 0,
When we
for Bessel
/W "^(y +
am
)
l)
"-
.t 1
(\rnxY
where f
-T{\-v)T{\)),
(6)
O-^-a
(1
IT
f (tv) cos mv dv
dt,
bm ~
This
19*24.
is
r(|-,)r(D J
"
<2)
*"*
dt.
./
ir
now prove that, when the function f{x) is restricted in a suitable manner, the function F(z) is bounded when [z|--oo, whatever be the value of arg. The reader will remember that the assumption that F(z) is bounded was made in 19"21 to secure
shall
We
We take
(1),
* When v is negative it is necessary to use a modified expression for the integrals ; cf. 19*3. t When p = 0, the expression for a flas * he modified by tbe insrtion of the term 2/(0), in consequence of the discontinuity in value of the expression on the right of (3).
19-24, 19-3]
SCHLOMILCH SERIES
first
629
and divide it into two parts, namely the where is the integer such that
N terms
1.
ir
N^
When n ^ AT
,
it
< N+
the terms of
n
\
(z)
do not exceed
n ~ l \z n jn
\
!,
and
therefore,
whwn n
< JV,
+n
When n ^ JV, we
1
(z)
n*-i\sW(n l)
z\ n+1
|*|.(-1)!
,
have
yfr
(?)
<n
sinh
1
n-
and therefore
l
IFMi^ W J_
'
'
Since
\z\" +l
\z
\
tends to zero as
as z
I j
00
it is
-*- 00 is
2 \Pn\
should be convergent
;
and
2 " + i|/()(0)|
is
convergent.
19*3.
series.
into
a generalised Schlomilch
Now that the forms of the coefficients in the generalised Schlomilch expansion have been ascertained by Filon's method, it is an easy matter to specify sufficient conditions for the validity of the expansion and then to
establish
it.
shall
prove*
is
as follows
\ < v < \;
and letf(x)
be defined arbitrarily
= 2vf(x) + xf'(x),
ir, ir).
exists
and
is
(II)
( ir,
it).
(III)
Ifv
i*i\*r if (*)-/(<>)}]** r J O
is absolutely convergent
*
when
is
a (small) number
by Nielsen, Handbuch der Theorie der Cylinderfunktionen (Leipzig, which he gives for the coefficients in the expansion seem to be quite inconsistent with those given by equation (2). t The effect of conditions (I) and (II) is merely to ensure the uniformity of the convergence of a certain Fourier series connected with h (x). X If v is positive, this Lipschitz condition is satisfied by reason of (II).
is stated
The expansion
630
[CHAP. XIX
+^
g^;
where
K=L
wAen
r<F^r(J) 2?
of a
is
[sin2v
f( U8in V-f(
M sin ^*>
m > 0;
the value
2I> + l)/(0)
on the right in the first equation of the system
(2).
We
it will
is
(4)
sr(*)=I> + l)/(0)
+ r^7>
by condition
equation (4)
(III).
f
r
sec2 " +1
'
secured
may
be omitted when v
/:o
cUf>
r( ^
which
is
valid only
when
v is positive.]
We
solution of (3),
proceed to verify that the function g (x) denned by (4) actually is a by taking g (x) to be defined by (4), substituting in the expres-
result to /(#).
The
IT
result of substitution is
2cosy7r
Jo Jo
[* [
in
C082*
sec^+i
*
<j>
[gin*-
<*9
{/(* sin
sin
<f>)
-/(0)}]
d<f>d0
+/(0).
^^
IT
Hence we have
**
f
to prove that
2"
JO
Jo
f ^cos
$ sec 2" +1
<f>
^r
d<P
[sin"
<f>
{f(x sin
sin
<f>)
-/(0)}] d<f>d0
=/(*)-/(0).
Replace
<f>
on the
left
by a new
sin
variable
i
x sni ^ s
n 9>
change the order of the integrations in the resulting absolutely convergent by a new variable ty defined by the equation integral, and then replace
cos
cos x sin ^.
^
19*3]
SCHLOMILCH SERIES
thus deduce that
631
We
/**/
X) "WW*** * {/( "Jo Jo ( sinfl-sin*xr> djc l n* rh* sin cos 2 " d r = [Sm" {/(a; Sm X) " /(0)1] 2 * (cos io i x X - cos 0)"+* d
sin
cos2 "
,,
^
is
'.Jit
rjir
[sin-
= ir( v +|)r(i-v){/(^)-/(0)},
and hence the formula to be established is evident; and defined by (4), then equation (3) is satisfied.
so,
when g(x)
Now, by
Fourier's theorem,
oo
g{x) = \a
+ 2
JB-l
where
\am
(5)
J
= I
7r
'-'
g
<7
(u) cos
mudu,
&m
(u) sin
mudu;
and
it is
when f(x)
( ir,
is 7r)
g (a:), and
therefore the
is
- ir + 8 ^ x ^ it - 8, where
cos2 *
;
an
number.
Replace
by
a;
sin
in the expansion of
g (x), multiply by
0,
which
we deduce
-T( v + 1) +.*
m and
b m given
^ a
_i_
W^f
and
The
values of
a.
by formula
(2).
by the Riemann-Lebesgue lemma, a m and bm 0(l/m) when m is large. This seems to be connected with the fact that when we come to deal with any Schlomilch series ( 1962) we are unable to make any progress without assuming that 26TO/m is convergent (or some equivalent hypothesis) this assumption will appear in 1962 to be
are both
;
necessary because the differential equation which Struve's function satisfies is not homogeneous, so that Struve's function is not of a type which occurs in
solutions of Laplace's equation or the
wave equation
632
[CHAP. XIX
quently seem to be reasons of a physical character for the limitations which have been placed on f(x) in order to ensure the existence of the Schlomilch
expansion.
[Note.
Just. as in 19*11,
is
if
all statements made in this section up to this point about convergence of series have to be replaced by statements about sumtnability (Cl).]
vf{x)-\-xf'(x)
not
satisfied,
then
There is one important consequence which follows from the fact that a m (1/m) when 2vf{x) + xf (x) has limited total fluctuation and bm are both in ( it, 7r), namely, that in the neighbourhoods of it and it, the general (l(m v+ %), and so the expansion repreterm of the Schlomilch expansion is
sents a continuous function
;
There are a few problems of Mathematical Physics (other than the problem mentioned in 19*1) in which Schlomilch series occur in a natural manner, and we shall now give an account of various researches in which Schlomilch
series are to
be found.
is
1+1
this series is convergent
e~ mz J (mp)
when p and
it is
cylindrical-polar coordinates,
z are positive, and, if p and z denote a solution of Laplace's equation at all points
= 0.
Various transformations of the series have been given by Whittaker*; thus, by changing to Cartesian coordinates (x, y, z) and using 2*21, we have
00
(1) v/
1+ %
,
e- mz
J
2
n= i
du
exp
>
{ (z
-.
done,
CO
we getf
1
fir
1+2
7W=l
e~ mz Jo (mp) r/
= 52-7T
_..
fill
.
ix cos u
(z v
+ ~r~. 2 + %y sin u 5
ix cos
+7T 2 -7s
27T OT=!
(2m)l
-r-^
u + iy sin
J_,r
VI
m) 2**" 1
du
where
(r,
coordinates
and
*
I
1}
B.2
...
Math. Ann.
Cf. 4-8
19'4]
SCHLOMILCH SERIES
633
Another transformation of the series, also given by Whittaker, is obtained from the expansion for 1/(1 e~ l ) in partial fractions; this expansion is
\-e-
^L_ = I + l4. 2
1
i
w=1
1
\t - 2miri
f
I
l
t
+ 2miri)
'
1+ I =
1M
e-Jo(mp) = i + s & T
?! \jj{(%nviri
+ 2
'
due to a set of unit charges (some positive and some negative) at the origin and at a
It follows that the series represents the electrostatic potential
set of imaginary points.
The reader may find it interesting to discuss the Lipschitz-Hankel integral of 13'2 as a limiting form of a series of Whittaker's type.
Some other series have been examined by Nagaoka* in connexion with a problem of Diffraction. One such series is derived from the Fourier series for the function which is equal to 1/V(1 a?) in the interval ( 1, 1).
The Fourier
(4)
series in question is
1
-r-z
^ = " +
*"
(- 1
+ A,
- A), where A
(also
due
Nagaoka)
I e%
<3 >
goaii
fa
Jj
w=*>
~b
igoxi
2 ?.
r J
'
(mv)
^^
J
interval
The series on the right in (5) converges uniformly throughout the (-1,1) and so we may take - 1 and 1 as limits of integration.
Hence,
/r\ (6)
for all values (real
r J
t (a) \
and complex) of a,
[1
_l.9 8 3 + **
- -j-
sina Ti
(-)"J,(m7r) l
a?
_m
A
(7) ' v
more general
v
result, valid
when
R (v + ) > 0, is
i
^/'^
* Journal of the Coll. of ScL, Imp. Univ. of Japan, iv. (1891), pp. 301322. formulae are quoted by Cinelli, Nuovo Cimento, (4) i. (1895), p. 152.
W.
B. P.
21
634
This expansion
fractions*.
[CHAP. XIX
as a sura of partial
obtainable
by expressing
Various representations of the integral on the left of (5) were obtained by Nagaoka the formula quoted seems to be the most interesting of them.
Finally
we
formula f
This
is
" cos(2wi-l)a?
?,
(2m-iy
ol -g(*-MX
ir.
,,
<<<)
.
by replacing
the variable
a;
by
a;
sin
from
to
\tr.
As an example of the
lies
sum
of the
sum of a Schlbmilch series when we shall take ir < x < 2tt, and Fourier series, we see that
7r, ir),
fare
sin (/)
^ T
O
fi*
")
\f(xsmd)d0
J arc sin {nix))
(JO
/"arc sin (nix)
'
^(Tr-2x8m6)d0
o
TTJ o
+
so that,
-/"""
(2# sin " 'w arc sin (ir/x)
- 3tt) d0,
when
-n-
<
a?
<
2-7T,
we have
!
S J".[(2m-1)}
(9)
To
rts
if
8"
19*41.
series.
We
(1)
shall
now prove
provided that
when x =
and diverges
to
oo
when x = ir.
series, and, in fact,
definitely
ir).
Modern Analysis,
7*4.
19*41]
It is easy to prove (1)
SCHLOMILCH SERIES
by using Parseval's integral; when
2
ft" il
635
M
)
is
a large
integer,
we have
1
M
+ 2
(-)m J (mx)
= -\
k+
M
2
(-) cos
a;
Jo
cos (# sin
_ (~) M [ x cos (^ + h) u
ir
cos
\u
'
du ^(x2
v?)
= 0(1),
as
itf -*- oo
,
is
applicable because
the integral
du
f Jo cos \u
'
\Z(ar*
a it )
exists
and
is
absolutely convergent
when
< x<
ir.
that
lim
\\+ 2
(-) J, (*)]
stated.
=0
when
this is the
theorem
It is easy to prove in
(i)
when <
$ J and
(ii)
when
>
and
< x ^ tt.
By
\
v> \)
r<v+i)
(-)*Jp(imB)
"",.-!
{\mxy
(rw * 8in
cos" tdt
r(* + *)r<*)Wo
cos**
<*
'
= o(l),
as Af -* co
,
*(a?-u*y
.'o
cosu
dw
is
exists
and
is
the case
when x and
v satisfy
The truth
If
of (2)
is
now
evident.
if v is
is
so large that v
In > \
the operator
d [ d xdx\ dx
* Cf.
Modern Analysis,
941.
636
[CHAP* XIX
may
The
effect of
Jv {mx)j{\mx)
2
by
s
;
and
when
{)
that
is
to say,
l<\\
(3)
y (-)"'/>
J5i
(i^r^x<
ir,
H
or
_ -
'
\ < v In ^ \ and
in this
;
(ii)
- In > \ and
^ x ^ 7r.
lii.
section are
pp. 582 587 two other papers by Nielsen on this subject were published at about the same time, Nyt Tidsskrift, x. B (1899), pp. 73 81 ; Ooersigt K. Danske Yidenskabernes
Selskabs, 1899, pp. 661
(1899),
665.
In the
first
made
sum of the
series in (2)
when
x>
ir
M
thus found that, when x
f* cos
(M+$)u
cos^it
^_ ur -j d
3r,
5w,
positive
and q
is
(2q-l)ir<x<(2q + l)ir,
then
(~)m Jv (mx)
2r(J)
(2tt-l) 2
7ry-*
(%mxy
xT{*
evident that,
The importance of Nielsen's formulae lies in the fact that they make it when a function f{x) is defined for the interval ( ir, ir), if the
number of points) the representation is not unique and there are an unlimited number of Schlomilch series which are equal to the function f(x) throughout the interval, except at a finite number of points, namely the points already specified together with the origin and (when
(except possibly at a finite
< v K. i)
coefficients
the end-points
ir.
represent
points
of the
interval
ir<x<ir,
(whenl
\ <
I> + 1)
*
|
^-x
(-) m J,(mx)
($mxy
'
Formula
(1)
60;
(1904), p. 101.
by
pp.
;
301329
for
more complicated
J The theorem is untrue when y>f cf. formula (3). any Schlomilch series other than the one given can represent a
when J<.r<|.
19*5]
is,
SCHLOMILCH SERIES
much deeper
character,
637
of course, of a
to this
and
it
We
;
shall
now
up
theorem
main
is
applicable to trigonometrical
19*5.
Theorems concerning
the convergence
of Schlomilch
series.
We shall now discuss the special type of Schlomilch series in which v = 0, and in which Struve's functions do not appear the object of taking this particular case is to avoid the loss of clearness due to the greater complication in the appearance of the formulae in the more general case. With a few
;
exceptions, the complications in the general case are complications in detail only; those which are not matters. of detail will be dealt with fully in
19-6 1962.
to
be considered
ao
is
a + 2 am J
m=\
(ma;),
We
of X,
shall first
condition that
am J
is sufficient to
ensure that
= o ( *Jm).
[Note.
.
If the origin is
n
m=o(l)
interval
is
obviously true.]
be called It
origin,
and
Throughout Ix we have
am J (mx)
and,
(cf.
7'3)
am I
asm-*-,
P (ma?, 0)^1,
Hence,
Q(mco,
0)^0.
all
m, (say
values exceeding
ra,,)
P(mx,0)>$,
at all points of/,.
\Q(mx,0)\^$
Now suppose
that am
is
not o (^/m);
we have
to
shew that
this hypothesis
leads to a contradiction.
* Cf. Modern Analysis, 9'69*632. t Sinee J (mx) is an even function of
x, the portion
may
638
If
[CHAP. XIX
number
am >
|
must
slm
whenever m is given any value belonging to a certain unending sequence * Let the smallest member of this sequence which exceeds both wh, raa m* m and 2ir/Li be called ?/.
,
Then
cos
all its
x ,
cos (w/a?
.
far) >
|
V3,
sin (ra/a?
\ir) < |
j
at all points of
J2
If
L% is
Z2 = \irjm-l.
Next
and
let
the smallest
member
mr which
,
exceeds both
m/
2tt/Z 2
be called mj.
Then
cos
all its
be a portion of 7a say
j
I3 such that
,
cos (rn^'x
.
\ir) ^ \/3,
|
sin
(tfij'a?
^7r) <
|
at all points of
7S
is
If
i3
is
= $w/m2
is
'.
By
which
1,,...
predecessor ; there
therefore a point
we have
j
cos
sin
I
when
(mX)\>\am \^/[^~ I)
x [P(mX,0)
.
|
cos
sin
(toX -
J*-)
>6 V
and
all
3-1 //2\
4
vUi'
that
this is inconsistent
points of
The
o {\Jm).
Om must be
and
sufficient condition
* It is supposed that mx<m <m3 <.... 2 t There are, in fact, at least two such portions of I\ ; in order that I8 mined, we take I2 to be that portion which lies on the left of the others.
may
be uniquely deter-
19*51]
SCHLOMILCH SERIES
639
^ a **/(
] .
cob (wmj
- iw) +
sin
(mx -
\ir)\
This theorem
(am m~S) 19*51.
is
J (mx)
by a function of
series.
which
is
= o (ra~ 2)
and So (m -2)
is
a convergent
%a + 2 am J
m-l
at any point at which the series
is
(mx),
Let
(1)
F(*)- o,*- 2
a J< mc>
.
Then
series
P (x)
will
whose sum
is f(x).
f(x) converges at
all points
of any
P (x)
of
x.
For am J
(195)
(ntx)
and therefore
a = o ( Vw).
Again, by 2*5
(5), for all real
values of x
lJ.(*)ki;
and consequently am J (mx)
Since
^ o / 1\
*)
is
convergent,
it is
convergent.
It is evident, moreover, not only that the convergence is absolute,
that
it is
but also uniform throughout any domain of values of the real variable x.
640
19*52.
[CHAP. XIX
We
,i\ (1)
shall
is the
if
= (x + a)T(x-r2a) + (x-a)T(a;-2a)-2xT(a;)
then
(2)
at
am =
It is easy to
o (Jm).
G- (x, a)
= \a x 2
,. l
4o2
[(x
(mx)]
follows that
lim
a +o4>m cP
(mx 2ma)
+ 2m(x + a) J (mx + 2ma) 2m (x a) Jo' (mx 2ma)] = x Jo" (mx) + J^ (mx)fm = xJ (mx).
Consequently the limits of the individual terms of the series defining Cr(#,a> are the individual terms of the series defining xf(x).
It is therefore sufficient to prove that the series for Ot(x, a) converges when x uniformly with respect to a in an interval including the point o has any value such that the series forf(x) is convergent.
It shall
x
;
is positive f,
and we
we
shall
a.
now prove
when - \x ^
\x.
By
observing that
-V{(2)}
and that the
series
% tg + V
< |a /^
3
fc
(g2g)}
S a m i w2
Jo
[x
(mx 2ma)
+ *J{x(x
2a)}]
* Since we are not assuming more than the convergence of / (x) at a single point, it is not permissible to infer from 19-5 that am must be o (Jm). (0, o) =0, the special t The functions under consideration are even functions of x ; and since
case in which
x=0
19*523
is
SCHLOMILOH SERIES
641
we
*".*,
W?
.
(mx)]
series,
each of which
is
uniformly convergent.
Now
[
series,
namely
.
.
^^ V(# +
it
{mx + 2wa) +
ftx
(mx)]
and write
in the form
,
.
.
//2x\[
sin
(mx ?r)"l
/sin
ma\ 2
If
2x\
'
cos
V
V
\wr/
8m (a? - 4a2)
ma
a
_ aw + am
where 4> (y)
is
|7r) cos
'
2wa
m*
/(2x\
V UJ-
4m^
'
sum
uniformly convergent.
The first two series are proved to be uniformly convergent, in connexion with the theory of trigonometrical series*; and the third is obviously uniformly convergent from the test of Weierstrass.
To
we observe
that,
by the
first
mean-value
theorem, numbers f
and
-1<0<1, -1<0,<1,
* Cf. Modern Analysis, 9'62, 9-621. It has been the general (bat not invariable) custom to obtain various properties of the series without establishing the uniformity of their convergence. The convergence of the series for f(x) is required to deal with the first serits the second series can be dealt with in consequence of the less stringent hypothesis that a m =o\ >/m).
;
The number
is
a function of a variable
which
642
for
[CHAP. XIX
<l>
(mx + 2ma)
(nix
2ma)
2mat)} dt
= 2ma
{<'
Jo
= 2ma
= 4m
Since 4>"(^)
2 2
4>mat
(mx
1
2moi6t) dt
Jo
a2 4>"
3/
( - 2ma6
is
).
= 0(l/y ) when
m=l
is
a.
Hence
G (x, o)
is
expressed as the
sum of six series each of which when \x < a < \x\ and therefore
lim
a--0
converges
G (x, a)
G (x, a),
;
is
is
equal to the
sum
i.e. it
for/O)
is
convergent
and this
is
Lemma
(\/m)
II.
We
that a m
of%
is sufficient to
ensure that
(x
a
for
all values
of
x.
As
in 1952,
we need
and we express
of six series each of which is easily seen to x < a < \x, by applying the theorems con 19*52.
Hence
lim [a
G (x, a)]
= lim(W)- \im-^~[(x + a)J (mx + 2ma)
+ (x - a) Jo (mx - 2ma) - 2xJ
{mx)\
a0
= 0,
and
this is the
lemma
to
be proved.
series.
19-54.
We
consideration
with v
=
**
0,
Cf.
two Schlomilch series of the type now under and with Struve's function absent) converge
Modrrv Analysis, 963.
19-53, 19*54]
SCHLOMILCH SERIES
643
and have the same sum-function throughout the interval (0, ir), then corresponding coefficients in the two series are equal. The formal statement of the theorem is as follows
all points
series, of the special type, which converge and are equal at of the closed interval (0, tr), with the possible exception of a finite number of points, must have corresponding coefficients equal, unless the end-
Two Schlomilch
points
and
7T
If these
may differ by a
constant
+2
(-T
(nix).
$a + 2 am J
m-l
(mx),
all
and
let
the
sum
values of x between
of this series be f(x), so that f(x) converges to zero for and nr, except the exceptional values.
(0, ir),
Let x
&
such
We
if
V(x)
is
interval ( 2 ).
Schlomilch series for fix), then P (x) is a linear function of log x in the This is the analogue of Schwarz' lemmaf.
If
= 1, orif0 = -l,andif
*()-^[F()-F(ft)-^^{F(6)-F(ft)j]
then
<f>(x) is
*(fc)-*(&)-o.
If the first term of
will
first
be some term of
(x) is not zero J throughout the interval (&, 2), there <f> point c at which it is not zero. Choose the sign of so that the
<f>
(c) is positive at
c,
so small that
<f>
(c) is
still positive.
Since
<f>(x) is
continuous in (&,),
positive.
positive since
<f>(c) is
Let
it
ft<c,<(r,.
Now by Lemma
j.
I (
<f>
1952)
(c,
+)
(c,
+ 2a)+(d - a)
4o
2
<ft
(ct
- 2a) - 2c,
(cQ
=^
^o
*
The
points 2
|2 themselves
may
be exceptional points.
P {x)
is
644
[CHAP. XIX
left
But
(Cj 4-
2a)
<f>
(cj),
<f>
(Cj
2a) ^
tf>
on the
must be
negative or zero.
term of (cc) must be zero throughout (f1} f2), that is to say that F(#) must be a linear function of log x and this is the theorem to be proved.
This contradiction shews that the
first
;
is
F (a?)
consists of a set of
segments
y= A
Now, by
log x
+ B.
and so these logarithmic curves y = 'F(x) cannot have
1951, F(#)
is
continuous in
(0, ir),
Lemma
II,
a-*-o
{_
la
is
2a
;
even when
an exceptional point
that
is
to say
fF'(+0) = F'(f-0).
Hence the constants
points,
A and B cannot be discontinuous at the exceptional and so they have the same values for all values of x in the
-n-).
interval (0,
Consequently,
when
<x<
2
ir,
H*
Make
x-*
;
- 2 -*-^
r=l
"'*
'-
= A\ogx + B.
Therefore
A \ogx + B
has a limit
when
when
^x %
m=\
ir,
if*
and the
series
on the
left
(0, ir), so
integrations
Replace x by x sin
0,
multiply by sin
0,
tt
nd integrate from
to \tt.
Then,
by 12
11,
&a x*-B=
2.
s am sin ma?
Hence, when
^ x ^ tt,
"
am
smmx =
-
D f^a x3 Bx.
19'6]
SCHLOMILCH SERIES
645
it is
Multiply by sin
mx and
7ra8 2m i*
integrate from
to
ir
[2m3
m
a m = (-)w a
.
'
Since dm
is
B = J9 n*a
Hence we must have
f(x) = a
w=l
From the
and at #= ir, it is evident that f(x) cannot be a series on the right at a? = convergent Schlomilch series at either point unless a is zero; and this proves
the theorem stated at the beginning of this section.
19*6.
Theorems concerning
the convergence
We shall now
{)
" am Jv (mx)
+ w H, (mx)
fr
i> + l).ti
when v<\,
-*- oo
QmwY
the condition that the
We
(m +
\)th
term of
at all points
of any
interval of values
of x
ensure that
am =
[Note. If the origin
is
o(ra" + *),
6w
= o(ir+ *).
is
m=0(l)
obviously true.]
is
unaffected
by a change
in the sign
is
the signs of
all
by considering an
We call
this interval /,
we
,
have,
by
,
1041
v
(4),
am Jv (mx) + bm
(mx)
*
.\ 5 ($mxy
= 71
rD/ [P (w*,
v) cos
Q (mx, v) sin (mx \vir \ir vm )] + b, n (n _1 where am = cm cos m b m = cm sin M We now suppose that a m and bm are not both o(w + *); we have to shew
.
),
r\
i/
number
must
exist such
>em v+i
whenever
is
wio,
3 ,....
646
[CHAP. XIX
We now
inequalities
some point
of
Ilf the
P(mX,v)>l \Q(mX,v)\<l, sin (mX \vir \tr m ) ^ \ |i/7r -\ir n m ) > \ */3, cos (mX
|
r)
w has
(mr
).
For values of
a m Jv
(ml) + bm H (mX)
qmxy
and, since v
small.
>m&-***>
is sufficient
\
is
is
This
am and
bm
must both be
o(ra"+*) if the
(m + l)th term
(as in
may
x {whether a point of
-l
Bin
Smx
(mx -
W
the
l-rr
- Vm )\ +
r(l + i)r(i) ]
,
19*61.
Let us take
< v<
a
and
let
sum
of the series
r(i/+l)
at
=i
| am Jv (mx) + bm ($mxy
(mx)
series is
m {)
Then
v_ *"
/
W ~8I>-|-2)
floff
5 a m J, (mx) + bm H (mx)
mti
m\(tmxy
F(a?) will be called the function associated with the Schlomilch series
The only
theorem that
(x^x"
is
a bounded function of
19-61, 19-62]
SCHLOMILCH SERIES
647
Again,
(2)
let
(x,a)
Then, just as in
Hm.(^)-./.(.)at
rfr+ )r(t)
j^
i
any point x
at
fy (x)
ao
is convergent,
provided that
(m -"'*) and
prove that
bm/m is convergent.
may
0,
ao
(to" + *),
X
m=l
bmfm is
We
order v {where
can now prove that, if two generalised Schl&milch series of the same \ < v < ^) converge and have the same sum-function at all
(it is
ir, ir) with the possible exception of a finite supposed that the origin and the points tt are not all exceptional points), and if the coefficients of the terms containing Struve's functions in the two series are sufficiently nearly equal f each to each, then all
number of points
series
be
_Joo__
I> + 1)
and
let
|
ti
am Jv (mx) + bm H.,(mx)
($mxy
(x),
the
sum
of this series be
( tt,
with a
finite
number
of exceptions.
The convergence
( 7r,
7r)
am = o (m"+*),
bm
= o (m"+i).
terms containing Struve's funcis
coefficients of the
two
mean
that 6m
to be interpreted to
L.
is
convergent.
We
*
now
for / (x).
It can
be proved J that
if
the interval
is
(,,
f2 )
is
The presence
of the series
on the right
equation satisfied by Struve's function. f This statement will be made definite immediately.
t It seems unnecessary to repeat the arguments already used in 19*54.
648
[CHAP. XIX
and the exceptional points (if any) are not internal points of the interval, then F,(#) is a linear function* of a? - 2" in the interval. It may then be shewn
that the exceptional points do Dot cause any discontinuity in the form of
r (x),
iT v (x) = Ax-*" + B,
(0<x%ir)
, >
(0>x>-tt)
A\
Now
p
replace
" K
a^ m = 8I> +
>
2)
0/cos2v
0,
to \ir.
0,
The
series for
Fr (#sin0)
so term-by-term integra-
F (X Sin 0)
Jo
00
cos "0
2
d0
JQ
- 2
fj *
"a
0) sin" +1
'
m=l JO
m
Y(\
.(m#)"
*
cos2"0
(Iff
_ g^T(\ v)
12r()
v)
g m sin mx +b m (l
cos mx)
T(i)
Zx
we deduce
a*a?
m*x
that
1
When we
/ 1X (1)
^
we
when 0^#^7r
and a similar
() -^r(, = 12-- r + i), ( f-,) equation may be obtained when O^x^ tt.
,
4a?
- 2"
(m v+*),
it is
term-by-term when
>*/>-; but
it
may
If
differentiate, twice or
on the
fails to
left
once as the case may be, the resulting series tends to a limit as x -* 0, but the resulting expression on the right
do so unless
infer that
is zero.
We
of
It
<e>\
A = 0,
and
in like
F (x) at the
now
(2)
Jr
% am Sm mX + km (1 - COS mx)
_ -^-BxTiv+l)
(^X3
.
when 7r $ x ^ ir.
*
When
v is zero
x~iv has
to be replaced
by logx.
19*7]
SCHIidMELCH SERIES
649
to it
;
ir
and then
6TO
= 0.
(-) m dm
This equation
is
}.
am
is
o(mv+l)
unless
and then
am
series for / (#)
= (-)m a
Hence the
must reduce
to
lr(r+i)
t1 m
ir,
a^y
is
J;
Now
not an exceptional point and cannot converge at that point unless a* is zero, so that am
- 7r
have therefore proved that, if the series 2 bm/m is convergent all the am and bm must vanish that is to say, the two Schlomilch series with which we started must have corresponding coefficients equal. And this is the theorem to be proved.
coefficients
;
We
We have therefore established for Schlomilch series in which \ < v < theorems analogous to the usual theorems concerning the representation of
null-functions
by trigonometrical
series.
19*7.
series
We
shall
now
namely
t=i
2 am Jv {jm a;),
2
t=i
bm
Jv (\m
a;),
(in
which v
after the
manner
of Riemann's investi-
The method
series just
These points of detail are very numerous and there is no special difficulty in discussing any but it is a tedious and lengthy process to set them out in full, and they do not seem to be of sufficient importance to justify the use of the space which they would require. The reader who desires to appreciate the details necessary in such investigations may consult the papers by C. N. Moore, Trans. American Math. Soc. x. (1909), pp. 391435; xn. (1911), pp. 181206; xxi. (1920), pp. 107156.
*
of
them
650
In the
if
[CHAP. XIX
195
that
the series
00
OmJ.ijmti),
2 bmJyQ^x)
associated function
OO
we
write
f(x)
= 2 am Jv (jm x\
m=l
is
F(x) =
It
mljm
2 5s/ (**>.
"
may be proved
is
that,
defining /(a?)
convergent, and
the expression
arranged as a series in which the rath term has am for a factor, then the with respect to a in an interval containing the point o = 0, and that its limit when a -* is ^""/(a?).
It
may
also
be proved
that,
whether the
series for
f(x) converges or
not,
is sufficient
to ensure that
The
number
ain2/
8
of
- rt
2
m=l! Jm+1"'
is
fm<**
a bounded function of a
proofs of the
lines of the
It
now
manner
19'54) that,
when/()
is
a null-
1),
then F(x)
satisfies
the differential
equation
and sof
where
^=
is
962.
>
B are
constants.
* Cf.
This equation
valid
when
< x ^ 1.
Modern Analysit,
19*7]
SCHLOMILCH SERIES
since
651
Now
v>-,
J,(jm^)/(jf>^y
so,
is
bounded when
O^a^l
whatever
be the value of
and
when*
<
, the series
converges uniformly
when
^x<
1,
by the
test of Weierstrass.
Hence F{x)
is
is
zero
when
v is
a continuous function of x in the closed interval and so positive and is zero in the case v = 0.
;
For any assigned value of n multiplying the series for F(x) by x+1 J,(j x) n does not destroy the uniformity of its convergence; and, when we integrate,
we
find that
J? 0) = j
=jn
2
J
{Ax*
J
+ Bar 9 ) x Jv (jn x) dx
~ (A + B) J: n)
*
B
[
'^rW
Now, when n
is large,
j''->i-V(),
for
an
is
= o (00 when
>
unless both
and then an
is
zero.
Hence a series of Bessel functions (in which v > |) cannot converge to sum zero at all points of the interval (0, 1), with the possible exception of a finite number of points (the origin not being an exceptional point} when
the
v
>
),
unless
all
We infer that two series of Bessel functions, in which v > \, cannot converge and be equal at all points of the interval (0, 1), with the possible exception of a finite number of points, unless corresponding coefficients in the
two
series are equal.
Dini's series
00
f(x)= 2
may be
dealt with in the
6m 7F (\JB a;)
associated function
is
defined
by the equation
^J^zl,
* When *>}, the convergence of the series for /(a)/*" at.r=0 is sufficient to ensure the uniformity of the convergence. t An exception might occur when v - 1= - ; but this is the trigonometrical case.
% The
series divided
origin.
is
H+v>0,
inserted.
652
[CHAP. XIX
(0, 1),
and
inferred that
when f(x)
that
is
a null-function throughout
then
constants
and
B exist such
F(x) =
bn
A + Bar*,
(l
- ) /* (\n) + JJ 2 (X.)] = v/^ (A*P + Bar*) xJv (X) efo - \ r^^+1 (\) + b
and
5=
f^p - ^.
<*)}]
J, +1 (X)=0(X-*),
so that, if
/-iOO = 0(\n-),
'
5^0,
n
~ 2T( V)
bn
and
this is inconsistent
= o (\/n),
since v
> .
J? is zero,
Hence
and therefore
6n [(l
) ^
2
(X.)
This equation
is
unless
A is zero, since /+, (\) is not zero and then bn is zero. We next consider what happens when H + v is zero or negative
assume the forms
oo
in these
Xp
+ 2
w=l
bmJyCKnX),
oo
6 / (\ x)
respectively.
+ 2 bm J, (\m x),
In the second of the two cases the previous arguments are unaffected by the first of the two cases needs more careful consideration because the initial term to be inserted in the associated
the insertion of an initial term
function
;
is
b x*
4(i/+l)'
and hence, when n
>
1,
l
bn JS (\)
= X
[Ax*
J
^f^ + Bar*} xJ
{\nx) dx
19-7]
SCHLOMILCH SERIES
and then that
653
Since bn
(^n)">
2?
= 0,
infer also that, as in the limiting case of series of Bessel functions, Dini's series of Bessel functions cannot represent a null-function throughout the interval (0, 1), and that if two of Dini's series (with the same v and H)
We
all
1),
number
are equal.
CHAPTER XX
THE TABULATION OF BESSEL FUNCTIONS
20' 1.
and
associated functions.
vii,
It is evident
VIII
and XV that a large part of the theory of Bessel Functions has been con-
structed expressly for the purpose of facilitating numerical computations connected with the functions. To the Mathematician such computations are
of less interest and importance* than the construction of the theories which make them possible; but to the Physicist numerical results have a significancef
fail
to convey.
it
As an application of various portions of the Theory of Bessel Functions, has been considered desirable to insert this Chapter, which contains an historical account of Tables of Bessel Functions which have been previously published, together with a collection of those tables which seem to be of the
greatest value for the present requirements of the Physicist.
The reader will not be concerned with the monotony and technical irrelevance of this Chapter when he realises that it can be read without the efforts required to master the previous chapters and to amplify arguments so ruthlessly condensed.
his
The first Tables of J (*) and J (x) were published by Bessel himself in memoir on Planetary Perturbations, Berliner Abhandlunyen, 1824 [1826],
x
pp. 46
52.
fo*r
decimals
These tables give the values of J (x) and J, (x) to ten places of a range of values of x from x = to x = 3'20 with interval O'Ol.
A short Table of J(x) and J Hx) to four places of decimals was constructed
by Airy,
the same
Phil.
Mag.
0*2.
(3)
xvm.
(1841),
p.
7; its
range
is
from
a?
to
x 100
with interval
2/j (x)/x,
of
scope.
The function Ji{x)jx was subsequently tabulated to six places of decimals by Lommel, und Phy. xv. (1870), pp. 164167, with a range from x-^0 to #=20*0 Table, with a Table of JJ (a?)/^2 was republished by, Lommel, 0*1 this interval with Munckener Abhandlungm, xv. (1886), pp. 312315.
,
London Math.
" in formulas unless I feel their arithmetical f Cf. Lord Kelvin's statement I have no satisfaction magnitude at all events when formulas are intended for definite dynamical or physical problems."
p. 76.
% Tram. Camb. Phil. Soc. v. (1835), p 291. A Table of 2JX (x)/x and its square, to four or five 1125 (with interval 15) places of decimals, in which the range is from to the circular measure of was given by Sohwerd, Die Beugung$er$cheitmngen (Mannheim, 1835), p. 146.
:
20 *1]
655
In consequence of the need of Tables of Jn (x) with fairly large values of for Astronomical purposes, Hansen constructed a Table of J (x) and (x) to six places of decimals with a range from x = /"i to x = 10-0 with
n and x
interval
was published in his Ermittelung der absoluten StOrungen und Neigung (Gotha, 1843). Hansen's Table was reprinted by Schlomilch* and also by Lommelf who extended it
;
01
this
toa: = 20.
These
tables,
J (x)
and
J (x)
x
Abhand-
from x =
to
15*50
by Gray and Mathews, A Treatise on Bessel Functions (London, 1895), pp. 247 266, and an abridgement of it is given in Table I infra, pp. 666697.
was reprinted in
full
A
to
Table of
(x)
and
/,
(x) to
x = 60 with
twenty-one places of decimals, from x = been constructed by Aldis, Proc. Royal Soc.
50 has been computed by
series, Phil.
lxvi. (1900),
p. 40.
l, 2, 3, ...,
The value
Mag.
of
(40)
A
pp.
Table of Ji(x) to six places of decimals from ^=201 to # = 41 with interval 0*1 or by Steiner, Math, und Naturwiss. Berichte aus Ungarn, xi. (1894),
372373.
The
B. A. Smith, Messenger, xxvi. (1896), pp. 98101; this 0) places of decimals of Neumann's functions (x) and
<
from x =
interval 01.
to
x =100 with
interval
01 and from
is
82;
whose range is from x = to x= 1550 with interval 002; a year later a table was published, ibid. 1915, p. 33, in which the values of F (x) and F< (x) were given to ten places of decimals for a range from x = to x = 6'0 with interval 0*2 and from x = 6*0 to x= 160 with interval 05.
(0
Shortly after the appearance' of Smith's Table, an elaborate table was constructed by Aldis, Proc. Royal Soc. lxvi. (1900), p. 41, of Heine's functions O (x) and Gx (x) to twenty-one places of decimals; the reader should be
far Math, und Phijs. n. (1857), pp. 158165. t Studien iiber die Bessel'schen Functionen (Leipzig, 1868), pp. 127 135. X Meissel's Table contains a misprint, the correct value of J (0-62) being +0-90618..., not + 0-90518.... An additional misprint was made in the reprint of the Table by Gray and Mathews.
* Zeitschrift
(5),
123
the
same
and
(,c).
656
[CHAP.
XX
reminded that these functions are equal to \irY (x) and |w Y t (x) reto x = 6'0 with interval 0*1. spectively. The range of Aldis' Table is from x =
decimals from
x 65
The Report for 1915, p. 33, contains a to x 15*5 with interval 0"5.
The
functions
;
(x)
and
in Table I infra
make
interpolation a trivial
is
impracticable.
By means of the
which have now been described, it is an easy matter to construct tables of functions whose order is any integer. Such tables of /", (x) were constructed by Hansen and reprinted by Schlomilch and Lommel after their Tables of J (x) and J (x). Subsequently Lommel, Miinchener Abhandlungen, XV. (1886), 316, published a Table of Jn (x) to six places of decimals, in which pp. 315 n = 0, 1, 2, ..., 20, and x - 0, 1, 2,..., 12; this Table is reprinted in Table IV 731. A Table of Jn (x) of practically the same scope was also infra, pp. 730 published by Meissel, Astr. Nach. cxxviii. (1891), col. 154 155.
much more
extensive Table of
it.
Jn (x) was computed by Meissel, but it He communicated it to Gray and Mathews 279. This table gives Jn (x) to pp. 267
when n = 0,
1, 2, ...
60,
and x =
211
0, 1, 2, .... 24.
214.
n (x) to 6*0
The
86 contains Tables of G
x
= 0,
1, 2,
= 60
to
160 with
interval 05.
{n)
34
36.
Some
Yn (x)
Nicholson, Proc. London Math. Soc. (2) XI. (1913), pp. 113
114.
is
Table of
Yn (x)
contained in
(x)
and
(x).
and 6,
(x)
computing Table I, a small part of the British Association Table of G (x\ was checked, and the last digits in it were found to be unreliable in about 5 / of the
entries checked.
20,1 J
657
Tables of log10 [s/i^irx) ^> (x) v ] to eight significant figures are given in the British Association Report, 1907, pp. 9497. The values assigned to v are 0, \, 1, 1, ..., 6, and the range of values of x is from 10 to 100
j
x=
For this range of values of x, the asymptotic expansion ( 7*51) gives so rapid an approximation that the Table is of less value than a table in which the values of x and the intervals
to
(interval 10)
and 100
1000 (interval
100).
Functions of the first kind with imaginary argument have been tabulated in the British Association Reports. The Report for 1896, pp. 99149, contained a Table of I (x) to nine places of decimals, its range being from to x = 5-100 with interval 0001. Table of /, (x) of the same scope had been
x=0
published previously in the Report for 1893, pp. 229 279; an abridgement of this (with interval 001) was given by Gray and Mathews in their Treatise,
pp.
282284.
Tables of
I (x) and 1^ (x) to twenty-one places of decimals have been conby Aldis, Proc. Royal Soc. lxiv. (1899), p. 218. The range of these Tables is x = to x = 60 with interval 01; Aldis also gave (ibid. p. 221) the values of I (x) and I (x) for x = 7, 8, 9, 10, 11.
structed
x
Extensive tables connected with I (x) and I\ (x) have been published by Andmg,Sechsstellige Tafeln der Bessel 'schen Funktionen imagindren Arguments
(Leipzig, 1911).
to
x= 1000
.
These tables give log ]0 1 (x) and log 10 /, (x)/x] from #=0 with interval 001. They also give the values of the functions
{
vX27nr) <r*
(x),
^(2irx) e~*Ix
.
(x),
log 10
{y/x /, (*)}
.
for values of
(interval 1),
to
= 500
(interval 01),
# = 50
to
#=200
(interval 10),
and
Table II infra, pp. 698 713, gives the values of e~ x IQ (x) and er* J, (a?);, these have been computed, for the most part, by interpolation in Aldis' Table.
constructed bv These give (x) and x (x) to twenty-one places of decimals for values of x from ^= to x = 60 with interval 01, and also to between seven and thirteen significant figures from x = 5*0 to x = 12*0 with interval 01.
Aldis, Proc.
The
Kn (x) were
Royal
219221.
The
values of
(x) in
and e^I^x).
to
By means of recurrence formulae, In (x) has been tabulated to twelve signiw = 0, 1, 2, ..., 11 over the range of values of x from x = x = 60 with interval 02. These Tables of In (x) were published in the
British Association Report, 1889, pp. 29 32, and reprinted by Gray and Mathews in their Treatise, pp. 285 288. An abridgement (to five significant figures) of these Tables has been given by Isherwood, who added to them
658
Tables of
[CHAP.
.
XX
0, 1, 2,
. .
of values of x from
x = 6'0 with
interval 0*2.
Lit.
published in the
no. 19.
1904,
Tables of e~ x In (x) and n (x) to seven places of decimals are given in Table IV infra, pp. 736739.
The
Meissel,
11,
. .
.
earliest
who
has calculated
Jm (n)
?i
n=10,
21, Astr. Nach, cxxix. (1892), col. 284; Meissel also calculated
Jn (1000)
= 1000,
999,
...,
n from n = l to 7? = 50 (interval 1), = w 50 to w=100 (interval 5), n = 100 to n = 200 (interval 10), n = 200 to n = 400 (interval 20), n = 400 to n = 1000 (interval 50), n = 1000 to n = 2000 (interval 100) and for various larger values of n, are given in the British
to six places of decimals for values of
96.
to seven places of decimals are given
Tables of
in Table
VI
infra, pp.
The
functions ber
A brief Table of ber (x) and bei (x), computed by Maclean, was published by Kelvin, Math, and Phys. Papers, ill. (1890), p. 493. Tables of J (xy/i) and y'2 J (x \fi) to twenty-one places of decimals have been constructed by Aldis, Proc. Royal Soc. lxvi. (1900), pp. 42 43; their range is from
.
x=
These are extensions of the Table of J (x *Ji) to nine places of decimals for the range from x = to x = 60 with interval 02 published in the British Association Report, 1893, p. 228, and reprinted by Gray and Mathews in their Treatise, p. 281.
to
x = 60 with
interval O'l.
x= 1, 2, 3,
p. 53.
...,30,
(a;), ker(#) and kei (x) to four significant figures for have been published by Savidge, Phil. Mag. (6) xix. (1910),
The
functions ber
(x),
and
tabulated to nine
to x
01
in the British
Association Report, 1912, pp. 57 68; and a Table of ker (x), kei (x), ker' (x) and kei' (x) of the same scope (except that only six or seven significant figures
were given) appeared in the Report for 1915, pp. 36 38. Tables of squares and products of the functions to six significant figures from x = to x= 100 with interval 02 were given in the Report for 1916, pp. 118 121.
The functions </ (+$)(#) have been tabulated to six places of decimals by Lommel, Milnchener Abh. xv. (1886), pp. 644 647, for n 0, 1, 2, ...,6 with x = 1 2, 14 50, and (in the case of functions of positive order) n = 7, 8,
. .
20-1]
659
;
34 with smaller ranges of values of x see and n = 15, Table V infra, pp. 740 743. A Table of the same functions to four places of decimals with n =0, 1, 2 and from a; = to x = 80 with interval 02 is given by Dinnik, Archiv der Math, und Phys. (3) xx. (1913), pp. 238240.
,
with x =
20,
. . . ,
Association Reports.
Functions related to J(n +h)( x) have recently been tabulated in the British The notation used is
s/Qirx)
.
J_ n . h (x) = Cn (x),
and the functions and their logarithms. In the Report and in the Report
12,..., 19.
for 1916, pp.
En (x) = \Cn (x) + iSn (x)\, tabulated are Sn (x), Cn (x), En (x),
2
Sn'(x), Cn'(x),
En'*(x),
88
97107,
Functions of order + J, , have been tabulated by Dinnik, Archiv der Math, und Phys. (3) xvni. (1911), pp. 337 338, to four places of decimals; the functions tabulated are T(l + J) Ji (x), T(l %)J$(x) from = to x = 8*0 with interval 0*2; and Dinnik has also tabulated I(x), I$(x), ibid.
326. All these have the range x = x = 8 with 02. The Tables of T(l + $)Ji(x) are extensive than Table III 714 729 with the exception of Dinnik's there
pp.
ibid. (3)
xxi. (1913),
interval
infra,
324
to
pp.
but,
tables,
exist
no
and f
, Airy's
miff)
cos
^rn-
(w3
dw
Jo
must be mentioned
0'2
;
Airy calculated by quadratures and by ascending series from 5'6 to + 5'6 with interval
= 4tom = 4is given in the Trans. Camb. a seven-figure Table from Phil. Soc. vi. (1838), p. 402, and a five-figure Table from ra = - 56 to m = 56,
vin. (1849), p. 599.
-
ibid.
Apart from the work of Euler described in 15 5 the earliest computation of the zeros of J (x) and */, (x) is due to Stokes, Trans. Camb. Phil. Soc. ix. (1856), p. 186 [Math, and Phys. Papers, n. (1883), p. 355]. Stokes gave the values of the first twelve zeros (divided by if) of J (x) and J, (x) to four places of decimals. In the same memoir he gave the first fifty zeros of Airy's integral, and the first ten stationary points of this integral.
The
Ann.
sci.
first
nine zeros of
ill.
;
87. Bourget's results are (1866), pp. 82 some corrections in his Tables have recently
Mag.
pp.
714.
660
[CHAP.
XX
Bourget's Tables have been reprinted so frequently that their authorship has been overlooked by the writers of the articles on Bessel Functions in the Encyclopiidie der Math. Wiss. and the Ewrgclopedie de.i Set. Math.
The first five zeros of J (x) and J2 (x) were given to six places of decimals by Lommel, Zeitschrift fur Math. und. Phys. xv. (1870), p. 167 and Munchener Abhandlungen, XV. (1886), p. 315.
x
The
The
pp. 153
first
ten zeros of
(x)
J
x
(x)
155;
they also
American Math. Soc. Hi. (1897), gave the values of J (x) and log \Ji{x)\ at these
respectively.
zeros to eight
The
this
computed
and the corresponding values of J (x) were to sixteen places of decimals by Meissel*, Kiel Programm, 1890; Table is reprinted by Gray and Mathews in their Treatise, p. 280.
first fifty
x
(x)
values
0, , 1, f, 2, f.
Dinnik in his Tables of functions of fractional order mentions the values of a few of the zeros of each function, while Airey, Phil. Mag. (6) xli. (1921), 205, has computed the value of the smallest zero of pp. 200 v {x) 'for small fractional values of v by Euler's method.
i.
(1899), pp.
Rayleigh, Proc. London Math. Soc. x. (1878), pp. 6 363364], has calculated that
7 [Scientific Papers,
has a
maximum when
ten zeros of
a?
computed
the
first
3xJ
(x)
2xJ
(x)
of decimals.
Lommel
In his memoirs on Diffraction, Munchener Abhandlungen, xv. (1886), has published tables connected with his functions of two variables,
but these tables are so numerous that a detailed account of them will not be given here. His Table of Fresnel's integrals (p. 648) to six places of decimals from x = to x 50'0 with interval 0*5 (with auxiliary tables for purposes of interpolation) must, however, be mentioned, and with it his Table of the first sixteen maxima and minima of these integrals.
* Jahrbitch iiber die Fortschritte der
Math. 1890,
p. 521.
of Meissel's table, the zeros of J\ (x) were recomputed (to ten places of decimals) for insertion in
Table VII,
p. 748.
20*2]
661
Lommel's form
was
Ji(t)dt;
J o
a different form was tabulated earlier by Lindstedt, Ann. der Physik und Ch&mie, (3) xvn. (1882), p. 725.
Defining the functions
/
M (x) and N (x) by the equations cos dt = M {x) cos x JV (x) sin x
t
2 %
2
,
J X
sin
Pdt =
2
,
and writing x =
{(y
of decimals from
and
The
if Hi 2(2) rf*
t
00
has been tabulated to four places of decimals by Struve, Ann. der Physik und Chemie, (3) xvn. (1882), pp. 10081016, from x = to 4'0 (interval 01), from
= 40
to
70
(interval
(x),
has
been published by Steiner, Math, und Naturwiss. Berichte aus Ungarn xi. (1894), pp. 366 367 this integral occurs in the problem of Diffraction by
a Circular Aperture:
No
Table of
is
697.
20*2.
to
Preliminary considerations on the magnitude and character of the tables be included in this book led to the following decisions
(I)
That space did not usually admit of the inclusion of more than seven
of
moderately large interval (such as 0*02), together with an auxiliary table to facilitate interpolation, would be more useful than a table with a smaller
interval (such as 0*01), occupying the
auxiliary, in
same space
and
its
662
(III)
[CHAP.
XX
computing
tables, calculations
number
apply to the auxiliary tables of angles which are entered in Tables I and III.
it is
tenths of a second of arc, because the differences per minute of arc in a seven-
may be
as large as
an error of a hundredth of a second does not affect the value of the sine by more than 0*00000005. Hence, for seven-figure accuracy, it was considered
adequate to compute to nine places of decimals the sines (or cosines) of the angles tabulated and then to compute the angles from Gifford's Natural Sines
(Manchester, 1914)
;
hundredth of a second of
and F, (x) from and J^ix) up to 1550 are taken from Meissel's Table J, while the values of Y (x) and Y {x) were computed partly by interpolation in Aldis' Table of G (x) and Gx (x) and partly from the asymptotic expansions of J 2 (x) + F 2 (x) and J^ (x) + Y? (x)
Table I consists primarily of Tables of
(x), (x),
1
(x)
1600 with
interval of 0*02.
The
values of
{x)
given in
7*51.
The
n
{1) auxiliary tables give the values of \H n w(x)\ and arg n (x) for and n=l. In these tables the first differences are sufficiently steady
H
is
with but
little
when x
about 10
H
(x)
W
j
and arg H (x) can consequently be computed by the reader for any value of x less than 16, with the exception of quite small values. The corresponding values of Jn (x) and Yn (x) can then be calculated immediately by the use of seven-figure logarithm tables.
The values
j|
of
Hn
ll)
{1)
and it was found that the by more than a unit in the tenth place so it is hoped that the number of errors in the last figure retained does not exceed about one in every thousand entries. tNo tables with a smaller interval have been published; the use of any tables with a larger interval and a greater number of decimal places would have very greatly increased the labour of constructing the auxiliary tables of angles, and the increased accuracy so obtained would be of no advantage to anyone using the auxiliary tables for purposes of interpolation. X I must here express my cordial thanks to the Preussische Akademie der Wissenschaften zu
*
The
tables were di. T erenced before removing the last three figures,
me
to
make use
of this Table.
The
idea of constructing the auxiliary tables grew out of a conversation with Professor Love,
in the course of
of
which he remarked that it was frequently not realised how closely Bessel functions any given order resemble circular functions multiplied by a damping factor in which the rate
of decay is slow.
I|
is
or
1.
20*2]
663
The
most
between the various functions tabulated may be expressed w (x) and arg n (a?) as the polar coordinates by regarding n
|
{1)
of a point in a plane
Jn (x)
and 7 (x).
= 800,
+ 01716508 = 0-2818259 cos 412 28' 40"60, + 02235215 = 02818259 sin 412 28' 40"60.
Table I also contains the values of Struve's functions (x) and These functions are included in Table I, instead of being contained
separate
Table,
to facilitate
is
(x).
in a
interpolation;
by
Hn (x) Yn (x)
The Tables
and Ho'(ar)
Ho"(#), Ho"'
10'41 (4),
it
the difference
varies sufficiently
when x
is
not small.
computed by calculating the values of (x) when x=\, 2, 3,..., and then calculating these values o( x from the differential equation 10-4(10) and the
it.
A few differential coefficients are adequate to calculate Ho (x) and Ho' (x) by Taylor's theorem for the values 0-5, 0'6, 07, ... of x. Interpolation to fiftieths of the unit is then effected by using Taylor's theorem in the same manner. This process, though it seems at first sight to be complicated and lengthy, is, in reality, an extremely rapid one when a machine* is used. It is very much more effective than the use of asymptotic expansions or the process suggested in the British Association Report, 1913, p. 116. As an example of the rapidity of the process, it may be stated that thewalues ofe~ x I (x) and e~ x Ix (x) in Table II took less than a fortnight to compute of course the time taken over this tabulation was appreciably shortened by the use of Aldis' Table as a framework for interpolation.
;
Table II consists of Tables of e-* I (x), e~x Ij (x), e? (x), and e x A", (.r), and a Table of e* is inserted, in case the reader should require the values of the functions'^ I (x), Ix (x), (x) and (x); the functions are tabulated
from
to
1600 with
interval 002.
is
Interpolation by differencing
first
four functions
The Table of e* was constructed with the help of Newman's Table of e~ x Trans. Camb. Phil. Soc. XIII. (1883), pp. 145241. Unlike the other Tables in this book, the Table of e* is given to eight significant figures J, and care
,
last digit
given
is
Interpo-
* The machine on which the calculations were carried out is a Marchant Calculating Machine, 10 x 9 recording to 18 figures. t These functions were not tabulated because tables of them are uusuited for interpolation.
its
appearance.
664
[CHAP.
XX
Newman's Table gives e~ x to a large number of places of decimals, but the actual number of significant figures in the latter, part of the Table is small and less than half of the Table of e* was constructed by the process of calculating reciprocals the rest was con;
structed from
of e13
and
e 16
Newman
These ex-
zero.
(x),
(*),
H^ (x)
and arg
\
H^ (x)
I,
and
e*
manner
already explained.
Table of
K$ (x)
is also
included.
order
( 8'43),
and
easily
this table
by means
J_j (x)
H^ (x)
cos {60
+ arg H^ (x)}.
Table
IV
values of x and
The
values
Jn (x), Yn (x), e~x In (x) and Kn (x) for various of Jn (pc) are taken from Lommel's Table t, with
;
but the remainder of Table IV, with the exception of some values of n (x) taken from Isherwood's Table f, is new they have been constructed in part by means of Aldis' Tables of functions of orders zero and unity.
some
corrections,
Lommel's Tablet of J (w+i )(a;) and Fresnel's integrals with some modifications and corrections. Table VI gives the values of Jn (n), Yn (n), Jn '(n), Yn (n) and n*Jn (n),
Table
is
'
'
n Y (n), n*Jn (n), n n ' (n) for n = 1, of the last four of the eight functions
!>
'
2, 3,
. . .
50.
is easy.
first
is
forty zeros of
Forty zeros of various cylinder functions of order one-third are also given. Table VIII gives the values of
X
li
from x
(t)dt,
I J
(t)dt,
= to 50 with interval 1, together with the first sixteen maxima and minima of the integrals. The former table of maxima and minima can be used to compute the coefficients (cf. 18'12) in certain Fourier-Bessel series
for
which v
* Trans. t I
= 0.
my
(1883), p. 245. cordial thanks to the Bayerische
Miinchen, to the Manchester Literary and Philosophical Society, and to the American Mathematical Society for permitting me to make use of these Tables. The non-existence of adequate trigonometrical tables of angles in radian measure has made it impracticable to check the last
digital in
TABLES
OF
BESSEL FUNCTIONS
666
/.(*)
(x)
!?(*)!
arg#(*)
90
HoW
o-ooooooo
+ i-ooooooo
0-02 0-04 O'OO o-o8 O-IO 0-12 oI4 0-16 0-18 0-20 0'22 0-24
+ + + + + + + + + +
o o o o o
o o o o o
68 64 61 5Q 56
+ + + + + + + + + +
+ + + + +
0-0127318 0-0254603 0-0381819 0-0508934 0-0635913 0-0762722 0-0889328 0-1015697 0-1141796 0-1267590
rh
018
0-20
0-22 0'2A 0-20 0-28 0-30
026
0-28
030
0-32 o-34 0-36 0-38 0-40
+ + + + +
o o o o o
9879366 9856518
9804958 9776262
4182414 3752907 3362914 3006375 2678500 2375444 2094070 1831788 1586436 1356190
1
45 44 42 41 39
50 44-63
6*41 37 46*91 4 27"57 32 53*32 13
+ + + + +
+ + + + +
o o o o o o o o o o
o o o o o
9745634 9713081 9678615 9642245 9603982 9563838 9521825 9477955 9432242 9384698
0-7627204 0-7205732 0-6805558 0-6424376 0-6060246 0-5711520 o-5376789 0-5054836 0-4744608 -4445i87
0-4155768 0-3875642 0-3604182 0-3340833 0-3085099
+ + + + + + + + + + + + + + + + + + + +
O32
o-34 0*36
02258999
0-2380565 0-2501497 0-2621765 0-2741336 O-2OOOI0O 0-2978265
038
0-40
0-42 o-44 0*46
139500
30 50 29 27 28 4 26 42 25 20
44*27 9*25
19*71 11*96 42*72
O3O95559
0-32I2033 0-3327655 0-3442396 0-3556226 0-36691 14 0-3781032 0-3891950 0-4001841 0-4110675 0-4218424
048 050
0*52 o-54 0-56
+ + + + +
+ + + + +
058 060
0-62
o o o o o p o o o o
o o o o o
16
064
o-66 o-68 0-70
0-72
oo-7
078
o-8o
0-82 0-84 o-86 o-88 0-90 0-92 0-94 0-96 0-98 I-OO
+ + + + +
+ + + + + + + + + +
10 55 45*35 39 16*05 to 8 23 3*05 7 7 5*4 5 5i 22*51 4 35 3 20 53*4J 37*48 2 5 34*o6 o 50 42*55 o2 3 57*6i
i 3 8 26*96 2 52 45*99
+ 0-4325061 + 0-4430558
+ 0-4534888 + 0-4638026 + 0-4739944
7% o>7
8388338 8312284 8234734 8155711 8075238 7993339 7910039 7825361 7739332 7651977
0-0120311 + 0-0056283
0-8415424 0-8326455 0-8240231 0-8156598 0-8075434 0*7996618 0-7920038 0-7845590 0-7773I77 0-7702706
+ + + + +
0-82
084
o-86 o-88 0-90
0-92
o o o o o
+ + + + +
6 < 5 20
+ + + + +
667
X
o
0-02 0-04 O-OO 0-08 O-IO
0-12 0-14 o-io T8 0-20 0-22 0-24 O-20 0-28
AM
O-OOOOOOO
Y^x)
00
IH^WI
00
arg rfi(*)
Hit*)
o-ooooooo
-90
9 58' 55-26 89 55 4I-64 89 50' 20*64 89 42; 54*34 89 33 25*29
+ + + + +
00099995
0-0199960 0*0299865 0-0399680
00499375
0-0598921 0-0698286 0-0797443 0-0896360 0-0995008
31-8598144
159643214
10-6758314 8-0377690 6-4591441
+ + + + +
+ + + + +
+ + + + +
3323825b
3-0416730 2-8071277 2-6091059 2-4396971 2-2931051 2-1649866 2-0520233 1-9516372 1-8617949 1-7808720
I-7075549 1-6407704 I-579633I
x
5-4097717 4-6625082 4-1038295 3-6706983 3-3253I40 3-0436375 2-8096547 2-6122883 2-4436328 2-2978968
2-1707415 2-0588527 i-959656i 1-8711218 1-7916282
1-7198647 1-6547603 1-5954320 1-5411449
1
- 89 21' 56*38 -89 8' 30*73 -88 53' 11*68 -88 36' 2*65 - 88 17' 7*17
- 87 56' 28*81 - 87 34' 11*11 - 87 10' 17*63 -86 44' 51*87 - 86 17' 57*28
01093358
0-1191381 0-1289046 0-1386325 0-1483188
030
0-32 o-34
036 038
0-40
0-42
+ + + + + + + + + + + + + + +
+ + + + + + + + + + + + + + +
" -
5! 85 84 84
49' 37-22
19' 55'oo
48' 53*80
43'
-83
16 36*72 6*78
+ + + + + + + + + + + + + + + + + + + +
+ + + + +
0-32 o-34
036 038
0-40
0-42 0-44 0-46 0-48 0-50
0-52 o-54 0-56
044
0-46 0-48 0-50
0-52 o-54 0-56 0-58 o-6o
0-62 0-64 o-66 o-68 0*70
02331540
0-2422685 0-2513105 0-2602774 0-2691665 0-2779752 0-2867010
5234063
4912830
- 83 8' 26*86 - 82 32' 39*76 - |i 55 48*18 X " 54;69 -8o 3 9' 1*79
K
79
I-4453258 1 -4028308
1
-79
-
3634193
1-3267657 1 -2925880
1-2606415 1-2307120 I-2020I22 1-1761767
058
o-6o
0-62 0-64 o-66 o-68 0-70
0295341
0-3038932 0-3123547 0-3207230 0-3289957
0-337I705 0345244 8 0-3532164 0-3610829 0-3688420
II5I2595
I-I2773I2 1-1054763 1-0843920 1-0643861 I-0453758
- 76 75 74 74 73
10 33*73
23' 42*38
078 080
0-82"
- 72 36' 12*87 - 71 48' 6*68 - 70 59' 25*22 - 70 io' 9*80 - 69 20' 21*73 - 68 30' 2*20 - 67 39' 12*39 - 66 47 53*42 -65 56 6*35 - 65 3' 52*21 64 63 62 6l 6o
ii' 11*97
0-84
0-3764916 0-3840292
086
o-88 0-90 0-92
03914529
0-3987603 0-4059495 0-4130184 0-4199649 0-4267871
09558093 09341970
0-9132475 0-8929069 0-8731266
+ + + + + + + + +
084
o-86 o-88 0-90
0-92 o-94 0-96
094
0-96 0-98 i-oo
+ + + +
08538622
0-8350735 0-8167241 0-7987806 0-7812128
0-9485066
+ 0-1697186
0-1767354 0-1838656 0-1911070 0-1984573
09347290
0-9215126 0-9088223 0-8966259
04334829 + 0-4400506
1 8'
098
I-OO
668
7oW
+ 07563321 + 07473390 + 0-7382212 + 07289813 + 07196220
+ + + + +
0-7101461
Y9 i*)
+ + + +
0-1037085 0-1188188
|ffJW
0-7634092
arg H^(*)
H
+ + + + +
+ + + + + + + + + + + + + + +
+ + + + +
(*)
X
I'02
1-02
:a
i-o8 I-IO
OI335943
0-1480406 0-1621632
07507255
0-7502I20
074386I4 07376671
0-7316228 0-7257225
7 48' 27*80 9 2' 1*68 io 15' 27*64 11 20' 46*01 12 41' 57*10
53
1
08
r-io
I-I2
112
i-i4 i-i6 i-i8 1-20
1-22 1-24 1-26 1-28
07005564
0-6908557 0-6810469 0-6711327 0-66^1163 0-6510004 0-6407880 0-6304822 0-6200860 0-6096023 0-5990343 0-5883850 0-5776576
+ 0-1759670 + 0-1894567
O7I99606
20 49*49
o7 I 433i7
07088309
07034533
0-6981944 0-6930499 0-6880157 0-6830879
130
1-32
+ + + + +
+ + + + +
XXI
1-28
1
0-2865354
0-2973645 0-3079127 0-3181824 0-3281758 0-3378951
35 52-28 48 4 ? 22
0' 11*36 12' 13*86
30
1-38
I
m
40
0566855
0-5559807 0-5450376 0-5340289 0-5229579 0-5118277
26 27 28 29 30
0-6885928 0-6946595
132
i-34
07005492 07062606
0-7117925
136
1-38
140
1-42 1-44 1-46
1 1
+ + + + +
O3824489
0-3905639 0-3984176 0-4060116 0-4I33476 0-4204269
0-4272512 0-4338219 0-4401404 0-4462083 0-4520270
31 59' 40*54 33 11' 22 "08 34 22' 59*82 35 34' 33*9o 36 46' 4*43 37 39 40 4i 42
57' 31*50
0-7171439 0-7223136
07273008 07321043
0-7367235
-48
50
52
+ 0-5006415 + 0-4894026
-f-
*% 1-58
i-6o
1-62
0-4781143
+ 0-4667797 + 0-4554022 + + + + + + + + + +
+ + + + + + + + +
0-4439850
+ + + + +
+ + + + +
+ + + + +
+ + + + +
!ts
1-68
04325313
0-4210446 0-4095280 0-3979849 0-3864185 0-3748321 0-3632292 0-3516128 0-3399864
170
1-72 1-74
43 45 46 42 48
53' 5**77
5, 7*26 16' 12*95
07605226 07638306
0-7669493
162
sa
1-68
27 /,
J 5*94
38' 16*30
07698781 07726168
170
176 178
i-8o
1-82 1-84 1-86 1-88 1 -90 1-92 1-94 1-96 1-98 2-00
+ 0-4728378 + o-47743i7
+ + + + + + + + + +
0-4817849 0-4858989 0-4897751 0-4934149 0-4968200 0-4999917 0-5029315 0-5056411
0-'j08l220
49 49' 14*10 51 0' 9*41 52 il' 2*31 53 21' 52*85 54 32' 4i*n 55 56 58 59 6o 6i 62 6 3 65 66
43' 27*14 54' 11*00
4' 52*75 15' 32*45
+ + + + + + + + + +
+ + + + +
m
it
1-82
1-72
4 0-3283532
26 10*14
36' 47' 57' 8' 18'
lit
i-88
1
90
05683310
0-5655192 0-5627481 0-5600168 o-5573243
0-5103757
192
i-94
07909681 07908588
196 198
2-00
669
X
1-02 I-04 1-06 1-08 I-IO
I-I2
ZiW
+ + + + +
0-4464882 0-4527939 0-4589660 0-4650027 0-4709024
-
{*)
K'wi
0-8848938 0-8735987 0-8627154 0-8522205 0-8420926
-
arg
IT* 1 '/
,
(x)
H
+ + + + + + + + + +
(*)
X
1-02 1-04 I -oo 1-08 I-IO
I-I2 1-14
59 58 57 56 55
59 56*99
3' 41 '07
+ 0-4766634 + 0-4822840
- 55
-54
116
1-18 I-20
1-22 1-24 1-26 1-28 i-3
+ 0-5129786 + 0-5175766
+ 0-5220232
30
32
0-7880056
07799589
0-7721533 0-7645772 0-7572200 0-7500717 0-7431229 0-7363647 0-7297888 0-7233873
50 18' 6*48 49 20' 12*33 48 22' 3*58 47 23' 40*70 -46 25' 4*12
_ 45 44 43 42
26' 14*26
+ 0-2858095 + 0-2942761
4-
0-3028191
+ 0-3114357 + 0-3201231 + + + + +
0-3288788 0-3376999 0-3465837 0-3555273 0-3645280
+ + + + +
- o-5344i33 - 0-5204287 - 0-5065597 - 0-4928008 - 0-4791470 - 0-4655936 - 0-4521367 - 0-4387723 - 0-4254973 - 0-4123086
-
27
n*54
27' 56*34
-41
" 4 39 38 37 36
42
150
152
1-54 1-56 1-58 i-6o
1-62 1-64 1-66 1-68 1*70 1-72 1-74 1-76 1-78 i-8o
+ + + + + + + + + +
0-7171528 0-7110785 0-7051576 0-6993840 0-69375 1 0-6882554 0-6828896 0-6776493 0-6725298 0-6675267
0-6626355 0-6578524 0-6531734 0-6485948 0-6441131
+ + + + +
+ + + + + + + + +
1-48
150
1-52
03992036
0-3861800 0-3732356 0-3603088 o-347578o
-35
-
05632079
0-5656003 0-5678298 0-5698959
27' 7'65 34 26' 15*44 33 25, 13*93 32 24' 3*39 31 22' 44*05
+ + + + +
+ + + + + + + + + + + + + + +
03348619
0-3222194 0-3096498 0-2971522 0-2847262
- 3O 2l' 16*17 - 29 19' 39*97 - 28 17' 55*68 - 27 16' 3*52 - 26 14' 3*70
- 25
06230473
0-6190823 0-6151939 0-6113796 0-6076371 0-6039642 0-6003588 0-5968189 0-5933424 0-5899276 0-5865726
-24
+ + + + +
+ + + + +
05139633
0-5234937 0-5330289 0-5425661 0-5521021
0-5616342 0-5711594 0-5806748 0-5901775 0-5996645
- 18 56' 46*45
- 17 53' 52*04 - 16 50 51-57 - 15 47' 45*18
194
1 -96
98
2-00
- M 44' 33*oi - 13 4i' 15*22 - 12 37' 51*93 - n 34' 23*27 - io 30' 49*38
+ + + + +
2-00
670
X
2-02 2-04 2-06 2-08 2-IO
;.w
+ 0-2123697 + 0-2008776 + 0-1894177
+ 0-1779931 + 0-1666070 + + + + +
0-1552625 0-1439626 0-1327106 0-1215095 0-1103623
Y9 (x)
+ + + + + + + + + +
0-5124038 0-5142080 0-5157900 0-5171513 0-5182937
|tf ?(*)l
(
arg
H J{x)
{1
HW
+ + + + +
0-7905626 0-7900800 0-7894119 0-7885590 0-7875222
X
2-02 2-04 2-06 2-08 2-IO 2-12 2-14 2-16 2-l8 2-20
2-22 2-2A 2-20 2-28 2-30
0-5546698 0-5520523 0-54947 10 0-5469250 o-5444i37 0-5419362 o-53949i7 0-5370796 o-534699i o-5323496
212
2-16 2-l8 2-20
2-22 2-24 2-26 2-28 2-30 2.32 2-34 2-36 2-38 2-40 2-42 2-44 2-46 2-48
+ 0-7863025 + 0-7849006
+ 0-7833178 + 0-7815550 + 0-7796135
6*53
+ + + + + + + + + +
-
+ 0-5206508 + 0-5203112
+ 0-5197675 + 0-5190215
00555398
0-0447786 0-0340921 0-0234828 0-0129538 0-0025077
0-0078527 0-0181247 0-0283057
+ 0-5180754 + + + + + + + + + + + + + + + + + + + +
+ + + + +
0-5169311 0-5155908 0-5140565 0-5123304 0-5104147
0-5300304 0-5277408 0-5254803 0-5232482 0-5210439 0-5188670 0-5167167 0-5145926 0-5124942 0-5104209
0-5083723 0-5063478 0-504347 1 0-5023696 0-5004149
12' 18*15
+ + + + +
07774943
0-7751986 0-7727279 0-7700834 0-7672665
0-7642787 0-7611214 0-7577962 0-7543047 0-7506485
0-7468293 0-7428488 0-7387088 0-7344112 0-7299577
33
43'
5*90 6*62
+ + + + + + + + + +
+ + + + + + + + + +
00383929
0-0483838 0-0582758 0-0680664 0-0777531 0-0873*34 0-0968050 0-1061654 0-1154123 0-1245434 0-1335565 0-1424494
0-1512198 0-1598658 0-1683852 0-1767759
250
252
2-54 2-56 2-58 2-60
2-62 2-64 2-66 2-68 2-70
0-5083116 0-5060233 0-5035522 0-5009004 0-4980704 0-4950645 0-4918851 0-4885347 0-4850157 0-4813306
9 53'
250
252
2-54
96 97 99 IOO IOI
102 103 104 io6 107
07253504
0-7205912 0-7156821 0-7106251
256
2-58 2*60 2-62 2-64 2-66 2-68 2-70
2-72 2-74 2-76
07054223
0-7000759 0-6945880 0-6889609 0-6831967 0-6772977
0-4774820 o-4734724
04693043
0-4649805 0-4605035
0-4558761 0-4511009 0-4461806 0-4411181
32'
278 280
2-82 2-84 2-86 2-88 2-90
04803025
0-4785907 0-4768970 0-4752209 o-473562i
0-4719204 0-4702955 0-4686871 0-4670950 0-4655187
0-4639582 0-4624131 0-4608831 o-459368t
01850360
04359160
0-4305772 0-4251045 0-4195008 0-4137689 0-4079118
108 2i' 4*79 109 30' 49*80 no 40' 34*04 111 50; 17*53 0' 0*27 1 13 114 115 Il6 117 118
19 121 122 1 23 124
1
+ + + +
+ + + + +
278
2-80
2-82 2-84 2-86 2-88 2-90
2 92
+ + + + + + + + + +
04019323
0-3958334 0-3896181
1*94 39*86
02531964
03832893
0-3768500
- 0-2600520
04578678
36' 29*87
+ + + + +
2-94 2-96
298 300
671
X
2-02 2-04 2-06 2-08 2-IO 2-12 2-14 2-l6 2-l8 2-20
7iW
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
0-5753554 0-5738267 0-5721393 0-5702942 0-5682921
0-5661342 0-5638212 0-5613543 0-5587345 0-5559630
-
(x)
l#7(*)l
arg
(x)
H
+ + + + + + + + + +
(*)
X
2-02 2-04 2-06 2-08 2-IO
2-12 2-14 2-16 2-18 2-20 2-22 2-24 2-26 2-28
- 9 27' 10*37
-8 23 26*38
I9 37 " ll ;^ - 6 15 43*88 - 5 45*59
"
0-0408645 0-0301393
00195045
0-0089616
0-0118422 0-0220999 0-0322594 0-0423191 0-0522773
+ 0-0014878 + + + + +
230
232
2-34 2-36 2-38 2-40
2-42 2-44 2-46 2-48 2-50
0-5531678 0-5504135
0-54770H
0-5450295 o-5423977
+ + + +
230
2-32 2-34 2-36 2-38 2-40 2-42 2-44 2-46 2-48
+ 0-0621324 + 0-0718828
+ 0-0815267
4-
0-0910627 + 0-1004889
6 7 8 9 IO
+ 0-7902603 + 0-7986950
+ + + + + + + + + +
0-1098039 0-1190059 0-1280934 0-1370647 0-1459181 0-1546522 0-1632654 0-1717560 0-1801226 0-1883635
^8475755
0-8554I22 0-863I542 0-8707993 0-8783453 0-8857900 0-893I3I4 0-9003674
250
2-52 2-54 2-56 2-58 2-60
252
2-54 2-56 2-58 2-60
2-62 2-64 2-66 2-68 2-70 2-72 2-74 276 2-78 2-80
2-82 2-84 2-86 2-88 2-90 2-92 2-94 2-96 2-98 3<x>
17 18 19 20 21
20' 48*57
2-62 2-64
2-66
2-68 2-70 2-72 2-74
+ + + + +
+ + + + +
28 29 30 31 32
276
2-78 2-80
2-82 2-84 2-86 2-88 2-90 2-92 2-94 2-96 2-98 3-oo
+ + + + +
03390590
+ + + + +
+ + + +
672
/(*)
(x)
\H l\x)\
0-4563820 0-4549103 o-4534528 0-4520090 0-4505787 0-4491618 o-447758i 0-4463673 0-4449893 0-4436239
0-4422708 0-4409300 0-439601 0-4382841 0-4369787
axgH
(x)
H
+ + + + +
(*)
302
3-04 3 06 3-o8 3-10 3-12
02667583
0-2733140 0-2797178 0-2859683 0-2920643
+ 03703033 + 0-3636522
125 46' 5*31 126 55 40*16 128 5' 14*42 129 i 4 ' 48*11 130 24' 21*23
J 3i
308 310
312 3-M
3-16
- 0-2980048
3M 3-16
3'i8 3-20
33' 53'-8o
+ 0-5267021
132 43' 25*82 133 52' 57*3i 135 2' 28*27 136 11' 58*71 137 1 38 139 140 141
J 43
+ + + +
+ + + + +
318
3'20
3. 22 3-24 3*26 3-28
322
3*24 3-26 3-28 3 30
03303193
0-3351416 0-3398009 0-3442963
0-3486272
+ + + + + + + + +
+ + + + +
330
332
3*34 3-36 3-38 3-4
332
3-34 3-36 3'38 3'4
03527931
0-3567934 0-3606277 0-3642956 0-3677967 0-3711306 0-3742972 0-3772963 0-3801277 0-3827914 0-3852873 0-3876155 0-3897760 0-3917690
'50*95
0-4149532
147
46 36*44
+ 0-4060080
+ + + + + + + + + +
0-3970279 0-3880161 0-3789757 0-3699095 0-3608208
0-3517124 0-3425876
3H2
3-44 3-46 3-48
350
352 354
3-5f>
04245308
o-4233432 0-4221655 0-4209974 0-4198389 0-4186898
0-4175501 0-4164195 0-4152980 0-4141855 0-4130817
148 56' 1*71 150 5' 26*56 I 5 I 14' 5o"98 152 24' 15*00 153 33' 38*6i 154 155 157 158 159
43'
350
352
3-54
3-58
360
3-62 3-64 3-66 3-68
+ + + + +
+ + + +
1*82
O3334492
0-3243003 0-3151440
370
372 374 376 378
3'8o
3-82
+ 0-1393962
0-1310727 0-1227424 0-1144086 0-1060743 0-0977426 0-0894167 0-0810994 0-0727939 0-0645032
+ + + + +
33*17 52*36
11*21
+ 0-2602179
3-72
+ 025IO933
+ 0-2419852 + 0-2328964 + 0-2238298
04087528
0-4076915
29*74 53 47*93
3'
0-4027318 0-4027456
3-86 3-88
04025986
0-4022918 0-4018260
0-4012023 0-4004218
390
3 92
+ + + + +
04066383
0-4055933 0-4045561 0-4035269 0-4025054 0-4014915 0-4004853
12 21
3 40
+ + + + + + + + + +
0-2147883 0-2057749 0-1967923 0-1878435 0-1789312 0-1700582 0-I6I2273 0-1524412 0-1437027 0-1350146
394
3 96
;
+ 0-0152362 + 0-0071319
- 0-0009343
398
4-00
- 0-0089594 - 0-0169407
180
2*36
394 396
3-98 4-00
673
J lit)
+ + + + + + + + +
0-3315626 0-3239979 0-3163677 0-3086746 0-3009211
Yi(*)
Ifl'I'WI
argtf^*)
44 45 47 48 49
51' 43 ? 8i 57 58*93
HW
X
302
3-4 3.06
308 310
312 314
3-16 3-i8
+ + + + + + + + + + + + + + +
+ + + + + + + + + +
1-0247251 1-0292003
1
0335340
320
3.22 3*24
0-3541583 0-3585314 0-3627483 0-3668084 0-3707113 o-3744565 0-3780436 0-3814723 0-3847421 0-3878529
50 23' 16*83
51. 29' 40*52
312
3-14
52 36' 53 42'
5*84.
0530469
316
3-i8 3-20
3276
1*25
54 49
1-0565124 1-0598303
326
3-28
330
332 334
3-3o 3'38
+ + + + + + + + + + + + + + +
+ + + +
55 55' 31*29 2' 2*86 57 58 8' 35*92 59 15 10*45 6o 21' 46*43 6i 28' 23*84 62 35' 2*65 63 41' 42^85
322
3-24
326
3-28
+ 1-0741863 + + + + + + + + + +
+ + + + + + + + + +
1-0766072 1-0788770 1-0809955 1-0829624 1-0847774
330
332
3-34 3-36 3-38 3-40
3*42 3-44
340
342
3'44 3-46
04419983
0-4406145
04392439
0-4378863 0-4365415 0-4352093
3-48 3'50
+ + + + + + + + + + + + + + + + + + + + + + + + +
+ + + + +
04338895
0-4325819 0-4312864 0-4300020 0-4287305 0-4274699 0-4262206
0-4249824 o-4237552 0-4225387
3H6
3-48 3-5 3-52 3\54 3-56 3-58 3-6o
3-62 3-64 3-66 3-68
352 354
3-56
358 360
362 364 3 1 3-68 370
7235'5i r 5i
73 42 43-18 74 49' 36 "04 75 56' 30-05 3' 25*21 77
+ + + + +
04213329
0-4201376
0-4189527 0-4177779 0-4166131 0-4154582 0-4143131
78 79 80 81 82
370
3-72 3-74
+ 0-0455712
83 45' 19*23
+ 0-0373359 + 0-0291307*
+ 0-0209582 + 0-0128210 + 0-0047218 - 00033369
- O-OI 13524 - 00193223 - 0-0272440
84*52 21*94
59 25*65 f5 87 6' 30*37 88 13' 36*07
89 20'
+ + + + +
376 378
3-80
3-82
3 'h
3 1i 3-88
390
3.92
04131776
0-4120516
4275
04109349
0-4098274 0-4087290
+ + + + + 1-08-12762 + + + + +
3*2
3-88 3 90
394 390
3-98 4-00
0-0351151 0-0429330
00506953
0-0583995 0-0660433
392
3-94 3 96 3-98 4-00
674
/oW
- 0-3957530 - 0-3942053 - 03925079 - 0-3906622 -
{*)
l# >)l
0-396534 0-3955643 0-3946015 o-3936457 0-3926967
o-39i7546 0-3908191 0-3898903 0-3889680 0-3880522
0:3871428 0-3862397 0-3853428 0-3844522 0-3835676
0-3826891 0-3818166 0-3809499 0-3800891
arg
H
',
(x)
H(*)
X
4-02 4-04 4-06
0-0248755 0-0327610
00405944
0-0483732 0-0560946
0-0637561 0-0713550
0-0.788889
+ + + + + + + + + +
408
4-10
4-12 4-14 4-i6 4-18 4-20
0-0863551 0-0937512
189 21' 58*54 190 31' 11*87 191 40' 24*95 Ig2 49' 37*78 l 193 58 50*36
195 196 197 198 1 99
8' 271 17' 14*81 26' 26*67
+ 0-0435095 + 0-0356255
+ 0-0278243 + 6-020IOOI + 0-0124793
03792341
0-3783848
.0-3775411
200 54' o-88 3' 11*83 202 203 12' 22*56 204 21 ' 33*08 205 30' 43*37
206 207 208 2IO 211
212 2I3 o 214 215 217
39' 53*45 49' 3*33
+ 00049399
-
4H8
4-50
4-52 4-54 4-5o 4-58 4-60
450
4-52 4-54 4-5 4-58 4-60 4-62
00650755
0-0714966 0-0778050 0-0839990 0-0900771
$
4-78 4-80
4-82
4-62
4-68
470
472 474 476
03670603
0-3662896 0-3655237
218 11' 219 20' 220 29' 221 38' 222 47'
223 225 226 227 228
57'
1% 4-68
4-70
4-72 4-74 4-76 4.78 4-80 4-82
02635093
0-2679693 0-2723038
03647625
0-3640061
02463592 O2404253
0-2344201 0-2283462 0-2222062 0-2160027 0-2097383
03632543
03625071
0-3617645 0-3610265 0-3602929 o-3595637
0-2765116
02805915
0-2845427 0-2883640 0-2920546
4-86 4-88
490
4-92 4-94 4-96
229 42' 34*17 230 51 40-33 232 O' 46*33 233 9' 52*17 234 18' 57-85 235 28' 2*37 236 37' 8*74 237 46' I3 ?95 23 '55 19*01 4' 23*93 240
498
5-00
02956136
0-2990401
0-3588389 0-3581185
0302333s 03054928
0-3085176
675
X
4-02 4-04 4-06 4-08 4-10
4-12
/W
-
Yx()
!<(*)!
0-4023226 0-4012845 0-4002545
arg rf\\x)
HxW
4-
X
4-02 4-04 4-06 4-08 4-10
01032733
+ + + + + + + + + + + + + + +
03992325
0-3982185
40-00 56*49
13*81
1-0670119
31-94 50*87
414 416
4-18 4'20
4'22 4-24 4-26 4-28
03972123
0-3962138
03952229 03942396
0-3932638
no
+ + + + +
+ + + + +
4-
430
4-32 4'34
03922953 03913340
0-3903799
in
112
"4
115
49*79
03894328
0-3884928
116
8 15*45 15 41*80
+ 0-3441256 + o-3397472
430
4-38 4-40
+ 0-3352606
+ 0-3306677 + 0-3259707
117 23' 8*84 118 30' 36*56 1 19 38' 4*96 120 45' 34*02 121 53' 3*74
0' 34*10 123 8' 5*10 124 125 15' 36*73 126 23' 8*99 127 30' 41*86
+ 1-0113128
1-0060764 + 1-0019350 + 0-9970906 + 0-9921451
4H2
4-44 4'42 4-48 4-5
+ + + + + + + + + +
+ + + + + + + + + + + + + + + + + + +
+ + + + + + + + + + + + + + + + + + + + + + + + +
+ + + + +
450
452
4-54 4-56 4-58 4-60
4-62
452
4*54
- 0-2363882
-.
0-2416027
02903542
0-2849015
450
4-58 4-60 4*62
02793644
0-2737452
0-2622702 0-2564190 0-2504952 0-2445013
'59*33
8' 35*18
x\ 4-68
470
472
4-74
0-2613006 0-2659284
+ 0-2680464
02704352
0-2748196 0-2790807
0-2832174 0-^872286 0291 1 133 0-2948707 0-2984999 0-3019999 0-3053702 0-3086098
11*59
#* 4-68
4-70
02384397
0-2323128 0-2261230 0-2198730 0-2135652
476
4-78 4-80
4-82 4-|4 4-86 4-88 4-90
139 54' 21*96 141 2' 1*65 142 9' 41 -8 7 143 17 22*60 i4425' 3*85
32' 45*6i *4 6 40 27*87 147 48' 10*62 148 55' 53*86 150 3 37"59
472
4-74 4-76 4.78 4-80
4-82 4-84 4-86 4-88 4-90
4-92 4-94 4-96 4-98 5-00
03117182
0-3146947
M5
03639221
0363 1 57
0-3623971 0-3616418 0-3608913 0-3601456 0-3594045
492 494
4-96 4-98 5-00
11' 21*79
*?',
6 "A7
676
X
5-02 5-04 5-06 5-08
7o(*)
Y
-
(x)
\H l\x)\
o-3552793 o-3545799 0-3538847 o*353!934 0-3525062
arg //(*)
13' 28*68 22' 33*30 31' 37*76
H(*)
-
X
5-02 5-04 5*o6 5-o8 5-10 5-12 5-14 5-16 5'i8 5-20 5*22 5-24 5-26 5-28 5-30 5-32 5-34 5-36 5-38 5-40
5-42 5-44 5-46 5-48 5-50
510
5-12 5-14
0-3114072 0-3141609 0-3167784 0-3192590 0-3216024 0-3238083 0-3258764 0-3278063 0-3295978 0-3312509 0-3327654 0-3341413 0-3353785 0-3364772 0-3374373
0-3382591 0-3389428 0-3394886 0-3398969 0-3401679
0-1885712 0-1917864 0-1948610 0-1977971 0-2005919 0-2032458 0-2057586 0-2081301 0-2103600 0-2124483
516
5-18 5-20
5-22 5-24 5-26 5-28
0-1375776 0-1307919 0-1239803 0-1171456 0-1102904 0-1034176 0-0965297 0-0896295 0-0827198 0-0758031
0-0688822 0-0619598 0-0550386 0-0481211 0-0412101
246 58' 50*32 248 7' 54*22 249 1 6' 57*98 250 26' 1 "61
251 35'
5*"
530
5'32 5'34 5-36 5-38
5'4<>
252 44' 8?47 253 53 n*70 2' 14*80 255 256 n' 17*76 257 2o' 20*60
- 0-2143949 - 0-2161998
03426550
0-3420271 0-3414027 0-3407816 0-3401639 0-3395496
258 29' 23*32 259 38' 25*91 260 47' 28*37 261 56' 30*72 5' 32*95 263
264 14' 35*05 265 23 37*04 266 32' 38*91 267 41' 40*66 268 50' 42*30
5H2 5H4
5-4^ 5-48 5 50
5-52 5.54 5-56 5-58
5 60
+ + + +
59' 43*83
8' 45*24
17 46*54
26' 47*73
35 48*81
5-52 5-54 5-56 5-58 5-6o 5-62 5-64 5-66 5-68 5*70
570
572 574 576 578
5-8o
5-82 5 'h 5 *fi 5-88 5-90
5*92 5-94 5'96 5-98 6-oo
+ + + + + + + + + + + + + + + + + + + +
0-33475"
0-33+1655 o-3335828
03330033
0-3324267 -33 I 853 0-3312824 0-3307146
286
2
572 574
5-76
578
5-80
5-82
03301498
0-3295878 0-3290286 0-3284723 0-3279188
0-3273681 0-3268201 0-3262749
l 53*83 f ll 288 23 53*68 289 32' 53 ? 44 290 41' 53*11 291 50' 52*68
<
- 0-2085942
- 0-2063827 - 0-2040527 - 0-2016058 - 01 990435
lit
5-88 5-9o 5 92 5-94 5-96 5-98 6-oo
03257324 03251925
26 50-05 35' 49 ? i6
677
X
5-02
/iW
-
(x)
i^J'wi
0-3586680
argH^x)
i565o'
160
1 62
Ht(*)
X
5-02
504
5-06 5-08 5-io
5-12
9 ? 8o
03579362
0-3572088 0-3564859 0-3557675
+ + + + +
504
5-06 5-08
510
512
5-14
516
5-i8
520
5-22 5-24 5-26 5-28 5-30
+ + + + + + + + + + + + + + +
03550534
0-3543437
03536383
0-352937 1
03522402
o-35i5474 0-3508587
+ + + + + + + + +
+ + + + + + + + + +
0-3440409
+ 0-7304068
03501742
o-3494936 0-3488171
526 528 53
532 534
5'3 5-38
532 534
5'36 5-38
5HO
5'42 5*44 5*4 6 5'48 5'50 5-52 5'54 5'5o
00376356
0-0307351 0-0238485 0-0169784 0-0101273
03459903
0-3457337 o-3453448 0-3448242 0-3441725
3'
24*96
54
542
5-44 5-46 5'48
+ 0-0032975 - 00035083
03433905
0-3424788 0-3414382
- 0-0102879
- 0-0170386 - 0-0237582 -
03441904
0-3435445 0-3429024 0-3422638
*K
J
0-6599812 0-6530372 0-6461209 0-6392347 0-6323810 0-6255623 0-6187809 0-6120390 0-6053391 0-5986835 0-5920743 0-5855138
550
552 554
5-56 5-58 5-60
0-3402696
00304443
0-0370944 0-0437062 0-0502774 0-0568050
0-3416288
03389739
o-33755i8
03409975
0-3403696 o-3397452
558
5-60
5-62
0336004s 03343328
03391243
5: 6 ; 45-72 14' 42*38 187 22' 39*38 188 30' 36*72 189 38' 34*39
186
+ + + + +
+ + + + + + + + + +
S3 568
5-7
5'72 5-74 5-7
- 03264245 - 03241477
0-3217534 0-3192429 0-3166176 0-3138787 0-3110277
0-3080661
00632886
0-0697241 0-0761099 0-0824437 0-0887233
03385069
0-3378928 0-3372821 0-3366748 0-3360708
190 46' 32*39 I9i 54' 30 ? 7i 2' 29*35 193 194 io' 28*30 18' 195 27*57 196 26' 197 34' 198 42' 199 50' 200 58'
6'
562
5
5-68
05790044
0-5725481 0-5661472
s&
570
572
5-74 5'7| 5-78
578
5'8o
5-82
0-3354700
03348725
0-3342782
03336871
0-3330991
0-5598038
05535201
0-5472981 0541 1399 0-5350476
580
582 54
03049952
0-3018166 0-2985318
02951424
0-2916501 0-2880563 0-2843629 0-2805715
0'2 766839
0-3325I43
03319326
0-33I3540 0-3307784 0-3302059
0-3296363 0-3290697 0-3285061 o-3279453 0-3273875
0-1367560
01424539
0-1480772
202 29*60 203 14' 30*98 22' 32*65 204 205 30' 34*61 206 38' 36*85
+ + + + + + + + + +
5-86 5'88
590
5 92
594
5-96
598
6-oo
5 1
678
X
6-02 6-04 6-o6 6-o8 6-io 6-12 6-14 6-i6 6-i8 6-20 6-22 6-24 6-26 6-28 6-30 6-32 6-34 2' 3 f 6-38 6-40
6-42
' 4
/(*)
Y
-
(x)
K'(*)l
0-3246553 0-3241208 0-3235889
arg H"H( X )
Ho(*)
- 0-1813339 - 0-1780085 - 0-1745809 - 0-1710529 - 0-1674264 -
X
6-02 6-04
6.-06
+ + + + +
+ + + + +
0-2846430 0-2809893 0-2772356 0-2733835 0-2694349 0-2653917 0-2612556 0-2570287 0-2527128 0-2483100
0-2438221 0-2392513 0-2345996 0-2298691 0-2250617
03230596
0-3225328 0-3220087
298 299 301 302 303 304 305 306 307 309
45*98
6-o8 6-io
6-12 6-14 6-16 6-l8 6-20
6-22 6-24 6-26 6-28 6-3 6-32 6*34 6-36 6-38 6-40
0-1823646 0-1873484 0-1922411 0-1970413 0-2017472 0-2063574 0-2108705 0-2152848 0-2195991 0-2238120
0-2279222 0-2319283 0-2358292 0-2396237 0-2433106 0-2468888 0-2503573 0-2537151 0-2569612 0-2600946
0-2631145 0-2660201 0-2688106 0-2714853 0-2740434
03214870
0-3209679
03204513
Q-3I9937 1
+ + + + + + + + + +
310 14' 33*77 311 23' 31*88 32' 29*90 31 313 41' 27*85 314 5o' 25*73
03164059
0-3159109 0-3154182 0-3149278
315 59' 23*52 8' 21*24 317 318 17' 18*88 26' 16*44 319 320 35' 13*94
321 44' 11*36 322 53' 8*71 324 2' 5*98 325 11' 3*18 326 20' 0*31
3 2 z: 28 ; 57-38 328 37; 54*37 329 46' 51*29 330 55; 48'M 4' 44*93 332
+ + + + + + + + + +
+ + + +
03144396
o-3i39537 0-3134701
03129887
0-3125094
6-42
%\%
6-48 6-50
6-52
i 6-58 6-6o
6-62 6-64 6-66 6-68 6-70 6-72
PI
03120324
o-3i 15575
0-3110848 0-3106143
tit
6.58 6-6o 6-62 6-64 6-66 6-68 6-70
6-72 7 7t 6-78 6-8o
00560834
0-0506665 0-0452073 0-0397080
03101458
0-3096795 0-3092152
+ 0-2764843
0-2788074 0-2810122 0-2830981 0-2850647 0-2869117
o-'2886385
03087530
0-3082929
03078348 03073788
0-3069248 0-3064727 0-3060227 0-3055746
0-3051285
333 *3' 41*65 334 22' 38*30 335 3i' 34*88 336 40' 31*40 337 49' 27*86 338 58' 2 4 *2S 7' 20*58 340 341 16' 16*84 342 25' 13*04 343 34 9 ? i8
00341 7 1
0-0285990 0-0229940
6$ 6-78
6-8o
6-82 6-84 6-86 6-88 6-90
+ + + + +
+ + + + +
03046843
0-3042421 0-3038017 0-3033633
344 43' 5*26 345 52' 1*27 34Z 57*22 348 9'53'i2 349 18' 48*95
350 35i 352 353 355
27' 44*73
+ + + + + + + + + +
003433 1
0-0401386 0-0459497 0-0517624 0-0575743 0-0633830
+ + + + +
679
X
6-02 6-04 6-o6 6-o8 6-IO
6-12 6-14 6-16 6-i8 6-20
/iW
-
Y
-
(x)
\H \\*)\
0-3268325 0-3262804 o-32573ii 0-3251846 0-3246409
&rgH
213 214 215 216 217
(x)
H,W
+ + + + + + + + + + + + + + +
+ + + + +
+ + + +
0-4729337 0-4677800 0-4627160 o-457743i 0-4528629 0-4480769 0-4433866 0-4387935 0-4342988 0-4299040 0-4256104 0-4214192 0-4I733I7 0-4133490 0-4094724
0-4057028 0-4020415 0-3984894 0-3950474 0-3917166
X
6-02 6-04 6-o6
26' 56*12 35' 0*27 43' 4*69 51' 9*37 59' 14*31
608
6-io
6-12 6-14 6-16
6- 1
02423358
0-2376649 0-2329166 0-2280930 0-2231961 0-2182281 0-2131910 0-2080869 0-2029180 0-1976865
03240999
0-3235616 0-3230261
03224932
0-3219630 0-3214354 0-3209104 0-3203880 0-3198682 0-3193509
0-3188362 0-3183239 0-3178142 0-3173069 0-3168020
7' 19*50 219 220 15' 24*94 221 23' 30*64 222 31' 36*59 223 39' 4 2 *7 8
6-20
6-22 6-24 6-26 6-28 6-30 6-32 6-34 6-36 6-38 6-40
6-42
47' 49*21 55' 55"89 4' 2*8l 12' 9*97 20' 17*35
01923944
0-1870440 0-1816375
0-1761771 0-1706650 0-1651035 0-1594949 0-1538413 0-1481451 0-1424086 0-1366341 0-1308238 0-1249802
23O 28' 24*98 23I 36' 32"83 232 44' 4*9i 233 52' 49*22 235 0' 57*75
236
2 3Z
x
t$
6-48 6-50
6-52 6-54
652
6-
54
03138232
0-313335 0-3128490 0-3123653 0-3118839
0-3114047 0-3169277 0-3104529 0-3099804
2 +2
5!
3 ;J
+ + + + +
%1 600
6-62 6-64 6-66 6-68 6-70
0095342
0-0893469
0-2877269 0-2895883 0-2913310 0-2929548 0-2944593 0-2958444 0-2971098 0-2982554 0-2992811 0-3001869 0-3009727 0-3016385 0-3021846 0-3026109 0-3029176
0-3031051 0303 1 734
03095099
0-3090417
nl 3; 45*88 248 38' 56*96 249 47' 8*24 250 55' I9*7i 3' 3i*38 252
2
+ + + + +
+ + + + +
6-72 6-74
00833346
0-0773076 0-0712681 0-0652187
03085756
0-3081116 0-3076^97 0-3071899
0-3067322 0-3062765 0-3058229 0-30537I3 0-3049217
0-3044741 0-3040285
676 678
6-8o
6-82 6-84 6-86 6-88 6-90
6-92 6-94
253 11' 43*25 2 54 19 55*31 255 28' 7*56 256 36 20*01 257 44' 32*64
6-72
t%
6-78 6-8o 6-82 6-84 6-86 6-88 6-90
258 52' 45*45 260 0' 58*45 261 9 11*64 262 17' 25*01 263 25' 38*56
+ + + + +
+ + + + +
696
6-98 7-00
03031230 0302954
0-3026672
03035849
0-3031432 0-3027035
58' 34*53
6'
48*96
698
7-00
680
X
7'02 7-04 7-06 7-08 7-10
7-12 7.14 7-16 7-18 7-20 7-22 7-24 7-26 7-28 7*3 7-32 7-34 7-36 7-38 7-40
7-42 7'44 7-46 7-48 7-50
7*5^ 7'54 7 3 7-58 7 '60
'
;.w
+ + + + + + + + + + + + + + + + + + + +
+ + + + +
0-3001128 0-3000264 0-2998204 0-2994953 0-2990514 0-2984893 0-2970096 0-2970128 0-2960996 0-2950707
-
y.w
0-0199002 0-0138600 0-0078314 0-0018167 + 0-0041818
!?(*)!
arg# oM
(
HoW
+ + + + +
0-0691861 0-0749812 0-0807662 0-0865383 0-0922958
X
7-02 7-04 7-06 7-08 7-10 7-12 7-14 7-16 7-18 7-20
7-22 7-24 7.26 7-28 7-30
356 12' 22*74 357 21' 18*17 358 30' 13-54 8*86 359 39 360 48' 4*12
36i 56' 59*33 5' 3f3 14' 54*49 364 49*59 365 23' 44*64 366 32' 39*63
41' 34-57 3f7 368 50' 29*46 369 59' 24*30 371 8' 19*09 372 17' 13*83
+ + + + + + + + +
+ + + + + + + + + + + + + + +
+ + + + +
26'
+ 0-1539905 + 0-1593927
7-32 7'34 7-36 7-38 7-40 7-42 7-44 7-40 7-48 7-50 7-52 7'54 7-56 7-58 7-60 7-62 7 6 7 7-68 7.70
0-0961173 0-1015007 0-1068292 0-1121007 0-1173133 0-1224652 0-1275545 0-1325793 0-I375379 0-1424285 0-1472494 0-1519988 0-1566751 0-1612765 0-1658016
379 380
+ + + + +
384 55' 12*72 ' 6?88 386 4 387 13' 1*00 21' 388 55*07 389 30 49*09 39! 391 392 394 395
39' 43*07 48' 37*00 57' 30*89
6' 24*74
0-2058008 0-2106753 0-2154865 0-2202325 0-2249115 0-2295219 0-2340620 0-2385299 0-2429241 0-2472429
+ 0-2483717
+ 0-2450508 + 0-2416407 + 0-2381429
+ 0-2345591
+ + + + + + + + + +
4-
+ + + + +
15
18*54
+ + + + +
772 774
7-76 7-78 7-80
7-82 7-84 7-86 7-88 7.90 7-92 7-94 7-96 7-98 8-oo
0-2308910 0-2271400 0-2233081 0-2193967 0-2154078 0-2113430 0-2072042 0-2029932 0-1987118 0-1943618
0-1899452
+ 0-1702488 + 0-1746164
+ 0-1789029
+ 0-1831070 + 0-1872272 + + + + +
0-1912620 0-1952101 0-1990701 0-2028408 0-2065209
0-2101093 0-2136046 0-2170058 0-2203118 0-2235215
0-2868715 0-2865021 0-2861341 0-2857676 0-2854024 0-2850386 0-2846763 0-2843152 0-2839556 0-2835973 0-2832404
7'72 7'74
r7 6
7-78 7-80
7-82
59 46*92
8' 40*47 17' 33*98
+ + + + + + + + + +
74
+ + + +
01 854639
0-1809198 0-1763147 0-1716508
+ + + + +
02828848 02825305
0-2821776 0-2818259
681
X
7-02 7-04 7-06 7-08 7-10
7-12 7-14 7-16 7-18 7-20
/iW
+ + + + + + + + + + + + + + + + + + + + + + + + +
+ + + + +
0-0013241 0-0073134 0-0132828 0-0192302 0-0251533
-
(x)
i^i'wi
0-3022656 0-3018297 0-3013957 0-3009636
argtf,
270 271 272 2 73
(x)
H^*)
+ + + + + + + + + + + + + +
0-3466429 0-3470978 0-3476699 0-3483585 0-3491631 0-3500830 0-3511175 0-3522659 0-3535275 0-3549013
X
7-02 7-04 7-06 7-08 7-10 7-12 7-14 7-16 7-18 7-20
03005333
0-3001049 0-2996704 0-2992536 0-2988307 0-2984096
0-2979903 0-2975727 0-2971569 0-2967429 0-2963306
275 56' 19*08 2 7Z 4' 34"69 278 12' 50*45 279 21 ' 6*37 280 29' 22*45
281 37' 38-69 282 45' 55*09 283 54' 11*65 2' 28*36 285
286
287 288 289 290 291
10' 45*22
0-3563867 0-3579825 0-3596880 0-3615021 + 0-3634239 0-3654523 0-3675862 0-3698244 0-3721659 0-3746094
0-2959200
029551 12
0-2951041 0-2946986 0-2942948
+ + + + +
+ + + +
- 0-2705132 - 0-2678124
0' 29*54 293 294 8' 47*44 295 17' 5*48 296 25' 23*67 297 33' 42-00
+ 0-3771537
0-3797976 0-3825396 0-3853786 0-3883131
0-3913417 0-3944630 0-3976756 0-4009779 0-4043684
Tb%
7-58 7-60
7-62 7>6 i 7-66 7-68 7-70
01545636
0-1592138
02907345
0-2903469
298 42' 0-46 299 50' 19*07 300 58' 37*82 302- 6' 56*70 303 15' 15*72
304 305 306 307 308
23' 34*87 31 54*16
+ + + + +
0-4187811
+ 0-4225888
+ + + + +
0-4264750 0-4304379 0-4344758 0-4385870 0-4427696 0-4470219 0-4513419 0-4557279 0-4601780 0-4646902
0-4692627 0-4738934 0-4785806 0-4833221 0-4881160
7-72 7-74
776
7-78 7-80
7-82
+ + + + +
+ + + + + + + + + +
0-2880537 0-2876768
O2873OI4
0-2869274 0-2865549
5' 12*63 310 311 i 3 '32?58 312 21' 52*65 313 30' 12*85 3M 38' 33*17
Tb
55*71 20 16*66
11
+ + + + + + + + + +
7-82
32453 4 I *62
326
2'
3-16
682
X
8-02 8-04 8-o6 8-o8 8-io 8-12 8-14 8-i6 8-i8 8-20 8-22 8-24 8-26 8-28 8-30
7oW
+ + + + +
0-1669299 0*1621542 0-1573255 0-1524459 0-1475175
Y*{*)
iO)i
0-2814756 0-2811266 0-2807789 0-2804324 0-2800873
0-2797434 0-2794007 0-2790594 0^2787192 0-2783803
0-2780427 0-2777062 0-2773710 0-2770370 0-2767042
0-2763725 0-2760421 0-2757128 0-2753847 0-2750578
0-2747321 0-2744075 0-2740840 0-2737617 0-2734405
argH^W
413 414 4 I 5 417 418
37' 33*75 46' 26*87 55' I9"94
4' 12*98
HW
+ + + + +
0-3049098 0-3077342 0-3104602 0-3130870 0-3156137 0-3180394 0-3203635 0-3225852 0-3247036 0-3267183
X
8-02 8-04 8-o6 8-o8 8-io
8-12 8-14 8-i6 8-i8 8-20
+ + + +
4-
13'
5*98
0-2479207
0-2522101 0-2541963 0-2560762 0-2578492 0-2595150
419 21' 58^95 420 30' .5 1*87 4 2I 39/ 44*76 422 48' 37*62
423 57' 30*43
+ 0-2501180 + + + + + + + + + + + + + + +
4-
+ + + + + + + + + + + + + + +
+ + + + +
+ + + + +
4-
41' 53*99
50' 46*59 59' 39*i6
8'
8-32
8-36 8- 3 8 8-40
8-42 8-44 8-46 8-48 8-50 8-52 5i 8-56 8-58 8-6o
+ + + +
4-
31*70
S 8-38 8-40
8-42 8-44 8-46 8-48 8-50
+ 0-0637345
0-0582992 + 0-0528534 + 0-0473994 + 0-0419393
436 35' 9 -ii 437 44 i;5J 438 52' 53-88 46*22 r 440 441 io' 38*53
03432414
o-3437576 0-3441653 0-3444644 0-3446550 o-3447373 0-3447114 o-3445775 0-3443357 0-3439865 o-343530i 0-3429669 0-3422972
+ + + + +
+ + + +
4-
02710333
0-2712837 0-2714251 o-27i4577 0-2713818 0-2711977 0-2709056 0-2705060 0-2699992 0-2693857
0-^678405
46
+ + + + +
+ + + + +
44-
8-52 8-54
I'
56
8
,
%1 8-6o
8-62 8-64 8-66 8-68 8-70
+ 0-0091717 + OOO37293
- 0-OOI70I9 - 0-007I200 - 0-OI25227
0-2715368
3'
51*69
870
872
74
+ + + +
4-
02712233
0-2709109 0-2705996 0-2702894
0-2699803 0-2696722 0-2693651 0-2690591 0-2687541
12' 43*77
876 878
8-8o
8-82 8-84 8-86 8-88 8-90
8-92 8-94 8-96 8-98 9-00
+ 0-2686660
4-
458
23' 39*35
47*5i
872
8-76 8-78 8-8o
8-82 8-84 8-86 8-88 8-90 8-92 8-94 8-96 8-98 9-00
0-2621493
+ 0-2607030
+ 0-2591558
O0652532
0-0703522 0-0754116 0-0804295 0-0854042 0-0903336
459 32' 31*17 460 41' 22*95 461 50' 14*71 462 59' 6*44 464 7' 58-14
+ + + +
4-
+ + + + +
+ + + + +
683
X
8-02 8-04 8-o6 8-o8 8-io
7iW
+ + + + + + + + + +
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
0-2374329 0-2401291 0-2427241 0-2452173 0-2476078
-
(x)
HCtol
0-2825517 0-2821961 0-2818419 0-2814889 0-2811373
0-2807871 0-2804381 0-2800905 0-2797441 0-2793991
aig^Jw
327 328 329 330 33i
10' 18' 27' 35' 43'
HiW
+ + + + + + + + + + + + + + + + + + + + + + + +
+ + + + +
0-4929604 0-4978531 0-5027922 0-5077756 0-5128012
0-5178671 0-5229711 0-5281111 0-5332850 0-5384907
X
8-02 8-04
806
8-o8 8-io
812
8-14 8-i6 8-i8 8-20
8-22 8-24 8-26 8-28 8-30
8.32 ' 3 4
332 52' 14*79 0' 37*12 334 335 8' 59*56 17' 22*11 336 337 25' 44 T 78
338 339 340 o 341 343 344 345 346 34 Z 348
34' 7 r55 42 3o r 42 50 53 ? 40 59' 16*49 7' 39*69
16' 2*99 24' 26*39 32' 49 ?90
812
8-14 8-i6 8-i8 8-20
8-22 8-24 8-26 8-28 8-30 8-32 '34 8-36 8-38 8-40 8-42
Pi 8.38
8-40
8-42
00589319 00534845
0-0480290 0-0425676
^l* 49 37*22
1*03
*K
00371023 00316333
0-0207046
- 0-0261687
349 58' 351 6; 352 14 353 23' , 354 3i 355 356 357 359 360
%\%
8- 8 4 8-50
0-2732640 0-2732981 0-2732244 0-2730432 0-2727548 0-2723596 0-2718580 0-2712504 0-2705372 0-2697190
0-2687964 0-2677699 0-2666402 0-2654079 0-2640737
00152452
8-52
tu
8-58 8-6o
8-62 8-64 8-66 8-68 8-70 8-72 8*74 8-76 8-78 8-8o
8-82 8-84 8-86 8-88 8-90
+ + + + + + + + + + + + + + +
00065038
0-0119084 0-0172953 0-0226640 0-0280110
361 22' 4*52 362 30' 29*39 363 38' 54*35 364 47' 19*41 365 55 44*57
+ + + + + + + + + +
+ + + + + + + + + +
872
74 8-76 8-78 8-8o
8-82 8-84 8-86 8-88 8-90
0033334 6 0-0386328
00439037
0-0491453
02702305
0-2699196 0-2696098
0-2693011 0-2689934 0-2686868 0-2683813 0-2680768
00543556
0-0595326 0-0646744 0-0697790 0-0748447 0-0798694
0-0848513 0-0897886 0-0946795 0-0995220 0-1043146
'
9*81
+ 0-262638^
+ 0-2611028 + 0-2594677
+ 0-2577339 + 0-2559024 + + + + +
0-2539740 0-2519497 0-2498306 0-2476176 0-2453118
20'
0*98
8-92
i* 9 4
+ + + + +
8-92
i* 896
9*
2' 12*41
8-98 9-6o
684
X
9-02 9-04 9-06 9-08 9-10
9-12 9-14 9-16 9-18 9-20
/oW
-
Y*(*)
l?(*)l.
Mgfl^*)
4?I o 472 473 474 475
j/
HoW
+ + + + +
+ + + + +
0-3175904 0-3152111 0-3127393 0-3101761 0-3075226 0-3047800 0-3019495 0-2990324 0-2960300 0-2929435
0-2897743 0-2865238 0-2831934 0-2797845 0-2762985
X
9-02
+ + + + + + + + + + + + + + +
7 75
<
904
9-06 9-08 9-10
9-12
0-2643003
47645'25;oi
477*54 16*39
479
3'
914
9-16 9-18 9-20
9-22 9-24 9-26 9-28 9-30
7*73
02631495
0-2628641
480 11' 59*05 481 20' 50*35 482 29' 41*62 483 38' 32*86 484 47 24*08 485 56 15-28 487 5' 6*44
488 489 490 491 492
13' 57*59 22' 48*70 31 '39*80 40' 30*87 49' 21*92
930
9.32
+ + + + + + + + + + + + + + +
+ + + + +
934
9-36 9-38 9-40
9-42 9-44 9-46 9.48 9*5
+ 0-2051510 + 0-2016573
+ 0-1980905 + 0-1944522 + 0-1907439 + + + + + + + + + +
0-1869673 0-1831240 0-1792157 0-1752440 0-1712106
932
9-34 9.36 9-38 9-40 9-42 9-44 9-46
0-2597850
02595104
0-2592367
496i5'54'92
02589638
0-2586918
948
9-50
952
9-54 9'56 9-58 9-60
9-62 9-64 9-66 9-68 9.70 9.72 9-74 9-76
49942'27;70 500 51' 18-58 502 0' 9*44 503 9' 0*28 504 17' 51*08
505 5o6 507 508
26; 41*87 35' 32*65 44' 23*40
952 954
9-50 9-58 9-60
9-62
+ + + + + + + + + + + + + + + + + + + +
+ + + + + + + + + +
+ + + + +
964
9-66 9-68 9.70
9-72 9-74 9-76 9-78 9-80 9-82 9-84 9-86 9-88 9.90
9.92
978
9-80
9-82 9-84 9-86 9-88 9.90
02552318
0-2549714 0-2547117
0099753s
0-0949498 0-0901178 0-0852597 0-0803773
02544529
0-2541948
5^:55; 8*57
02539375
0-2536810 0-2534253
01 590296
0-1540700 0-1490847 0-1440757 0-1390449 01 339943 0-1289259 0-1238417 01 187437
9.92
994
9-96 9-98 io-oo
02531703
0-2529161 0-2526626 0-2524100 0-2521580
02459358
0055671
522 39'2I*II 523 48' 11-56 524 57' 1*99 526 5 52*39 / 52 7 i 4 42*78
+ + + + +
994
9-96 9-98 io-oo
685
AW
9-02
Yt(*)
\rf
M\
arg
H l\x)
{
H,W
+ + + + +
+ + + + +
0-7532504 0-7579051 0-7625048
9-02
904
9-06
908
9-10
9-12 9-14 9-16 9-18 9-20
9. 22
+ + + + + + + + + + + + + + +
+ + + + +
384 io' 38*94 3|5 19' 5-56 386 27' 32*26 387^35'59'o 4 388 44' 25*90
389 52' 52*84 391 i' 19-86 392 9' 46 -95 393 18' 14*12 394 26' 41*37
904
9-06
07670477 07715322
908
9-10
9-12 9-14 9-16 9-18 9-20
9-22 9-24 9-26 9-28 9-30 9.32 9-34 9-36 9-38 9-40
+ 0-1319221 + 0-1363164
4-
0-1406474
+ 0-1449133 + 0-1491128 + + + + +
o- 1532443
0-2141618 0-2108375 0-2074370 0-2039620 0-2004139 0-1967943 0-1931047 0-1893468 0-1855221 0-1816322
395 35' 8*70 396 43' 36*11 397 52' 3*59 399 o' 31*15 400 8' 58*78
401 402 403 404 4 05
17' 26*49 25' 54*27
+ + + + + + + + + +
+ + + + + + + + + + + + + + +
+ + + + +
07971234
0-8011549 0-8051156 0-8090043 0-8128195
0-8165598 0-8202240 0-8238107 0-8273187 0-8307469
932
9'34 9-36 9-38 9'4
9-42 9-44
+ + + + +
+ + + + + + + + + + + + + + +
+ + + + + + + + + + + + + + +
+ 0-1728314 + 0-1765251
+ 0-1801413 + 0:1836785
+ 0-1871357
946 948
9'5
+ + + + + + + + + + + + + + + + + + + +
+ + + + + + + + +
406 59' 46*14 408 8' 14*28 409 16' 42*50 410 25' 10*79 4" 33' 39-15
412 42' 7*58 413 5o' 36*08 416
9H2
9-44 9-46 9-48 9-50
9-52
9*5j
952
9-54 9-50 9-58 9-60
9-62 9-64 9-66 9-68
970
9.72 9-74
4^57*49
976 978
9-80
9-82
972
9-74 9-76
978
9-80
9-82 9-8 4 9-86 9-88 9-90
9 'h
02543235
0-2540646
0-2538064 0-2535491 0-2532925 0-2530367 0-2527816
429 49' 22*16 52*66 6' 21*22 432 433i4'5o"8 4 434 23' 20*52
43<> 57'
+ + + + + + + + + +
08864944
0-8876322
9-94
996
9.98 IO-OO
686
X
10-02 10-04 IO-OO 10-08 IO-IO IO-I2 10-14 IO-IO 10-18 10-20
IO-22
/.(*)
Y.ix)
!?(*)
arg H^(x)
H(*)
X
10-02 10-04 10-06 10-08 IO-IO IO-I2 10-14 10-16
+ 0-0506850 + 0-0456806
4-
0-0406838
5304i'i 3 ^2
53i 50' 4'i3 532 58 54'4i
+ 0-1136338 + 0-1085142
0-1033867 0-0982533 + 0-093II62
4-
4-
4-
1018
10-20
IO-22 10-24 IO-26 IO-28 IO-30 IO-32
IC24
IO-2D IO-28
IO3O
IO-32 10-34 10-36
0-2494350 0-2491536 0-2487732 0-2482942 0-2477108 0-2470416 0-2462690 0-2453994 0-2444335 0-2433718
0-2422148 0-2409633 0-2396178 0-2381792 0-2366482
4-
0-2494357
4-
02491925
0-2489501 0-2487084 0-2484674
00192978
0-0242303 0-0291435
00340355
0-0389045 0-0437486
1038
10-40
10-42 10-44 10-46 10-48 10-50
36' 6*31
OO369I92
0-0485659
00533546
0-0581128
02467995
0-2465640 0-2463291 0-2460949
0-2458614 0-2456285 0-2453963 0-2451648 0-2449339
0-2447037 0-2444741 0-2442451 0-2440168 0-2437891
0-2435621 0-2433357 0-2431099 0-2428847 0-2426602
00628386
0-0675304
55i2o' 16*43 552 29' 6*41 553 37 5<>"36 554* 46' 46*30 555 55 36 ? 22
557 558 559 560 561 562 563 565 566 567
1050
10-52 10-54 10-56 10-58 IO-OO
10-62 10-64 io-66 io-68 10-70
1052
10-54 10-50 10-58
1060
10-62 10-64 io-66
4'26?i2
13 16*00 22' 5*87 30' 55*72 39' 45*55
48' 35*37
1068
10-70
10-72 10-74 10-76 10-78 io-8o
10-82 10-84
- 0-0354157 - 0-0399295
- OO443948 - 0-0488103
- 0-0531741
01326384
0-1365537
568 32' 44*20 569 41' 33*91 570 50' 23*61 57 I 59'i3 ? 30 573 8' 2*97
574 575 576 577
16' 52*62 25; 42 -26
1086
io-88 10-90
10-92 10-94 10-96 10-98 II'OO
01973650
Q-I943357 0-1912343 0-1880622
01 404073
0-1441977 0-1479234 0-1515832
0-1551757 0-1586996 0-1621537 0-1655367 0-1688473
0-2424363 0-2422130
02419903
0-2417683 0-2415468
57852'n*07
580
581
1084
io-86 io-88 10-90
10-92 10-94 10-96 10-98 II-OO
0*64
OO973349
0-1009630 0-I045209 0-1080073 0-III42IO
687
7iW
10-02 10-04 IO-OO 10-08 IO-IO
IO-I2 10-14 IO-IO 10-18 10-20 IO-22 IO-24 IO-26 IO-28
^iM
+ + + + +
0-2495809 0-2500465 0-2504122 0-2506782 0-2508444
0-2509111 0-2508783 0-2507464 0-2505154 0-2501858
!<(*)!
0-2525274 0-2522739 0-2520212 0-2517692 0-2515180
arg
7w
+ + + + +
-1-
H,M
0-8923738 0-8928155 0-8931574 0-8933996 0-8935423
[0-02 [0-04 to-oo 10-08 [O-IO
+ + + + +
00234135 00183955
44ii4'i9'93 442 22' 50*04 443 31' 20*21 44439'5o'44 445 48' 20*73
446 56' 51*08 448 5' 21-49 449 13, 51 "95 450 22 22*48
+ + + + +
4-
08935856
0-8935295 0-8933744 0-8931204 0-8927679 0-8923172 0-8917685 0-8911224
45i3o'53 ?o6
452 453 454 456 457
39' 23*69 47' 54*39 56' 25*14
13' 26*80
+ + + +
+ + + + +
IO3O
IO-32 10-34 10-36
+ 0-2497578
0-2492319 + 0-2486082 + 0-2478874 + 0-2470699
02495353
0-2492907 0-2490469
0-2488038 0-2485614 0-2483197 0-2480787 0-2478385
4,55*94
08903793
0-8895395
- 0-0362001 - 0-0410586
1038
10-40
10-42 10-44 10-46 10-48
+ + + + + + + + + +
+ + + +
45 8 21' 57*72 459 30' 28-69 460 38' 59*71 461 47' 30*79 462 56' 1*93 464 4'33 ? i2 465 13 4*36 466 21 35*65 4673o' 7-00 468 38' 38*40 469 47o 472 473 474
47' 9 r 8 5 55' 4i*35 4' 12*91 12 44*52 21' 16*18
+ + + + +
+ + + + + + + + + +
1050
10-52 10-54 10-56 10-58 10-60
10-62 10-64 io-6o io-68 10-70 10-72
0-2475989 0-2473600 0-2471218 0-2468843 0-2466475 0-2464114 0-2461760 0-2459412 0-2457071 0-2454736
00879333
0-0924107 0-0960431 0-1012287
+ + + + +
+ + + + + + + + + +
475 29' 47*89 476 38 19-65 477 46 51*46 55 23*32 *lK 4o 3 '55'23
481 12' 27-19 482 20'59-20 4 8 329'3i T 25 44 38 3 T 36 48546'35 ? 5i
55; 7-71 4fto 488 3 39-96 489 12 12-26 490 20' 44*61 491 29' 17*00
+ 0-8634870
+ o-86li 1 10 + 0-8586529
+ 0-8561136 + 0-8534944
+ + + + +
0-8507964 0-8480203 0-8451689 0-8422418 0-8392408
1074
10-76 IO-78 io-oo
10-82 10-84 io-86 io-88
01 383343
0-1421666
02431755
0-2429492 0-2427236 0-2424986 0-2422742 0-2420505 0-2418273 0-2416048 0-2413829 0-2411616 0-2409410
[0-72 [o-74 10-76 [0-78 co-8o CO-82 [0-84 to-86 co-88 [0-90
1090
10-92 10-94 10-96 10-98 ii-oo
01459354
o- 1496394 0-1532774 0-1568479 0-1603497
+ 0-8231749 + + + + +
0-8197592 0-8162792 0-8127366 0-8091327 0-8054691
+ + + + +
492 37' 49 -43 493 46 21*92 49454'54 ?45 496 3 27*03 497 11 59 T 65
10-92 10-94
[096
[0-98 :i-oo
688
X
-
7.W
0-1676238 0-1639968 0-1603109 0-1565675 0-1527683
-
nw
0-1720845 0-1752470 0-1783338 0-1813437 0-1842758
Ifl^WI
0-2402308 0-2400135 0-2397968 0-2395807 0-2393652
arg H(x)
HoW
-
II-02
5^45,' 8*29
57-77 5fto 53 588 2 47*23 589 11' 36^69 5902o' 26*13
WZt
n-o8
II-IO
11-12 11-14 ii-i6 ii-i8 II-20
II-02
\\&
11-08 II-IO 11-12
02391503
0-2389359 0-2387222 0-2385090 0-2382963
0-2380843 0-2378728 0-2376618 0-2374514 0-2372416 0-2370323 0-2368236 0-2366155 0-2364078 0-2362008
1130
11-32
22*44 11*78
1*10 50*41
-24
39*70
01531801
0-1552926 0-1573169
\m
1 1 -38
4o
57' 28*98
6' 18*25 15' 7*51 23' 56*75
01592522
0-1610982 0-1628541
%%
11-42
11-38 II-40
H-42
\\.\\ 11.48 11-50
11-52
0-2198634 0-2359942 0-2213425 -0-2357882 0-2227306 0-2355828 0-2240273 0-2353779 0-2252321 0-2351735
n-44
11-46
1 1
-48
1150
n-52
\m
11-58
n-6o
11-62
614 25' 41*04 615 34' 30*17 616 43' 19*28 6i 7 52' 8*39 619 0' 57*48
mt
11-58
n-6o
11-62 11-64 11-66
w-it u-68
11-70
11-72 11-74 11-76 11-78 II-OO
02315362
0-2319050
C2321806
0-2323620 0-2324526 0-2324481 0-2323513 0-2321618
0-2318798 0-2315056 0-2310396 0-2304820 0-2298332
02331584 02329598
0-2327616
n-68
11-70
11-72
02325639
0-2323668 0-2321701
62553'5i"76 2 40*77 627 628 11' 29*75 629 20' 18*73 63O 29' 7*70
631 37' 56-65 632 46 45*59 6 33 55 34*52 635 4 23*44 636 13' 12*34
n-74
11-76
\\t
11-82
11-82
Hit
n-88
1 1 -90
+ + + + +
-+
wit
n-88
11-90
11-92
11-92
+ 0-0296200
0-0341710 + 0-0387007 + 0-0432074 + 0-0476893
n-94
1 1
-96
11-98 I2-00
02308073
0-2306146
02304223
0-2302306
W-tt
11-98 12-00
689
X
II-02 -
7iW
0-1798496 0-1828371 0-1857467 0-1885774 0-1913283
0-1939984 0-1965863 0-1990926 0-2015150 0-2038531
0-2061063 0-2082738 0-2103549 0-2123488 0-2142550
i\0)
+ + + + +
o- 1 60002
\H
(x)\
1
arg
H
20' 29' 37' 46' 54'
(x)
HW
X
\1%
11
08
II-IO
11-12
+ + + + +
I I 1 1
-02
04
n-o6 n-o8
II'IO
11-12
III4 IIl6
11-18 II-20
11-22 II-2 4 II-26 11-28 II-30
+ + + +
+
02394131
0-2391972 0-2389819 0-2387671
3'i6*34
ii' 49^27
+ + + + +
+ + + + + + + + + +
+ + + + + + + + + +
11-26 11-28
11-32
n-34
11-36 11-38
1140
11-42 11-44 11-46
11
515 28' 47*65 516 37' 21*01 ? 5 I Z45'54 42 518 54 520 3' 1*36
n-32 n-34
11.36 11-38 II- 4 o
27%
-48
1150
11-52
+ + + + + + + + + +
+ + + + +
521
n'34*89
52220' 8*46 523 28' 42*07 524 37' 15*72 525 45 49"4i
526 54' 23*14 2' 56*91 528 529 11' 30*72 530 20' 4*58 531 28' 38*47
+ + + + + + + + + +
+ + + + + + + + + + + + + + + + + + + +
1142
11-44 11-46
1 1
-48
11-50
11-52
n-54
1 1 -56
02301055
0-2308300
1158
ii-6o
02314617
0*2320005 0-2324463 0-2327992 0-2330591 0-2332261 0-2333002 0-2332817 0-2331707 0-2329674 0-2326720 0-2322847
0-2354086 0-2352034 0-2349987 0-2347946 o- 2 3459io 0-2343879 0-2341854 0-2339833 0-2337818 0-2335809
0-2333804 0-2331804 0-2329810 0-2327821 0-2325837 0-2323858 0-2321884 0-2319915 0-2317951 0-2315993
n-54
11-56
n-58 n-6o
11-62 ll-6 4
1162
\!U
n -68
11-70
11-72
532 37' 12*39 533 45' 46-3 534 '54 20*37 536 2 54*41 537 ii' 28*49 538 20' 2*61 539 28' 36*77 540 37' 10*96 54i45'45'i9 542 54 I9 ?46
544 2'53'77 545 11 28*11 5462o' 2*49 547 28' 36*91 548 37' 11*36 549 45' 45-85 550 54, 20*37 552 2 54*93 553 1 1' 29*53 554 20 4*16
n-66 H-68
11-70
11-72 11-74 11-76 11-78 II-OO
11-82 Ii.84
"'%
+ 0-0067849 + 0-0021209
- 0-0025199 - 0-0071598 - 0-0117890
-
ml
11-82
wit
n-88
11
90
0-0164057 0-02IOOOI 0-0255944 0-0301628 0-0347115 0-0392388 0-0437430 0-0482223 0-0526749 0-0570992
1192
l!U
11-98 I2'00
- 0-2280528 - 0-2270334
02314039
0-2312090 0-2310146 0-2308207 0-2306273
n-94
11-96 11-98 I2-00
690
X
12-02 12-04 I2-00 12-08 I2-IO 12-12 12-14 I2-I6 12-18 12-20
12-22 12-24 12-26 12-28 12-30 12-32 12-34 12-36 12-38 12-40
;.w
+ + + + + + + + + +
0-0521447 0-0565718 0-0609690 0-0653346 0-0696668
-
(x)
i^Cwi
02300393
0-2298485 0-2296581 0-2294682 0-2292788
axgH
(x)
HoW
-
X
12-02 12-04 12-06 12-08 I2-IO
0-2240513 0-2227778
02214173
0-2199706 0-2184384 0-2168214 0-2151204
643 6' 5*52 644 M' 54-35 645 23 43-16 646 32' 31*96 647 41' 20*74
02133362
0-2114698 0-2095218 0-2074933 0-2053852 0-2031985 0-2009341 0-1985931
0-1961765 0-1936855 0-1911210 0-1884844 0-1857766 0-1801527 0-1772390 0-1742591 0-1712143
6485o' 9*52 649 58' 58*29 651 7'47'-4 652 16' 35-78 653 25' 24^51
654 34' 13 -23 655 43 i*94 656 51' 50*64 658 o' 39*32 659 9' 28-00 66o 18' 16*67 66i27' 5-32 662 35' 53*96 663 44' 42*60 664 53' 31*22
666'
2' 19-83
12-12 I2-IA 12-16 12-18 12-20 12-22 12-24 12-26 12-28 I2-30 12-32 12-34 12-36 12-38 12-40
+ + + + +
+ + + + + + + + + +
+ + + + + + + + + +
- 0-1829990
667ii' 8*43
668 19' 57*02 669 28' 45*60 7037'34* I 7
67I 672 6 74 675
46' 22*73 55' 11*28
3' 59 -82 12' 48*35 21 '36*86
01320567
12-42 12-44 12-46 12-48 12-50 12-52 12-54 12-56 12-58 12-60
12-62 12-64 12-66 12-68 12-70
12-72 12-74 12-76 12-78 12-80
01533737
0-1565195 0-1595977 0-1626073
0-1655471 0-1684160 0-1712131 0-1739374 0-1765879
o-io8ni2
0-1048412 0-1015142 0-0981318 0-0946953 0-0912064 0-0876664
676
678 679 680 682 683 684 68 5 686 687
02231054
0-2229313
0-1969764 + 0-1988424
0-200626I
0-2\023269
688 58' 29*91 690 7' 18-31 691 16' 6*70 692 24' 55*08 69333'43''45
- 0-0654413 - 0-0615076 - 0-0576954 - 0-0537665 0-0498024 - 0-0458049 - 0-0417755 - 0-0377160 - 0-0336280 - 00295 1 33
12-82 12-84 12-86 12-88 12-90 12-92 12-94 12-96 12-98 13-00
1300
+ + + + +
- 00947977
17' 45*17
691
X
12-02 I2-04 12-00 12-08 I2-IO 12-12 12-14 12-16 12-18 12-20 12-22 12-24 12-26 12-28 I2-30 12-32 12-34
Ji(*)
Y1 (x)
-
I^WI
0-2304343 0-2302419 0-2300499 0-2298584 0-2296674
arg
^V)
+ + + + + + + + + + + + + + + + + + + +
H^*)
0-5794489 0-5750722 0-5707290 0-5664209 0-5621495
X
12
555 28' 38*83 556 37' 13*53 557 45 48-27 558 54 23*04 560 2' 57*84
561 11' 32*68 5622o' 7 55 563 28' 42*46 56437'i7 ? 40 565 45' 52^38
"-
02
12-04 12-06
1208
I2-IO
12-12 12-14 I2-IO 12-18 12-20 12-22 12-24 12-26 12-28 I2-30 I2-32
02139578
0-2120846 0-2101303 * X 0-2080958 0-2059820
"*
00912733
0-0953682 0-0994184
0-5579164 0-5537234 0-5495718 0-5454634 0-5413995 0-5373819 0-5334119 0-5294911 0-5256209 0-5218027
0-5180381 0-5143282 0-5106746 0-5070786 0-5035415
569n'37-50
570 20' 12*61 57i28' 4 7?75
I236
12-38 12-40
12-42 12-44 12-46 12-48
57237'22"93
573 574 576 577 45 58-13 54 33/37 3' 8*64 11 43*95
20' 19*28 28' 54*65
1234
12-36 12-38 12-40
12-42 12-44 12-46 12-48 12-50 12-52 I2 \54 12-56 12-58 12-60 12-62 12-64 12-66 12-68 12-70
1250
12-52 12-54 12-56 12-58 12-60
46
+ + + + +
+ + + + + + + + + + + + + + + + + + + + + + + + +
01522036
0-1487423
12-62
12
64
02241505
0-2239734 0-2237967 0-2236204 0-2234446 0-2232692
0-2230942 0-2229196 0-2227454 0-2225716 0-2223982
595 29' 13*04 48*86 596 37' ' 59 Zo 46 2 4*72 59855' 0*60 600 3' 36*51 6oi 12' 12*45 602 20' 48*43 603 29' 24*43
o-4559337 0-4535740
04512923
0-4490893 0-4469659
1280
12-82 12-84 12-86 12-88 12-90
01955054
0-1974561
- 00953513 - 0-0912483
-
01993253
0-2011125 0-2028170
0-2044382 0-2059758 0-2074291 0-2087978 0-2100814
12-92
1294
12-96 12-98 13-00
12-92
1294
12-96 12-98
04302435
1300
692
AW
+ + + + + + + + + + + + + +
+ + + + +
0-2082899 0-2095684 0-2107612 0-2118679 0-2128882
-
(x)
i*C(*)i
arg
(x)
H(*)
-
00253735
0-0212104 0-0170257 0-0128211 0-0085984
1302
13-04 13-06 13-08 13-10
13-12 13-14 13-16 13-18 13-2 13-22
1320
1322 1324 1326 1328 1330
13-32
0-0526132 0-0482867
00439475
0-0395973 0-0352379
7H 54' 36"i4
713 3 24*35 714 12 12*56 7i5 2i' 0*76 716 29*48*96 717 718 719 721 722
38' 47' 56' 5 13'
+ + + + +
+ + + + + + + + + +
0-0170219 0-0213229 0-0256282 0-0299359 0-0342443 0-0385517 0-0428564 0-0471565 0-0514504 0-0557363 0-0600126 0-0642775 0-0605292 0-0727662 0-0769866
1324
13-26 13-28 13-3
I3-32 13-34
!334 1336
I3-38 13-40
1336
I3-38 13-40
49*78
1342
J3-44 13-46 I3-48 13-5
+ + + + +
+ 0-OI28262
+ 0-0171605
+ 0-0214816 + 0-0257876 + 0-0300770 + + + + +
0-0343480 0-0385990 0-0428282
1342
I3-44 I3-46
49 2*29 72757'5o"39
729 6' 38*49 730 15' 26*58 731 24' 14*66 732 33' 2*73 733 41 '50*80
734 735 737 738 739
50' 38"86
1348
I3-50
+ 0-2141858
+ 0-2132981
+ 0-2123263 + 0-2112712 + 0-2101332
OO47034I
0-05I2I50
0-0553693 0-0594954 0-0635916 0-0676565 0-0716883
+ + + + + + + + + +
+ + + + +
0-0811889 0-0853714
1352
13-54 I3-56 I3-58 13-60 13-62 13-64 13-66 13-68 13-70
13-72 13-74 13-76 I3-78 13-80 13-82 13-84
00895324
0-0936702 0-0977832 0-1018698 0-1059283 o- 1099573 01 139550 0-1179199
0-1218505 0-1257452 0-1296025 0-1334209 0-1371989
1364
13-66 13-68
1370
13-72
+ + + + +
+ + + + +
1374
13-76
+ + + + +
01943356
+ + + + + + + + + + + + + +
02140945
0-2139407
0-2137871 0-2136339 0-2134010
+ + + + +
01 409350
0-1446278 0-1482758 0-1518777 0-1554320
6 ^f 13-88 1390
,
1396 1398
14-00
+ + + + +
01737085
0-1710735
02133284
0-2131762
2'
38*76
1392
13-94
37 50*45
+ 0-1691466 + 0-1724429
1396 1398
14-00
7 3
693
/iW
-
Y
-
(x)
itf'l'wi
axgH^x)
612 38' 13*45
U
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
(x)
1304
13-06 13-08 13-10
13-12 13-14 13-16 13-18 13-20
6i346'49 ? 7i
614 55' 25*99 616 4' 2*30 1 2' 38*63 61
00445140
0-0401645
00358056
0-0314391 0-0270667
618 21' 14*99 6i929 , 5i*38 620 38' 27*80 621 47' 4*24 622 55' 40*71 624 4' 17*21 625 12' 53-73 626 21' 30*28 6273o' 6*85 628 38' 43*45 629 630 632 633 634
47' 20*08 55' 56 ? 73 4 33 ? 4i 13 10*12 21 '46*85
1320
1322
13-24
1322
13-24 13-26
1328
13-30
00095532
- 0-0051775
1326
13-28 13-30
+ + + +
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
0-0165990
0-0209219 0-0252301 0-0295218 0-0337954 0-0380493
1332
13-34
1336
I3-38 13-40
1350
13-52 13-54 13-56 I3-58 13-60
635 30' 23*60 636 39' 0*38 6 3747'37* I 9 638 56' 14*02 640 4' 50*87
6 4 I I 3'27 ? 75 64222' 4*66
0-4230776 0-4236901 0-4243876 0-4251699 0-4260365 0-4269869 0-4280206 0-4291371 0-4303358 0-4316161
0-4329775 0-4344191 0'4359404 0-4375406 0-4392190
1342
13-44 I3-46 I3-48
1350
1352
13-54
6433o'4i-59
644 39' 18*54 645 47 55*52
1356
I3-58 13-60 13-62 13*64 13-66 13-68
1362
13-64 13-66 13-68 13-70 13-72 13-74 13-76 13-78 13-80
13-82 13-84 I3 1f I3'88 13-90
1370
1372
13-74 13-76 I3-78
02149936
0-2148374 0-2146815 0-2145260 0-2143708 0-2142159 0-2140614 0-2139072
652 39' 37*89 653 48' 15*04 654 56 52*21 656 5 29*41 657i 4 ' 6*62 658 659 66o 661 662
664 665 666 667 668
22' 43*86
1380
13-82 13-84 13-86 13-88
+ 0-4508850 + 0-4530871
+ o-45536o3 + 0-4577036 + 0-4601160
1390
1392
13-94
1392
13-94
1396 1398
14-00
OI333752
+ 0-4625965
+ + + +
0-4651439 0-4677571 0'4704350 0-4731766
02137533 02135998
0-2134466
694
X
14-02 14-04 14-06 14-08 14-10
I 4 -I2
;(*)
Y
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
+ + + + + + + + +
(x)
i^I'wi
02130243
0-2128727 0-2127214 0-2125705 0-2124198
0-2122695 0-2121195 0-2119699 0-2118205 0-2116715
axgH
757 758 760 761 762
(x)
Ho(*)
4-
X
14-02 14-04 14-06 14-08 14-10
I 4 -I2
+ + + + +
+ + + + +
0-1304974 0-1337454 0-1369354 0-1400660 0-1431362 0-1461449 0-1490909 0-1519731 0-1547905 0-1575421 0-1602268 0-1628437 0-1653919 0-1678704 0-1702783
0-1726147 0-1748790 0-1770701 0-1791875 0-1812302 0-1831977 0-1850893 0-1869042 0-1886420 0-1903019
0-1918835 0-1933862 0-1948095 0-1961530 0-1974163
46' 38^36
0-1756839
763 30' 37*76 764 39' 25*62 765 48' 13*47 76657' 1*32 768 5' 49-16
+ + + + + + + + + + + + + + +
+ + + + +
1424
14-26 14-28
1430
14-32
+ + + + + + + + + + + + + + + + + + + +
+ + + + +
0-1380914 0-1347629 0-1313851 0-1279596 0-1244877 0-1209709 0-1174107 o- 1 1 3808 s 0-1101658 0-1064841 0-1027650 0-0990100 0-0952206 0-0913984 0-0875449 0-0836617 0-0797504 0-0758127 0-0718500 0-0678641
0-0638565 0-0598280 6-0557827 0-0517198 0-0476418 0-0435503 0-0394470 0-0353334 0-0312113 0-0270823
0-0229481 0-0188102 0-0146704 0-0105303 0-0063915
0-2115227 0-2113743 0-2112262 0-2110784 0-2109310 0-2107838 0-2106369 0-2104904 0-2103441 0-2101982
0-2100525 0-2099072 0-2097621 0-2096174 0-2094729
769 14' 36*99 770 23' 24*81 771 32' 12*63 77 2 4i' 0*44 773 49 48*24
774 770 777 778 779
58' 36*03 7 23*82
J434
14-36 I4-38 14-40
14-42 14-44 14-46 14-48 14-50 I4-52 14-54 14-56 I4-58 14-60 14-62 14-64 14-66 14-68 14-70 14-72
16 11*60
24' 59*38
33 47*15
?K ?K 785
o;
8 58 6 , 7' 45 ll -89 1
786 26' 33*61 787 35' 21*33 7844 9 -o 5 789 52' 56*76 791 1' 44*46
792 10' 32*15 ' x 73 I g / 9;84 794 28' 7*52 79536'55 ? 20 796 45' 42*87
+ + + + + + + + + + + + + + + + + + + +
02396194
0-2408236
0-2419474 0-2429903 0-2439521
0-1985989 0-1997005 0-2007208 0-2016595 0-2025163 0-2032910 0-2039834 0-2045933 0-2051206 0-2055652
0-2062060 0-2064022 0-2065157 0-2065464 0-2064946 0-2063603 0-2061436 0-2058449
O2448324 O2456309
0-2463475 0-2469820 0-2475341 0-2480038 0-2483909 0-2486955 0-2489173 0-2490565 0-2491132 0-2490872
0-2487881 0-2485152 0-2481604 0-2477238
M-74
14-76 14-78 14-80
14-82 14-84 14-86 14-88 14-90 14-92 14-94 14-96 14-98
+ + + + + + + + + +
-
1480
14-82
+ 0-2059270
+ 0-0022558
0-0018753 0-0060002 0-0101171
1500
00142245
+ + + + +
0-2065069
02063688
0-2062309 0-2060934
+ 0-2489788
8n4o'
1*99
02054643
02059561
+ + + +
1492 1494
14-96 14-98
1500
695
X
14-02 14-04 14-00 14-08 14-10 14-12 14-14 14-16 14-18 14-20 14-22 14-24 14-26 14-28 14-30
7i(*)
Y1 (x)
-
i<'wi
0-2132937 0-2131411 0-2129889 0-2128370 0-2126855
0-2125342
arg
1
]
(x)
HiW
+ + + + + + + + + + + + + + +
0-4759804 0-4788455 0-4817705 0-4847542 0-4877954
0-4908927 0-4940449 0-4972506 0-5005085 0-5038172 0-507I753 0-5105814 0-5140341 0-5175320 0-5210736
X
14-02 14-04 14-06 14-08 14-10
+ + + + +
+ + + + + + + + +
+ + + + + + + + + +
4
02123833
0-2122327 0-2120824 0-2119325
01392332
0-1359159
0-1325491 0-1291344 0-1256731 0-1221667 0-1186166
5;i7 f7532' 6 7 4o' 42*77 677 49' 20*39 678 57' 58 -03 68o 6' 35*69
0-2117828 0-2116335 0-2114845 0-2113358 0-2111874 0-2110394 0-2108916 0-2107442 0-2105971 0-2104502 0-2103037 0-2101575 0-2100116 0-2098660 0-2097207
0-2095757 0-2094310 0-2092866 0-2091425 0-2089987 0-2088552 0-2087120 0-2085691 0-2084265 0-2082842
1430
1432
14-34 14-36
1432 1434
14-36 14-38 14-40
14-42 14-44 14-46 14-48 14-50 14*52
+ + + + +
+ + + + +
0-5246575 0-5282821 0-5319460 0-5356477 0-5393857 0-543I583 0-5469642 0-5508016 0-5546691 0-5585651
1438
14-40
14-42 14-44 14-46 14-48 14-50 14-52 J4-54 14-56 I4-58 14-60
6935o' 9*28 694 58' 47 -22 696 7' 25*18 697i6' 3"i6
698 699 700 701 702
704
24' 41*14 33' 19*15 41' 57*18
1454
14-56 14-58 14-60
14-62
+ + + + + + + + +
50 35*23
59' 13*31
+ + + + +
+ + + + + + + + + + + + + + +
0-5624879 0-5664361 0-5704080 0-5744019 0-5784163 0-5824496 0-5865001 0-5905662 0-5946463 0-5987387
0-6028418
1$
14-68
02035352
0-2042513
0-2048851 0-2054365 0-2059054 0-2062910 0-2065956
1470
14-72 14-74 14-76 14-78 14-80 14-82 14-84
14-62
1S&
14-68 14-70 14-72
+ + + + + + + + + +
+ + + + +
0-2081422 0-2080004 0-2078590 0-2077178 0-2075709 0-2074363 0-2072960 0-2071560 0-2070163 0-2068768 0-2067377 0-2065988 0-2064602 0-2063219 0-2061838
7095i' 2*16 7 IO !59'40-36 8' 18*59 712 713 16' 56*84 714 25' 35*10
715 716 717 719 720
721
34' 13*39 42' 51*69 51' 30*01
0' 8*36 8' 46*72
OOO69539
0-6110734 0-6151987 0-6193280 0-6234599 0-6275926 0-6317244 0-635853$ 0-6399791 0-6440987 0-6482109 0-6523142 0-6564068 0-6604873
M'74
14-76 14.78 14-80
14-82 14-84 14-86 14-88 14-90
i486
14-88
1490
1492
14-94 14-96
0-2068167 0-2069553 0-2070113 0-2069849 0-2068702 0-2066853 0-2064124 0-2060577 0-2056215 0-2051040
1498 1500
+ + + +
17' 25*09 72226' 3*49 723 34' 41*90 724 43' 20*33 725 5 I' 58*79
+ + + + +
1500
696
/oW
-
nw
4 0-205002I 4- 0-2044585 + 0-2038339 4 0-2031287 4- 0-2023432
1*0*)
argH
815
{x)
Ho(*)
0-2472058 0-2466066 0-2459265 0-2451659 0-2443252
X
I5-02 I5-04 15-06 I 5 -o8 I5-IO 15-12 I5-I4 15-16 I5-I8 I5-20
15-22 I5-24 I5-26 15-28
15-02 15-04 15-06 15-08 15-10 15-12 15-14 15-16 15-18 15-20 15-22 15-24 15-20 15-28 15-30 15-32 15-34 15-36 15-38 15-40
15-42 15-44 15-46 15-48 15-50
0-2058191 0-2056823 0-2055459 0-2054097 0-2052737 0-2051381 0-2050027 0-2048675 0-2047327 0-2045981
6; 24*71
4444-
4-
820 50' 22*47 82i59' 10*01 23 7'57 ? 54 824i6'45?o6 825 25' 32*57 826 34' 20*08
7-58 l 828 51' 55*08 830 0' 42*58 831 9'3o''o7
2
+ 0-1901815
4-
0-2376242
0-2347721 0-2332329
ll<
+ 0-2362361
44-
+ 0-2316191
4 0-2299315 0-2281707 4- 0-2263377 4- 0-2244331 + 0-2224578
44-
I53O
15-32 15-34
0-0773477 0-0810521 0-0847192 0-0883477 0-0919362 0-0954833 0-0989876 0-1024478 0-1058626 0-1092307 0-1125507 0-1158213 0-1190418 0-1222103 0-1253260
0-1283875 0-1313938 0,I 34343 8 0-1372363 0-1400702
+ + + 4
4-
832 18' 17*55 8 33 27' 5 ? 03 834 35, 52 -50 835 44, 39*96 8 3653 27*42
838
2;
1536
I5-38 I5-40
0-1792950
0-1751045
+ 0-1772338
4-
+ 0-1729078 + 0-1706449
4 0-1683167 + 0-1659242 + 0-1634685
4-
14*87
839n' 2*32
840 19' 49*76 841 28; 37*20 842 37' 24*63
84346' 1 2 "06 844 54' 59-48 846 3' 46*89 ?47 I2 '34 ? 30 848 21' 21*71 8493o' 9*11 85038'56 ?50 5I f o4 ^ 43 ? 9 852 56' 31*27 854 5'i8 ? 6 5
85 5
i4
'
444-
4-
I5H2
15-44 I5-46
I5H8
I5-50
I5-52 15-54 I5-50 I5-58 I5-60
15-62
4-
4444-
0-1609506 + 0-1583715
15-62 15-64 15-66 15-68 15-70 15-72 15-74 15-76 15-78 15-80 15-82
+ + + + +
4-
4-
0-1963278
+ 0-1935784 4- OT907714
44-
0-1879079 0-1849893
0-1789911 0-1759141 0-1727868 0-1696105
15-66 15-68
1570
15-72 15-74 15-76 I5-78 15-80 15-82 15-84
4-
6*02
+ 0-l820l66
4444-
mt
15-88 15-90
15-92 15-94 15-96 15-98 16-00
444-
44444-
OI530407
0-1495986
0-1425972 0-1390409
^is 1590
15-88 15-92 15-94 15-96
44-
4-
+ 0-1461170
4-
4-
+ OO993533
4-
OI995474
0-1994228
4-
1598
16-00
+ 0-1354493
697
/iW
15-02 15-04 15-06 15-08 15-10 15-12 15-14 15-16 15-18 15-20 15-22 15-24 15-26 15-28
Y
+ + + + + + + + + + + + + +
(x)
1*0*)
arg
(x)
\
HiW
+ + + + +
0-6645540 0-6606052 0-6726395 0-6766552 0-6806508
I5-02 I5-04 15-06 I5-08 I5-IO
+ + + + + + + + + + + + + + + + + + + +
' 37*25
0-2003130 0-1992373 0-1980838 0-1968530 0-1955454 0-1941618 0-1927027 0-1911688 0-1895608 0-1878794
0-1861255 0-1842998 0-1824032 0-1804363 0-1784003
+ 0-6846246 + 0-6885753
+ 0-6925011 + 0-6964006
4-
7-04 1 9 45-65
18' 24-28
1530
15-32 15-34 15-36 15-38
738=27' 2*93 739 35 4**59 740 44 20*27 741*52 58*97 1 37-69 743
744 10' 16*42 745 i8' 55*17
+ + + + +
+ + + + + + + + + +
+ + + + +
07I545I2
0-7191616
0-7228353 0-7264711 0-7300674 0-7336229
1540
15-42 15-44 15-46 15-48 15-50 15-52 15-54 I5-56 I5-58 15-60 15-62 15-64 15-66 15-68 15-70 15-72 15-74 15-76 15-78 15-80 15-82 15-84 15-86 15-88 15-90
15-92 15-94 I5-96 15-98 16-00
+ + + + + + + + + +
+ + + + + + + + + + + + + + + + + + + +
+ + + + +
1532
r 5-34 15-36 I5-38 15-40
746 27 33*94 747 36' 12*72 748 44 51-52 749 751 752 753 754
53' 30 -34
07371363
0-7406061
+ 0-1762958
+ 0-1741239 + 0-1718855
+ 0-1695816 + 0-1672132 + + + + +
0-1647812 0-1622868 0-1597310 0-1571149
O' 1 544396
0-2033475 0-2032153 0-2030834 0-2029518 0-2028204 0-2026892 0-2025584 0-2024278 0-2022974 0-2021673
I5-42
I5'4^
07440312
0-7474101 0-7507416 0-7540245
15 -4 I5-48 15-50
1552
15-54 I5-56 I5-58 15-60
15-62 15-64 15-66 15-68 15-70
+ 0-1517062 + 0-1489160
+ 0-1460700 + 0-1431695 + 0-1402157
0-2020374 0-2019078 0-2017784 0-2016493 0-2015204 0-2013918 0-2012634 0-2011353 0-2010074 0-2008798
0-2007524 0-2006252 0-2004983 0-2003717 0-2002452
45 56-46 54 35*5<>
3'i4''56
11' 53 ''64 20' 32 "73
+ 0-1372099 + 0-1341533
+ 0-1310471 + 0-1278927 + 0-1246913 + + + + + + + + + +
0-1214444 0-1181532 0-1148192 0-1114436 0-1080279
+ + + + + + + + + +
1572
15-74 15.76 15-78 15-80
O7989870
O-0OI27I2 0-8034880 0-8056365 0-8077161
21'
6-8i
778 29' 46 -03 779 38' 25*26 7o 47' 4*50 7i 55 43*76 783 4 23-04
+ + + + +
0-8097259 0-8116653
08135336
0-8153300 0-817054!
23
698
ex
X
o
0-02 0-04 0-06 0-08 O-IO 0-12 0-14 0-16 0-18 0-20 0-22 0-24 0-26 0-28
e-*I x {*)
o-ooooooo
0-0098025 0-0192196 0*0282657 0-0369542 0-0452984
0-0533111 0-0610043 0-0683899 0-0754792 0-0822831
K
00
(x)
e*
K^x)
00
1
gX
-ooooooo
-ooooooo
0-02 0-04 o-66 0-08 O-IO
9-2102792 8-0076794 7-1036124 6-3987260 5-8333860 5-3696274 4-9821285 4-6533504 4-3707591 4-1251578 3-9096449 3-7189398 3-5489328 3-3963772 3-2586739 3-1337176 3-0197845 2-9154495 2-8195242 2-7310097
2-6490599 2-5729537 2-5020724 2-4358821 2-3739200
2-3157825 2-2611160 2-2096099 2-1609894 2-1150113
1-1274969 1-1502738 1-1735109 1-1972174 1-2214028 1-2460767 1-2712492 1-2969301 1-3231298 1-3498588 1-3771278 1-4049476 1-4333294 1-4622846 1-4918247 1-5219616 -5527072 1-5840740 1-6160744 1-6487213
1
030
0-32 o-34 0-36 0-38 0-40
0-42 0-44 0-46 0-48 0-50
0-8122587 0-7979961 0-7841375 0-7706698 0-7575806 0-7448578 0-7324896 0-7204648 0-7087725 0-6974022
0-0888122 0-0950766 0-1010861 0-1068501 0-1123776 0-1176774 0-1227579 0-1276272 0-1322931 0-1367632 0-1410447 0-1451446 0-1490697 0-1528263 0-1564208
0-1598592 0-1631473 0-1662906 0-1692946 0-1721644
0-1749051 0-1775214 0-1800181 0-1823995 0-1846700
1-6820276 1-7160069 1-7506725 1-7860384 1-8221188 1-8589280 1-8964809 1-9347923 1 '9738777 2-0137527 2-0544332 2-0959355 2-1382762 2-1814723 2-2255409
0-72 o-74 0-76 0-78 o-8o 0-82 o-8 4 o-86 o-88 0-90
0-92 0-94 0-96 0-98 I-OO
2-2704998 2-3163670 2-3631607 2-4108997 2-4596031 2-5092904 2-55998I4 2-6116965 2-6644562 2-7182818
699
ex
X
1-02 1-04 i-oo
e~* I
(x)
e-*I x {x)
0-2088796 0-2097912 0-2106473 0-2114501 0-2I220I6
0-2I29039 0-2135587 0-2141680 0-2147335 0-2152569
0-2I57398 0-2161838 0-2165905 0-2169613 0-2172976
e*K
(x)
ex
{x)
e*
X
1-02 1-04 1-06 1-08 I-IO I-I2 1-14 1-16 1-18 1-20 1-22 1-24 1-26 1-28
I
108
I-IO
I-I347579 1-1253024
1 -ii 60860
1-6136817 1-5920135 1-5711046 1-5509141 1-5314038 1-5125382 1-4942846 1-4766120 x-4594919 1-4428976 1-4268038 1-4111872 1-3960258 1-3812990 1-3669873
1-3530725 1-3395375 1-3263660 1-3135429 1-3010537
112
i-i6 i-i8 1-20
1-22 I-2A 1-26 1-28 i-3
3H556I35
3-5254215 3*5966397 3-6692967
'
30
132
1-34 i- 3 6 1-38 1-40
1-42 1-44 1-46 1-48 1-50
1-52 i'54 l- 5 6 1-58 l-6o 1-62 1-64 1-66 1-68 1-70 1-72 1-74 1-76 1-78 i-8o 1-82 1-84 1-86 1-88 1-90 1-92 1-94 1-96 1-98 2-00
1-0141239
1 -00741 36
1-0008375
O2183243
0-2185076
0-2186638 0-2187941 0-2188994 0-2189809
09943910
0-9880700
0-9818703 o-975788i 0-9698197 0-9639615 0-9582101
0-9525622 0-9470148 0-9415648 0-9362095 0-9309460 0-9257717 0-9206842 0-9156810 0-9107597 0-9059181
0-9011541 0-8964656 0-8918506 0-8873072
40552000
4-1371204 4-2206958 4-3059595 4-3929457 4-4816891
4-5722252 4-6645903 4-7588212 4-8549558 4-9530324 5-0530903 5-1551695 5-2593108 5-365556o 5-4739474
5-5845285 5-6973434 5-8124374 5-9298564 6-0496475
1-32 1-34 1-36 1-38 1-40 1-42 1-44 1-46 1-48 1-50
O2190394
0-2190759 0-2190913 0-2190865 0-2190623 0-2190195 0-2189589 0-2188812 0-2187872 0-2186775 0-2185528
0-2184138 0-2182610 0-2180952 0-2179167 0-2177263
152
1-54
156
1.58 l-6o
1-62 1-64 1-66 1-68 1-70
11548123
1-1460392 r-1374494 1-1290368 1-1207955 1-1127201 1-1048054
1-0970461 1-0894376 1-0819752 1-0746544 1-0674709
o*33i53
0-3288719
0-3266691
08828335
0-8784278 0-8740882 0-8698132 0-8656012 0-8614506
0-2175244
03245058
0-3223810 0-3202938
02173115
0-2170881 0-2168548
o-2i66iij9
03182432
0-3162282 0-3142481 0-3123020 0-3103890 0-3085083
02152693
1-0604208
1
0535000
700
Table
ex
X
2-02 2-04 2 OO 2-08 2-IO 2-12 2-14 2-16 2-l8 2-20
e~*h{x)
0-3066592 0-3048408 0-3030525 0-3012935 0-2995631
er*I x (x)
0-2149779 0-2146797 0-2143750 0-2140643 0-2137477
e*
(x)
"x
if,W
ex
X
2-02 2-04 2-00 2-o8 2-IO 2-12 2-14 2-16 2-18 2-20
0-8377564 0-8339966 0-8302875 0-8266281 0-8230172 o-8i94537 0-8159366 0-8124650 0-8090377 0-8056540
0-8023128 0-7990133 o-7957545 o-792535$ 0-7893561
0-9964584 0-9906463 0-9849292 0-9793046 0-9737702 0-9683236 0-9629626 0-957685I 0-9524890 0-9473722
o-94 2 3329
90250135
9-2073308 9-3933313 9-5830892 9-7766804 9-9741825
10-1756743 10-3812366 10-5909515 10-8049029 11-0231764 11-2458593 11-4730407 11-7048115 11-9412644 12-1824940
12-4285967 12-6796710 12-9358173 13-1971382
232
250
2-52 2-54 2-56 2-58 2-60 2-62 2-64 2-66 2-68 2-70
2-72 2-74 2-76 2-78 2-80
134637380
I3-7357236 14-0132036 14-2962891 14-5850933 14-8797317
15-1803222 15-4869851 15-7998429 16-1190209 16-4446468
0-2570776 0-2559892 0-2549146 0-2538534 0-2528055 0-2517706 0-2507484 0-2497387 0-2487412 0-2477557
0-2467820 0-2458198 0-2448690 0-2439292 0-2430004
2-82 2-84 2-86 2-88 2-90 2-92 2-94 2-96 2-98 3-00
16-7768507
I7-H57655
17-4615269 17-8142732 18-1741454
18-5412875 18-9158463 19-2979718 19-6878167 20-0855369
01979955
0-1976053 0-1972157 0-1968267
07021551
0-6999479 0-6977616
3o
701
e~*h(x)
0-2420822 0-2411745 0-2402772 0-2393899 0-2385126
0-2376451 0-2367871 0-2359385 0-2350991 0-2342688
<?-*/,(*)
e*
KQ (x)
ex
{x)
ex
X
3-02 3-04
o- 1964383
20-491292
20905243
21-327557 21-758402 22-197951 22-646380 23-103867 23-570596 24-046754
306
3-o8 3-10
3-12
07940457
0-7910030 0-7879938 0-7850176 0-7820736 0-7791613 0-7762803 o-7734299 0-7706096 0-7678189 - 765057 3 0-7623243 0-7596194 0-7569422 0-7542922 0-7516690 0-7490721 0-7465010 0-7439555
01945076
0-1941240 0-1937412 0-1933594 0-1929786
0-1925988 0-1922200 0-1918423 0-1914657 0-1910902 0-1907158 0-1903425 0-1899704 0-1895995 0-1892299
314
3-16 3-18
318 320
24532530
25-028120 25-533722 26-049537 2 -575773 27-112639
27-660351 28-219127 28-789191 29-370771 29-964100
320
3*22 3-24 3-26 3-28 3-30
330
332
3*34 3-36 3-38
0-6749978 0-6730377 0-6710948 0-6691687 0-6672592 0-6653660 0-6634890 0-6616278 0-6597823 0-6579523
0-6561375 0-6543377 0-6525527 0-6507823 0-6490263
3HO
3H2
3-44 3-46 3-48 3-5 3-52 3\54 3-56 3-58 3-60
3-62 3-64 3-66 3-68
332
3-34 3-36 3-38
340
3-42 3*44 3-46 3-48 3-5o
3-52 3\54 3-56
02242325
0-2235140 0-2228024 0-2220978 0-2214000 0-2207089 0-2200243 0-2193462
0-2186745 0-2180091 0-2173498 0-2166966 0-2160494
0-2154081 0-2147726 0-2141429 0*2135187 0-2129001
07414350
o-738939i 0-7364675
33115452
33-784428 34-466919 35-163197 35-873541 36-598234 37-337568 38-091837 38-861343 39-646394 40-447304 41-264394 42-097990 42-948426 43-816042 44-701184
07340199
o-73i5957 0-7291947 0-7268165 0-7244607
358 360
3-62 3-64 3-66 3-68 3-7o 3-72 3-74 3-76
370
372 374 376 378 380
3-82 3-84 3-86 3-88
378 380
3-82 3-84 3-86 3-88 3-90 3-92 3-94 7 3-96 3-98 4-00
390
392 394 396
3-98 4-00
0-1817657 0-1814251 0-1810860 0-1807482 0-1804119 0-1800769 o-i797433 0-1794111 0-1790803 0-1787508
0-6227419 0-6212038 0-6196771 0-6181617 0-6166573 0-6151640 0-6136814 0-6122096 0-6107484 0-6092977
06999345
0-6978232 0-6957302 0-6936555 0-6915988
45-604208 46-525474 47-46535I 48-424215 49-402449 50-400445 51-418601 52-457326 53*517034 54-598I50
702
Table
e-*I
(x)
e~ x h{x)
(x)
e*K x {x)
0-6796219 0-6776840 0-6757623 0-6738564 0-6719662
0-6700914 0-6682318 0-6663872 0-6645575 0-6627424 0-6609418 o-659i553 0-6573830 0-6556246 0-6538798
ex
x
4-02 4-04 4-06 4-08 4-10
4-12 4-14 4-16 4-18 4-20 4-22 4-24 4-26 4-28
4-02 4-04 4-06 4-08 4-10 4-12 4-14 4-16 4-18 4-20 4-22 4-24 4-26 4-28 4-30
4-32 4-34 4-36 4-38 4-40
0-1784228 0-1780961 0-1777709 0-1774470 0-1771245 0-1768033 0-1764836 0-1761652 0-1758482 o-i755325
0-1752182 0-1749053 o-i745937 0-1742835 o-i739746
0-1736671 0-1733609
64071523
65-365853 66-686331
05953390
Q-5939955 6-5926611 o-59i3357 0-5900192 0-5887114
430
4-32 4*34 4-36 4-38 4-40 4-42 4*44 4-46 4-48 4-50 4-52 4*54 4-56 4-58 4-60
01 730560
0-1727525 0-1724502
4-42 4-44 4-46 4-48 4-50 4-52 4'54 4-56 4-58 4-60 4-62 4-64 4-66 4-68
90017131
9I-835598 93-690800 95-58348o
06339521
0-6323708 0-6308010 0-6292426 0-6276955 0-6261595 0-6246345 0-6231203 0-6216169
0-6201241 0-6186418 0-6171699 0-6157082 0-6142566
97-5M394
99-484316
101-494032 103-544348 105-636082 107-770073 109-947172
112-168253 114-434202 116-745926 1 19- 104350 121-510418
123-965091 126-469352 129-024202 131-630664 134-289780
470
4-72 4-74 4-76 4-78 4-80 4-82 4-84 4-86 4-88 4-90
4-92 4-94 4-96 4-98
5
4-f4
41^ 4-88
4-90
4-92 4-94 4-96 4-98
0183933s
0-1835408
00
M5-474382
148-413159
500
703
hi*)
5-02 5-04 5-06 5-08 5-10 5-12 5-14
e~ x h(x)
(x)
ex
{x)
0-1637083
o- I 634455
151-41130 154-47002 I57-59052 160-77406 164-02191 I67-33537 170-71577 174-16446 177-68281 181-27224
184-93418 188-67010 192-48149 196-36988 200-33681
5-i8 5-20
22
24 26 28 30
5-32 5'34 5-36 5-38
5HO
5-42 5'44 5'46 5-48 5-5
5-52 5'54 5-56 5-58 5-60 5-62 5-64 5-66 5-68
536
5-38 5-4 5-42 5*44 5-46 5-48 5-50
5-52 5*54 5-56 5-58 5-60 5-62 5-64 5-66 5-68 5-70
5-72 5-74 5-76
01 754991
0-1751585 0-1748199 0-1744833
OT579393
0-1577010
o-i574637 0-1572274 0-1569920 0-1567576 0-1565242 0-1562918 0-1560603 0-1558298 0-1556002 o-i5537i6
0,I 55i439
0-1741486
T738i59
0-1734850 0-1731561 0-1728291
0-1725039 0-1721806 0-1718591 0-I7I5394 0-1712215 0-1709054 0-1705911 0-1702785 0-1699676 0-1696584 0-1693509 0-1690451 0-1687410 0-1684385 0-1681377 0-1678384 0-1675408 0-1672448 0-1669503 0-1666574
57
572 574 576 578
5-8o
5-82 5-84 5-86 5-88 5-90
5-92
578 580
5-82
5-8 4
0-1540193
- I 53797i
594
5-96 5-98 6-oo
37993493
387-61012 395-44037 403-42879
704
Table
ex
e-*I
(x)
e-'I^xy
0-1518372 0-1516237 0-1514111 0-1511994 0-1509885
e*
(x)
ex
(x)
ex
x
6-02 6-04 6-o6 6-o8 6-io
6-12 6-14 6-i6 6-i8 6-20
0-1663661 0-1660763 0-1657880 0-1655012 0-1652159 0-1649321 0-1646498 0-1643689 0-1640894 0-1638114
0-5411776 0-5401848 o-539i973 0-5382151 o-537 2 382 0-5362666 0-535300I o-5343387 0-5333825 o-53243i3
0-5314851 0-5305438 0-5296075 0-5286761 o-5277494
411-57860 419-89303 428-37544 437-02919 445-85777 454-86469 464-05357 473-42807 482-99196 492-74904
502-70323 512-85851 523-21894 533-78866 544-57191
0-4970946 04963 1 30
0-495535J 0-4947608 0-4939902 0-4932232 0-4924597 0-4916998 0-4909434 0-4901905
0-4894411 0-4886950 0-4879524 0-4872132 0-4864773
6-22 6-24 6-26 6-28 6-30 6-32 6-34 6-36 y 6-38 6-40 6-42 6-44 6-46 6-48 6-50
6-52 6-54 6-56 6-58 6-6o
0-1621727 0-1619044 0-1616374 0-1613717 0-1611073 0-1608443 0-1605825 0-1603220 0-1600628 0-1598048
0-1595481 0-1592927 0-1590385 0-1587855 o-i585337 0-1582831 0-1580336 0-1577854 o-i5753 8 4 0-1572925
6-42 6-44 6-46 6-48 6-50 6-52 6'54 6-56 6-58 6-6o 6-62 6-64 6-66 6-68 6-70 6-72
73509519
749-945 10 765-09499 780-55094 796-31911 812-40583
0-5135486 0-5126982 0-5118520 0-5110099 0-5101719 0-5093380 0-5085080 0-5076821 0-5068602 0-5060421 0-5052280 0-5044178 0-5036114 0-5028088 0-5020099
t%
6-78 6-8o 6-82 6-84 6-86 6-88 6-90
01563204
0-1560802
0-1558411 0-1556031 0-1553662 0-1551304 0-1548956
04704235
0-4697612 0-4691016
01433663
0-1431852
6-82 6-84 6-86 6-88 6-90 6-92 6-94 6-96 6-98 7-00
692
6-94 6-96 6-98 7-00
0-4980716
705
ex
X
7'02
t-Itix)
0-1421120 0-1419354 0-1417594 0-1415840 0-1414093
0-1412352 0-1410617 0-1408889 0-1407166 0-1405450
e*K
(x)
e*K x {x)
0-4972948 0-4965217 o-49575 21 0-4949860 0-4942235
ex
1118-7866 1141-3876 1164-4452 1187-9685 1211-9671 1236-4504 1261-4284 1286-9109 1312-9083 I339-4308
1366-4891 1394-0940 1422-2565 1450-9880 1480-2999
T1 t
7-20
7-22 7-24 7-26 7-28 7'30
7SO
7'52 7-54 7'56 7-58 7-60 7-62 7 7 tt 7-68 7 7o
.
04499419
0-4493624 0-4487851 0-4482101 0-4476372 0-4470665 0-4464981 o-44593i7 o-4453676 0-4448056 0-4442457 0-4436879 0-4431322 0-4425786 0-4420271 o-44i4776 0-4409302 0-4403848 0-4398414
7-72 7'74 7.76 7-78 7-80 7-82 7-84 7-86 7-88 7.90 7.92 7-94 7.96 7-98 8-oo
04393001
0-4387607 0-4382234 0-4376880 o-437i545
04663178
0-4656789 0-4650426 0-4644089 o-4637777 0-4631491
7-82 7-84 7-86 7-88 7.90 7-92 7-94 7-96 7-98 8-oo
0-1436179 0-1434318
04366230
706
Table
ea
X
8-02 8-04 8-o6 8-o8 8-io
8-12 8-14 8-i6 8-i8 8-20 8-22 8-24 8-26 8-28
88-
e~* /(*)
e~* I^x)
e*
(x)
ex
{x)
ex
X
8-02 8-04 8-o6 8-o8
OI 33993 2
0-1338443 0-1336960 0-1335481 o- 1334007
0-1332538 0-1331073 0-1329613 0-1328158 0-1326708 0-1325262 0-1323821 0-1322384 0-1320952 0-1319524
0-1318101 0-1316683 0-1315269 0-1313859 0-1312454
0-4360935 o-4355658
04350401
o-4345i63 o-4339944 o-4334743 0-4329562 o-4324398 0-4319254 0-4314127
0-4625230 0-4618994 0-4612783 0-4606597 0-4600436 o-4594299 0-4588186 0-4582097 0-4576033 o-4569992 o-4563974 o-455798i 0-4552010 0-4546063 0-454 l 39
0-4534238 0-4528359 0-4522504 0-4516670 0-4510859
32292332
3294-4681
810
8-12 8-14 8-i6 8-i8 8-20 8-22 8-24 8-26 8-28 8-30
01423299
0-1421488 0-1419683 0-1417885 0-1416094
3361-0207 3428-9179 3498-1866 3568-8547 3640-9503 3714-5024 3789-5403 3866-0941 3944-1944 4023-8724
3o
32 34 8-36 8- 8 3 8-40
8-42
832
8-38 8-40
8-42
l$t
8-48 8-50
8-52 8-54 8-56 8 l"5 8-6o
8-62 8-64 8-66 8-68
S:
8-48 8-50
8-52
5l f 8-58 8-6o
04234513
0-4229682 0-4224868 0-4220071 0-4215289
53241055
5431-6596
5541-3864,
870
872
8-74
01293234
0-1291892 0-1290555 0-1289222 0-1287892 0-1286567 0-1285246
0-4420808
876 878
8-8o
8-82 8-84 8-86 8-88 8-90 8-92 8-94 8-96 8-98 9-00
04415356
0-4409923 0-4404511 0-4399119
04177616
0-4172978 0-4168355
0-4163747 0-4159155 -4 I 54578 0-4150016 0-4145468
- I
01273532
0-1272250
707
ea
e-*I 9 {x)
0-1348051 0-1346512 -i344978 0-1343450 0-1341927
e-*IM
0-1270971 0-1269697 0-1268426 0-1267159 0-1265895
(x)
ex
K,[x)
ex
X
9-02 9-04 9-06 9-08 9-10 9-12 9-14
9-02
904
9-06 9-08 9-10
9-12
0-4118497 0-4114053 0-4109623 0-4105207 0-4100806 0-4096419 0-4092045 0-4087686 0-4083341 0-4079010
0-4074692 0-4070388 0-4066098 0-4061821 0-4057558
04320519
o-43!5427
914
9-16 9-18 9-20
9-22 9-24 9-26 9-28
916
9-18 9-20
9-22 9-24 9-26 9-28 9-30
01255920
0-1254689 0-1253462
0-1252239 0-1251018 0-1249802 0-1248589 0-1247379
0-1246173 0-1244970 0-1243771 0-1242575 0-1241382
04275321
0-4270385
930
932
9-34 9-36 9-38
940
9-42 9'44 9-46 9-48 9-50
01319684
0-1318240 0-1316801 0-1315367 0-1313938
0-4053308 0-4049071 0-4044848 0-4040638 0-403644 0-4032257 0-4028087 0-4023929 0-4019784 0-4015651
0-4011532 0-4007425 0-4003331 0-3999249 0-3995180 0-3991123 0-3987078 0-3983046 0-3979026 o-39750i8 0-3971023 0-3967039 0-3963067 o-3959i67 o-3955!59
932
9-34 9-36 9-38 9-40 9-42
944
9-46 9-48 9-50
9-52
04231502
0-4226716 0-4221945
0-4217191 0-4212452 0-4207730
01312513
0-1311092 0-1309677 0-1308266 0-1306859 0-I305457
952
9*54 9-56 9-58 9-60
9-62 9-64 9-66
0-1240193 0-1239008 0-1237825 0-1236646 0-1235470 0-1234298 0-1233128 0-1231962 0-1230800 0-1229640
954
9-56 9-58 9-60
9-62 9-64 9-66 9-68 9-70
04203023
0-4198332
968
9-70
9.72
01298514
0-1297139 0-1295768 0-1294401 0-1293039 0-1291681
0-1290328 0-1288978 0-1287633 0-1286292 0-1284955
0-1283623 0-1282294 0-1280970 0-1279650 0-1278333
0-4193656 0-4188996 0-4184351 0-4179721 0-4175107 0-4170508 0-4165924 0-4161355 0-4156801 0-4152261
o-4i47737 0-4143227 0-4138731
974
9-76
978
9-80
9-82
972
9-74 9-76 9-78 9-80
9-82 9-84 9-86 9-88 9-90 9-92 9-94 9-96 9-98 io-oo
i69354 I 4
17326-6317 17676-6529 18033-7449 18398-0507 18769-7160 19148-8894 I9535-7227 I9930-3704
03951223
o-3947 2 99 o-3943386 o-3939485 o-3935596
9*4
9-86 9-88 9-90
01219348
0-1218220 0-1217096 0.12 15974 0-1214855 0-1213739 0-1212627
04134250 04129784
0-4125332 0-4120894 0-4116471 0-4112061 0-4107666
708
Table
ex
X
10-02 IO-04 10-06 10-08 IO-IO IO-I2 10-14 IO-IO 10-18 10-20 IO-22 IO-24 IO-26 IO-28 IO-30
e~*I
(x)
e~*I x {x)
0-1211517 0-1210411 0-1209307 0-1208206 0-1207109 0-1206014 0-1204922 0-1203833 0-1202747 0-1201664 0-1200584 0-1199500 0-1198432 0-1197360 0-1196292 0-1195226 0-1194162 0-1193102 0-1192044 0-1190990
0-1189938 0-1188888 0-1187842 0-1186798 0-1185757 0-1184718 0-1183682 0-1182649 0-1181619 0-1180591
e*
(x)
e'K^x)
0-4103284 0-4098917
ex
X
10-02 IO-04
0-1277021 0-1275713 0-1274409 0-1273109 0-1271813 0-1270521 0-1269233 0-1267948 0-1266608 0-1265392
0-3912498 0-3908688 0-3904889 0-3901101 o-38973 2 4 o-3893558 0-3889803 0-3886059 0-3882325 0-3878603
0-3874891 0-3871189 0-3867498 0-3863818 0-3860149
04094563
0-4090223 0-4085897
IOO6
10-08 IO-IO IO-I2 10-14 IO-IO 10-18 10-20 IO-22 10-24 IO-26 IO-28 IO-30 IO-32 io-34 10-36 10.38 10-40 10-42 10-44 10-46 10-48 10-50
27446-666 28001-126 28566-786 29143-874 29732-619 30333-258 30946-030 31571-181 32208-961 32859-626 33523-434 34200-652 3489I-55I 35596-4o8 36315-503 37049-124 37797.566 38561-128 39340-114 40134-837
40945-615 41772-771 42616-637 43477-550 44355-855
IO32
10-34 10-36 IO- 3 8 10-40 IO-42 10-44 10-46 10-48 IO-50 IO-52 10-54 10-56 10-58 io-6o
10-62 10-64 IO-66 10-68 10-70 10-72 10-74 10-76 10-78 io-8o
0-4018482 0-4014378 0-4010286 0-4006207 0-4002140 0-3998085 0-3994043 0-3990013 o-3985995 0-3981989
0-1245480 0-1244266 0-1243056 0-1241850 0-1240647 0-1239448 0-1238252 0-1237059 0-1235870 0-1234685
0-1233503 0-1232324 0-1231149 0-1229977 0-1228808
0-1227642 0-1226480 0-1225322 0-1224166 0-1223014 0-1221865 0-1220719 0-1219577 0-1218438 0-1217302
0-1179566 0-1178544 0-1177524 0-1176507 0-1175492 0-1174480 0-1173471 0-1172464 0-1171459 0-1170458 0-1169458 0-1168462 0-1167467 0-1166476 0-1165487 0-1164500 0-1163516 0-1162534 0-1161554 0-1160570
Q-3977995 0-3974013 0-3970043 0-3966084 0-3962137 0-3958202 o-3954 2 79 0-3950367 0-3946467 o-3942578
0-3938701 o-3934835
03930980
0-3927137 0-3923305 0-3919484 0-39I5673 0-3911874 0-3908086 0-3904309
1098
II-OO
1 1
709
ex
"*/,(*)
-"A(*)
0-1159603
e*K
(x)
e*Kt(x)
ex
X
02 04
II-02
0-I2l6l69
OI2I5039
01158631
0-1157662 0-1156694
VSot n-o8
II-IO
11-12 11-14 ii-i6
03889309
0-3885586
0-3881873 0-3878171 0-3874480 0-3870799 0-3867128
01155730
0-1154767 0-1153807
03721430
03718151
0-3714881 0-3711619 0-3708367 0-3705122
61083-680 62317-652 63576-552 64860-883 66171-160 67507-906 68871-656 70262-956 71682-362 73130-442
74607-775 76114-952 77652-576 79221-262 80821-638
11
1 1
1112
01152849
0-1151894 01 1 5094
1118
II-20
11-22
A-lt n-i8
II-20
11-22
II-2 4 11-26 11-28
Wit
11-28 11-30
01146211
0-1145272 01 144335 0-1143401 0-1142468 0-1141538
03852548 03848929
II30
11-32
1132 "34
11-36 11-38
01 198463
0-1197382 0-1196303 0-1195228
199547
03685833
0-3682648
03845320
0-3841721 0-3838132
82454343
84120-031 85819-368 87553-035 89321-723
03679470
0-3676301 0-3673140
z&
1138 1140
1142
11-44 11-46 1 1 48 11-50
-
03834553
0-3830984
1140
11-42
n-44
11.46 11-48
1150
n-52
3669987
03827425
0-3823875 0-3820336 0-3816806 0-3813286 0-3809775 0-3806275
03666843
0-3663706 0-3660578 o-3^57457 o-3654344
"54
11-56 11-58
1152
xx-lo 11-58 11-60
11-62 11-64
03651240
0-3648143 0-3645054 o-364i973
03802783
0-3799302 0-3795830 0-3792367 0-3788914 0-378547
1160
11-62 11-64
n-66 n-68
11-70
11-72
01180473
0-1179440
0-3638900
03635834
0-3632777 0-3629727 0-3626684 0-3623650 0-3620623 0-3617603 o-36i459i 0-3611587
03782035 03778610
o-3775i94 0-3771787 0-3768389
n-66 n-68
11-70
n-74
11-76 11-78 II-OO
11-82
03765001
0-3761621
0-3758251
1172
W%
11-78 II-OO 11-82 11-84
133252353
01173298
0-1172284 0-1171272 0-1170263 0-1169256 0-1168252 0-1167251 0-1166252
01122513
0-1121630 0-1120750 0-1119871 01 1 18995 0-IIl8l20 0-1117248 0-1116378 0-1115509 0-1114643
'Ait n-88
11
135944229
1 38690-48
<>
03754890
o-375i537 0-3748194 o-3744859
o-374i533 0-3738216 o-37349o8 0-3731608
-90
03599643 03596676
o-35937i6 0-3590763 ' 358 7i 0-3584880 0-3581949
n-86 n-88
11-90
11-92
j;u
ii- 9 8
1192 n-94
1 1
-96
01165256
0-1164262
12-00
03728318
11-98 I2-00
710
Table
X
12-02 I2-04 12-00 12-08 I2-IO
e-I
(x)
e-*Ii(x)
e*
(x)
e*K x {x)
0-3725035 0-3721762 0-3718497 0-3715240 0-3711992 o-37o8753 0-3705522 0-3702299 0-3699085 0-3695879
0-3692681 0-3689492 0-3686311 0-3683138 o-3679973
ex
X
12-02 I2-04 I2-00 12-08 I2-IO 12-12 12-14 I2-IO
0-1113779 0-1112916 0-1112056 0-1111197 0-1110341 0-1109487 0-1108634 0-1107783 0-1106935 o- 1 106088
0-1105243 0-1104400 0-1103559 o-i 102720 0-1101883
0357902s
0-3576108 o-3573i99 o-357o 2 96 0-3567401 0-3564513 0-3561631 o-3558757
o-ii58353
0-H57377
0-1156404 0-1155432 0-1154464
03555890
0-3553029
I2l8
12-20
12-22 12-24 12-26 12-28 1 2 30
12-32
12-22 12-24 12-26 12-28 I2-30 12-32 12-34 12-36 12-38 12-40 12-42 12-44 12-46 12-48 12-50 12-52 12-54 12-56 12-58 12-60
o- 1 o- 1
0-3550176 o-3547329 0-3544489 0-3541656 0-3538830 o-3536oio o-3533i98 0-3530392 o-35 2 7592 0-3524800
0-1101048 0-1100215 0-1099383 0-1098553 0-1097726 0-1096900 0-1096076 0-1095253 0-1094433 0-1093614
1234
12-36 I2- 3 8 12-40
12-42 12-44 12-46 12-48
03522014
o-35!9234 0-3516461 0-35I3695 0-3510935
0-3508182 0-3505435 0-3502694 o-349996o o-3497233 0-3494512 0-349I797 0-3489088 0-3486386 0-3483690
25271054
257815-63 263023-85 268337-29
1250
1252
12-54 12-56 12-58 12-60
12-62 12-64 12-66 12-68 12-70 12-72 12-74 12-76 12-78 I2-80 12-82 12-84 12-86
0-H37433
0-1136509 0-1135587
0-1134668 0-1133750 0-1132835 0-1131922 0-1131011 0-1130103 0-1129196 0-1128292 .0-1127390 0-1126490 0-1125592 0-1124697 0-1123803 0-1122912 0-1122023
0-3645687 0-3642617 o-3639555 0-363650 o-3633453 0-3630414 0-3627383 0-3624359 0-3621342 0-3618333
0'36i5332
29655857
302549-45 308661-35 314896-72 321258-06 327747-90
12-62 12-64 12-66 12-68 12-70 12-72 12-74 12-76 12-78 I2-0O 12-82 12-84 12-86 12-88 12-90 12-92
33436885
341123-55 348014-70 355045-06 362217-45
03612337
0-3609351 0-3606371
03472968
0-3470303
03603399
0-3600434 o-3597477 o-3594527 0-359I584 0-3588648
36953473
376999-82 384615-73 392385-48 400312-19
1288
12-90
12-92 12-94
12 12
1294
12-96 12-98
1300
43365302 44241339
96 98
I3-00
1 1
711
ex
X
13-02 13-04
*"*/(*)
e-'I^x)
e*KQ [x)
0-3441388 o-3438795 0-3436208 0-3433626 o-343io5i
0-3428481 0-3425917
o-34 2 3359
*Ki{*)
0-3571182 0-3568296 o-35654!7 o-3562544 o-3559679
ex
X
13-02
1306
13-08 13-10
13-12 13-14 13-16 13-18
1304
13-06 13-08 13-10
13-12 I3-I4 13-16 13-18 13-20
OIII2379 OIIII5I5
0-III0652 0-II09792 0-II08934
1320
13-22 13-24 13-26 13-28
0-3420807 0*3418260
1330
1332 1334
13-36 I3-38
01065307
- io 64552
0-3415719 0-3413184
03542629
o-35398ii o-3537oo
03410654
0-3408130
03534195
o-353i398 0-3528606 0-3525821 0-3523043 0-3520272 0-35I7506
o-35i4748
1322
13-24 13-26 13-28 I3-30
0340561
0-3403098
59719561
609259-77 621567-63 634124-13 646934- 29 660003-22
1340
1342
3-44 13-46 I3-48
J
1332
x
03400591
0-3398089 3395593 0-3393 102
3-34
01057824
0-1057084
03390616
0-3388137 0-3385662
Q-35H995
0-3509250
01056346
0-1055609 0-1054874
1350
I3-52 13-54 I3-50 13*58
03383193 03380729
0-3378271 0-3375818 o-337337i 0-3370928
03506510
0-3503777
0-3501051
1360
13-62
x 3-52
03498330
0-3495616 0-3492909
03368491
0-3366060 o-3363633 0-3361212 o-3358796 0*3356385 o-335398o o-335i579
03490207
0-3487512 0-3484823 0-3482140 0-3479463 0-3476793
1364 1366
13-68
I37O
I372
13-74 13.76
1364
13-66
1368
13-70
13-72 J3-74 13-76
1378 1380
13-82 13-84 13-88 I3-90
13-92 13-94 I3-96 I3-98
0-3474128
03349184
-3346794 0'3344409
0-347H70 03468818
0-3466172
03463532
0-3460897 0-3458269 0-3455647
1378 1380
13-82
1386 on
03342029
0-33396^4 0-3337285 o-333492o 0-3332560
1384 1386
13-88
03453031
0-3450420
1390 1392
13-94
I400
01076937
0-1076153
03330206
0-3327856
0332551
0-3323171 0-3320836
1396 1398
14-00
712
Table
X
14-02 I4-04 14-06 14-08 14-10
14-12 14-14 14-16 14-18 14-20
14-22 14-24 14-26 14-28 14-30
e-*h(x)
0-1075370 0-1074589 0.1073810 0-1073032 0-1072256 0-1071482 0.1070710 0-1069939 0-1069169 0-1068402
e-*Ii(*)
e"
(x)
*!(*)
0-3434881
e*
X
14-02 14-04 14-06 14-08 14-10 14-12 14-14 14-16 14-18 14-20 14-22 14-24 14-26 14-28
0-34323"
0-3429747 0-3427189 0-3424637 0-3422090 o-34 r 9549 0-3417013 0-3414484
1226898-5 1251683-5 1276969-2 1302765-7 1329083-3 I355932-5 1383324-2 1411269-2 I439778-7 1468864-2
J 498s37-2 1528809-7 I559693-7 1591201-6 1623346-0
0-1032814 0-1032126 0-1031438 0.1030752 0-1030067 0-1029384 0-1028702 0-1028021 0-1027342 o- 1 020663
0-1025987 0-1025311 0-1024637 0-1023965 0-1023293
0-1022623 0-1021954 0-1021287 0-1020621 0-1019956
0-1019292 0-1018630 0-1017969 0-1017309 0-1016650 0-1015993 0-1015337 0-1014682 0-1014029 0-1013377
0-3306930 0-3304629
03302333
0-3300042 o-3297755
o-34"959
0-3409441 0-3406927 0-3404420 0-3401918 o-339942i
o-3295474 0-3293197 0-3290924 0-3288657 0-3286394 0-3284136 0-3281882 0-3279633 0-3277389 0-3275149
0-3272914 0-3270684 0-3268458 0-3266236 0-3264019
0-3261807 o-3259599 o-3257396 0-3255197
1430
1432
14-34 14-36 I4-38 14-40
14-32 14-34 14-36 14-38 14-40 14-42 14.44 14-46 14-48 14-50 14-52 14-54 14-50 14-58 14-60 14-62 14-64 14-66 14-68 14-70 14-72 14-74 14-76 14.78 14.80
0-1060067
18303175
1867292-4 1905014-2 I943498-0 1982759-3
1442
14-44 14-46 14.48 14-50 14-52 14-54 i4-5o I4-58 14-60 14-62
01059319
0-1058572 0-1057827 0-1057084
03365034
0-3362617
03253002
0-3250812 0-3248626 0-3246445 0-3244268 0-3242096 0-3239928 0-3237764 0-3235604 o-3233449 0-3231298
0-3229152 0-3227010 0-3224872 0-3222738 0-3220608
01049736
0-1049009 0-1048284
0-3360206 o-3357799 o-3355398 0-3353002 o-335o6i 1 0-3348226 o-3345845 o-3343469 0-3341098 o-3338733
o-3336372 O-33340I7 0-3331666 0-3329320 0-3326979
:&*
14-68 14-70
01047561
0-1046839 0-1046119
1480
14-82
01042539
0-1041827 0-1041117 0-1040408 0-1039701 01 038995
i 4 -88
1490
1492 1494
14.96 14.98
0-3218483 0-3216362
03214245
0-3212132 0-3210024
03317665
0-33I5349
32690174
1500
713
X
1502 1504 1506
15-08 15-10 15-12 15-14 15-10 15-18 15-20
e-*h(x)
0-1038291 ** An 0-1037588 0-1036087 0-1036186 0-1035488
e-'I^x)
0-1003109 0-1002478 0-1001847 0-1001218 0-1000590
0-0999964 0-0999338 0-0998714 0-0998090 0-0997468
e*K
(x)
*K
{x)
e*
03207019
0-3205819
03203723
0-3201631 o-3i99543
03303839
0-330I552 0-3299269 0*3296990 0-3294717 0-3292448
1502
15-04 15*06 15-08
1510
15-12 15*14 15-10 15*18 15-20
0-3I97459 0-3I95379 0-3193303 0-3191231 0-3189164 0-3187100 0-3185040 0-3182985 '3 l8o ^23 0-3178885
0-3176841 03 1 74801 0-3172766
1522 1524
15-26
1528
15-30
0-1031325 0-1030636 0-1029949 0-1029203 0-1028578 0-1027895 0-1027213 0-1026532 0-1025853 0-1025175
1522
15-24
1526
15-28
1530
1532
15-34 15-30 I5-38
1532
15-34 i5-3f I5-38 15-40
I5-42
J 5-44
00993146 00992533
0-0991921 009913 10 0-0990701 0-0990092 0-0989485 0-0988879 0-0988274
03170734
0-3168705
0-3166681 0-3164661 0-3162644 0-3160632 0-3158623
1540
1542
15-44 15-40 I5-48 I5-50
15-46 I5-48
1550
I5-52 15-54
1556 1558
15-60
15-62 5 -66 15-68 15-70
1
O3252392
0-3250210
O324303I
0-3245857 0-3243687 0-3241522 0-323936I
1552
15-54 I5-56 I5-58
1560
15-62 15-64 15-66 15-68 15-70 15-72 15-74 15-76 15-78 i5-8o 15-82 I5 'ii 15-86 15-88 15-90
1564
00982877
0-0982283
03138743
0-3136776 0-3134812
O32372O4
0-3235052 0-3232903 0-3230759 0-3228620 0-3226484
0-1014507 0-1013851
01013197
0-1012544 0-1011892
0-1011241 0-1010592 0-1009944 0-1009297 0-1008651
1580
15-82
03132852 03130896
0-3128943
0-3126994
O3224353
0-3222226 0-3220I03 0-3217085 0-3215870 0-3213760
03125049
0-3123107 0-3121169 0-3119235
00976399
0-0975816 0-0975235 0-0974654 0-0974075
1592
15-94
1596
15.98 16-00
00973496
O32I I654
0-3209552 0-3207454 0-3205360
1592
15-94 15-96 15-98 16-00
714
X
o
0-02 0*04 o-oo
/1/3W
O-OOOOOOO
-
*W*)
00
*(*)
00
-90
86 23' 4*72 84 8' 47*63 82 12' 15*40 8o 25' 14*29 78 44' 23*54
77^ 7; 75 34 4 74 72 36 9' 71
OO8
o-io
0-I2 0-14 O-IO
+ + + + + + + + + + + + + + +
+ + + + +
0-2412455 0-3038819 0-3477275 0-3825227 0-4117819 0-4372223 0-4598264 0-4802143 0-4988049 0-5158967
- 3-8181573 - 2-9641628 - 2-5398832 - 2-2665744 - 2-0682566 1-9140102 I -7884275 1-6828031 1-5917752 1-5118289
008
o-io
0-12 0-14 0-16 0-18 0-20 0-22 0-24 0-26 0-28
Ol8
0-20
0-22 0-24 O-20 0-28
" -
29795927
2-7996089 2-6511003 2-5256038 2-4175728
030
0-32 o-34 0-36 0-38 0-40
0-42 0-44 0-46 0-48
030
0-32 o-34 0-36 0-38 0-40
0-42 0-44 0-46 0-48
1-1665964 1-1227224 1-0813409 1-0421378 1-0048529 0-9692681 0-9351991 0-9024892 0-8710041 0-8406278 0-8112601 0-7828134 0-7552112 0-7283861 0-7022788
4 -57
57 4i 35 T i3
050
0-52 o-54 0-56 0-58 o-6o 0-62 0-64
+ + + + + + + + + +
56 24' 23*72
55 7' 3"oi 53 5i 15*93 52 35' 15*65 5i 19 35*54
5 oo
4'
1-7608136 1-7252429 1-6918274 1-6603536 1-6306366 1-6025156 I-575850I 1-5505163 1-5264049 1-5034188
1-4814718 1-4604863 1 4403931 1-4211296
1
050
0-52 o-54 0-56 0-58 o-6o 0-62 0-64 o-66 o-68 0-70
M'M
066
o-68 0-70
0-72
+ + + + +
+ + + + + + + + + + + + + + +
0-6768367 0-6520129 0-6277661 0-6040589 0-5808580 0-5581337 0-5358591 0-5140100 0-4925646 0-4715032
4026393
074
0*76 0-78 o-8o
3201469
082
0-84 o-86 o-88 0-90
0-92 0-94 0-96
082
0-84 o-86 o-88 0-90
0-92 o-94 0-96
098
I-OO
03523093 03335201
0-3150111 0-2967741 0-2788016
098
I-OO
To compute
by 6o c
715
7i/sW
1-02
I I
Yu*(*)
<(*)
o-7749574 0-7678666 0-7609657 0-7542466 0-7477012 0-7413222 0-7351026 0-7290360 0-7231162 o-7i73372
-
-OA
+ 0-7296524 + 0-7281940
+ 0-7265045 + 0-7245872 + 0-7224452 + + + + +
0-7200818 0-7175000 0-7147030 0-7116937 0-7084752
I-OO
08
I-IO
I-I2
0-2610869 0-2436239 0-2264069 0-2094308 0-1926912 0-1761839 0-1599051 0-1438514 0-1280198 0-1124076
0-0970123 0-0818317 0-0668639 0-0521072
08
I-IO
I-I2
2 I-IO
1-18 1-20
1-22
I-I
1-1309205 1-1216703 1-1126469 1-1038412 1-0952444 1-0868482 1-0786451 1-0706275 1-0627885 I-055I2I5
I -0476204 1-0402790 1-0330918 1-0260535 I-OI9I588
\\l Il8
1-20
1-22 1-2A 1-26 1-28 I-30 1-32 i-34
t 1-20 1-28
I I 1
*' 2
30
32
+ + + + +
+ + + + + + + + + +
0-7050506 0-7014229 0-6975950 0-6935699 0-6893506 0-6849400 0-6803413 0-6755573 0-6705909 0-6654453
0-6601234 0-6546281 0-6409626 0-6431297 0-6371326 0-6309743 0-6246578 0-6181862 0-6115625 0-6047900 0-5978715 0-5908104 0-5836090 0-5762725 0-5688020
00375600
4i7'47'4i
3 7
7*45
1-34
36
138
1-40
i 56' 30*14
0 45' 55*39 o 24' 36*88 i 35' 6*76 2 45' 34*33 3 55' 59-6 7 6' 22*83 5 6 16' 43*91 7 27' 2*94 8 37' 20*02
136
1-38 1-40
1-42 1-44 1-46 1-48 1-50
I
42 44 42 48
50
1 52
+ + + + +
+ + + + +
+ + + + + + + + + +
+ + + + +
i-6o
1-62 1-64 1-66 1-68 1-70
+ + + + + + + + + + + + + + + + + + + + + + + + +
0-1089100 0-1210079 0-1329039 0-1445980 0-1500900 0-1673799 0-1784675 0-1893528 0-2000354 0-2105152
0-2207919 0-2308653 0-2407351 0-2504011 0-2598629
0-2691204 0-2781733 0-2870212 0-2956640 0-3041014
9 10 12 13 14
47; 35/18
52
06135590
0-6100034 0-6065083 0-6030722 0-5996933 0-5963702 0-5931013 0-5898852
0-5867204 0-5836056 0-5805395 0-5775209 o-5745485
0-5716212 0-5687379 0-5658974 0-5630987 0-5603409
72 74
.76
:g
82
21 28' 33*69 22 38' 31*26 23 48' 27*48 24 58' 22*40 26 8' 16*05
09053239
0-9005961
05213588
0-5130449 0-5046236 0-4960979 0-4874713
0-4787471 0-4699285 0-4610189 0-4520215 0-4429398
% 88
90
92
1-96 1-98 2-00
27 28 27' 29 37' 3o 47
31
57'
7'
18'
08959423
0-8913605 0-8868489 0-8824057
0-8780291 0-8737176 0-8694694
186 188
1-90
1-92 1-94 1-96 1-98 2-00
+ + + + +
03123332 03203591
0-3281790 0-3357927 0-3432000
33 34 35 36 37
16
26' 36' 46'
08652832
0-8611573
= =
- 0-0757500.
+0-5551971.
716
X
2-02 2*04 2-06 2-08 2-IO
2*12 2-XA 2-l6 2-l8 2-20
/j/W
+ + + + + + + + + + + + + + + + + + + +
0-4337771 0-4245367 0-4152219 0-4058363 0-3963830
Vi/sW
+ + + + + + + + + +
+ + + + +
0-3504008 o-3573949 0-3641824 0-3707631 0-3771371 0-3833043 0-38926^7 0-3950165 0-4005657 0-4059065
0-4110411 0-4159696 0-4206923 0-4252096 0-4295216
*!>)
X
2-02 2-04 2-06 2-08 2-IO 2-12 2-14 2-IO 2-l8 2-20
38 55' 5**34
40
o-5324394
05301033
o-5277974 0-5255212 0-5232740
0-52IP55 1 0-5188641 0-5167002 0-5145631 0-5124521
230
2-32 2-34 2'36 2-38 2-40
2-42 2-44 2-46 2-48 2-50
50 32' 0*49 51 41' 33*06 525i' 4*91 o' 36*06 54 55 10' 6*52 56 19' 36-32 5'46 2 5Z 9; 5 S 38 33;95 59 48' 1*81 6o 57' 29*06
6' 55*70 62 63 1 6' 21*75 64 25' 47*23 65 35 12*14 66 44' 36*50
08056325
0-8023043 0-7990173 0-7957706 0-7925634 0-7893949 0-7862643 0-7831710 0-7801140 0-7770928 0-7741066 o-77ii547 0-7682366 o-76535i5 0-7624989 0-7596781
+ 0-4336289
+~o-43753i8 + 0-4412307 + 0-4447262 + 0-4480187
0-2084181
+ 0-1983209 + + + + +
+ + + + + + + + + + + + + + +
-
+ + + + + + + + + + + + + + + + + + + + + + + + +
67 54'
0*31
69
2-52 2-54 2-56 2-58 2-60 2-62 2-64 2-66 2-68 2-70
278
2-80
2-82 2-84 2-86 2-88 2-90 2*92 2-94 2 96 2-98
0-0881509 0-0783000 0-0684899 0-0587238 0-0490046 0-0393353 0-0297189 0-0201583 0-0106564 0-0012161 0-0081598 0-0174085 0-0267073 0-0358733 0-0449638
07356309
0-7330983
280
2-82 2-84 2-86 2-88 2-90 2-92 2-94 2-96 2-98
0-4717009 0-4707537 0-4696281 0-4683256 0-4668480 0-465197 0-463374 1 0-4613813 0-4592203 0-4568930
59"2i
0-7305919 0-7281111
07256555
0-7232247 0-7208183
91 92
93 94 28'
300
0*28 95 37' I3 ? 8l
300
To compute
functions of order
by 6o
717
X
3-02
AaW
- 0-0539763 - 0*0629080 - 0-0717564 - 0-0805188 - 0-0891928
-
Ym(*)
+ 0-4544013 + 0-4517471
+ 0-4489323 + 0-4459590 + 0-4428292
hJJw
a*g #i/3
,/.(*)
0-7068697 0-7046231 0-7023978 0-7001936 0-6980101
X
3-02
304
3'o6 3-08
96 46' 26*99
97 55' 39*85 99 4' 52*38 IOO 14 / il59 101 o 23' 16*49
102 32' 28*08 103 41' 39*37 104 50 50*36 1*05 106 9' 11*46 107
304
3*o6
308
3'io
3'12 3-14
310
04517223
0-4502888 0-4488687 0-4474619 0-4460682 0-4446874 0-4433192
3M 316
3-i8
312
320
3'22 3'24 3-26 3-28 3-30 3'32
0-0977759 0-1062656 01 1 46595 0-1229552 0-1311505 0-1392429 0-1472303 0-1551105 0-1628813 0-1705405
0-1780862 0-1855162 0-1928286 0-2000215 0-2070929
0-2140411 0-2208642 0-2275605 0-2341283 0-2405659
+ + + + + + + + + +
0-439545I
04361086
0-4325221 0-4287877 0-4249076
316
3-i8 3-20
3*22 3-24
04419636
0-4406202 0-4392890 o-4379697
0-4366621 o-435366i 0-4340816 0-4328083 0-4315461
108 18' 21*58 109 27' 31*42 110 36' 40*99 in 45' 50*29 112 54' 59*32
4' 8*09 1 1 115 13' 16*61 22' 116 24*87 117 31' 32*89 118 40' 40*66
326
3-28
330
3'32
334
3-36 3'38
+ + + +
334
3-36 3-38 3-40
340
3-42
+
+ + + + + + + + + + + + + + +
0-4302948
344
3-46
04290543
0-4278244 0-4266049 o-4253958 0-4241969 0-4230080 0-4218290 0-4206598 0-4195001 0-4183500 0-4172093 0-4160778 -4 I 49554 0-4138420
0-4127375 0-4116417 0-4105546 0-4094760 0-4084058
119 49' 48*19 120 58' 55*48 *22 8' 2*54 123 17' 9*37 124 26' 15*97 125 35' 22*35 1 26 44' 28*51 127 53' 34*46 2 40*19 129 130 ii' 45*72
131 132 133 134 135
20' 51*03 29' 56*15
342
3-44 3-46 3-48 3-50
3-52 3-54 3-56 3-58 3"6o
06582379
0-6564241 0-6540254 0-6528413 0-6510719 0-6493168
360
3*62 3-64 3-66 3-68 3'7
0-2468718 0-2530444 0-2590821 0-2649836 0-2707474 0-2763722 0-2818568 0-2871997 0-2924000 0-2974564
362
3*64 3-66 3-68 3-70
372
3'74
- 0-^023678
370
3-78 3'8o
3-82 3-4 3 -86 3-88 3'90
03071333
0-3117518 0-3162224 0-3205442 0-3247164 0-3207383 0-3326092 0-3363283 0-3398952
+ 0-2809376 + 0-2740767
4 0-2671440
+ 0-2601423 + 0-2530746 + + + +
+ + + +
+-
6' 14*63 137 138 15' 18*78 139 24' 22*73 140 33' 26*51 141 42' 30*10
372 374
3.76 3-78 3-8o 3-82
M2
51' 33*52
34
392
3-94 3 96 3-98 4-00
03991361
0-3981444
148 36' 48*01 149 45' SO'4 1 1 50 54' 52*65 J52 l 3 5 Jl?3 153 12' 56*65
392
3-94
3-96 3'98 4-00
x cos 213 12' 56*65 = -0-3330932. x sin 213 12' 56*65 = -o-2i8ioo8.
718
/1/3W
4-02 4-04 4-06 4-08 4-10
4-12 4-14 4-16 4-18 4-20
YutW
+ + + + + + + + + + + + + + +
0-1718183 0-1641895
tfSw
arg
H m (x)
01565333
0-1488527 0-1411506
0-1334301 0-1256940 0-1179455 0-1101873 0-1024224
03942503
0-3932945 0-3923455 0-3914034 0-3904679 0-3895391 0-3886169 0-3877012 0-3867919 0-3858890
408
4-10
60
61 62 63 64
16'
7,
0-3748770
o-3759693 0-3769073 0-3776914 0-3783220 o-3787997 0-3791248 0-3792981 0-3793201 0-3791916 0-3789I3 1
25 34 43
4-20
4-22 4-24 4-26 4-28 4-3o 4-32 4'34 4-36 4'38 4-40
4-22 4-24
426
4-28 4-30
03849923
0-3841018
0-3832175 0-3823392 0-3814669 0-3806006 0-3797400 0-3788853 0-3780363 -377 I 93 o-3763553 o-375523i
432
4-34 4'3 4-38 4-40
4-42 4-44 4-46 4-48 4-50
+ + + + +
< K K
46
u
55,
75 u 4 7 13
05903159 05890079
0-5877086 0-5864178 0-5851356 0-5838617
05825961
0-5813387 0-5800895 0-5788483 0-5776151 0-5763897
0-5751721 0-5739622 0-5727599 0-5715652 0-5703779
0-3746963 o-373875o
03730591 03722485
0-3714431
83 84 85 86 87
6'
03598033
0-3573972
0-3706429 0-3698478 0*3690578 0-3682729 0-3674929 0-3667178 0-3659476 0-3651822 0-3644216 0-3636657 0-3629145 0-3621679 0-3614258 0-3606883 0-3599553 0-3592267 0-3585026 0-3577827 0-3570672 o-3563559
-
90
J& 4-68
4-70
4-72 4'74 4-76 4-78 4-80 4-82 4-84 4-86 4-88
4-72 4-74 4-76 4-78 4-80 4-82 4-84 4-86 4-88 4-90
4-92 4*94 4-96 4-98 5-00
03434221
0-3402493
0-5691980 0-5680255
05668601
0-5657019 0-5645508
98 99
3'
29*09
12' 26*17
03369544 03335393
0-3300057 0-3263554
-
200 2l' 23*16 201 30' 20*06 202 39' 16*87 20 3 48' I3-58 204 57' IO*2I
05634067
0-5622696 0-5611393 0-5600158 0-5588991
490
4-92 4-94 4-96 4-98 5-00
03225903
0-3187124 0-3147234 0-3106254 0-3064205
6'
6*75
41' 52*06
To compute
functions of order
1/3, increase
the phase
719
/1/3W
5'02
-
*W*)
- 0-1876576 - 0-1932947
-
*!>>
**S
, /s
(*)
^l/3(*)
504
5-06 5-o8
0-3021105 0-2976976
0-3556489
02931838
0-2885714 0-2838623 0-2790589 0-2741632 0-2691776 0-2641042 0-2589454 0-2537034 0-2483807 0-2429794 0-2375020
510
512 5-M 516
5-18
03549460 03542473
0-3535527 0-3528621
211 50' 48*18 212 59' 44*22 8' 40*17 214 215 17' 36*05 216 26' 31*84
217 35' 27 T 56 218 44' 23*20 219 53' 18*76 221 2' 14-25 222 II' 9*67 223 20'
5*OI 29' 0-27 37' 55 ;46
05523364 05512650
.0-550I998
0-5491408 0-5480878
508 510
5-12
520
522
5-24 5-26 5-28
03494688
0-3488018
03481386
o-347479i 0-3468234 0-3461714
"
02527435
0-2569696
530
532
5'34 5-3
02319509
02263285
0-2206371 0-2148793
46 50*59 55' 45 T 6 4
4'
538
5*40
5-42 5 44
02090575
0-2031741
40-62
13' 35 ? 53
5H6
5-48 5'50 5-52 5'54 5-56 5-58
01790751
0-1729216 0-1667216 0-1604777 0-1541924 0-1478684 0-1415082
0-1351143 01 286894 0-1222361
234 49' 14*49 235 58 9*07 237 7' 3-58 238 15' 58-02
23924'52-40
240 33' 46 : 72 241 42 40*97 242 51' 35*16 o 29*30 244 245 9' 23-37
560
5-62 5*64 5'66 5-68
552
5*54 5-56 5-58 5-60
5-62 5-64 5-66 5-68 5-70
5-72 5-74 5-76
05235899
0-5226772 0-5217693 0-5208661 0-5199676 0-5190737
o-"57569
0-1092543
0-1027311 0-0961898 0-0896330 0-0830632 0-0764830 0-0698951 0-0633019 0-0567061 0-0501102
570
5'72 5-74
576
5-78 5-8o
5-82 5-84 5-86 5-88 5*90
2'
V78
5-80
5-82 5-84 5-86 5-88
5
03237134
0-3243639 0-3248832
00435167
0-0369283
03252714
0-3255288
0-3256557-
05129432
0-5120850 0-5112312 0-5103816
0-5095362 0-5086951 0-5078581 0-5070252 0-5061964
260
5'
"06
90
592
5*94 5-96 5-98 6oo
00303473
0-0237764 0-0172181 0-0106747
0-3256526
03255199 03252580
(6-oo)
5*92 5-94
J-v*
7'
+0-2763441.
y_i/s (6-)
7'
12*96= -0-1718736.
720
/1/3W
6-02 6-04 6-o6 6-o8 6-io
*!/(*)
#1/3 (*)
argH, /3 (*)
269 16' 5*89 270 24' 5877 27 l0 33' 5i ? 6o 272 42' 44*39 273 5i 37 1
+ + + +
"
+ + + + +
6' 29*81 275 276 9' 22*46 277 18' 15*05 278 27' 7*60
2 79 36' 2
O-II
+ + + + +
+ + + + + + + + + +
0-3I56255 0-3I5I335
286 29' 14*22 287 38' 6*42 288 46' 58*58 289 55' 50*70
291
4 42*77
o-4934659 0-4927018 0-4919413 0-4911843 0-4904308 0-4896807 0-4889341 0-4881909 0-4874510 0-4867145 0-4859814 0-4852510 0-4845250 0-4838017 0-4830817
0-4823648 0-4816511 0-4809406 0-4802333 0-4795290
tit
6-48 6-50
6-52
292 13' 34*80 293 22' 20*80 294 3 1 ' l8 *75 295 40 10*67 29649' 2*54
29 ? 299 300 301 302
57' 54-38
6-42
t
6-48 6-50 6-52 6-54 6-56 6-58 6-6o 6-62 6-64 6-66 6-68 6-70 6-72 6-74
670
672
+ + + + + + + + + + + + + + +
0-1719453 0-1768080 0-1815856 0-1862766 0-1908793 0-1953922 0-1998139 0-2041429 0-2083778 0-2125171
0-3098690 0-3094033 0-3089398 0-3084783 0-3080189 0-3075615 0-307I062 0-3066528 0-30620I5 0-3057522 0-3053048 0-3048593 0-3044159 0-3039743 0-3035347
*7j 67
309 26' 30*66 310 35' 22*08 3" 44' 13*47 312 53' 4*83 1' 56*15 314
3 1 5*
1 ! 47*44 316 19 38*70 317 28' 29*92 318 37' 21*11 319 46' 12*26
676
6-78 6-8o 6-82 6-84 6-86 6-88 6-90
6-92 6-94 6-96 6-98 7-00
6-82 6-84 6-86 6-88 6-90 6-92 6-94 6-96 6-98 7-00
55'
3*38
21 36*55
30' 27*54
To compute
by
6o.
721
X
7-02 7-4 7-06 7 08 7-10
/i/j()
*W*)
-
ff Siw
afg
l, 3
e?K*z(x)
+ + + + +
+ + + + +
0-3009362
03005095
0-3000847 0-2996617 0-2992404
326 39' 18*50 3 2 Z48' 9'42 328 57' 0*32 330 5'5i"i9 331 14 42*03
332 333 334" 335 336
23' 32' 41' 50' 58'
7'
7 02
7-20
7-22 7*24 7-26
+ + + + + + + + + + + + + + + + + + + + + + + +
+ + + + +
0-2967496
02963405 02959330
0-2955273 0-2951232 0-2947207
04603654
0-459745
0-4591271 0-4585116 0-4578986
f-28
730
7.32 7-34 7-36 7-38 7-40 7-42 7*44 7-46 7-48
732 734
7'36 7*38 7-40
7-42 7'44 7-46 7-48 7'50
0-0818961 0-0761070
02943199
0-2939207
00703033
0-0644874
- 0-0586615
02935231
0-2931272
343 51' 59*28 345 49*77 346 9 40*23 8' 30*66 347 1 348 27' 21*07
349 36' 11*45 3 5 o 45' 1*80 351 53' 52*12 353 2' 42*42 354 11 32*70
04572881
0-4566801
+ 02896766
00353042
0-0294630
750
7-52 7*54 7-56 7-58 7-60 7-62 7-64 7-66 7-68 7.70 7-72 7-74 7.76 7-78 7-80 7-82
0-0236256
00177943
0-0119713 0-0061589
02903995
0-2900160 0-2896341 0-2892536
0-2888746 0-2884971 0-2881211 0-2877465 0-2873734
0-2870018 0-2866315 0-2862627
00003594
0-0054250 0-0111920 0-0169396 0-0226654 0-0283672
04530819
0-4524905 0-4519013 0451 3145 0-4507299
46 53*54 55 43 ? 69
rU
7-68 7-70
7'72 7V74 7.76 7-78 7-80 7-82
02868525
0-2859699
+ + + + +
4' 33*81 361 362 13' 23*91 363 22' 13*98 3643i' 4*03 365 39' 54*05
0-4501476 0-4495676
04489898
0-4484142 0-4478408
+ + + + + + + + + + + + + + +
00340430
0-0396905 0-0453076 0-0508922 0-0564422
02858954 02855294
0-2851648 0-2848017 0-2844399 0-2840794 0-2837204 0-2833627 0-2830063 0-2826513 0-2822976
366 48' 44-05 367 57' 34*o3 369 6' 23-98 370 15' 13*91 371 24' 3-82
372 373 374 375 377
32' 53*70 4i 43*56 T 50, 33 40
04472697
0-4467007
04461339
o-4455692
04450067
o-4444463 0-4438880 o-44333i8 0-4427777 0-4422257
o-44i6757 0-4411278 0-4405819 0-4400381 o-4394962
+ 0-2783532
+ 0-2767042 + 0-2749490 + 0-2730886 + 0-2711241
7-86 7-88
790
7.92 7-94 7-96 7-98
59 23-22
8'
13*01
2*7.8
800
+ + + + +
378i7'
02819453
T 3Z9 2 5 52 53 380 34' 42-25 381 43' 31*96 382 52' 21*64
800
J-ila (8-oo)
y_i/( 8 o)
=0-2819453 x cos 442 52' 21*64 = +0-0349823. =02819453 x sin 442 52' 21*64= +0-2797667.
722
X
8-02 8-04 8-o6 8-o8 8-io 8-12 8-14 8-i6 8-i8 8-20 8-22 8-24 8-26 8-28 8-30 8-32 8-34 l % 8.38 8-40
/1/3W
+ + + + +
0-2572095 o- 2 545467 0-2517890 0-2489377 0-2459942
Y V3 (x)
+ + + + +
+ + + + + + + + + +
0-1146236 0-1196012 0-1245187 01 293743 0-1341664
0-1388931 0-1435528 0-1481438 0-1526645 0-1571133
!>)
ar g
H m (x)
**i/s(*)
X
8-02 8-04 8-06 8-o8 8-io
1' 11*30 384 385 10' 0*94 386 18' 50-56 387 27' 40*16 388 36' 29*74
o-43 8 9564
+ + + + + + + + + +
+ + + + +
0-2429598 0-2398360 0-2366242 0-2333259 0-2299425 0-2264758 0-2229271 0-2192981 0-2155904 0-2118057
45' I9 ? 29
04352324
0-4347081 0-4341857
0-1614886 0-1657888 0-1700125 0-1741581 0-1782243 0-1822096 0-1861127 0-1899322 0-1936668 0-1973152
0-2008763 0-2043488 0-2077315 0-2110234 0-2142234
830
8-32 8-34 8-36 8-38 8-40 8-42 8-44 8-46 8-48
+ + + + + + + + + + + + + + + + + + + +
401 13' 33*76 402 22 23*10 4033i' 12*42 40 4 4o' 1*73 405 48' 51*01
04310905
0-4305810 0-4300733 0-4295675 0-4290634
0-4285611 0-4280605 0-4275617 0-4270646 0-4265693
+ + + + + + + + + +
+ + + + +
57' 40*27
6'
850
8-52 8-54 8- 6 5 8-58 8-6o
8-62 8-64 8-66 8-68 8-70
I't 8-6o
8-62 8-64 8-66 8-68 8-70
8-72 8-74 8-76
0-2173304 0-2203434 0-2232615 0-2260837 0-2288092 0-2314370 0-2339664 0-2363966 0-2387270 0-2409567
0-2430852 0-2451118 0-2470360 0-2488572 0-2505751
0-4260757 0-4255838
04250936
0-4246051 0-4241183 0-4236331 0-4231496 0-4226678 0-4221876 0-4217090
0-4212321 0-4207568 0-4202831 0-4198109
2*83
418 25' 51*92 419 34' 40*99 420 43' 30*04 421 52' 19*08 1' 8*10 423
424 9' 57*10 42 5 18' 46*08 426 27' 35*04 427 36' 23*99 428 45' 12*93
4 2954' 1*85 2' 50*75 431
878
8-8o
8-82 8-84 8-86 8-88 8-90
8-92 8-94 8-96 8-98 9-00
+ + + + +
+ + + + + + + + + +
+ + + + +
+ + + + + + + + + +
04193404
0-4188715 0-4184041 o-4i79383 0-4I74740
0-2521890 0-2536987
02551038
0-2564039 0-2575988
0-2586882 0-2596720
02673365
0-2670371 0-2667388 0-2664414 0-2661450 0-2658496
04170113
0-4165501 0-4160905 0-4156323 o-4i5i757 0-4147206
02605500
0-2613221 0-2619882
898
9-00
To compute
by 6o
723
X
9-02 9-04 9-06
/1/3M
+ 0-0398497 + 345485 + 0-0292452 + 0-0239420 + 0-0186408
+ 00133439 4- 0-0080532 + 0-0027709 - 0-0025010 - 0-0077604
-
^i/sto
H%{*)
1
*#,(*)
X
9-02
9 08
9-10
9-12
+ + + + +
+
0-2625482 0-2630023
0-4142670 0-4138148
4 # 3&"3& 57 25*06
6' 13*75
04133642 04129150
0-4124673
914
9-16 9-18 9-20
9-22 9-24 9-26 9-28
02635213
0-2632346 0-2629487
914
9-16 9-18 9-20
9-22 9-24 9-26 9-28
04102503
0-4098112 0-4093735 0-4089372
00130053
0-0182336
00234434
0-0286326
930
9.32
00337991
0-0389411 0-0440566 0049 1 435 0-0541999
02593402
0-2572281
02615334
0-2612531 0-2609737 0-2606951 0-2604175 0-2601407
04085023
0-4080687
93O
9.32 9-34 9-36 9-38
00592239
0-0642137 0-0691672 0-0740826 0-0789581 0-0837918
458 34' 19*84 459 43 8*37 460 51' 56*89 462 0' 45*40 463 9' 33*9
04063483
0-4059216
940
942
9*44 9-46
+ 0-2518062
+
0-2598648
944 946
9-48
0-2502055
02595898
0-2593157 o- 2 590424 0-2587700
95
9-52 9*54
464 18' 22*37 465 27' 10*83 466 35' 59*28 4Zl 44' 47*72 468 53' 36*15
24*56 470 471 11 12*96 472 20' 1*35 473 28 49-72 474 37' 38-o8
475 46' 26*43 476 55 14*77 478 4 3*09 479 12' 51*41 480 21' 39*71
2'
04054962
0-4050722
04046495
0-4042281 0-4038080
948
9-50
9-52 9*54
00885819 00933266
0-0980241 0-1026727 0-1072706
02407732
0-2386075
950
9-58 9-60
9-62
02363512 02340052
6-2315707 0-2290489 0-2264409 0-2237479 0-2209712
956 958
9-60
9-62
964 966
9-68 9-70
9.72 9-74 9-76 9-78 9-80
01118161
0-1163076 01 207433 0-1251217 0-1294411
+ + + + + + + + + +
+ + + + + + + + + +
964
9-66
9-68 -9-70
9.72 9-74 9-76 9-78
01 336999
0-1378966 0-1420297 0-1460977 0-1500990
02553050
0-2550442 0-2547842
48 1 482 4 83 484
4 8'
4*53
03984637
0-3980614 0-3976602
0-3972603 0-3968616
980
982
9-8 4
990
992
9*94
03964641
0-3960677 0-3956726
9-86 9-88
990
9-92
996 998
io-oo
01852836
0-1816090 0-1778693 0-1740662 0-1702011
03952786
0-3948858 0-394494 2 0-3941037 o-3937i44
02527323
0-2524792 0-2522270
9 94 9 96
998
io-oo
= =
-0-2404711.
-00761059.
724
X
10-02 10-04 10-06 10-08 IO-IO
Jxnix)
-
*W*)
+ 0-1662759 + 0-1622921
+ 0-1582514 + 0-1541556 + 0-1500064
hSw
arg
H m (x)
etKvzi*)
X
10-02 IO-04 10-06
0-1893250 0-1924230 0-1954380 0-1983690 0-2012150 0-2039750 0-2066482 0-2092336 0-2117305 0-2141379
0-2164551 0-2186814 0-2208160 0-2228584 0-2248078
498 42' 30*90 499 51' 19*00 501 o' 7*09 8' 55*17 502 503 17' 43-24
03933263
o-3929393 o-3925534 0-3921687 0-39I785 1
IO08
IO-IO
IO-I2
+ + + + + + + + + +
0-3914026
03910213
0-3906411 0-3902619 0-3898839
ICI4
IO-IO 10-18 10-20
IO-22 IO-24 IO-26 IO-28 IO-30 IO-32 10-34 10-36
02497452
0-2495010 0-2492576 0-2490148 0-2487728 0-2485314
0-2482908 0-2480509 0-2478116 0-2475731 0-2473352
IO3O
IO-32
0-3895070
03891311
0-3887564 0-3883827 0-3880101 0-3876386 0-3872681 0-3868987 0-3865304 0-3861631
514 45 43-35
515 54' 3i"3 3' J 9 : 2s 517 518 12' 7?i8 519 20' 55'iii 520 29' 43^03
38' 30*93 522 47' 18-82 56' 6*70 523 525 4 '54 r 58 526 13' 42^44
1034 1036
10-38 10-40 10-42 10-44 10-46 10-48
+ 0-1013503
+ 0-0967006 + 0-0920212 + 0-0873142 + 0-0825814
1038
10-40
10-42
1050
1052
io-54 10-56 10-58 10-60
10-62 10-64
+ + + + + + + + + + + + + + +
52
03857968
0-3854316 0-3850674 0-3847043
1044
10-46
03843422
0383981
0-3836210 0-3832620 0-3829039 0-3825468
0-3821908 0-3818357 0-3814816 0-3811285 0-3807764
1048 1050
1052 1054 1056 1058
10-60
10-62 10-64
00342593
0-0293704 0-0244790 0-0195870 0-0146963 0-0098090
53o48'53 ? 8i 53i57'4i-62
533 6' 29*43 534 15' I7 T2 3 535 24' 5*02 536 32 52 ? 8o 537 41' 40*57
1066
io-68 10-70
10-72 10-74 10-76 10-78 io-8o
10-82 10-84
0-2429945
02447631
0-2445332 0-2443040 0-2440754 0-2438474
0-2436201 0-2433935 0-2431674 0-2429420 0-2427172
02433049
0-2435175 0-2436325 0-2436501 0-2435703 0-2433934 0-2431198 0-2427495 0-2422830 0-2417207 0-2410629 0-2403100 0-2394626 0-2385212
1066
10-68 10-70 10-72 IO-74 10-76 10-78 10-80
10-82
+ 0-0049268 + 0-0000518
- 0-0048141 - 0-0096692 00145113 - 0-0193387 - 0-0241495 - 0-0289417 - 0-0337136 - 00384633 -
538 50' 28*33 541 8' 3-83 542 16' 51*56 543 25' 39-29
544 34' 27*00 545 43 i4 r 7T 2" 4 I 546 52 0' 50*10 548 549 9' 37 ? 7
03804253
0-3800751 o-3797259 o-3793777
539 59 16-08
03790304
0-3786841
1086
to-88 10-90
10-92
03783387
o-3779942 0-3776507 o-3773o82 0-3769665 0-3766258 0-3762860 0-3759472
1084
io-86 10-88 10-90
10-92
1094
10-96
1098
II-OO
02405031
18'
25*46
1094
10-96 10-98 II-OO
44 48*43 53 36-07
03756092
by
6o.
To compute
725
X
II-02 II-04
1 1
/1/3W
-
^i/s(*)
<iw
arg
H U3 {x)
n
2*23*69
#
-oo
11-08 II-IO
556
11*31 557 558 19' 58-93 28' 559 46*54 560 37' 34*13
08
II-IO
11-12 11-14 ii-i6 ii-i8 II-20 11-22 11-24 11-26 11-28 II-30 11-32
11-12 11-14
02392033
0-2389887 0-2387747 0-2385613 0-2383484
0-2381361 0-2379244
n-i6
11-18 II-20
11-22 11-24 11-26 11-28 II-30 11-32
00973689
0-1016210 0-1058249 0-1099790 0-1140819 0-1181319 0-1221276 0-1260674
3-56*88
!2' 44*44 21' 32*00
- 0-2II2I89
O2OIO343
0-1982963 0-1954839 0-1925985 0-1896412 0-1866134
02377132
0-2375026 0-2372926
567 30' 19*55 568 39' 7*09 569 47' 54*62 570 56' 42*15 5' 29*67 572 573 14' 17*18 574 23' 4*68 575 3i' 52*17 5?6 40' 39 T 66 577 49' 27*14 578 580 581 582 583
58' 14*61
7'
03719509
0-3716236 0-3712972 0-3709716 0-3706469
0-3703230 0-3700000 0-3696779 0-3693565 0-3690360
"34
- o- 1 299499 - 0-1337736
02370831
0-2368741 0-2366658 0-2364579 0-2362506
V' 3 * 1138
11-40
11-42
?
1
"34 II36
II- 3 8
11-40
II-42
'44
11-46 11-48
1150
11-52
0-1484535 0-1519029 01 554055 0-1587801 0-1620853 0-1653202 0-1684833 0-1715738 0-1745904 0-1775320
2*08
0-3687164 0-3683975
n-44
11-46 11-48
15' 49*53
03680795
0-3677623
02352223
0-2350182 0-2348147 0-2346117 0-2344092 0-2342072
03674460
0-3671304 0-3668157 0-3665018 0-3661886 0-3658763
0-3655648 0-3652541 0-3649441 0-3646350 0:3643266 0-3640190 0-3637122
n-50
1152 "54 1156
1 1
"54
11-56
, 1158 n-6o
584 42' 11*86 5 8 5 5' 59-29 586 59' 46*70 8' 34*11 588 589 17' 21*52
590 591 592 593 595 597 598 599 6oo
26' 8*92 34' 56"3i 43' 43*69 52' 1 "7
1'
-58
n-6o
11-62 11-64
'-
01803977
0-2340058
02338049
0-2336044
02334046
0-2332052 0-2330063 0-2328070 0-2326101 0-2324127
n-66 n-68
11-70
11-72 11-74 11-76 11-78
1 1
18*44
u-74
11-76 11-78 11-80
11-82
OI935528
0-1959428
01982505
0-2004750 0-2026158 0-2046720 0-2066430 0-2085282 0-2103269 0-2120386
0-2136627 0-2151988 0-2166464 0-2180050 0-2192744
03634062 03631009
0-3627964 0-3624927 0-3621897 0-3618875 0-3615861 0-3612854
0-3609854 0-3606862 0-3603878 0-3600901 0-359793I
02322159
0-2320195
-So
02318237
0-2316283 0-2314335
02312391
6oi 54' 2*51 2' 49*84 603 604 "' 37 -i 605 20' 24*46 606 29' 11*76
'
11-82 11-84
n-86 n-88
1 1
-90
1192 "94
11-96 11-98 I2-00
02310452
0-2308518 0-2306589 0-2304664 0-2302745
4*20*90
13'
1198
12-00
612
8*17
J-va (12-00) =0-2302745 x cos 672 ^-i/3( r2 ')=o 23 02 745 xsn 672
-
726
X
-
/1/3W
0-0658652 0-0613901 0-0568980 0-0523905 0-0478696
-
Vi/sW
0-2204540
H, >)
*Kv{*)
613 21' 55*43 42*69 29*93 17*17 4*41
X
12-02 12-04 I2-O0 12-08 I2-IO
12-12 12-14 12-16
02215437
0-2225430 0-2234519 0-2242700
0-2256333 0-2261782 0-2266320 0-2269945 0-2272658 0-2274458 0-2275347 0-2275326 0-2274397
o-3594968
I2l8
12-20
12-22 12-24 12-26 12-28 I2-30 12-32 12-34 I2-36
I2- 3 8
- 0-2249971
619 5' 620 14' 38*86 621 23 25*08 622 32' 13*29 6234i' 0*50
b^M
03580265
o-3577346 o-3574434 o-357i529
1218
12-20
12-22 I2-24 12-26 12-28 I2-30
03568631
o-356574i 0-3562857 o-355998o 0-3557110 o-3554 2 48
624 49' 47*70 625 58' 34-89 7' 22*08 627 628 16' 9*26 629 -24' 56*43
6 30 33' 43"6o 631 42' 30*76 6325i' 17*92 6 3< 0' 5*07 8' 52*21 635
12-40
I2-42
+ + + + + + + + + + + + + + + + + + + +
12-32
1234
I2-36 12-38 12-40
I2-42 12-44 12-46 12-48 12-50
03540036
0-35372I5 0-3534400 o-353i59i 0-3528790 o-3525995
1244
12-46 12-48 12-50
12-52 12-54 12-56 12-58 12-60 12-62 12-64 12-66 12-68 12-70
0-2249866 0-2242647
02234544
0-2225562 0-2215705
640
642 643 644 645 646
0-0469744
00513356
0-0556694 0-0599741 0-0642479 0-0684894 0-0726967 0-0768684 0-0810028 0-0850983
0-2204979
i'34 ? 95
10 22*05 19' 9*15 27' 56*24 36' 43*33
02193390
0-2180943 0-2167645 0-2153502
1252
12-54 12-56 12-58 12-60
12-62 12-64 12-66 12-68 12-70 12-72 12-74 12-76 12-78 12-80
6 4Z 45' 3o-4i
03509363
0-3506614 0-3503871 0-3501135 0-3498405
12-72 12-74 12-76 12-78 I2-80 12-82 I2-84 12-86 12-88 12-90 12-92
+ 0-0891534 + 0-0931665
+ 0-0971361 + 0-1010607 + 0-1049388
02233176
0-2231429 0-2229687
0-2227948 0-2226214 0-2224484 0-2222757 0-2221035 0-2219317 0-2217602 0-2215892 0-2214186 0-2212483 - 1/3,
01967305
o- 1 944399
653 29' 25*73 6 54 38' I2'7 655 46' 59 ? 82 656 55 46-86 658 4' 33-89
659 66o 661 662 663 664 666 667 668 669
13' 20*92 22' 7*94 30' 54*96
03495681 03492964
0-3490254 0-3487549 0-3484851
+ + + + + + + + + +
39'4 I,r 97
48' 28*97 57' I5 T97
6'
1282 1284
12-86 12-88 12-90
12-92
1294
12-96 12-98 13-00
2*96
1294
12-96
1298 1300
To compute
functions of order
by 6o
727
7i/ 3
^i/.W
- 0-1676415 - 0-1645982 - 0-1614938 - 0-1583297 - 0-1551071
-
"!>)
13-02
1304
13-06 13-08 13-10
13-12 13-14 13-16 13-18 13-20 13-22 13-24 13-26 13-28
+ + + + + + + + + + + + + + + + + + + +
+ + + + +
[310
[3-12
03442514
o-34399i9 0-3437330 o-3434747 0-3432169 o-3429597 0-3427031 o-342447i 0-3421917 0-3419368 0-3416825 0-3414288 0-3411756 0-3409230 0-3406709 0-3404194 0-3401685 o-3399i8i 0-3396683 0-3394I90
0-3391702 0-3389221 0-3386744 0-3384273 0-3381808 o-3379347 0-3376892 0-3374443 o-337i999 0-3369560 0-3367126 0-3364698 0-3362275 o-3359857 o-3357444
1330
13-32
0*35
1334 1336
13-38
1340
13-42
01932433
0-1950745 0-1968252 0-1984946 0-2000822 0-2015875
0-2030101 0-2043496 0-2056054 0-2067772 0-2078647
52' 54*90
[332
^3-34 C3-36 [3-38 [3-40
[3-42
[
1344
13-46 13-48 13-50
3H4
9*92
11' 56*78
+ + + + +
1370
13-72
705
E370
[3-72 '3-74 [3-76 C3-78
1374 1376
13-78 13-80
13-82 I3-8 13-86 13-88
7M
7X5
02147445
0-2145891 0-2144341 0-2142795 0-2141251 0-2139712
23' 39*10
32' 25*89
52*31
[380
[3-82
1390
1392
13-94 13-96 13-98 14-00
+ + + + + + + + + +
0-3355037
03352635 03350237
o-3347845 o-3345459 0-3343077 0-3340700
32*99
19*76 6*52 53*28 40*03 26*78
+ + + + +
03338329
0-3335962 o-33336oo
^-i/s (14-00) =0-2132061 x cos 786 51' 26*78 xsin 786 51' 26*78
= =
+00837943.
+0-1960494.
728
X + + + + +
Ju( x)
0-2109787 0-2101933 0-2093250 0-2083743 0-2073417 0-2062277 0-2050330
Y lla (x)
+ + + + + + + + + +
0-0296652 0-0338559 0-0380272 0-0421773 0-0463046
!>)
ars
Hm (*)
o' 13*52
X
14-02 14-04 14-06 1 4 08 14-10
14-02 14-04 14-06 14-08 14-10 14-12 14-14 14-16 14-18 14-20 14-22 14-24 14-26 14-28
728
729
0*26
730i7'47"oo
731 26 33*73 732 35 20*46
733 44' 7 ? i8 734 52 53-90 1 40*62 736 737 10' 27*33 19' 14*04 738
+ + + + +
+ + + +
02037580
0-2024034 0-2009699
0-1978687 0-1962026 0-1944604 0-1926429
00625547
0-0665445
0-0705021 0-0744259 0-0783145 0-0821664 0-0859801
0-2122989 0-2121488 0-2119990 0-2118495 0-2117004 0-2115516 0-2114031 0-21 12549 0-2III070 0-2I09594
0-2I08I2I 0-2106652 O-2IO5I05 0-2I0372I 0-2I02261
0-33I9536
03317209
0-3314887 0-3312570 0-3310257
+ 0-1994581
1430
14-32 14-34 14-36 14-38 14-40 14-42 14-44 14-46 14-48 14-50 14-52 14-54 14-56 14-58 14-60
+ + + + + + + + + +
+ + + + +
0*75 36 47 ? 45 45 34 T i4 54 20*83 ? 3 7 52
11' 54-21 20' 40*89 29' 27*57
r
28'
03301057
0-3298769
0-3296485 0-3294206
1430
1432
14-34 14-36 I4-38 14-40
14-42 14-44 14-46 14-48 I4-50
+ + + + + + + + + +
03291932
0-3289663 0-3287398 0-3285138 0-3282883 0-3280632 0-3278386 0-3276145 0-3273908 0-3271676 0-3269448 0-3267225 0-3265006
0-3262792 0-3260583 0-3258378 0-3256177 0-3253981
0-1730454 0-1705256
7555'47 T 58
752 4 34 r 24 753 13 20*90 754 22' 7*55 755 30' 54 -20 756 39' 40*85 757 48 27*49 750 57, I4 ? i3 6 0*77 760 761 14' 47 ? 40 762 763 764 765
23' 34*03 32' 20*65 41' 7*27 49 53" 8 9 58' 40*51
7' 6'
+ + + + +
+ + + + +
+ + + +
+ + + + + + + + + + + + + + + + + + + +
I4-52 14-54 14-50 I4-58 14-60 14-62 14-64 14-66 14-68 14-70
14-72 14-74 14-76 14-78 14-80 14-82 14-84 14-86 14-88
0-2086394 0-2084969 0-2083548 0-2082I29 0-2080713 0-2079300 0-2077889 0-2076482 0-2075077 0-2073676
0-2072277 0-2070881 0-2069488 0-2068097 0-2066710
766
768 769
M-74
14-76 14-78 14-80
14-82 14-84 14-86 14-88 14-90 14-92 14-94 14-96 14-98 15-00
+ + + + + + + + + +
+ + + + +
770 25'
0-3251789 0-3249602 0-3247419 0-3245240 0-3243006 0-3240896 0-3238731 0-323657 0-3234413 0-3232261
0-3230112 0-3227969 0-3225829 0-3223694 0-3221562
773 51' 20*12 775 0' -7i 776 8 53*30. 777 17,39*88 778 26 26*46
779 35' 13-04
1490
14-92 14-94 14-96 14-98 15-00
02065325
0-2063943 0-2062564 0-2061187 0-2059813
780 43' 59*61 781 52' 46*18 783 i'32"75 784 io' 19*31
by
To compute
6o
729
/1/3W
+ + + + +
+ + + + +
0-0859558 0-0821423 0-0783610 0-0744336 0-0705416 0-0666265 0-0626899 0-0587336 0-0547589 0-0507677 0-0467614 0-0427416 0-0387101 0-0346684 0-0306181
>W*)
+ 0-1870386 + 0-1885953 + 0-1900745 + 0-1914758 + 0-1927988
<W
arg
1
H m {x)
5"8 7 27 52*42 36' 38-97 45' 25*52 54 12*07
2'
<?K VZ X
{
7 8 5 i9'
03219435
0-3217313 0-3215194 0-3213080 0-3210970 0-3208864 0-3206762 0-3204664 0-3202570 0-3200481
+ + + + +
58*61
1524
15-26
1528 1530
4 + + + +
+ + + + +
+ 0-1990697 + 0-1998342
+ 0-2005178 + 0-2011203 + 0-2016418
02043540
0-2042201 0-2040865
02039531
0-2038200 0-2036872
796 46' 51*28 797 55' 37"8o 799 4 24-32 8oo 13 10*84 8oi 21' 57*35
802 30' 803 39' 804 48' ^557; 807 5
43*86 30;37 16*87 3*37 49*87
03198395
0-3196314
03194237
0-3192164
03190094
0-3188029 0-3185968 0-3183911 0-3181857 0-3179808
0-3177763 0-3175721 0-3173684 0-3171650 0-3169621 0-3167595 0-3165573
I53 2 1534
I5-36 15-38 15-4
+ + + + + + + + + + + + + + +
+ + + + +
+
0-2020820 0-2024408 0-2027184 0-2029145 0-2030294 0-2030630 0-2030154 0-2028867 0-2026772 0-2023869
0-2020162 0-2015652
1532
15-34 15-36 I5-38 I5 40 15-42 15-44 I5-46 I5-48 I5-50
15-52 15-54 15-56 I5-58 15-60 15-62 15-64 15-66 15-68 15-70 15-72 15-74 15-76 I5-78 15-80
15-82 15-84 15-86 15-88
02035546
0-2034223 0-2032902 0-2031584 0-2030269 0-2028956 0-2027646 0-2026338
0-2025032 0-2023730 0-2022429 0-2021132 0-2019836
1542
15-44
I 5 - 4
+ 0-0062274 + 0-0021626
- 0-0018978 - 0-0059522 - 0-0099990
-
1548
15-50
808 14' 36*36 809 23' 22*85 8io 32' 9*34 81 1 40' 55*83 812 49 42*31 813 815 816 817 818
58' 28*79
02010343
0-2004237 0-1997338
0-1989651 0-1981178 0-1971924 0-1961894 0-1951093
16'
03163555
0-3161541 o-3i5953i
1562
15-64 15-66
1568
15-70
15-72 15-74 15-76 15-78 15-80 15-82 15-84
02013397
0-20I2II7 0-20I0839
819 42' 21-15 820 51' 7*61 82i59 , 54 t07 8' 40*52 823 824 17' 26-98
825 26' 13-43 826 34' 59*88 827 43' 46*32 828 52' 32*76 1' 19*20 830
0-3I57524
03155522
o-3i535 2 3 0-3151528 03 ! 49537
01939526
0-1927199 0-1914118 0-1900289 0-1885717
0-1870411 0-1854377 0-1837622 0-1820154 0-1801980
0-3M7549
0-3I45565 o-3i43586 0-3141609 0-3139637 0-3137668
0-31357.03
0057391
0-0611960 0-0649716 0-0687165
0-0724292 0-0761082 0-0797522 0-0833597 0-0869294
+ + + + + + + + + + + + + +
O2OO9563
0-2008290 0-20070I9
0-200575I 0-2004485 0-200322i 0-200I960 0-200070I
o- 1999445 0-1998191 0-1996939 0-1995690 0-1994442
1586
15-88 15-00
1590
15-92 15-94 15-96 15-98 16-00
1592
15-94
1596 1598
16-00
0-0904599
00939498
0-0973977 0-1008025 -1041627
836 53' 2' 44*18 838 839 11' 30*60 840 20' 17*01 841 29' 3*42
0-3127879 0-3125932
03123989
0-3122050 0-3120114
Y-va
W.
B. F.
= =
- 0-1993773. -00051662. 24
730
Jn (x)
/.(*)
/.w
+ + + + +
+ + + + +
0-0012490 0-0049034 0-0111659 0-0197347 0-0306040
0-0436651 0-0587869 0-0758178 0-69^5863 0-1149035
;.w
+ + + + + + + + + +
0-0000208 0-0001663 0-0005593 0-0013201 0-0025637 0-00^3997 0-0009297 0-0102460 0-0144340 0-0195634
0-0256945 0-0328743 0-0411358 0-0504977 0-0609640
/(*)
o-5
+ + + + +
o-i
+ + + +
0-2
03
0-4
05
o-6
o-6
11 09
I-O
i-i 1-2
+ + + + + + + + + + + + + + +
+ + + + +
+ + + + +
+ + + + +
+ + + + +
11 09
I-O
I-I
+ + + + + + + + + + + + + + + + + + + + + + + + +
+ + + + +
+ + + + + + + + + + + + + + + + + + + +
+ + + +
i-6
XI
1-9 2-0
2-1
2-2
23
2-4 2-5 2-6
0-3746236 0-3950587 0-4139146 0-4309800 0-4460591 0-4589729 0-4695615 0-4776855 0-4832271 0-4860913 0-4862070 0-4835277 0-4780317 0-4697226 0-4586292
0-4448054 0-4283297 0-4093043 0-3878547 0-3641281
0-3382925
0-0404526 0-0476471 0-0555957 0-0643070 0-0737819 0-0840129 0-0949836 0-1066687 0-1190335 0-1320342
0-1456177 0-1597218 0-1742754 0-1891991 0-2044053
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
21
2-2
23 24
2-5
u
2-9
30
3-i
+ + + + +
+ + + + + + + + + +
2-6
n
2-9
30
31 32 33
+ 0-3264428
0-3430664 0-3587689 0-3733889 0-3867701
0-3987627 0-4092251 0-4180256 0-4250437 0-4301715
0-4333147 o-4343943 o-433347o 0-4301265 0-4247040
3-2
33 34
3-5
34 35
3-6 3-8 3'9 4-0
4*1
3-6
37
3-8 3'9 4-o
4'i 4-2
+ + + + + + + + + + + + + + +
0-0896796 0-0994854 0-1098400 0-1207178 0-1320807 0-1439079 0-1561363 0-1687200 0-1816009 0-1947147
0-2079912
37
+ + + + +
+ + + + +
03105347
0-2810592 0-2500861 0-2178490
0-1845931 0-1505730 0-1160504 0-0812915 0-0465651
+ + + + +
+ + + + +
42
4*3
44 45
4-6
4-6
4-9 5-o
02213550
0-2347252 0-2400168 0-2611405
U 49
50
731
Jn (x)
/#(*)
'/,(*)
7iW
+ 0-440051 + 0-576725 + 0339059
- 0-066043 - 0-327579 - 0-276684 - 0-004683
/.(*)
/.(*)
JM
+ + + + +
0-000250 0-007040 0-043028 0-132087 0-261141 + 0-362087
+ 0-765198 + 0-223891
- 0-260052 - 0397150 - 0-177597 + 0-150645
3 4 5
+ + + + +
6
2
+ + + + + +
0019563
0-128943 0-309063 0-430171 0-364831 0-114768
+ + + + + +
-
0-002477
0033996
0-132034 0-201129
2
H
0391232 0357642
4
>
6
I 9 IO
II
+ 0-300079 + 0-171651
- 009033^ - 0-245936 - 0-171190
+ 0-234636
+ 0-245312 + 0043473
- 0-176785 - 0223447
9
IO
ii
12
+ 0-047689
+ 0-157798
0-105357 0-265471 0-219603 0-015040 + 0-182499
+ 0-347896 + 0-185775
- 0-055039 - 0-234062 - 0238286 - 0-073471
12
X + + + +
JM
0-000021 0-OOI202 0-011394 0-049088 + 0-131049
;,w
+ + + + + + + + + + +
0-000002 0-000175 0-002547 0-015176 0-053376 0-129587
,/bW
/.(*)
7ioW
7nW
3 4 5 b
+ 0245837
+ + + + + + + + + + +
+ + + + +
+ + + +
0-000013 0-000195 0-001468 0-006964
+ + + +
0-000002 0-000037 0-00035I 0-002048
0-008335 0-025597 0-0622I7 0-I23II7 0-20IOI4 0-270412
2 3
4
S
2 9 10
11
+ 0-339197 + 0-337576
+ 0-204317 - 0-014459 - 0-201584
0233584
12
0243725
+ + + + + +
+ + + + + +
+ + + + + +
I Q IO
11
12
/(*)
/(*)
/(*)
/(*)
/(*)
/(*)
4
5
6
7 8
_
+ 0-000003 + 0-000030 + 0-000205 + 0-OOIOI9
+ + + +
0-003895
+ 0-000006
+ o-oooooi + + + + + +
0-OOOOI2 0-000078 0-000393 0-001567 0-005110 0-013991
+ + + + + +
0-000002 0-000019 0-OOOII2 0-000506 0-001856 0-005698
4
5
6
2 8
9
10
11 12
+ + + + + +
0011957
0-030369 0-065040
9 IO
II
12
732
/(*)
/(*)
I 9 IO
ii
12
+ + + + + +
_
+ + + + +
o-oooooi 0-000007 0-000043 0-000199 0-000759
_
+ + + +
0-000002 0-OOOOI2 0-000059 0-000251
I 9 IO
II
12
nw
- 0-2881947 - 00259497
ViW
- 0-1750103 - 0-3026672 - 0-1580605
2 (x)
nw
+ 0-3282489 + 0-2680006 + 0-0265422
- 0-2050949 - 0-2513627
- 0-0914830
Y*{*)
X
6
I
+ 0-2298579
- 0-0605266 - 0-2630366 - 0-2267557 - 0-0058681
+ 0-2235215
9
IO
II
+ 0-2499367
+ 0-0556712
- 0-1688473 - 0-2252373
+ + + +
9 IO
II
12
+ 0-1986120 + 0-2157208
+ 0-1290061
- 0-2485118 - 0-1512177
12
X
6
VfW
+ + + +
- 0-1970609 0-0637022 0-2564011 0-2851178
(x)
(x)
V.to
1-1052194 0-6114352 0-3876699 0-1999469 + 0-0010755
-
Y
-
(x)
9
IO
II
0-1354030
2-2906609
6
I 9
10
11
09921953
0-5752760 0-3727057 o- 1992993
12
- 0-0892528 - 0-2298179
+ 0-1673728 - 0-0402973
+ 0-1786071
+ 0-2614047
- 0-0120492
+ 0-1590189
12
X
-
Y 10 (x)
5-7667633 1-9399240 0-9067010 0-5454645
lt (x)
Y lt (x)
- 56-3168097 - 12-3614737 - 3-7448595 - 1-5064942 - 0-7849097 - 0-4987558 - 0-3385583
Y13 (x)
- 208-3353554 - 37-83I7507 - 9-543ioi8 - V1778801 - i-3634543
- 0-7396546 - 0-4799704
X
6 I 9 10
11 12
6 I 9 10
11
03598142
12
- 0-1983240 - 0-0228763
733
Yn (x)
Y*(*)
(-x)
YAx)
-
Y2 (*)
- 127-6447832 - 32i57 x 446 - 14-4800940 - 8-2983357 - 5-4413708 3- 8 927946
X
o-i 0-2
OI
0-2
04
o-5
- 5099-3323786 - 639-8190662 - 190-7748150 - 81-2024845 - 42-0594943 - 24-6915728 - 15-8194791 - 10-8146466 - 77753605 - 5-8215176
o-6
11 09
I-O
i-i 1-2
+ 0-0056283 + 0-0882570 + + + + +
0-1621632 0-2280835 0-2865354 0-3378951 0-3824489
o-6
SI
0-9
i-o
i-i
i-5
11304119
1-0223908 0-9321938
45072313
3-5898996 2-9296706 2-4419696
20735414
1-7896705 1-5670362
1
1-6
U
2-0
2-1 2*2
+ + + + +
+ + + + + + + + + +
i-6
3895534
U
1-9 2-0
2-1 2-2
1-2458651 1-1277838
23
2-4 2-5
- 0-4266740 - 0-3813358
-
23
2-4 2-5
2-6
n
2-9 3.0
+ + + + + + + + + + + + + + +
+ + + + +
03364356
05770644 05385416
0-5012882 0-4649097 0-4291009
30
31 32
31 32 33
+ 0-3431029
+ 0-3070533 + 0-2690920
+ 0-2296153 + 0-1890219
34 35
3-6
0-3496295 0-3707113 0-3878529 0-4010153 0-4101884 0-4153918 0-4166744 0-4141147 0-4078200 0-3979257
33
3*4
0393631 03583353
35
3-6
3 9 4-o
4-1 4-2
+ + + +
-
0-1477100 0-1060743 0-0645032 0-0233759 - 0-0169407 0-0560946 0-0937512 o- 1 295959 0-1633365 0-1947050
03230993
0-2878581 0-2525864 0-2172943 0-1820221
13 39 40
4-i 4-2
4-3 4-4
45
4-6
+ + + + +
43
4-4 4-5 4-6
4-9 5'0
+ 0-2737452 + 0-2445013
+ 0-2135652
+ 0-3424796 + 0-3534308
+-
0-3612893
+ 0-1812467 + 0-1478631
+ 0-3660328 + 0-3676629
+ + + + +
4-9
50
734
Yn (z)
Y
7
Y<(x)
Y>{x)
Y*(*)
(x)
X
o-i
o-i
0-2 o-3
o- 4
'5
- 24461484-502 - 765856-775 - 101169-657 - 24113-576 - 7946-301 - 3215-6142 - I499-9983 - 776-6983 - 435-6898 - 2604059 - 163-88133 - 107-65135 - 7332353 - 5I-5I9I3 - 37 I 903i
- 293476652667 - 2295654722 - 134639666 - 18024776 - 3794296 - 1063795-33 - 363572-8o - 143672-96 - 63445-75 - 3058896 - 15836-229 - 8696-433 - 5018-701 - 3021-772 - 1887-397 - 1217-2798 - 807-6135 - 549-47I7 - 382-366 - 271-5480 - 196-43901 - I44-5I734 - 107-97306 - 81-82481 - 6282986 -
0-2
03
o-4 o-5 o-6
o-6
%l
o-g
I'O
i-i
- 243-02293 - 13263406 - 78-75129 - 49-88983 - 3327842 - 23-153427 - 16-686187 - 12-391145 - 9-443I93 - 7-361972 -
- 53350-547 - 21295-913 - 9629-977 - 4791-108 - 2570-780 - 14666768 - 8804084 - 55 1 -6360 - 358-5506 - 240-5734
- 165-96960 - ii7'35239 - 84-81625 - 62-52037 - 46-91400 -
Zl 09
i-o
11
1-2 i-3 1-4 i-5
1-2
i*5
i-6
5856365
4-7437I7 3-905897
n
i-9
3264432
2-765943
2-3733331 2-0603205 1-8079562 1-6023566 I-433I973
1-2926953 1-1749076 1-0752421 0-9901112 0-9166828
2-0
2-1
i-6
n 19
2-0
2-1
2-2
8-011973 6-546165
23 24
2-5
5-4M324
4-529576 3-830176
2-2
23
2-4
25
2-6
2-6
n
2-9 3 -o
U
2-9
30
31 32 33
3-4
31
32 33
3'4
35
3-6
2555924
1-1494603
1-0581497 0-9790651 0-9100926 0-8494985 0-7958514
35
3-6
n 39
4-o
4-1 4-2 4*3 4-4 4-5
- 0-6215621 - 0-5859520
37
3-8
39
4-o
4-1 4-2
03404998
0-3110929 0-2815701 0-2519027 0-2220872 0-1921423
05963194
- 1-0612100 - 0-9846543
-
43 44
4-5
029425
4-6
XI
4-9 5-o
50
735
Yn (x)
*\o(*)
Y a (x)
- 410842855308 2 - 1606575569 2 - 62798159 '2 - 6302655 '2 - 1060819 [2] - 24768540-5 - 7250160-1 - 2504646-8 - 982142-7 - 425674-6
Y9 (x)
- 65731922082781:3
X
O-I 0-2
OI
0-2
o- 4
1 2850308895 [3
o-5
- n83i335i320 4 5[5; - 11563671430IV - 200810527 5 - H330366[5; - I2i9636[5 - 197581500 2 - 42447194 2 - 11213538 '2] - 3469552 '2 - 1216180 [2] - 47J64393 - 19884570 - 8993478 - 4318759 - 2183993
1
03 04
o-5
o-6
09
i-o
i-i
- 659430617 - 165354373 - 49949263 - 17396869 - 6780205 - 2894495-9 - 13323432 - 653392-5 - 338225-9 - 183447-3 - 103635-01 - 60684-92 - 36684-77 - 22816-93 - I4559-83
-
10
i-i
1-2 i-3
i-5
- 200085-33 - 100577-97 - 53495'9i - 29859-17 - I7375-I3 - 10485-229 - 6533582 - 4188-852 - 2754-916 - 1853922 - 1273-8144 - 891-9608 - 635-4947 - 460-0405 - 3379597
i-6
1554086
636012-7 362658-9 213405-5 129184-5
i-6
n
1-9 2-0
2-1 2-2
20
2-1 2-2
XI 19
23 24
2*5
23
2-4 2-5 2-6
29
3-o
n
2-9
30
3-i 3-2
31
3-2
33
3-4
35
3-6
33792355
259-23861 200-77848 156-90853
12367548
-
33 34 35
3-6
n
41
4-2
3*9 4-o
467-7617
3633271
284-4645 224-4160 178-3306
39 40
41
4-2
43
4'4 4-5
^144157
- 5-329II4
43
4-4
45
4-6
4-6
49
5-o
Si
49
50
The numbers
points.
first
in
are the numbers of digits between the last digits given and the decimal integral part of F10 (o-i) is a number containing 19 digits of which the
14 are given.
736
e~*I z {x)
e-x h{x)
o-i 0-2
03
o- 4
05
o-i
0-2
03
0-4
05
o-6
00254458
0-0316770 0-0379022 0-0440151 0-0499388
09
I-O
i-i
00081553
0-0099497 0-0118547 0-0138406 0-0159110 0-0180231
11 09
I-O
i-i 1-2
00556193
0-0610206 0-0661209 0-0709088
00013479
0-0017471 0-0022045 0-0027189 0-0032885
0075381
0-0795406 0-0833947
0-OOOI468 0-0002072 0-0002826 0-0003746 0-0004843 0-0006I29 0-00076I 0-0009298 0-OOIII92 0-0013298
0-0015615 0-OOl8l42 0-0020879 0-00238I9 0-0026960
0-0201679 0-0223299
0-0039110 0-0045834
i-6
00869539
0-0902306
00244955
0-0266527 0-0287912
00053023
0-0060642 0-0068654
n
1-9 2-0
2-1 2-2 2-3 2-4 2-5
2-0
2-1
00932390
00959939 00985103
0-1008034 0-1028881 0-1047787
0-1064892 0-1080327 0-1094217 o- 1 106680 0-1117825
0-0309022 0-0329781
00350127
0-0370010
25
2-6
00389387
0-0408227 0-0426507 0-0444207 0-0461318
0-0122829
0-0030293
2-6
n 29
30
31 32 33
00132534
0-0142344 0-0152228 0-0162159
0-OI72II2 0-0182063 O-OigigiO 0-0201876 O-02II70O 0-022I447 0-023II02 0-0240654 0-0250090 0-0259400
OOO33813
0-0037511 0-0041380 0-0045409
U 29
30
3-i
00477833
34
3'5
00493750
0-0509071 0-0523802 0-0537949
00551523
00564535
0-0576999 0-0588928 0-0600338 0-0611243
0-0621661 0-0631607 0-0641096 0-0650147 0-0658774
32 33 34 35
36
3 "Z 3-8
39 40
41
4-2
36
n
41
4-2
3-9
00092443
0-0097611 0-0102826 0-0108082 0-0113371 0-0118685
40
43
4*4 4-5
01178323
0-1179905 0-1181048 0-1181791 0-1182166
43
4'4
45
4-6
47
4-8
49 50
OO336449
0-0344190
46
XI
4-9
00145403
50
737
K n (x)
K
3 (x)
Ko
(*)
iW
9-8538448 4-7759725 3-0559920 2-1843544 1-6564411
1-3028349 1-0502835 0-8617816 0-7165336 0-6019072
K*(x)
O'l 0-2
o-3
o- 4
o-5
0-4
05
o-6
o-6
n
0-9
I-O
i-i
SI
0-9 I-O
i-i 1-2
i-6
i-6
:i
1-9 2-0
2-1
01654963
0-1459314 0-1288460
02093625
0-1826231 0-1596602 0-1398659
XI 19
2-0
2-1
01138939
0-1007837 0-0892690
2-2
23
2'4 2*5 2-6
00791399
0-0702173 0-0623476 0-0553983
o-5373847
04485459
o-3762579 0-3170382 0-2682271 0-2277714 0-1940711 0-1658685 0-1421668 0-1221704
0-1052398 0-0908577 0-0786032 0-0681323 0-0591618
0-0514581 0-0448273 0-0391079 0-0341649 0-0298849
2-2
23
2-4 2-5
n
2-9
00492554
0-0438200 0-0390062 0-0347395 0-0309547 0-0275950 0-0246106 0-0219580
3 -o
3'i
2-6 2 .y 2"8
29 30
31 32
00356341
0-0316429 0-0281169 0-0249990 0-0222394 0-0197950 0-0176280
3*2
33 34
3*5
33
3*4 3'5
00195989
0-0174996 0-0156307
00323071
0-0284968
3-6
P 39
40
4*i 4-2
00251593
0-0222321 0-0196614 0-0174014
00139659
0-0124823 001 1 1597
00157057
0-0139993
39
4-0
4-1
00124835
0-0111363 0-0099382 0-0088722 0-0079233 0-0070781
43
4*4
45
4-6
0-0154123 0-0136599
0-012,1146
42 43
4'4 4'5
00057304
0-0051321 0-0045972 0-0041189 0-0036911
0-0063250
00136993
0-0120691 0-0106415 0-0093900 0-0082918
4-6
&
00056538
0-0050552 0-0045212 0-0040446
00075380
0-0067036 0-0059643 0-0053089
rl
4*9 5-o
738
K n (x)
K,(x)
X
t (x)
K,{x)
(x)
O-I 0-2
03
o- 4
o-5
7522451
359-502336 191-994207 111-175708 68-456722 44-232416
29-708098 20-596272 14-661702 10-672824 7-918871
5 -973 ! 29 4'5 7 567
460608047990 3594005995
20991 1239 27938248
O-I 0-2
03 4
o-5
5837^2
1621619-74 548248-34 2I3959-70
93I5505
44207-02
1-4 i-5
2-0
2-1
i-6
U
1-9 2-0
2-1
2-2
23
2-4 2-5
0-6312432 0-5232937 o-43576i5 0-3643764 0-3058512 0-2576343 0-2177299 0-1843662 0-1568967 o-i337274
2-6
4320732
3-43I763
n
2-9
30
31
3-2
31
3-2
33 34 35
3-6
27418356
2-2026750 1-7786158 1 -4430764 1-1760828
33
3-4
35
3-6
40
4-i 4-2
U 39
43
4'4 4'5
0-1142604 0-0978523 0-0839814 0-0722228 0-0622288 0-0537I39 0-0464423 0-0402191 0-0348822 0-0302965
0-0263491
37
3-8 3-9 4-0
4-1
08235478
0-6790539 0-5616138 0-4658258 0-3874361 0-3230800 2700847 0-2263181
42 43 44 45
4-6 4*7 4-8 4*9
4-6
u 49
50
00229453
0-0200054 0-0174623 0-0152591
01317219
0-1116385 0-0948088 0-0806716
50
2 1
739
Kn (x)
*
10 (*)
iW
644889647992 [2]
25i6402998[2]
9 (x)
o-i 0-2
03
o- 4
98ouoi7[2]
9787687[2] i6 3 68 3 8[2]
18574295846304(5]
1 81 23852594^]
o-i 0-2
3i38 5 92i2[5]
1 7630197(5]
03
0-4 o\5
o*5
5243719^]
1012785182 25 1 904 1 04 75383634 25977933 1 000504
42I5494-70 1912706-01
o-6
SI
0-9 i-o
i-i
o-6
64885309O]
16998902J2] 52ioi47[2] i8o7i33[2]
S3
0-9
i-o
i-i
1-2
14254429
74475'03 40774-60 23240-24
92346336
470026-62
I "5
25035361
138717-80 79569-40 47059-44 28600-23 17810-48 11337-247 7361-331 4866 694 3270-797 2231-581 1543-7592 1081-6417 766-8400 549-6277 397-9588
i-6
I -7
1-8
1-9 2-0
2*1
3343496
2188-117
1
19
2-0
2-1 2-2
459981
99I-34I3 683-9099 478-7011
2-2
23
2-4 2*5
3395354
243*7750i 1 76994 A 129-84408 96-17151 71-86762
54-I5I9I 41-11923 31-44899 24-21577 18-76452
23
2-4
25
2-6 2 '1 2-8 2-9
2-6
3 29
30
3-i
5059530
3507-654 2459-620
30
31
3-2
32 33
3-4
290-87739 214-49139
1743-1174
12476333
901 3053 656-7945 482-5358
357-24I3 266-3991 200-0162 I5I-I457 114-9141 87-87352 67-56482 52-22047 40-56082 31-65296
15947366
119-48709 90-17775
35
33 34 35
3-6
36
U 39
40
4'i
14-627050 II-465773
9035174
7-155283 5-693179
n 39
40
41 42
42 43 44
4*5
9333122
7-430286 5-938798 4-764583 3-836264
43 44
4-5
46
49 50
The numbers
points.
first
24-812255 I9-533I25
15 439946
4-6
I0539552
0-8664794 0-7143624
3099405
2-512278
12-252049 9-758563
4-9
50
in [ ] are the numbers of digits between the last digits given and the decimal For example, the integral part of K10 (o-i) is a number containing 19 digits of which the
14 are given.
740
n + i) (x)
/*<*>
0-671397 + 0-513016 + 0-065008
-1
J-l(*)
+ 0-431099 - 0-234786 - 0-456049 0-260766 + 0-101218
+ 0-312761 + 0227356 - 0-041045 - 0-242326 - 0-211709
/|
(*>
J -I
(*)
J%
/-*
+ 0-240298 + 0-491294
+ 0-477718 + 0-185286 - 0-169651
- 0-327930
- o- 199052 + 0-075931 + 0-254504 + 0-197983
3 4 5
+ + + + +
2 3
4
5
6
7 8
+ 0-038889
+
9 IO
ii 12
6
I
+ 0-196659
+ 0-234314
+ 0269886 9 + 0-164179 10
- 0-066647 11 - 0-22I2I2 12 - 0-175790 13 + 0-016554 14 + 0-l8l2I2 15
0240569 + 0-001064
+ 0-194364 + 0-200012 + 0-029158 - 0-156506
- 0-I9I025 - 0'053240
13 14 15
+ 0-240472
0-107392 - 0-108427 - 0213323 - OI23534
H-
+ 0-072423
- 0-137671 - 0-2I42S6 - o- 100880
16 17 IO 19 20
21
20
22 23 24 25
0-102303
+ 0-069085 + 0-158173
+ OOI4793
- 0-123217 - 0-145193 - OO35665 + 0-102148 + 0-143181
+ + + +
26
3
29 30
3i 32 33
+ 0-146854
+ 0-040849 - 0-098326 - 0-143930
- 0-057900
+ 0-IOI229 - 0-044859
- 0-145147 - OI I0835
+ 0-053672
- 0-081454 - 0-138826 - 0-068982 + 0-06I230
32 33 34 35
+ 0-086489
43 44 45
+ 0-121365 41 + 0-04IIOI 42
- 0-074495 - 0-119958 - 0-055643
4-
43 44 45
46
% 49
50
+ 0-053148 - 0-104983 - 0-I02622 + 0-115797 - 0-011925 - 0-00699I + 0-071879 + 0-0900I2 + 0-092968 - 0-036481 + 0-108013 + 0-106479 - 0-109477 + 0-027428 + 0-023037
0-057689 46
+ 0-II6253 + 0-068096
- 0-040876 49 - 01 10530 50
741
J n + ^ {x)
(
AW
+ 0-007186 + 0-068518 + 0-210132
+ 0-365820 + 0-410029 + 0-267139
- 0-003403 - 0-232568 - 0-268267 - 0099653
J -I
/.(*)
2
/-#(*>
+ 90-079718 + 5-034028 + 1-269137 + 0-625147 + 0-332945 + 0-054598
- 0-193887 - 0-304868 - 0-217577 + 0-004188
X
I
2 3
4 5
+ 0-000807 + 0-015887
+ 0-077598
+ O-I9930O +' O333663
4 5 6
6
7 8
+ 0-384612 + 0-280034
+ 0-047I22
- 0-183879 - 0-266416 - 0-151943
9 IO
ii 12 13
I 9 10
11 12 13
15
+ 0-064567 + 0213437
+ 0-183031 + OOO7984
- 0-l6l920 - 0-187495 - OO55OO5 + 0-II65I9 o-i8oiii
-i
14 15 16
16
\l 19
20
21 22 23
+ + + +
- 0-158507 0-014610 0-165146 0-164856 0-021518 - 0-133507 - 0-164423 - 0-049581 + 0-103998 + 0-159426
\l 19
20
21 22 23
+ 0-086555
- 0-077008 - 0-166640 - 0-107753 + 0-042601
24 25
+ 0-155133
+ 0049333 - 0-095706
- 0-151235 - 0-071438 + 0-068766
24 25
26
+ 0-071548
- 0-076458 - 0-151096 - 0-088575 + 0-050802
4-
3
29 30
3i
+ 0149737 + 0-121443
- 0-012033 - 0-130821 - 0-129349 - 0013372 + 0-110760 + 0-132602 + 0-035744 - 0090154
- 0-132005 - 0-054717 + 0-069461 4- 0-128176 + 0-070464
-
26
11 29 30
3i
0-140218
32 33 34 35
+ 0-101394
- 0-027012 - 0-127357 - 0-110507
+ 0-088450
- 0-043448 - 0-132704 - 0-101062
32 33 34 35
36
11 39 40
4i
+ + + +
36
11 39 40
4i
42 43 44 45
- 0-119045 - 0-032729 + 0-081099 + 01 19045 + 0-048542 - 0-064303 - 0-116541 - 0-062195 + 0-047346 + 0-111781
0049057
42 43 44 45
46
49 5
+ 0-098712
- 0-000443
+ 0-092836
0-010367 0-102038 0-099715 0-007388
46
%
49 50
742
oiJ (n + ^(x)
hi(*)
7-V
(*)
/-V
W
+ + + + + + + + + +
-
/-vW
+ 8681-738496 + 110-346069 + 10-117087 + 2-283448 + 0-924903
2 3 4 5
+ 0-000074 + 0-002973
+ 0-022661 + 0-082606 + 0-190564 + + + +
0-309779 0-363446 0-285580 0-084388 - 0-140121
3 4
5
0085579
0-183316 0-291096 0-345551 0-287020 0-112283
6
7 8
9 IO
ii 12
6 I 9 10
11
.
- 0-253757 - 0-186414
13 14 15
12
J3
14 15 16
16 17 io 19
+ 0-067428
- 0-113872 - 0-192649 - 0-109663
+ 0-208276 + 0-113813
- 0-062725 - o- 1 80008 - 0-147369
0-010308
0165924
11 19
20
21 22
+ 0-059532
+ 0-170603 + 0-132919 - 0-015626
- 0-144405 - 0-144089
20
21 22
23
24
+ 0-178483 + 0-098451
- 0-059244 - 0-160681 - 0-122994
+ 0041567
o- 106000
25
23 24 25
26
3
29 30
31
+ 0-047975
- 0-089606
- 0-144100 - 0-070242
+ + + +
26
11 29 30
3i
32 33 34 35
+ on 17599
+.
- 0-139905 - 0-045989 + 0-085043 + 0-137843 + 0-068414 - 0-059088 - 0-131075 - 0085598 + 0-034271 + 0-120760
32 33 34 35
36 ii 39 40
4i
+ 0-128214 + 0-038110
- 0-083440 - 0-127550 - 0-056866
0-120357
+ 0-017176
- 0-098362 - 0-124012 - 0-039312
36 11 39 40
41 42 43
42 43 44 45
+ 0-108276 + 0-006668
- 0-098498 - 0-113059 - 0-025987
+ 0-098102
-
44 45
46
%
49 50
46
%
49 50
743
Jn + ^
(z)
(*)
/*<*>
0-000008 0-000207 0-001974 0-010243
/j/to
+ o-oooooi + 0-000034
;v
7V
/v
(*)
2 3 4 5
+ 0-000005 + o-oooooi + 0-000434 + 0-000086 + 0-000015 + 0-000002 + 0-002887 + 0-000727 + 0000165 + 0-000034
0-012324 0-037852 0-089213 0-167162 0-252556
4
5
+ 0-087406 + 0-035199 + + 0-177161 + 0-088535 + I + 0-275940 + 0-171837 + 9 + 0-330196 + 0263308 + IO + 0-286089 + 0-316850 +
6
ii 12
+ + + + +
+ + + + +
+ 0-133432
- 0-068653 - 0-214523 - 0-218661 - 0-081213
13 14 15
+ 0-283766 + 0-305116 + 0243253 + + 0-149630 + 0-280630 + 0-294700 + - 0-040059 + 0-162139 + 277030 + - 0-192766 - 0-015412 + 0-171849 + - 0-222722 - 0-171205 + 0-005862 +
- 0-112842
16
17 IO 19
20
0155322
+ + + +
- 0-221691 - 0-137449
+ 0-024269 - 01 30704
- 0-209985 - 0-169805 - 0-032875
+ 0-185304 16 + 0-040241 17
-
0-II2207
- 0-20I228
18 19
0-179418 20
/y M
+ 0-000007 + + + + +
0-000063 0-000402 0-001846 0-006568 0-018837 0-089695 0-152818 0-221379 0-269315
7?f
(*)
5 6
7 8
+ o-oooooi
W
4-
/v <*>
J* (*)
/v w
+ + + + + + + + + + + + + +
O-OOOOOI 0-000005 0-000039 0-000212 0-000898
+ + + +
o-oooooi 0-000009 0-000058 O-OOO20O
9 10
11
+ + + + +
+ + + + +
0-000003 0-000024 0-000152 0-000718 0-002683 0-008237 0-021263 0-046907 0-089312 0-147378
0-210215 0-255927 0-257478 0-196122 0-076893
+ + + + +
0-000355 0-001288
6
7
+ 0-044768
12 13 14
*5
+ + + +
+ + + + +
0-001086
0003532
0-009760 0-023297
0003955 13
0-010469 14 0-024193 15
0-049201 16 0-088443 17 o- 140592 18 0-196755 19 0-239451 20
0048533
+ + + + +
744
X
o-5 I-O
i\5
\\]j^{t)dt
+ + + + +
+ + + + +
0-550247 0-721706 0-779084 0-753302 0-670986 0-56I020 0-452047
*/V*w*
+ + + + +
0-092366 0-247558 0-415^48 0-5628^9 0-665787
0-711685 0-700180 0-642119 0-556489 0-465942
*/V 4
+ + + +
(/)<*<
ij'ji
+ + + + +
w*
255
26-0 26-5 27-0 27-5 28-0 28-5 29-0 29*5
2-0 2-5
3-o 3'5
4 -o
4'5 5-o
O368193
0-325249 0-328457
0-372439 0-443274 0-522202
+ + + + + + + + + +
+ + + + + + + + + +
300
30-5 31-0 3i-5
+ + + + +
+ + + + + + + + + + + + + + + + + + + +
+ + + + + + + + + +
+ + + + + + + + + + + + + + +
5S
6-o 6-5 7-o 7-5 8-o 8-5 9-o 9-5 io-o
io- 5
+ + + + +
O59OI 16 0-631845
0-63930I 0-6I2868 0-560804 0-496895 0-436964
320
32-5
+ + + + +
+ + + + +
33-o
335
34-o 34-5 35-o
0532930
0-55749 0-567709 0-561313
0-540094 0-509417 0-476871 0-450396 0-436345
0-395087
ii-o
1 1 -5
O38039O
0-395149 -434557 0-488146
12-0 12-5
355
36-0 365^ 37-o
375
38-0 38-5 39-o
I30
I3'5 14-0 14-5
+ 0-542511
+ 0-584583 + 0-604721 + 0-598871 + 0-569335
I50
15-5 16-0
+ + + + +
+ + + + + + + + + + + + + + + + + + + +
395
40-0
0437971
0-454670 0-482187
0513690
0-541464
165
17-0 17-5 18-0 18-5 19-0 19*5 20-0 20-5
2 I-O
+ + + + +
405
41-0 4i'5 42-0 42*5 43-o 43'5 44-0 44'5
+ 0-443897 + 0-439006
+ + + + +
0-449025 0-471341 0-500382 0-529002
+ + + + +
+ + + + + + + + + +
0-558799 0-561608
0549384 0525282
0-495309 0-466829 0-446755 0-439878 0-427720 0-468209
0-496215
+ 0-427837
+ 0-465971 + 0-511332
+ 552774 + 0-580389
-
45o
45'5 46-0 46-5 47-0 47-5
0550239
+ 0-587849 + 0573842
+ 0-542266 + 0-501167 + 0-460707
+ 0-559004 + o-5533oi
+ 0-534676 + 0-507802 + 0-479313
0524837
0-547099
0557650 0554044
0-537309 0-511657
23O
23-5 24-0 24-5
250
+ + + + +
0-400311 + 0-484658
+ + + + +
0483428 0459523
0-445722
745
*/>-i>
0-02 0-04 o-oo 0-08 o-io
dt
/>i<"
+ + + + +
dt
#=(-)1-570796
//-i><"
+ 0-779893
+ + + +
0-321056 0-640807 0-380389 0-605721 0-404260 0*588128 0-417922 0-577121 0-427036
0-569413
0-12 0-14 O-IO 0-18 0-20 0-22 0-24 O-20 0-28 0-30
0-32 o-34 0-36 0-38 0-40
+ + + + + + + + + +
+ + + + + + + + + +
0-2759976 0-2979565 0-3183378 '3374 I 86 0-3554002 o-3724338 0-3886365 0*4041012 0-4189028 0-4331026
0-4467517 0-4598932
4712389
7-853982 IO-995574 14-137167
+ + + + +
+ + + + +
26703538
29-845130
32-986723 36-128316 39-269908 42-411501 45 553093
+ + + + + + + + + +
04725635
0-4847941 0-4966121
0433666
0-563631 0-438767 0-559088
048 050
0-52 o-54 0-56 0-58 o-6o
0-62 0-64 o-66 o-68 0-70
+ 0-0714853 + 0*0765575
48-694686
+ 0-442848
+ 0-0817309
+ 0-0870016 + 0*0923658 + + + +
4-
+ + + + + + + + + +
x = n*
3-I4I593 6-283185 9-424778 12-566371 15*707963
*/>!><<
+ + + + +
4-
0-713972 0-3434I5
06045308
0-6126537 0-6205324 0-6281731 0-6355815
+ + + + +
+ + + + + + + + + + + + + + +
O62894O
0-387969 0-60036l
0-584942
OT503961
0-1565683 0-1627958 0-1690757 o -1 754054 0-1817820
+ 0-408301
+ 0-420516 + o-574957 + 0-428877
28274334
31*415927
34-557519 37*699112 40-840704 43*982297 47-123890
50*265482
+ + + + +
0-1882030 0-1946656 0-2011673 0-2077055 0-2142775 0-2208809 0-2275131 0-2341717 0*2408543 0*2475583
+ + + + +
0558096
+ 0-443747
+ + + + +
07079643
0*7127261
07173059
0*7217059
'
'
746
Jn(n)
* /
()
/' (n)
Jn' (n)
0-3251471
03554045
0-3682342 0-3755633
03803908
0-3838497 0-3864704 0-3885364 0-3902143 0-3916089 0-3927897 0-3938047 0-3946882
02335836
9 10
ii 12
9 10
11
13 14 15
03954655
0-3961557 0-3967734 0*3973300 0-3978347 0-3982948
00651336
0-0624879 0-0600969 0-0579234 0-0559374
12 13 14 15
16
II 19
16
20
21
\l 19
00541 141
0-0524332 0-0508777 0-0494332 0-0480874 0-0468301 0-0456522 0-0445460 0-0435048 0-0425226
03987163
0-3991041
20
21 22
22 23 24 25
03994624
0-3997946 0-4001035 0-4003917
0-4006614 0-4009143 0-4011521 0-4013762 0-4015877
23 24 25
26
3
29 30
31 32 33
00415942
0-0407151 0-0398812 0-0390889 0-0383350
26 27 28 29 30
3i
34 35
0-4472450 0-4472476 0-4472500 0-4472523 o-4472544 0-4472564 0-4472583 0-4472600 0-4472616 0-4472632
0-4017879
04019775
0-4021576 0-4023288 0-4024918
0-4026472 0-4027956
00376165
0-0369309 0-0362758 0-0356491 0-0350489 0-0344734
32 33 34 35
36 3Z 38 39 40
41
36
38 39 40
4i
04029374
0-4030732
04032033
0-4033281 0-4034479
42 43 44 45
0403563
0-4036738 0-4037805
00319156
0-0314594 0-0310192 0-0305941 0-0301833 0-0297861
42 43 44 45
46
42 48 49 50
04038833
0-4039824 0-4040781 0-4041705 0-4042599
01239383
0-1230719 0-1222291 0-1214090
46 47
48 49 50
For values of n exceeding 50, the following approximations may be used with seven-figure accuracy: ~\ 1 000586 92885 r 1213 -| 0-44730 73184 F
J,,' (w)
L 0-41085 01939 r L
225 2J
23 1 3i5onaJ
"'
"
L
r
i4625n 2 J
947
"I
0-08946 14637
Hi
L "69300aJ
747
n
i
-V-W
07812128
0-6174081 0-5385416 0-4889368 '453&948
-n*Yn
(n)
Y'
()
*f
YH
'
()
n
1
2 3 4 5
02615525
0-2297650 0-2060642 o- 1 876060 0-1727588 0-1605149
2 3
4
5
6 I 9
IO
II
6 I 9 10
II 12 I3
12 13
03385583
0-3296303 0'32i5755 0-3142546
M 15
16
!S 19
0-7751458 0-7751049 0-7750711 0-7750426 0-7750184 0-7749976 0-7749796 o-7749638 o-7749499 0-7749374
01502159
0-1414121 0-1337852 0-1271029 0-1211915
0-7429809 0-7412092
07396683 07383135
0-7371112
!4 15 16 11
0-3075580
03013982
20
21
0-7360358 0-7350670
07341890
0-7333887 0-7326559 0-7319817
19
20
21
22 23 24 25
02650095
0-2615652 0-2582933 o-255i79i 0-2522100 0-2493744 0-2466622 0-2440643
07313591 07307820
0-7302453
07297446
0-7292763
22 23 24 25
26
26
3
29 30
31
11 29 30
31
32 33 34 35
02415724
0-2391794 0-2368784
0-2346635 0-2325292 0-2304705 0-2284828 0-2265620
0-2247042 0-2229059 0-2211637 0-2194748 0-2178364
32 33 34 35
36
U
39 40
4i 42 43 44 45
36
00653451
0-0641718 0-0630496 0-0619751
07255720
0-7253259 0-7250904
$ 39
40
41 42 43 44 45
07248647
0-7246483 0-7244405
00572157
0-0563695 0-0555539 0-0547671 0-0540074
46
49 50
For values of accuracy.
46
02117381
0-2103166
07233475
0-7231873 0-7230324
H 49
50
"1
00532735
n exceeding
may
YM '(n).
The
225*J
001016 59059 f_ +
_*
l4625^_r
947 "| 6930on8J
23
[
0-15495 18004
rt
V3
Jn
(n)
and
748
,y
h,
Vi, n,
of
0*0,
(x),
J1
(x),
Y x (x)
n
Jo,n
yo,n
JUH
3-83I7060 7-OI55867 IO-I73468I I3-32369I9 I6-470630I
I9-6I58S85 22-7600844 25-903672I 29-0468285 32-I896799
yi,n
3 4 5
133610975
16-5009224 19-6413097 22-7820280 25-9229577 29-0640303
32-2052041 35-3464523 38-4877567 41-6291045 44-7704866
3 4 5
6
2 8
9
IO
ii 12
27H93479I
30-6346065
6 I 9 10
II 12
13 J4 15
33-7758202 36-9170984 40-0584258 43-I9979I7 46-3411884 49-4826099 52-6240518 55-7655 Io8 58-9069839 62-0484692 65-1899648 68-3314693 71-4729816 74-6145006 77-7560256
447593I9O
47-90I4609
5I-0435352 54-I855536
13 14 15
16 \l 19 20
21
16 \l 19 20
21 22
526145508
55-7565449 58-8984962 62-0404111
65-1822951 68-3241522 71-4659861 74-6077996 77-7495953
573275254
60-4694578 63-6II3567
66-7532267 69-89507I8 73 0368952 76-I786996 79-3204872
22 23 24 25
23 24 25 26 11 29 30
3i 32
26 11 29 3
3?
934583692
96-6000923 99-7418072 102-8835147 106-0252153 109-1669097
112-3085985 115-4503820 118-5919607 121-7336349 124-8753051
96-6052680
32 33 34 35
997468199
102-8883743 106-0299309 109-1714896 112-3130503 115-4546127 118-5961766 121-7377421 124-8793089
33 34 35
36
36
11
39 40
I233O447O5
I26-446I387
11 39 4
749
Ji.n
y,
Ja n
VI356223
8-4I7244I 11-6198412
3 4 5
I479595I8
17-9598195
21-1169971 24-2701123 27-4205736 30-5692045 33-7165195
4 5 6
6 i 9 IO
ii
1 9 10
11 12 13
J4 15
12 13 14 15
36-8628565 40-0084467 43-I534538 46-2979967 49-4421641 52-5860235 55-7296271 58-8730158 62-0162224 65-1592732
35-2897939 38-4357335 41-5810149 44-7257771 47-8701227 51-0141287 54-1578545 57-3013461 60-4446401 63-5877658
66-7307471 69-8736034 73-0163509 76- 1 59003 79\30i57 I 3
82-4440651 85-5864927 88-7288612 91-8711766 95-OI3444I
462438744
49-39I4980
16 11 19 20
21
54-1116156
525383976
55*6846964 58-8304911 6I-9758587 65-I2086l2
68-2655491 71-4099642 74-554I409 77-6981084 80-8418910
16
11 19
572576516
60-4032241 63-5484022 66-6932417
20
2r 22 23 24 25
22 23 24 25
68-3021898 71-4449899 74-5^76882 77-7302971 80*8728269 84-0152867 87-1576839 90-3000252 93-4423160 96-5845614
69-8377884 72-9820804 76-1261492 79-2700214 82-4137195 85-5572629 88-7006678 91-8439487 94-9871177 98-1301857
101-2731621 104-4160552 107-5588722 110-7016197
1
26 11 29 30
31
26 27 28 29 30
3i 32 33 34 35
997267657
102-8689327 106-0110655 109-1531673 112-2952406
1
32 33 34 35
99-7016848 102-8446186
105-98^74728
IO9I302542
II2-272969I
138443033
36
15-4372877 118-5793107
I2I-72I3II<;
11 39 40
12^-8632917 128-0052530
36 3Z 38 39 40
750
y* n j5
,
n,
y6, of
,
J4 (*), F4 (*),
J,
(a;),
F5 (*)
J4.H
y*,*
H,n
.Vs,
877M838
12-3386042 15-7001741 18-9801339 22-2177999
9 10
ii
22-4248106 25-0102671
2V43034II
28-6266f83 3I-8lI7l67 3^-98878l3 38-I598686
287858937
31-9546867 35-1185295
38-2786681 41-4359606 44-5910182 47-7442881 50-8961052
9 10
11
12 13 14 15
493203607
52-4715514
397436277
42-9082482 46-0696791 49-2285437 52-3853121
539630266
57-II73028 60-2702451 63-4220540 66-5728919 69-7228912
72-8721613 76-0207934 79-1688641
12 13 14 15
16 17 ie 19
20
21
55-5403458
16
586939271
61-8462803 64-9975855 68-1479890
71-2976113 74-4465520 77-5948946 80-7427095 83-8900562
\l
19
20
21 22
22 23 24 25
697875753 729340384
76-0800980 79-2258022 82-3711919
82316438O
85-4635703 88-6I03082 91-7566925
23 24 25 26 27 20 29 30
31 32
26 27 28 29 30
31 32 33
87-0369859
901835423
93-3297633 96-4756819 99-0213268
102-7667232 105-9118934 109-0568569 112-2016312 115-3462317 118-4906725 121-6349657 124-7791228 127-9231536 131-0670674
949027585
98-0485369 101-1940546
I043393353
107-4843998 IIO-6292667 II3-7739523 II6-9I847I3
34 35
33 34 35
36
3Z
38 39 40
1295298972
36
37.
3 39 40
751
;
y ly 3 n of
,
Jv 3
(x),
J_ ll3
{x)
+ J vz
to
},
(z),
J_ 1/3
(a;)
zeros,
^3.
ri/3 (x)sin
120
J 11*,'
ym
13530196
4-4657883 7-6012412 10-7402128 13-8803575
SH
0-8477186
39441020
7-0782997 10-2169407 J3-3569532
9 10
11
24350I925
27-4914601
30632794I
9
10
11
34'2977437
12 13 14 15
374391666
40-5806158 43-7220857 46-8635719
50-0050715 53-1465821 56-2881019 59-4296294 62-5711634
337741762
36-9I5594I 40-0570394 43-1985061 46-3399899
421513485
45-2928269
48-4343202 51-5758256 54-7I734 10 57-8588648 61-0003956 64-1419325 67-2834747 70-4250213 73-5665718 76-7081259 79-8496829 82-9912426 86-1328048 89-2743691 92-4I59353 95-5575032 98-6990728 101-8406437 104-9822160 108-1237894
111-2653639 114-4069394 117-5485159 120-6900931 123-8316712
12 13 14 15
16 \l 19 20
21 22 23
49-4814874
16
526229964
55-7645I47 58-9060410 62-0475740
\l 19 20
21 22
65-7127030 68-8542475
-9957961 75-I373484 78-2789040
71
24 25
77-7553U2
80-8968692 84-0384298 87- 1 799926 90-3215576 93-463I244
23 24 25
26
29 30
31
81-4204625 84-5620234
26 II 29 30
3i 32 33
877035867
90-8451519 93-9867191 97-1282878 100-2698581 103-4114297 106-5530025 109-6945765
112-8361516
32 33 34 35
96-6046929 99-7462629
IO28878343
106-0294070 109-1709808
112-3125557
II 5454i3i5
34 35
36
11 39 40
"5-9777275
119-1193044 122-2608821 125-4024605
36
11 39 40
752
of
(t)dt
lf'Y
J
(t)dt
lj o
(t)dt
and
hjj
0-02 0-04 O'OO 0-08 O-IO 0-12 0-14 O-IO 0-18 0-20 0-22 0-24 O-20 0-28
+ + + + +
Jo.
i/VoO
+ + + +
at
+ + + + +
0-0599280
01234500
0-1370979 0-1498103 0-1617001 0-1728544
00698858
0-0798295
00897573
0-0996672
+ -735 22 8
0-3344230 0-6340842 0-3845594 0-6028269
0-4064156 0-5864441 0-4192836 0-575991 0-4279931
030
032
o-34 0-36 0-38 0-40
0-42 0-44 0-46 0-48
+ + + + +
- 0-2196416
- 0-2274894
-
+ + + + +
-h
0-2349040
+ 0-5685888 + 04343856
!
05629957
49-4826099
- 0-2606471 - 0-2661901
- 0-2714049 - 0-2763037 - 0-2808977
02069333
050
0-52 o-54 0-56 0-58 o-6o
+ + + + + + + + + +
0-2542004
O2635IO3
0-2727682 0-28I9722 0-29II206
% = yro,n
*/l
y W
(It
0893577O
3-9576784 7-0860511 IO-2223450 I3-36I0975
16-5009224 19-6413097 22-7820280 25-9229577 29-0640303
32-2052041 35-3464523 38-4877567 41-6291045 44-7704866
- 0-320929I
02996358
0-3025960 0-3053111 0-3077877 0-3100319 0-3120498
0-3138471 0-3154290 0-3168010 0-3179678 0-3189344
+ 0-I920I49
- 0-1474447
-;-
0-1237411
01 085949
078
o-8o
0-82 0-84 o-86 o-88 0-90
+ + + + +
47-9118963
+ 0-0576118
0-92
094
0-96 0-98 I-OO
may
Y
i
j
if
(t)dt
= inx{Y (x)U
'(x)+
Y^H,^)}.
BIBLIOGRAPHY*
ADAMOFF, A. the asymptotic representation of the cylinder functions Jv (2) and /' (2) for large values of the modulus of z. Petersburg, Ann. Inst, polyt. 1906, pp. 239265. [JahrbuZh itber die Fortschritte der Math. 1907, pp. 492493.]
On
Note on the Function Phys. Math. Soc. of Japan,
Km (x),
(3)
AICHI, K.
the Solution of the Modified Bessel's Equation.
Proc.
xxm.
219 224.
The Vibrations of Circular Plates and their Relation to Bessel Functions (Feb. 15, 1911). Proc. Phys. Soc. xxm. (1911), pp. 225232. The Oscillations of Chains and their Relation to Bessel and Neumann Functions. Phil. Mag. (6) xxi. (1911), pp. 736742. Tables of Neumann Functions G n (x) and n (x). Phil. Mag. (6) xxu. (1911), pp. 658 663. The Asymptotic expansions of Bessel and other functions. Archiv der Math, und Phys. (3) xx. (1913), pp. 240244. The Vibrations of Cylinders and Cylindrical Shells. Archiv der Math, und Phys. (3) xx. (1913), pp. 289294. Tables of the Neumann functions or Bessel functions of the second kind. Archiv der Math, und Phys. (3) xxn. (1914), pp. 3043. Bessel and Neumann Functions of Equal Order and Argument. Phil. Mag. (6) xxxi. (1916), pp. 520528. The Roots of Bessel and Neumann Functions of High Order. Phil. Mag. (6) xxxil.
(1916), pp.
714.
Bessel Functions of Equal Order and Argument. Phil. Mag. (6) xxxn. (1916), pp. 237 238. The Numerical Calculation of the Roots of the Bessel Function n (x) and its first derivative Jn (x). Phil. Mag. (6) xxxiv. (1917), pp. 189195. The Addition Theorem of the Bessel Functions of Zero and Unit Orders. Phil. Mag. (6) xxxvi. (1918), pp. 234242. The Lommel- Weber 12 Function and its Application to the Problem of Electric Waves on a Thin Anchor Ring (Dec. 7, 1917). Proc. Royal Soc. xciv. (1918), pp. 307314. Bessel Functions of small Fractional Order and their application to problems of Elastic Stability. Phil. Mag. (6) xli. (1921), pp. 200205.
'
AIRY, SIR
GEORGE
B.
On
Trans.
Soc. v. (1835), pp. 283291. On the Intensity of Light in the neighbourhood of a Caustic (May 2, 1836 March 26, 1838). Trans. Camb. Phil. Soc. vi. (1838), pp. 379402. On the Diffraction of an Annular Aperture (Dec. 4, 1840). Phil. Mag. (3) xvur. (1841),
Camb. Phil.
pp.
110. Supplement to a Paper, On the Intensity of Light in the neighbourhood of a Caustic (March 24, 1848). Trans. Camb. Phil. Soc. vin. (1849), pp. 595599.
AKIMOFF,
M.
:
Transcendantes de Fourier-Bessel a plusieurs variables (July 10, 1916 June 25, 1917 Dec. 24, 1917). Comptes Rendus, clxiii. (1916), pp. 2629; clxv. (1917), pp. 2325, 11001103.
;
iiber
In the case of a few inaccessible memoirs, references are given to abstracts in the Juhrbuch die Fortschritte der Math, or elsewhere.
754
d2 u
(June
16, 1898).
dy
/,
n2 \
203223.
On
1899).
JH (x Kfi)
(June
15,
3243.
P.
ALEXANDER,
ANDING,
Sechsstellige Tafeln der Besselscken
Expansion of Functions in terms of Linear, Cylindric, Spherical and Allied Functions (Dec. 20, 1886). Trans. Edinburgh Royal Soc. xxxin. (1888), pp. 313320.
E.
ANGER,
Untersuchungen
iiber die
T.*
Function Ikh mit Anwendungen auf das Kepler'sche Problem. Neueste Schriften der Naturforschenden der Ges. in Danzig, v. (1855), pp. 1 29.
ANISIMOV,
The
pp.
1
33.
generalised form of Riccati's equation. Warsaw University, 1896, [Jahrbuch iiber die Fortschritte der Math. 1896, p. 256.]
V. A. Proceedings of
Sur
de
pp.
APPELL, P. E. l'inversion approchee de certaines integrates reelles et sur 1'extension de l'equation Kepler et des fonctions de Bessel (April 6, 1915). Comptes Rendus, clx. (1915),
419423.
AUTONNE,
Sur
la nature des integrates algdbriques
L.
7,
1883).
Comptes
Rendus, xcvi. (1883), pp. 13541356. Sur les integrales algebriques de l'equation de Riccati (Feb. cxxvin. (1899), pp. 410412. BACH, D. De Pintegration par les series de l'equation
Comptes Rendus,
d2 y
dx*
n \ dy __ x dx~ y
47 G. F.
'
Ann.
set.
68.
W.
I
BAEHR,
Sur
les
cos
(.r
cos >)<&> =0 et
cos(#cos)sin2 G>rfo>=0
(April, 1872).
Archives Neerlandaises,
351
358.
BALL,
L.
DE.
Ableitung einiger Formeln aus der Theorie der Bessel'schen Functionen (June 6, 1891). Astr. Nach. cxxvin. (1891), col. 1 4.
BARNES, E. W. the homogeneous linear difference equation of the second order with linear coeffi71. cients. Messenger, xxxiv. (1905), pp. 52 On Functions denned by simple types of Hypergeometric Series (March 12, 1906). Trans. Camb. Phil. Soc. xx. (1908), pp. 253279. The asymptotic Expansion of Integral Functions denned by generalised Hypergeometric Series (Dec. 3, 1906). Proc. London Math. Soc. (2) v. (1907), pp. 59116.
On
BASSET, A. B. of finding the potentials of circular discs by means of Bessel's functions (May 10, 1886). Proc. Camb. Phil. Soc. v. (1886), pp. 425443. On the Potentials of the surfaces formed by the revolution of Limacons and Cardioids about their axes (Oct. 25, 1886). Proc. Camb. Phil. Soc. vi. (1889), pp. 219. A Treatise on Hydrodynamics (2 vols.) (Cambridge, 1888). On the Radial Vibrations of a Cylindrical Elastic Shell (Dec. 12, 1889). Proc. London Math. Soc. xxi. (1891), pp. 5358. On a Class of Definite Integrals connected with Bessel's Functions (Nov. 13, 1893). Proc. Camb. Phil. Soc. vin. (1895), pp. 122128.
On a method
* See also under Bourget and Cauchy.
BIBLIOGRAPHY
BATEMAN,
H.
755
Certain definite integrals connected with the Legendre and Bessel functions. Messenger
xxxiii. (1904), pp. 182188. generalisation of the Legendre polynomial (Jan. 1, 1905). Proc. London Math. Soc (2) in. (1905), pp. 111123. The inversion of a definite integral (Nov. 8, 1906). Proc. London Math. Soc. (2) iv. (1906), pp. 461498. On an expansion of an arbitrary function into a series of Bessel functions. Messenger, xxxvi. (1907), pp. 3137. The Solution of Linear Differential Equations by means of Definite Integrals (Jan. 25, 1909). Trans. Camb. Phil. Soc. xxi. (1912), pp. 171196. The History and Present State of the Theory of Integral Equations. British Association Report, 1910, pp. 345424. Notes on integral equations. Messenger, xli. (1912), pp. 94 101, 180 184. Some equations of mixed differences occurring in the Theory of Probability and the related expansions in series of Bessel's functions. Proc. Int. Congress of Math. i. (Cambridge, 1912), pp. 291294. Electrical and Optical Wave-motions (Cambridge, 1915).
BAUER,
Von den
Coefificienten der
G.
Reihen von Kugelfunctionen einer Variablen. Journal fur Math. lvi. (1859), pp. 101121. Bemerkungen iiber Reihen nach Kugelfunctionen und insbesondere auch iiber Reihen, welche nach Producten oder Quadraten von Kugelfunctionen fortschreiten mit Anwendun<* auf Cylinderfunctionen (July 3, 1875). Miinchener Sitzungsberichte, v. (1875), pp. 247 272
BECKER,
iiber die Fortschritte
J.
Die Riccatische Differential-Gleichung. Programm Karlsbad, 1908 (25 der Math. 1908, p. 395.]
pp.).
[Jahrbuch
BELTRAMI,
E.
Intorno ad un teorema di Abele e ad alcune sue applicazioni. R. 1st. Lombardo Rendiconti, (2) xin. (1880), pp. 327337. Intorno ad alcune serie trigonometriche (June 17, 1880). R. 1st. Lombardo Rendiconti,
(2) xin. (1880), pp.
402413.
Sulle funzioni cilindriche (Jan. 14, 1881). Atti della R. Accad. delle Sci. di Torino, xvi (188081), pp. 201205. Sulla teoria delle funzioni potenziali simmetriche (April 28, 1881). Bologna Memorie (4) ii. (1880), pp. 461505.
BERNOULLI, DANIEL.
Correspondence with Leibniz 16971704. [Published in Leibnizens Oes. Werke, Dritte Folge (Mathematik), in. (Halle, 1855).] Notata in praecedens schediasma 111. Co. Jacobi Riccati, Actorum Eruditorum quae Lipsiae publicantur Supplementa, vm. (1724), pp. 73 75. Solutio problematis Riccatiani propositi in Act. Lips. Suppl. Tom. vm. p. 73. Acta Eruditorum publicata Lipsiae, 1725, pp. 473 475. Exercitaticm.es quaedam mathematicae (Venice, 1 724), pp. 77 80. Theoremata de oscillationibus corporum filo flexili connexorum et catenae verticaliter suspensae. Comm. Acad. Sci. Imp. Petrop. vi. (1732 33) [1738], pp. 108 122. Demonstrationes Theorematum suorum de oscillationibus corporum filo flexili connexorum et catenae verticaliter suspensae. Comm. Acad. Sci. Imp. Petrop. vn. (1734 35)
[1740], pp.
162179.
BERNOULLI, JOHN.
Methodus generalis construendi omnes aequationes differentiales primi gradus. Acta Eruditorum publicata Lipsiae, 1694, pp. 435437. \0pera, I. Lausanne aud Geneva, 1742,
p. 124.]
BERNOULLI, NICHOLAS
Correspondence with Goldbach. See under Fuss.
(the younger).
BESSEL,
F.
W.
Analytische Auflosung der Keplerschen Aufgabe (July 2, 1818). Berliner Abh. 181617 [1819], pp. 49 55. [Abhandlungen, herausgegeben von R. Engelmann, I. (1875), pp. 17
20.]
Ueber die Entwickelung der Functionen zweier Winkel u und u' in Reihen welche nach den Cosinussen und Sinussen der Vielfachen von u und u' fortgehen (June 21, 1821). Berliner Abh. 182021 [1822], pp. 5660. [Abhandlungen, n. (1876), pp. 362364.]
756
Untersuchung des Theils der planetarischen Storungen welcher aus der Bewegung der Sonne entsteht (Jan. 29, 1824). Berliner Abh. 1824 [1826], pp. 152. [Abhandlungen, I. (1875), pp. 84109.] Beitrag zu den Methoden die Storungen der Kometen zu berechnen (Sept. 24, 1836). Astr. Nnch. xiv. (1837), col. 148. [Abhandlungen, I. (1875), pp. 2954.]
BINET,
Note sur
l'integrale
/
J.
P.
M.
J a
u dye~y i
qy
(May 24,
1841).
962.
BOCHER, M.
kind (Jan. 1892). Annals of Math. vi. (1892), pp. 8590. On some applications of Bessel's functions with pure imaginary index (Feb. 11, 1892). Annals of Math. vi. (1892), pp. 137160*. On certain methods of Sturm and their application to the roots of Bessel's functions (Feb. 1897). Bulletin American Math. Soc. ill. (1897), pp. 205213. An elementary proof that Bessel's functions of the zeroth order have an infinite number of real roots (Feb. 25, 1899). Bulletin American Math. Soc. v. (1899), pp. 385388.
Bessel's functions of the second
On
4, 1901).
Bulletin
American Math.
333340.
BOHMER,
Uber die Zylinderfunktionen (Nov. pp. 3036. Uber das Airysche
26, 1913).
P. E.
Sitz.
BOHREN,
Integral t (Oct. 6, 1902).
A.
G.
BOOLE,
On
224.
ill.
(1843), pp.
216
On a
pp.
general method in analysis^Jan. 18, 1844). Phil. Trans, of the Royal Soc. 1844,
225282.
Treatise
BOURGET,
Note sur une formule de M. Anger (Aug.
p. 283.
J.
7, 1854).
Memoire sur
celeste.
sci.
les
le
Journal de Math.
33
54.
Memoire sur
mouvement
vibratoire des
95.
membranes
Ann.
BRAJTZEW, J. R. Fourier-Besselschen Funktionen und deren Anwendung zur Auffindung asymptotischer Darstellungen von Integralen der linearen Differentialgleichungen mit rationalen Koeffizienten. Warschau Polyt. Inst. Each. 1902, nos. 1, 2. [Jahrbuch uber die Fortschritte der Math. 1903, pp. 575 577.]
Uber
die
BRASSINNE,
Sur diverses equations
differentielles
(sic).
E.
l'equation de Ricatti I'Acad. R. des Sci. de Toulouse, (3) iv. (1848), pp. 234 236. Sur des equations differentielles qui se rattachent a l'equation de Riccati. Journal de Math. xvi. (1851), pp. 255256.
Mem. de
Summation
Bull.
American
BRENKE, W. C. of a series of Bessel's functions by means of Math. Soc. xvi. (1910), pp. 225 230.
an integral (Nov.
27, 1909).
BRIDGEMAN,
On
(1908), pp.
P.
W.
22, 1908).
Phil.
Mag.
(6) xvi.
947948.
* See also ibid. p. 136. t This paper, which might have been mentioned in 6*4, contains the formula 6*4 (1) the author does not give the formula 6-4 (2).
but
BIBLIOGRAPHY
757
BRUNS, H. Ueber die Beugungsfigur des Heliometer-Objectivea (Oct. 15, 1882). Astr. Nach. civ. (1883), col. 18. BRYAN, G. H. On the waves on a viscous rotating cylinder (June 4, 1888). Proc. Camb. Phil. Soc. vi. (1889), pp. 248264. Wave Motion and Bessel's Functions. Mature, lxxx. (1909), p. 309.
BURKHARDT, H. F. K. L. Trigonometrische Reihen und Integrale (bis etwa 1850). Encyklopadie der Math. Wiss.
II. 1
(Leipzig,
LUTTERWORTH,
On
30, 1913).
S.
the Evaluation of Certain Combinations of the Ber, Bei and Allied Functions
Proc.
(May
Phys
Soc.
294 *97.
C.
Sci. (Soc. Helvetique), (4) xiv. (1902), pp.
CAILLER,
Sur
350.
1904). 368.
les fonctions
347
Note sur une operation analytique et son application aux fonctions de Bessel (March, M4m. de la Soc. de phys. et cPhistoire naturelle de Geneve, xxxiv. (1902 5), pp. 295
CALLANDREAU,
Calcul des transcendantes de Bessel
.
0.
l.(n + l)" ".1.2.(+l)(+2) 1.2...L pour les grandes valeurs de a au moyen de series semiconvergentes.
(2) xiv. (1890), pp. Sur le calcul des
nK)
(rr
(it
,
(it
t
i
'"J'
Bidletin des Sci. Math. Bulletin
110114. polynomes
CARLINI, F. Ricerche sulla convergenza della serie che serva alia soluzione del problema di Keplero* (Milan, 1817).
CARSLAW,
Some Multiform
H.
S.
Solutions of the Partial Differential Equations of Physical Mathematics and their Applications (Nov. 10, 1898). Proc. London Math. Soc. xxx. (1899), pp. 121 161. The Green's function for a wedge of any angle and other Problems in the Conduction of Heat (Oct. 30, 1909). Proc. London Math. Soc. (2) vm. (1910), pp. 365374. The Scattering of Sound Waves by a Cone. Math. Ann. lxxv. (1914), pp. 133 147. The Green's function for the equation V 2 u 2 =0 (April 28, 1913; March 20, 1916). Proc. London Math. Soc. (2) xm. (1914), pp. 236257 (2) xvi. (1917), pp. 8493.
CATALAN,
Note sur l'integrale
[Reprinted,
f
**
E. C.
Sur
Journal de Math. v. (1840), pp. 110114. \ x-) (1 Jo /h la Soc. R. des Sci. de Likge, (2) xn. (1885), pp. 26 31.] Pequation de Riccati (March 4, 1871). Bulletin de VAcad. R. de Belgique, (2) xxxi.
(Feb. 1840).
cos -
-^r +
(ije
dx
Mem. de
(1871), pp.
C. Le Paige (Rapport de M. Catalan). Bulletin de VAcad. R. de Belgique, (2) xli. (1876), pp. 935939. Application d'une formule de Jacobi (Nov. 1868). Me'm. de la Soc. R. des Sci. de Liege, (2) xn. (1885), pp. 312316. CAUCHY, A. L. Resume d'un m^moire sur la m^canique celeste et sur un nouveau calcul appele calcul des limites (Lu a l'Acad. de Turin, Oct. 11, 1831). Exercices dCAnalyse, n. (Paris, 1841), pp. 48-^112. [Oeuvres, (2) xn. (1916), pp. 48112.]
Memoire sur la convergence des series (Nov. 11, 1839). pp. 587588. [Oeuvres, (1) iv. (1884), pp. 518520.]
Comptes Rendus,
ix. (1839),
Translated into German by Jacobi, Astr. Nach. xxx. (1850), Werke, vn. (1891), pp. 189245.]
col.
197
254.
[Get. Math.
758
Considerations nouvelles sur la theorie des suites et sur les lois de leur convergence (April 20, 1840). Comptes Rendus, x. (1840), pp. 640656. [Oeuvres, (1) v. (1885), pp. 180
198.]
et generate pour la determination numerique des coefficients que rendeveloppement de la fonction perturbatrice (Sept. 14, 1840). Comptes Rendus, XI. (1840), pp. 453475. [Oeuvres, (1) v. (1885), pp. 288310.] Note sur le developpement de la fonction perturbatrice (Sept. 21, 1840). Comptes Rendusr xi. (1840), pp. 501511. [Oeuvres, (1) v. (1885), pp. 311321.1 Methodes propres a simplifier le calcul des inegalites periodiques et seculaires des mouvements des planetes (Jan. 11, 1841). Comptes Rendus, xil. (1841), pp. 84 101. [Oeuvres, (1) vi. (1888), pp. 1634.] Note sur une transcendante que renferme le developpement de la fonction perturbatrice relative au systeme planetaire (Oct. 4, 1841). Comptes Rendus, xin. (1841), pp. 682687. [Oeuvres, (1) vi. (1888), pp. 341346.] Note sur la substitution des anomalies excentriques aux anomalies moyennes, dans le developpement de la fonction perturbatrice (Oct. 25, 1841). Comptes Rendus, xni. (1841), pp. 850854. [Oeuvres, (1) vi. (1888), pp. 354359.1
Methode simple
le
ferme
le calcul
Comptes Rendus,
xvm.
(1844), pp.
625643. [Oeuvres,
(1)
vm.
(1893), pp.
168
Sur la transformation des fonctions implicites en moyennes isotropiques, et sur leurs developpements en series trigonometriques (May 22, 1854). Comptes Rendus, xxxvm. (1854), pp. 910913. [Oeuvres, (1) xn. (1900), pp. 148151.] Sur la transformation des variables qui ddterminent les mouvements d'une planete ou m^me d'une comete en fonction explicite du temps, et sur le developpement de ses fonctions en series convergentes (June 5, 1854). Comptes Rendus, xxxvm. (1854), pp. 990 993. [Oeuvres, (1) xn. (1900), pp. 160164.] Sur la resolution des equations et sur le developpement de leurs racines en series convergentes (June 26, 1854). Comptes Rendus, xxxvm. (1854), pp. 1104 1107. [Oeuvres, (1) xn. (1900), pp. 167170.] Sur une formule de M. Anger et sur d'autres formules analogues (July 15, 1854). Comptes Rendus, xxxix. (1854), pp. 129 135. [Oeuvres, (1) xil. (1900), pp. 171177.]
CAYLEY,
A.
Sur quelques formules du calcul integral. Journal de Math. xn. (1847), pp. 231 240. [Collected Papers, I. (1889), pp. 309316.] On Riccati's equation (Sept. 29, 1868). Phil. Mag. (4) xxxvi. (1868), pp. 348351. [Collected Papers, vii. (1894), pp. 912.] Note on the integration of certain differential equations by series. Messenger (Old Series), v. (1869), pp. 7782. [Collected Papers, vm. (1895), pp. 458462.] Proc. London Math. Soc* v. (1874), pp. 123124. [Collected Papers, IX. (1896), pp. 1920.] CHALLIS, H. W. Extension of the Solution of Riccati's Equation (Oct. 5, 1864). Quarterly Journal, vn. (1866), pp. 5153.
CHAPMAN,
S.
the general theory of summability with applications to Fourier's and other series. 52. Quarterly Journal, xliii. (1911), pp. 1
On
CHESSIN,
Note on the General Solution (1894), pp. 186187.
A.
S.
of BessePs Equation.
On the expression of BessePs Functions in Univ. Circulars, xiv. (1895), pp. 2021.
Form
Note on Cauchy's Numbers. Annals of Math. x. (1896), pp. 12. On the relation between Cauchy's numbers and BessePs Functions (July 1, 1898). Annals of Math. xn. (1899), pp. 170174. On some relations between Bessel functions of the first and of the second kind (Oct. 20, 1902). Trans. Acad. Sci. of St Louis, xn. (1902), pp. 99108. Sur Pequation de Bessel avec second membre (Oct. 27, 1902). Comptes Rendus, cxxxv. (1902), pp. 678679. Sur une classe d'equations differentielles redactibles a Pequation de Bessel (May 11, 1903). Comptes Rendus, cxxxvi. (1903), pp. 11241126.
* See
BIBLIOGRAPHY
CHREE,
pp. 329
C.
759
Longitudinal vibrations of a circular bar. Quarterly Journal, xxi. (1886), pp. 287298. On the Coefficients in certain Series of Bessel's Functions. Phil. Mag. (6) xvn. (1909),
331.
CHRISTOFFEL,
Zur Abhandlung: "Ueber
die Zahler
E. B.
briichen" pag. 231 des vorigen Bandes* (March, 1860). pp. ye) y^,
CINELLI, M.
Diffrazione per aperture fatte sopra superfici curve. 155. pp. 141
II
Nuovo Cimento,
(4)
i.
(1895),
CLEBSCH,
(Jeber die Reflexion pp. l7u ^u2.
R. F. A.
30, 1861).
On
Bessel's Functionst.
349.
Second order. Quarterly Journal, xx. (1885), pp. 250260. Second order. Quarterly Journal, xxi. (1886), pp. 183 192.
Equations of the Second Order (Dec. 24, 1861 Jan. 15, 1862: Messenger (Old Series), I. (1862), pp. 118124, 164173, 241247.
;
A New
COTTER,
Method of Solving Legendre's and
J.
R.
Bessel's Equations and others of a similar Irish Acad. % xxyii. (1909), pp. 157161.
A
.
CRAWFORD,
L.
nx=
n
.
a 3 5...(2n
.
-r 1) \sin
(~ ^-V" x dxj
sin**-*
x and
CRELIER,
1895.]
Sur quelques proprtetes des fonctions Besseliennes tirees de la theorie des fractions continues (June, 1895). Ann. di Mat. (2) xxiv. (1896), pp. 131163. {Dissertation, Bern, L '
'
e espeee n (x) (Dec. 1896). la fonction Besselienne de 96. [1898], pp. Sur les fonctions besseliennes 0*(x) et n {x) (Sept. 6, 1897
Sur
61
CURTIS, A. H.
On
and
the integration of Linear and Partial Differential Equations (Nov. 24, 1854) Dublin Math. Journal, ix. (1854), pp. 272290.
Camb.
CURZON, H. E. J. Generalisations of the Hermite functions and their connexion with Bessel functions (Nov. 10, 1913). Proc. London Math. Soc. (2) xin. (1914), pp. 417440.
DATTA, A. a generalisation of Neumann's Expansion in a Series of Bessel Functions (Feb. 29, 1920). Bulletin Calcutta Math. Soc. xi. (1921), pp. 2334. On an extension of Sonine's Integral in Bessel Functions|| (Jan. 6, 1921). Bulletin Calcutta Math. Soc xi. (1921), pp. 221230.
On
t Clifford died March 3, 1879. xxi., as stated in the Jahrbuch fiber die Fortschritte der Math. 1908, p. 383. See footnote + on p. 27. This paper, which deals with the corrected form, given in 13-46 (10), of Nicholson's integral, and with various related integrals, was published after Chapter xin had been passed for Press.
||
760
Naherungsformeln fur die Zylinderfunktionen fur grosse Werte des Arguments und unbeschrankt veranderliche Werte des Index (Dec. 1908). Math. Ann. lxvii. (1909), pp. 535558. Semikonvergente Entwickelungen fur die Zylinderfunktionen und ihre Ausdehnung ins Komplexe (Feb. 5, 1910). MUnchener Sitzungsberichte, XL. (1910), no. 5.
DE LA VALLEE POUSSIN,
C. J.
Integration de l'equation de Bessel sous forme finie (Jan. 26, 1905). Sci. de Bruxelles, xxix. (l 6re partie) (1905), pp. 140143.
Ann. de la Soc.
On
1917).
NICHOLSON, J. W. DENDY, A. the Influence of Vibrations upon the Form of Certain Sponge Spicules
Proc.
AND
(May
11,
Royal
Soc.
lxxxix.
(1917), pp.
573587.
DINI, U.
Serie di Fourier e altere rappresentazioni analitiche delle funzioni di (Pisa, 1880).
una
variabile reale
DINNIK,
Uber
A.
die Darstellung einer willkiirlichen Funktion durch BessePsche Reihe. Kief. Polyt. 85. [Jahrbuch iiber die Fortschritte der Inst. (Engineering Section), 1911, no. 1, pp. 83 Math. 1911, p. 492.]
J\ und
(3)
337338.
und Phys.
(3)
J \
und Ja-
(3)
324326.
xxn.
(1914), pp.
226227.
DIXON, A. C. a property of Bessel's Functions. Messenger, xxxil. (1903), pp. 7 n The expansion of x in Bessel'sFunctions. Messenger, xxxn. (1903),
On
8.
p. 8.
DONKIN, W.
On
the Equation of Laplace's Functions,
Soc. cxlvii. (1857), pp.
F.
Phil. Trans, of the
Royal
4357.
DOUGALL,
The determination
J.
of Green's function by means of Cylindrical or Spherical (March 9, 1900). Proc. Edinburgh Math. Soc. xvni. (1900), pp. 3383. Theorem of Sonine in Bessel Functions with two Extensions to Spherical (Dec. 13, 1918). Proc. Edinburgh Math. Soc. xxxvn. (1919), pp. 3347.
Harmonics
Harmonics
BOIS REYMOND, P. D. G. Die Theorie der Fourier'schen Integrale und Formeln (June
(1871), pp.
DU
26, 1871).
Math. Ann.
iv.
362390.
EARNSHAW,
An
essay towards
S.
an
entirely
ELLIS,
R. L.
the Integration of certain Differential Equations (Nov. 1840 and Feb. 1841). Camb. Math. Journal, n. (1841), pp. 169177, 193201. On the Method of Least Squares (March 4, 1844). Trans. Camb. Phil. Soc. vm. (1849),
pp. 204219.
On
EMDE,
Zur Berechnung der
der Math,
F.*
Archiv
und
ENNEPER,
Ueber
ein bestimmtes Integral.
A.
Math. Ann.
360
365.
EPSTEIN,
845846.]
S. S.
Die vier Rechnungsoperationen mit Bessel'schen Functionen nebst einer geschichtlichen Einleitung (Bern, 1894, 58 pp.). [Jahrbuch Uber die Fortschritte der Math. 1893 94, pp.
BIBLIOGRAPHY
ERMAKOFF, W.
Ueber die Cylinderfunctionen (May,
1872).
761
Math. Ann.
v. (1872), pp.
639640.
ESCHERICH,
Zur Bessel'schen
p. 142.
Differential-Gleichung.
und Phys. m.
(1892),
Uber
eine Naherungsformel.
und Phys.
EULER,
Lettre de
pp.
M. Euler a M. de
la
Grange (Jan.
1,
L. 1760).
110.
Recherches sur l'integration de l'equation
ddz _
d(l
ddz dx2
b dz
x dx
xx
Misc. Taurinensia, ill. (1762 65), pp. 60 91. De integratione aequationum differentialium.
61)
[1763], pp. 154169. De resolutione aequationis dy+ayydx = bxm dx. Novi Comm. Acad. Petrop. ix. (1762 63) [1764], pp. 154 169. Deniotu vibratorio tympanorum. Novi Comm. Acad. Petrop. x. (1764) [1766], pp.
243260.
Institutionum Calculi Integralis,
II.
(Petersburg, 1769).
v.
De
oscillationibus
pars
(1781)
[1784], pp.
157177.
De perturbatione motus chordarum ab earum pondere oriunda. Acta Acad. Petrop. v. pars 1 (1781) [1784], pp. 178190. Analysis facilis aequationem Riccatianam per fract.onem continuam resolvendi. Mem. de I' Acad. R. des Sci. de St. Petersbourg, vi. (1818), p) 1229.
209
248.
in der
Nahe
einer
FAXEN,
Expansion in
series of the integral
/
H.
e~ x
(tt ~'i)
Vdt
J v
FELDBLUM, M. The theory of Riccati's equation and applications of the function which satisfies it. Warsaw University, 1898, nos. 5 and 7 1899, no. 4. \Jahrbuch iiber die Fortschritte der
;
FERIET, K. DE.
Sur les fonctions hypercylindriques (June pp. 14641466.
13, 1921).
A method
VI.
FIELDS, J. C. of solving Riccati's Equation (April 8, 1886). Johns Hopkins Univ. Circulars
p. 29.
(188687),
(May
19, 1886).
(1886
FILON, L. N. G. of Expressing Solutions of Laplace's Equation in Terms of Operators involving Bessel Functions. Phil. Mag. (6) vi. (1903), pp. 193213. On the expansion of polynomials series of functions (May 10, 1906). Proc. London Math. Soc. (2) iv. (1906), pp. 396430.
On a New Mode
FORD, W." On
Math. Soc.
w.
B. F.
B.
the possibility of differentiating term-by-term the developments' for an arbitrary function of one real variable in terms of Bessel functions (June, 1902). Trans. American
iv. (1903), pp.
178184.
25
762
On
by the
,
series
a<30
M| a. g+1.
ye Quarterly Journal, xix. (1883), pp. 292 337. The expression of Bessel functions of positive order as products, and of their inverse powers as sums of rational fractions. Messenger, l. (1921), pp. 129149.
"
La
FOURIER,
J. B. J.
[Translated by A. Freeman,
Cam-
bridge, 1878.]
FREEMAN,
Proc.
A.
allied to
Note on the value of the least root of an equation Camb. Phil. Soc. in. (1880), pp. 375377.
Ja (z) =
(April 19 1880)
Memoire sur
R. des Sci.
v.
la diffraction
(1866), pp.
FRULLANI,
Sopra
la
G.
(Feb. 4,
dipendenza fra i differenziali delle funzioni e 1818). Mem. soc. ital. (Modena), xvm. (1820), pp. 458517.
FUSS,
Correspondance mathe'matique
Steele*, n. (Petersburg, 1843).
et
P.
H.
ce'lebres
physique de quelques
geometres
du
xviii*'"
GALLOP, E. G. distribution of electricity on the circular disc Journal, xxi. (1886), pp. 229256.
The
GASSER,
A.
Ueber die Nullstellen der Besselschen Funktionen (July, 1 904). Mittheilunqen der Naturf. Ges. in Bern, 1904, pp. 92135.
GEGENBAUER,
Note
berichte,
L.
tiber die Bessel'schen Functionen zweiter Art (Feb. 8, 1872). Wiener Sitzunqslxv. (2) (1872), pp. 3335. Zur Theorie der Bessel'schen Functionen zweiter Art (July 4, 1872). Wiener Sitzunqsberichte, lxvi. (2) (1872), pp. 220223. Note uber bestimmte Integrale (Feb. 6, 1873). Wiener Sitzunqsberichte, lxvii. (2) (1873). \ /\ /> pp. 202204. Uber die Functionen nm (June 13, 1873). Wiener Sitzungsberichte, lxviii.
pp. 357
367.
(2) (1874),
die Bessel'schen Functionen (March 19, 1874). Wiener Sitzunqsberichte, lxx. (2) (1875), pp. 616. Uber einige bestimmte Integrale (June 18, 1874). Wiener Sitzunqsberichte, lxx. (2) * (1875), pp. 433443. Uber einige bestimmte Integrale (June 17, 1875). Wiener Sitzunqsberichte, lxxii. (2) v ' (1876), pp. 343354. Uber die Bessel'schen Functionen (June 22, 1876). Wiener Sitzunqsberichte, lxxiv (2) ' (1877), pp. 124130. Zur Theorie der Bessel'schen Functionen (Jan. 18, 1877). Wiener Sitzunqsberichte, *. lxx v. (2) (1877), pp. 218 222. Uber die Functionen Cn v {x) (April 12, 1877). Wiener Sitzungsberichte, lxxv. (2) (1877), \ / \ /> 905. pp. 891
Uber
w
w
'
Das Additionstheorem derjenigen Functionen welche bei der Entwicklung von "* nach den Naherungsnennern regularer Kettenbruche auftreten. Wiener Sitzunqsberichte, lxxxv.
(2) (1882), pp.
491502. die Bessel'schen Functionen (Oct. 11, 1883). (1884), pp. 9751003.
Uber
* This work contains a number of letters from Nicholas Bernoulli (the younger) to Goldbach, which Bernoulli's solution of Biccati's equation is to be found. The reader should notice that in Daniel Bernoulli's letters to Goldbach (ibid. pp. 254, 256, 259), the equation described in the table of contents as Riccati's equation is really the linear equation; Biccati's equation is mentioned on p. 260.
in
BIBLIOGRAPHY
pp.
763
Zur Theorie der Functionen Cn" (x). Wiener Akad. Denkschriften, xlviii. (1884), 293316. Uber die Bessel'schen Functionen (March 10, 1887). Wiener Sitzungsberichte, xcv. (2) (1887), pp. 409410.
Einige Satze iiber die Functionen
pp. 425
766.
CH'(x).
480.
1891).
Bemerkung zu der von Herrn Elaas gegebenen Theorie der elektrischen Schwingungen in cylindrischen Drahten. Monatshefte fiir Math, und Phys. iv. (1893), pp. 379 380. Uber die zum elektromagnetischen Potentiale eines Kreisstromes associierte Function. Monatshefte fur Math, und Phys. iv. (1893), pp. 393401. Eine Integralrelation. Monatshefte Mr Math, und Phys. v. (1894), pp. 53 61. Bemerkung iiber die Bessel'schen Functionen. Monatshefte fiir Math, und Phys. vni. (1897), pp. 383384. Notiz iiber die Bessel'schen Functionen erster Art. Monatshefte fiir Math, und Phys. x. (1899), pp. 189192. Quelques proprtetes nouvelles des racines des fonctions de Bessel. Mem. de la Soc. R. des Set. de Liege, (3) n. (1900), no. 3. [Letter on a paper by H. M. Macdonald.] Proc. London Math. Soc. xxxii. (1901), pp. 433436. tJber eine Relation des Herrn Hobson (May 22, 1902). Wiener Sitzungsberichte, cxl (2 a)
(1902), pp. 563572. On integrals containing functions of Bessel. [Letter to Kapteyn.] Proc. Section of ScL, K. Akad. van Wet. te Amsterdam, iv. (1902), pp. 584588.
GENOCCHI, A. Studi intorno ai casi d' integrazione sotto forma finite. Mem. delV Accad. delle Sci. di Torino, xxni. (1866), pp. 299362. Sur Pequation de Riccati (Aug. 13, 1877). Comptes Bendus, lxxxv. (1877), pp. 391394.
GIBSON, G. A. Proof of the Binomial Theorem with some Applications (Dec. 12, 1919). Edinburgh Math. Soc. xxxviil (1920), pp. 69.
Proc.
GILBERT,
L. P.
Recherches analytiques sur la diffraction de la lumiere. Mem. couronnes de VAcad. R. des Sci. de BruxelUs, xxxi. (1863), pp. 1 52.
GIULIANI,
G.
Sopra la funzione P (cosv) per n infinite. Oiornale di Mat. xxn. (1884), pp. 236 239. Sopra alcune funzioni analoghe alle funzioni cilindriche. Oiornale di Mat. xxv. (1887) pp. 198202. Alcune osservazioni sopra le funzioni spheriche di ordine superiore al secondo e sopra altre funzioni che se ne possono dedurre (April, 1888). Giornale di Mat. xxvi. (1888), pp. 155171.
GLAISHER, On On
Riccati's equation
J.
W.
L.
(March 15, 1871). Quarterly Journal, xi. (1871), pp. 267 273. the Relations between the particular Integrals in Cayley's solution of Riccati's Equation (May 12, 1872). Phil. Mag. (4) xliii. (1872), pp. 433 138.
On
1517. Notes on definite integrals. Messenger, n. (1873), pp. 72 79. On a Differential Equation allied to Riccati's (Oct. 11, 1872). Quarterly Journal, xil. (1873), pp. 129137. Sur une Propriete' de la Fonction e^ x Nouvelle Corr. Math. n. (1876), pp. 240 243, 349350. On a Formula of Caucby's for the Evaluation of a class- of Definite Integrals (Nov. 6, 1876). Proc. Camb. Phil. Soc. in. (1880), pp. 512. On certain Identical Differential Relations (Nov. 9, 1876). Proc. London Math. Soc. vm. (1877), pp. 4751.
1872, pp.
Proc.
(1878),
pp.
469470.
764
Example illustrative of a point in the solution of differential equations by series. Messenger, vin. (1879), pp. 2023. On a symbolic theorem involving repeated differentiations (May 19, 1879). Proc. Camb.
Phil. Soc. in. (1880), pp. 269271. On Riccati's Equation and its Transformations satisfy them (June 16, 1881). Phil Trans, of the [Proc. Royal Soc. xxxn. (1881), p. 444]
and on some Definite Integrals which Royal Soc. 172 (1881), pp. 759828.
GORDAN,
See under Hermite.
P.
GRAF, J. H. Ueber die Addition und Subtraction der Argumente bei BesseFschen Functionen nebst einer Anwendung (March, 1893). Math. Ann. xliii. (1893), pp. 136 144. Ueber einige Eigenschaften der Bessel'schen Function erster Art, insbesondere fur ein grosses Argument. Zeitschrift fur Math, xxxvm. (1893), pp. 115 120. Beitrage zur Auflosung von linearen Differentialgleichuugen zweiter Ordnung mit linearen Coefficienten so wie von Differentialgleichungen zweiter Ordnung denen gewisse bestimmte Integrale geniigen (March, 1894). Math. Ann. xlv. (1894), pp. 235262. Relations entre la fonction Besselienne de l re espece et une fraction continue (May, 1894). Ann. di Mat. (2) xxm. (1895), pp. 4565. Ableitung der Formeln fiir die Bessel'schen Functionen bei welchen das Argument ein
4,
1896).
5962.
;
Einleitung in die Theorie der Besselschen Funktionen. Von J. H. Graf und E. Gubler Bern, 1898, 1900). Beitrag zur Auflosung von Differentialgleichungen zweiter Ordnung denen gewisse bestimmte Integrale geniigen (May, 1902). Math. Ann. lvi. (1903), pp. 423 444.
Hefte
Treatise
GREENHILL, SIR A. GEORGEL On Riccati's Equation and Bessel's Equatipn. Quarterly Journal, xvi. (1879), pp. 294 298. On the Differential Equation of the Ellipticities of the Strata in the Theory of the
Figure of the Earth (April 8, 1880). Quarterly Journal, xvn. (1880), pp. 203207. Determination of the greatest height consistent with stability that a vertical pole or mast can be made, and of the greatest height to which a tree of given proportions can grow
(Feb.
7, 1881). Proc Camb. Phil. Soc. iv. (1883), pp. 6573. The Bessel-Clifford Functiou (March 14, 1919). Engineering, cvn. (1919), p. 334. The Bessel-Clifford Function and its applications (Aug. 11, 1919). Phil. Mag. (6) xxxvin.
(1919), pp.
501528.
GRUNERT,
^,
t-1
J.
1
A.
- =(-l) '"
,
1.3...(2t'-l)
fur
z=conz. Archiv
und Phys.
104109.
GUBLER, E.t Die Darstellung der allgemeinen Bessel'schen Function durch bestimmte Integrale (Sept. 1888). Zurich Vierteljahrsschrift, xxxni. (1888), pp. 130 172. Verwandlung einer hypergeometrischen Reihe im Anschluss an das Integral
Ja {x)e- hx xc ~
/:
dx.
Inaugural-dissertation, Zurich, 1894 (38pp.). [Graf and Gubler, Einleitung in die Theorie der Besselschen Funktionen, n. (Bern, 1900), pp. 110 135, 156.] Ueber ein discontinuierliches Integrale (Dec. 1895). Math. Ann. xlviii. (1897),.pp. 37 48. Beweis einer Formel des Herrn Sonine (Dec. 1896). Math. Ann. xlix. ( 1897), pp. 583 584. Ueber bestimmte Integrale mit Bessel'schen Functionen (Oct. 1902). Zurich Vierteljahrsschrift, XLVll. (1902), pp. 422428.
GUNTHER,
Bemerkungen pp. 292297.
iiber
S.
Cylinderfunctionen.
*
und
Phys.
lvi.
(1874),
See also under Sir Joseph John Thomson. t See also under Graf.
BIBLIOGRAPHY
GWYTHER,
The employment
765
R. F. of a geometrical construction to prove Schlomilch's series, and to aid in its development into a definite integral. Messenger, xxxm. (1904), pp. 97 107.
HADAMARD,
Sur
l'expression asymptotique
J.
de
la fonction
85.
HAENTZSCHEL,
Ueber die functionentheoretisehen Zasammenhang zwischen den Lame'schen, Laplace'schen und Bessel'schen Functioned. Zeitschrift fiir Math. xxxi. (1886), pp. 25 33. Ueber die Fourier Bessel'sche Transcendeute (Nov. 20, 1887). Zeitschrift fiir Math. xxxm. (1888), pp. 185186. HAFEN, M. Studien iiber einige Probleme der Potentialtheorie. Math. Ann. lxix. (1910), pp. 517 537.
HAGUE,
B.
Note on the Graphs of the Bessel Functions of Integral Order. Proc. Phys. Soc. xxix. (1917), pp. 211214.
HALL,
The
Besselian Function.
A.
The Analyst,
I.
(1874), pp.
8184
On Fluctuating Functions (June 22, 1840). Trans. R. Irish Acad. xix. (1843), pp. 264321. On the Calculation of the Numerical Values of a certain class of Multiple and Definite
pp. 375382. H. Die Cylinderfunctionen erster und zweiter Art (Dec. 15, 1868). Math. Ann. I. (1869), pp. 467501. Bestimmte Integrale mit Cylinderfunctionen +. Math. Ann. vm. (1875), pp. 453 470. Die Fourier'schen Reihen und Integrale fiir Cylinderfunctionen t (May 16, 1869). Math. Ann. viu. (1875), pp. 471-494.
Integrals (Sept. 29, 1857).
Phil.
HANKEL,
HANSEN, P. A. Ermittelung der absoluten Storungen in Ellipsen von beliebiger Excentricitat und Neigung, I. Schriften der Stermoarte Seeberg (Gotha, 1843). [Memoire sur la determination des perturbations absolues dans les ellipses d'une excentricite et d'une inclinaison quelconques. Par M. Hansen. Traduit de l'AUemand par M. Victor Mauvais (Paris, 1845).] Eutwickelung des Products einer Potenz des Radius Vectors mit den Sinus oder Cosinus eines Vielfachen der wahren Anomalie in Reihen. Leipziger Abh. II. (1855), pp. 181 281.
HANUMANTA
On a
RAO,
G. H.
C. V.
137.
HARDY,
:
General theorems in contour integration with some applications. Quarterly Journal, xxxn. (1901), pp. 369384. Notes on some points in the integral calculus, xvm. Messenger, xxxv. (1906), pp. 158 166. Further researches in the Theory of Divergent Series and Integrals (May 13, 1908). Trans. Camb. Phil. Soc. xxi. (1912), pp. 1 48. On an Integral Equation (Feb. 20, 1909). Proc. London Math. Soc. (2) vn. (1909),
pp. 445 472. On certain definite integrals whose values can be expressed in terms of Bessel's functions. Messenger, xxxvm. (1909), pp. 129132. On certain definite integrals considered by Airy and Stokes. Quarterly Journal, xli.
(1910), pp.
226240. Notes on some points in the integral calculus, xxvu. Messenger, XL. (1911), pp. 44 51. Notes on some points in the integral calculus, xxxv. Messenger, xlii. (1913), pp.
8993.
On
the expression of a
number
as the
sum
of two squares.
Proc.
problem (April
22, 1915).
London Math.
pp. 125.
Notes on some points in the integral calculus, xlvii. Messenger, xlviii. (1918), pp. 81
*
88.
letter by Hamilton on Bessel functions is published in Sir G. G. Stokes, Memoir and 135. Scientific Correspondence, i. (Cambridge, 1907), pp. 131 Hankel died Aug. 29, 1873. These memoirs were composed from materials found among his t
papers.
766
On
Royal
the Solution of Linear Differential Equations (June 10, 1847). Phil. Trans, of the
Soc. 1848, pp. 3154. Riccati's Equation (April 4, 1865).
On
258.
Mag.
(6)
HARGREAVES,
an Asymptotic Theorem
HARNACK, A*
Ueber die Darstellung einer willkurlichen Function durch die Fourier-Bwsel'schen Functionen (Dec. 12, 1887). Leipziger Berickte, xxxix. (1887), pp. 191 214; Math. Ann. xxxv. (1889), pp. 4162.
On Harmonic
420.
HARTENSTEIN,
Integration der Differentialgleichung
J.
s
H.
fur elliptische
^ + ^=i /
K.
und parabolische
und
199.
HATTENDORF,
See under Riemann.
HAVELOCK, T. H. Mathematical Analysis of Wave Propagation in Isotropic Space of p Dimensions (March 16, 1904). Proc. London Math. Soc. (2) n. (1904), pp. 122137.
T. cylindrical function. Nyt Tidsskrift, xxiu. B (1912), pp. 8690. [Jahrbuch fiber die Fortschritte der Math. 1912, p. 555.] 8 8 On the Integrals Texp Lr? p6 q6d6 (Dec. 1920). Tdhoku Math. Journal, xx.
HAYASHI,
On a
Neumann's
Y?
(1922),
pp. 107114.
HEAVISIDE,
O.
Electrical Papers, i., n. (London, 1892). On Operators in Physical Mathematics (Dec. 15, 1892 ; Lll. (1893), pp. 504529; Liv. (1893), pp. 105143. Electromagnetic Theory +, n., in. (London, 1899, 1912).
June
8, 1893).
HEINE, H.
Ueber
die Zahler
E.
(Sept. 1859).
Journal fur Math. lvii. (1860), pp. 231 247. Die Fourier-Besselache Function (June, 1868). Journal fur Math. lxix. (1869), pp.
141.
128
Theorie
und Anwendunqen
C.
(2
Bande)
(Berlin, 1878,
transcendante n Ann. di Mat. (2) in. (1870), p. 83. Extrait d'une lettre de Monsieur Ch. Hermite a Monsieur Paul Oordan (June 9, 1873). Journal fUr Math, lxxvi. (1873), pp. 303 311. Extrait d'une lettre a M. E. Jahnke (Nov. 25, 1900). Archiv der Math, und Phys. (3)
la
.
Sur
HERMITE,
I.
(1901), pp.
2021.
HERTZ,
Uber
H.
die Induktion in rotierenden Kugeln. Dissertation, Berlin, Werlee, I. (Leipzig, 1895), pp. 37 134]
March
15, 1880.
[Oes.
449455.
[Oes. Werlee,
I.
und Chemie,
* Harnack died April 3, 1888. t This work consists of a series of articles first published in The Electrician, Nature and elsewhere in and after 1894, with numerous additions. X See also under Christoffel. This note contains a statement of Carlini's formula, which Hermite apparently derived from Poisson's integral.
BIBLIOGRAPHY
HERZ,
N.
(Sept. 21, 1883).
Astr.
767
Bemerkungen zur Theorie der Bessel'schen Functionen cvn. (1884), col. 1728.
evil. (1884), col.
Nach.
Note, betreffend die Entwicklung der storenden Krafte (Mar. 30, 1884). Astr. Nach.
429432.
HILB, E. willkurlicher Funktionen nach Eigenfunktionen (Sept. 11, 1917). Math. Zeitschrift, I. (1918), pp. 5869. Uber die Laplacesche Reihe (March 15, 1919 ; Nov. 17, 1919). Math. Zeitschrift. v. (1919), pp. 1725 ; vm. (1920), pp. 7990.
Zur Theorie der Entwicklungen
De
ubi a,
pp.
2237.
dxV + a + by+cy^O, functiones rationales ipsius x (May 24, 1840). Journal fur Math. xxv. (1843),
* HOBSON, E. Systems of Spherical Harmonics (June 11, 1891). Proc. London Math. Soc. xxn. (1891), V " pp. 431449. On the Evaluation of a certain Surface- Integral, and its application to the Expansion, in Series, of the Potential of Ellipsoids (Jan. 12, 1893). Proc. London Math. Soc. xxiv (1893), pp. 8096. On Bessel's Functions, and Relations connecting them with Hyper-spherical and Spherical Harmonics (Dec. 14, 1893). Proc. London Math. Soc. xxv. (1894), pp. 4975. On the most general solution of given Degree of Laplace's Equation (May 9, 1895) Proc. London Math. Soc. xxvi. (1895), pp. 492 494. Note on some properties of Bessel's Functions (Jan. 14, 1897). Proc. London Math. Soc. xxvni. (1897), pp. 370375. On the representation of a function by series of Bessel's functions (Dec 10 1908) Proc. London Math. Soc. (2) vn. (1909), pp. 359388.
'
L. A. J. W. Uber komplexe Integraldarstellungen der Zylinderfunktionen. Archiv der Math, Phys. (3) xviii. (1911), pp. 116.
HOPF,
UND SOMMERFELD,
HORN,
J.
und
Ueber lineare Differentialgleichungen mit einem veranderlichen Parameter (Dec 1898). Math. Ann. lii. (1899), pp. 340362.
30.
HURWITZ,
A.
Ueber die Nullstellen der Bessel'schen Function (June 2, 1888). Math. Ann xxxm (1889), pp. 246266. Ueber die Wurzeln einiger transcendenten Gleichungen. Hamburqer Mittheilunqen,
II.
(1890), pp.
p. 115.]
HYMERS,
Treatise loO?j
J.
IGNATOWSKY, W. VON.
Uber die Reihenentwicklungen mit Zylinderfunktionen (May 13, 1911). Archiv der Math, und Phys. (3) xviii. (1911), pp. 322327. Uber Reihen mit Zylinderfunktionen nach dem Vielfachen des Argumentes (Dec 1913) Archiv der Math, und Phys. (3) xxm. (1915), pp. 193219.
ISELI, F. Die Riccati'sche Gleichung. Dissertation, Bern, 1909 (42 pp.). [Jahrbuch Uber die Fort*r , L schrUteder Math. 1909, p. 369.]
ISHERWOOD,
J.
G.
Tables of the Bessel Functions for pure imaginary values of the 1904). Manchester Memoirs, xlviii. (1903 4), no. 19.
argument (April
.
26.
JACKSON,
XXI. (1903), pp.
F.
H.
Proc. Edinburgh Math. Soc.
6572.
On
Royal
Soc.
Generalised functions of Legendre and Bessel (Nov. 16, 1903). Trans. Edinburgh y xh. (1905), pp. 128.
* See also
under Gegenbauer.
768
generalization of Neumann's expansion of an arbitrary function in a series of Bessel functions (Nov. 26, 1903). Proc. London Math. Soc. (2) j. (1904), pp. 361366. Theorems relating to a Generalisation of the Bessel- Function (March 21, 1904). Trans.
Edinburgh Royal Soc. xli. (1905), pp. 105118. Note on a theorem of Lommel (May 13, 1904). Proc. Edinburgh Math. Soc. xxil. (1904), pp. 8085. The application of Basic numbers to Bessel's and Legendre's functions (June 2, 1904
Dec.
1,
1904
London Math.
192220;
Proc.
(2) ill.
(1905), pp.
120, 2123.
1904).
Edinburgh
Trans.
Royal
Soc. xxv. (1904), pp. 273276. Theorems relating to a Generalization of Bessel's Edinburgh Royal Soc. xli. (1905), pp. 399408.
JACKSON, W.
On
H.
the diffraction of light produced by an opaque prism of finite angle (Nov. 11, 1903). Proc. London Math. Soc. (2) I. (1904), pp. 393414.
JACOBI, C. G. J * Formula transformationis integralium definitorum (July 9, 1835). Journal fur Math. xv. (1836), pp. 126. [Ges.Math. Werke, vi. (1891), pp. 86 118.] Formule pour la transformation d'une classe d'integrales definies, Journal de Math. I. (1836), pp. 195 196. Versuch einer Berechnung der grossen Ungleichheit des Saturns nach einer strengen Entwicklung. Astr. Nach. xxvm. (1849), col. 6580, 8194. [Ges. Math. Werke, vn. (1891),
pp. 145174.] Uber die annahernde Bestimmung sehr entfernter Glieder in der Entwickelung der elliptischen Coordinaten, nebst einer Ausdehnung der Laplacefschen Methode zur Bestimmung der Functionen grosser Zahlen. Astr. Nach. xxvm. (1849), col. 257270. [Ges. Math. Werke, vn. (1891), pp. 175188.]
JAHNKE, P. E. E.+ Funktionentafeln mit Formeln und Kurven. Von E. Jahnke und F. Emde (Leipzig, 1909). Uber einige, in der elektromagnetischen Strahlungstheorie auftretende, bestimmte Integrate. Archie der Math, und Phys. (3) xxm. (1914), pp. 264267.
Sur
caise,
les
JAMET, E. V. equations anharmoniques (Sept. 13, 1901). Comptes Rendus de V Assoc. FranSci.
xxx. (Ajaccio) (1901), pp. 207228. Sur les equations anharmoniques. Ann. de la Fac. des pp. r 121.
JEKHOWSKY,
B.
Les fonctions de Bessel de plusieurs variables exprimees par des fonctions de Bessel d'une variable (Feb. 28, 1916). Comptes Rendus, clxii. (1916), pp. 318319. Sur la fonction generatrice des fonctions de Bessel a plusieurs variables. Bulletin des
math. (2) xli. (1917), pp. 5860. les fonctions de Bessel a deux variables (1921), pp. 13311332.
Sci.
Sur
(May
30, 1921).
JOHNSON, W. W.
On
the Differential Equation
g +y + py+<2 =0.
2
JOLLIFFE,
The expansion
functions.
A. E.
of the square of a Bessel function in tht form of a series of Bessel Messenger, xlv. (1916), p. 16.
Sur
les fonctions
xxvm.
(1895), pp. 221225. Sur les ondes lumineuses sphenques et cylindriques (Dec. XXVIII. (1895), pp. 226244.
*
BIBLIOGRAPHY
KALAHNE,
tJber die Wurzeln einiger Zylinderfunktionen
769
A.
Gleich-
ungen (March
1,
1906).
und gewisser aus ihnen gebildeter Zeitschrift fur Math, und Phys. liv. (1907), pp. 5586.
KAPTEYN,
Nouvelles formules pour representer
xvi. (1892), pp.
W*
Jn _Ax).
Bulletin des Sci. Math. (2)
(1893). V
la fonction
4144.
12, 1892).
Recherches sur
(1893), pp.
les fonctions
sci.
91120.
definies con tenant des fonctions
'
de Bessel.
Archives Ne'er-
landaises, (2) vi. (1901), pp. 103116. A definite integral containing Bessel's functions (June 29, 1901). K. Akad. van Wet. te Amsterdam, iv. (1902), pp. 102103.
pp.
Sur un developpement de M. Neumann. Nieuw Archief voor Wiskunde, (2) h *\ i vi. (1905). \ 4955. Einige Bemerkungen iiber Bessel'sche Functionen. Monatshefte fur Math, und Phys. xiv. (1903), pp. 275282. The values of some definite integrals connected with Bessel functions (Nov. 26, 1904). Proc. Section of Sci., K. Akad. van Wet. te Amsterdam, vn. (1905), pp. 375 376. On a series of Bessel functions (Dec. 24, 1904). Proc. Section of Sci., K. Akad. van Wet te Amsterdam, vn. (1905), pp. 494500. A definite integral of Kummer (Sept. 30, 1905). Proc. Section of Sci., K. Akad. van Wet. te Amsterdam, vyi. (1906), pp. 350 357. On an expansion of an arbitrary function in a series of Bessel functions. Messenger, xxxv. (1906), pp. 122125. Sur la sommation d'une serie infinie. Nieuw Archief voor Wiskunde, (2) vn. (1907), pp.
The quotient c-f two successive Bessel Functions (Dec. 30, 1905 ; Jan. 27, 1906). Proc. Section of Sci., K. Akad. van Wet. te Amsterdam, viii. (1906), pp. 547 549, 640 642. Sur le quotient de deux fonctions besseliennes successives. Archives Neerlandaises, (2) xi. (1906), pp. 149168. Recherches sur les fonctions cylindriques. Mem. de la Soc. R. des Sci. de Liege, (3) vi.
numerique de la
serie
I
*~
(a2
l q
.
+0V)2
Mem. de la Soc. R. des Sci. de Liege, (3) vi. (1906), no. 9. On some relations between Bessel's Functions (Feb. 24,
Akad. van Wet.
te
969.
1912).
K.
KELVIN, LORD.
the waves produced by a single impulse in water of any depth or in a dispersive medium (Feb. 3, 1887). Proc. Royal Soc. xlii, (1887), p. 80; Phil. Mag. (5) xxin. (1887) "pp.352 255. -{MaiK. and Phys. Paper*, iv. (1910), pp. 303^306.] Ether, Electricity and Ponderable Matter (Jan. 10, 1889). Journal of Inst, of ElectHcal Engineers, xvni. (1890), pp. 437. [Math, and Phys. Papers, in. (1890), pp. 484515.]
On
KEPINSKI,
TJber die Differentialgleichung
o 22
S.
m+1
x
dz
cvr
dx*
(Jan. 1905).
n dz _ x dt~
198205.
w. b. F.
26
770
Ueber den inducirten Magnetisrnus eines unbegrenzten Cylinders von weichem Eisen
(June, 1853).
376.'
graphendrahten (Oct. 29, 1877). Berliner Monatsberichte, 1877, pp. 598 611. Ueber die Transversalsohwingungen eines Stabes von veranderlichem Querschnitt (Oct. 27, 1879). Berliner Monatsberichte, 1879, pp. 815 828; Ann. der Physik und C/iemie,
501512.
Ueber die elektrischen Strdraungen in einera Kreiscylinder (April 26, 1883). Sitzungsberichte, 1883, pp. 519 524.
Berliner
te
probability problem (Sept. 30, 1905). Proc. Section of Sci., K. Akad. van Wet. Amsterdam, vm. (1906), pp. 341350. An integral theorem of Gegenbauer (Feb. 27, 1909). Proc. Section of Sci., K. Akad.
Wet. te
A local
KLUYVER,
J.
C.
van
Amsterdam,
749
755.
J. C. C. A. Die Entwicklung willkiirlicher Funktionen in Reihen die nach Bessel'schen Funktionen fortschreiten (July, 1903). Archie der Math, und Phys. (3) vn. (1903), pp. 123133. Die Theorie der Integralgleichungen und die Darstellung willkiirlicher Funktionen in der mathematischen Physik. Math. Ann. lxiii. (1907), pp. 47s7 524.
KNESER,
KNOCKENHAUER,
KONIG,
K.
W.
Ueber die Oerter der Maxima und Minima des gebeugten Lichtes nach den Fresnel'schen Beobachtungen. Ann. der Physik und Chemie, (2) xli. (1837), pp. 103 110.
J.
Ueber die Darstellung von Functionen durch unendliche Reihen (Sept. 1871). Math. Ann. v. (1872), pp. 310340. Ueber Reihenentwicklung nach Bessel'schen Functionen (1880). Math. Aim. xvn. (1880), pp. 8586.
KOESTLER, W.
Beitrage zu Reihenentwickelungen nach Besselsche Zylinderfunktionen. Dissertation, Bern, 1907 (110 pp.). [Jahrbuch fiber die Fortschritte der Math. 1908, p. 535.]
KOPPE, M. Die Ausbreitung einer Erschutterung an der Wellenmaschine darstellbar durch einen neuen Grenzfall der Bessel'schen Functionen. Programm (96). Andreas Realgymn. Berlin, 1899 (28 pp.). [Jahrbuch Uber die Fortschritte der Math. 1899, pp. 420 421.]
E. E. Sur l'integration de ^equation de Riccati par des integrales definies. Journal fiir Math. XII. (1834), pp. 144147. Ueber die hypergeometrische Reihe
1 1
KUMMER,
+ l.y J
|
a.g
.
|
a(" + l)/8 09 + l)
"*"
1.2.y(y+l)
"
ra
,
^ + -Journal
fiir
De
integralibus
quibusdam
Note sur
l'integration
de l'equation
T-f=^m -y-
Journal
fiir
286288.
Sur
l'iutegration de l'equation
-^=3?" .y.
Journal de Math.
iv. (1839),
pp. 390
391.
Sur
le
204233.
J. L. DE. 1770). Hist, de V Acad. R. des Sci. de Berlin, xxv. [Oeuvres, in. (1869), pp. 113138.]
1,
LAGRANGE,
LAMB,
On On
XIII. (1882),
H.
Proc.
London Math.
Soc.
Soc.
5166.
(May
11, 1882).
Proc.
London Math.
xm.
(1882), pp.
189212.
BIBLIOGRAPHY
On
Royal
Electrical Motions in a Spherical Soc. clxxiv. (1883), pp. 519549;
771
14, 1883).
Conductor (March
the Induction of Electric Currents in Cylindrical and Spherical Conductors (Jan. Proc. London Math. Soc. xv. (1884), pp. 139149. Note on the Induction of Electric Currents in a Cylinder placed across the Lines of Magnetic Force (June 12, 1884). Proc. London Math. Soc. xv. (1884), pp. 270274. On the Motion of a Viscous Fluid contained in a Spherical Vessel (Nov. 13, 1884). Proc. London Math. Soc. xvi. (1885), pp. 2743. problem in Resonance illustrative of the Theory of Selective Absorption of Light (Jan. 8, 1900). Proc. London Math. Soc. xxxn. (1901), pp. 1120. Problems relating to the Impact of Waves on a Spherical Obstacle in an Elastic Medium (March, 8, 1900). Proc. London Math. Soc. xxxn. (1901), pp. 120150. On Boussinesq's Problem (Feb. 13, 1902). Proc. London Math. Soc. xxxrv. (1902), pp.
10, 1884).
On
276284.
Deep-water Waves (Nov. 10, 1904). Proc. London Math. Soc. (2) n. (1905), pp. 371 [Presidential Address.] On the theory of waves propagated vertically in the atmosphere (Dec. 6, 1908). Proc. London Math. Soc. (2) vii. (1909), pp. 122141.
On
400.
See Wangerin,
LAMBERT,
A.
LANDAU, E. G. H. die Gitterpunkte in einem Kreise (May 8, 1915 ; June 5, 19i5 ; June 4, 1920). Nachrichten von der K. Get. der Wis*, zu Obttingen, 1915, pp. 148160, 161 171 ; 1920, pp. 109134. Zur analytischen Zahlentheorie der definiten quadratischen Formen (Uber die Gitterpunkte in einem mehrdimensionalen Ellipsoid) (May 20, 1915). Berliner Sitzungtberichte {Math.-phys. Klatse), xxxi. (1915), pp. 458 476. Uber Dirichlets Teilerproblem (July 3, 1915). Miinchener Sitzungtberichte, 1915, pp.
Uber
''.#*
DE.
le refroidissement
317328.
LAPLACE,
Memoire sur
la diminution
P. S.
de
la
de
la terre.
Conn, des Tern*, 1823 [published 1820], pp. 245257. Traitdde Mecanique Celeste, v. (Paris, 1825 and 1882).
(3)
I.
(1875), pp.
373398.
LEBEDEFF,
WERA MYLLER.
in einer parabolischen Randwertaufgabe
Uber die Anwendung der Integralgleichungen Math. Ann. lxvi. (1909), pp. 325330.
LEBESGUE,
Remarques sur
l'equation
V.
A
xi. (1846), pp.
338340.
LEFORT,
F.
Expression numerique des integrates d^finies qui se presentent quand on cherche les termes generaux du developpement des coordonnees d'une planete, dans son mouvement elliptique. Journal de Math. xi. (1846), pp. 142152.
LEIBNIZ,
See Daniel Bernoulli.
G.
W.
LE PAIGE, C*
Note sur l'equation xy"+iy'-xy^0. pp. 10111016.
Bull, de VAcad. R. de Belgique, (2) xli. (1876),
LERCH, M.
Mittheilungen aus der Integralrechnung. Monatshefte fttr Math,
pp. 105112.
und
Phys.
I.
(1890),
Betrachtungen uber einige Fragen der Integralrechnung. Rozpravy, 'Uber Fortschritte der Math. 1896, pp. 233234.]
* See also under Catalan.
v. (1896), no.
23
772
Die Beziehungen der begrenzten Ableitungen mit komplexen Zeiger zu den Besselschen Funktionen und ihren Nullstellen (Nov. 29, 1911). Sitz. der Berliner Math. Oes. xi. (1911),
pp.
35.
LINDSTEDT, A.
Zur Theorie der Fresnel'schen Integrate (Aug. xvn. (1882), pp. 720725.
1882).
J.
-
(3)
LIOUVILLE,
Sur
la classification
des transcendantes et sur rimpossibilite d'exprimer des racines de certaincs Equations en fonctions finies explicites des coefficients (June 8r 1835). Journal de Math. n. (1837), pp. 56105; in. (1838), pp. 523547. Sur l'integration d'une classe d'^quations diflerentielles du second ordre en quantity finies explicites (Oct. 28, 1839). Comptes Rendus, ix. (1839), pp. 527530; Journal de Math. iv. (1839), pp. 423456. Remarques nouvelles sur l'equation de Riccati (Nov. 9, 1840). Comptes Rendus, xi. (1840), p. 729; Journal de Math. vi. (1841), pp. 113.
Sur
l'integrale
J
cos
de Math.
6973.
LIPSCHITZ,
Ueber ein Integral der Differentialgleichung
fiir
R. O. S.
^+
R
~-+/=0
LOBATTO,
Sur l'integration des equations
(April, 1837).
LODGE,
pp. 494
A.
Note on the Semiconvergent Series of Jn (x). British Association Report, York, 1906,
498.
LOMMEL,
Beugung dea
E. C.
J.
VON.
Archiv der Math, und Phys. xxxvi.
Lichts.
385419. Methode zur Berechnung einer Transcendenten. Archiv der Math und Phys. xxxvn. (1861), pp. 349360. Zur Integration linearer Differentialgleichungen. Archiv der Math, und Phys. XL. (1863), pp. 101126.
Studien fiber die BesseVschen Functionen (Leipzig, 1868).
Integration der Gleichung *" + i
jr^n +y=0.
Math. Ann.
II.
(1870), pp.
624635.
schriftfiir
Ueber der Anwendung der Bessel'schen Functionen in der Theorie der Beuguug. ^ZeitMath, und Phys. xv. (1870), pp. 141169. Zur Theorie der Bessel'schen Functionen (Dec. 1870). Math. Ann. in. (1871), pp. 475
487.
(Jan. 1871).
Math. Ann.
103
Ueber eine mit den Bessel'schen Functionen verwandte Function (Aug. 1875). Math. Ann. ix. (1876), pp. 425 444. Zur Theorie der Bessel'schen Functionen (Oct. 1878). Math. Ann. xiv. (1879), pp. 510
536.
(Oct. 1879).
Die Beugungserscheinungen einer kreisrunden Oeffnung und eines kreisrunden Schirmchens theoretisch und experimentell bearbeitet. Miinchener Abh.xv.(l 884 86), pp. 233 328. Die Beugungserscheinungen geradlinig begrenzter Schirme. Miinchener Abh. xv. (188486), pp. 531663.
The headings
BIBLIOGRAPHY
LORENZ,
Sur
le
773
L.
developpement des fonctions arbitraires au moyen de fonctions donnees. Tidsskriftfor Mathematik, 1876, pp. 129144. [Oeuvres Scientifiques, II. (1899), pp. 495513.] Theorie de la dispersion. Ann. der Phi/sik und Chemie, (3) xx. (1883), pp. 1 21. [Oeuvres Scientifiques, I. (1898), pp. 371396.] Sur la lumiere reflechie et refractee par une sphere transparente. K. Danske Videnska-
62.
[Oeuvres Scientifiques,
I.
LOVE, A. E. H. of Electric Waves by a Dielectric Sphere (Feb. 9, 1899). Proc. London Math. Soc. xxx. (1899), pp. 308 321. The Transmission of Electric Waves over the Surface of the Earth (Dec. 19, 1914). Phil. Trans, of the Royal Soc. ccxv. 131. (1915), pp. 105
The Scattering
MACDONALD,
pp.
H. M.*
Proc.
11,
1897).
;
London Math.
Zeroes of the Bessel Functions (April 7, 1898 Jan. 12, 1899). Proc. London Math. Soc. xxix. (1898), pp. 575584; xxx. (1899), pp 165179. The Addition Theorem for the Bessel Functions (April 5, 1900). Proc. London Math. Soc. xxxn. (1901), pp. 152157. Some Applications of Fourier's Theorem (Dec. 11, 1902). Proc. London Math. Soc. xxxv. (1903), pp. 428443. The Bending of Electric Waves round a Conducting Obstacle (Jan. 21, 1903 ; May 12, 1903). Proc. Royal Soc. lxxi. (1903), pp. 251258 lxxii. (1904), pp. 5968. Note on the evaluation of a certain integral containing Bessel's functions (Dec. 6, 1908). Proc. London Math. Soc. (2) vn. (1909), pp. 142149. The Diffraction of Electric Waves round a Perfectly Reflecting Obstacle (Aug. 13, 1909). Phil. Trans, of the Royal Soc. ccx. (1910), pp. 113144. The Transmission of Electric Waves around the Earth's Surface (Jan. 10, 1914). Proc. Royal Soc. xc. (1914), pp. 5061. Formulae for the Spherical Harmonic n ~ m (/*) when 1 - ft is a small quantity (Feb. 6, 1914). Proc. London Math. Soc. (2) xm. (1914), pp. 220221. [Extract from a letter to Prof. Carslaw, Oct. 17, 1912.] Proc. London Math. Soc. (2) xm.
;
problems (Jan.
14, 1915).
Proc.
London Math.
pp.
410427.
The Transmission
Royal Soc. xcn.
of Electric
Earth's Surface
(May
Proc.
McMAHON,
On
On
pp. 23
(1894), pp.
J.
5761.
the Roots of the Bessel and certain Related Functions. Annals of Math,
30.
The Expression for a Rational Polynomial in a Series of Bessel functions of nth Order as required in certain Cases of Dirichlet's Problem. Proc. American Assoc. 1900, pp. 42
43.
MacROBERT,
The Modified
(Jan. 14, 1921).
Bessel Function
Kn
(z)
Asymptotic Expressions for the Bessel Functions and the Fourier-Bessel Expansion Proc. Edinburgh Math. Soc. xxxix. (1921), pp. 1320.
MAGGI,
Sulla storia delle funzioni cilindriche.
vol
Rend, del R.
1st.
(1880), pp.
384390.
MALMSTEN,
cos ax dx /* -jz -^
C. J.
(1841), pp. 65
K' Svenska
V.
74.
774
De
18, 1849).
108115.
Thebremes sur
Math. Journal,
l'integration
de l'Equation -r\
+ ~j
C.
='(
^m +Zi)
V-
v. (1850), pp.
180182.
MALTEZOS,
Sur la chute des corps dans le vide et sur certaines fonctions transcendantes. Nouvelles Annates de Math. (4) xi. (1902), pp. 197204.
MANFREDIUS,
De
construetione
G.
MARCH,
H. W.
Uber die Ausbreitung der Wellen der drahtlosen Telegraphic auf der Erdkugel 50. 1911). Ann. der Physilc und Chemie, (4) xxxvn. (1912), pp. 29
(Oct. 21,
MARCOLONGO,
diconto, (2)
R.
6,
1889).
Napoli Ren-
m.
(1889), pp.
9699.
MARSHALL, W.
of Computing the Roots of Bessel's Functions (Oct. 1909). Annals of Math. (2) xi. (1910), pp. 153160.
On a New Method
MATHEWS,
See under Gray.
G. B.
On
ALL, R. H. D. the Diffraction Pattern near the Focus of a Telescope (Feb. 22, 1897). Proc. Camb.
259269.
MAY
MEECH,
Integration of Riccati'a Equation.
pp.
L.
W.
I.
Annals of Math.
F. G.
(1886), pp.
97103
in. (1887),
4749.
MEHLER,
Ueber die Vertheilung der statischen Electricitat in einem von zwei kugelkalotten begrenzten Korper (July, 1867). Journal fur Math, lxviii. (1868), pp. 134 150. Ueber eine mit den Kugel- und Cyhnderfunctionen verwandte Function und ihre Anwendung in die Theorie der Electricitatsvertheilung. '[EUnng Jahresbericht, 1870.] Math. Ann. xviii. (1881), pp. 161194. Ueber die Darstellung einer willkurlichen Function zweier Variablen durch Cylinderfunction* (Dec. 1, 1871). Math. Ann. v. (1872), pp. 135140.
und
Notiz uber die Dirichlet'schen Integralausdrucke fur die Kugelfunctionen P" (cos S) iiber eine analoge Integralform fur die Cylinderfunction J(x)^ (Dec. 2, 1871). Math.
v. (1872),
Ann.
pp. 141144.
Iserlohn
Programm, 1862.
II.
(1902), pp.
396
Tafel der Bessel'schen Functionen 7 und i"*1 von -0 bis =15-5 (Nov. 8, 1888). Berliner Abh. 1888. (Math. Abh. I.) Ueber die Besselschen Functionen Jk und kl Programm, Oberrealschule, Kiel, 1890. [Jahrbuch Uber die Fortschritte der Math. 1890, pp. 521 522.] Einige Entwickelungen die Bessel'schen /-Functionen betreffend (May 3, 1891). Astr.
410J
Nach. cxxvu. (1891), col. 359362. Beitrag zur Theorie der allgemeinen Bessel'schen Function (June 11, 1891). Astr. Nach. CXXVIH. (1891), col. 145154. Abgekurzte Tafel der Bessel'schen Functionen Ikh (July 10, 1891). Astr. Nach. cxxvin.
(1891), col. (1891), col.
154155.
7,
1891).
Astr. Nach.
cxxvm.
435438.
* The headings of the pages of this paper are "Ueber die Cylinderfunction J(x)." t The headings of the pages of this paper are "Notiz iiber die Functionen P" (cos a) und
J (x)."
BIBLIOGRAPHY
Neue Entwickelungen uber
cxxix. (1892),
col.
775
Astr.
Nach.
281284. Weitere Entwickelungen uber die Bessel'schen Functionen (May 2, 1892). Astr. Nach. cxxx. (1892), col. 363368. Uber die Absoluten Maxima der Bessel'schen Functionen. Programm, Oberrealschule,
Kiel, 1892 (11 pp.).
[Jakrbuch
478.]
MELLIN, R. HJ. Abriss einer einheitlichen Theorie der Gamma und der hypergeometrischen Funktionen (June, 1909). Math. Ann. lxviil (1910), pp. 305337.
MICHELL,
The Wave Resistance
of a Ship* (Aug.
9,
J.
H.
Phil.
1897).
Mag.
(5)
MOLINS,
Sur
l'integration
H.
6, 1876).
de l'equation
differentielle
Mem. de VAcad.
vm.
(1876), pp.
167189.
MOORE,
Note on the roots of Bessel functions. Annals of Math. (2) IX. (1908), pp. 156 162. The summability of the developments in Bessel functions with applications (Sept. 1908). Trans. American Math. Soc. x. (1909), pp. 391 135.
On the uniform convergence of the in Bessel functions (Oct. 30, 1909). Trans. American Math. Soc. xn. (1911), pp. 181 206. continuous function whose development in Bessel's functions is non-summable of certain orders (Sept. 4, 1916). Bulletin American Math. Soc. xxiv. (1918), pp. 145 149. On the summability of the developments in Bessel's functions (Sept. 4, 1917). Trans. American Math. Soc. xxi. (1920), pp. 107 156.
developments
11,
MORTON,
The Value
W.
B.
of the Cylinder Function of the Second 1900). Nature, lxiii. (1901), p. 29.
Kind
for
MURPHY,
On
ill.
R.
(May
24, 1830).
MYLLER-LEBEDEFF, W.
See under Lebedeff.
NAGAOKA,
Diffraction
Coll.
H.
Surface. Journal of the
322.
301
nach Quadraten und Producten der FourierBessel'8chen Functionen (Nov. 1869.) Leipziger Berichte, xxi. (1869), pp. 221 256. [Math. Ann. in. (1871), pp. 581610.] Ueber Producte und Quadrate der Bessel'schen Functionen. Math. Ann. n. (1870), p. 192. Uber die nach Kreis-, Kugel- rind Cylinder-functionen fortschreitenden Entwickelungen
(Leipzig, 1881).
Ueber gewisse particulare Integrate der Differentialgleichung insbesondere uber die Entwickelung dieser particularen Integrale nach Kugelfunctionen (March 8, 1886). Leipziger Berichte, xxxviii. (1886), pp. 75 82.
AF= F
* This paper contains some Tables of Bessel functions t See also under Schlafli.
776
F. E. Eiiie Verallgemeinerung der Zylinderfunktionen. Dissertation, Ealberstadt, [Jahrbuch iiber die Fortschritte der Math, 1909, p. 575.]
NICHOLSON,
The Asymptotic Expansions
(1907), pp.
J.
W*
High Order.
Phil.
Phil.
(6)
of Bessel Functions of
Mag.
(6) xiv.
697707.
Equal Argument and Order.
Parallel
On
pp. pp.
Bessel Functions of
Mag.
xvi. (1908),
271279.
On On
pp.
the Inductance of
Two
Wires (June
12, 1908).
PhU. Mag.
255275.
the Relation of Airy's Integral to the Bessel Functions. PhU. Mag. (6) xvm. (1909)* 617. The Asymptotic Expansions of Bessel Functions. Phil. Mag. (6) xix. (1910), pp. 228 249. On the Bending of Electric Waves round a Large Sphere, I. PhU. Mag. (6) xix. (1910), pp. 516537. The Approximate Calculation of Bessel Functions of Imaginary Argument. PhU. Mag. (6) xx. (1910), pp. 938943. The Scattering of Light by a Large Conducting Sphere (March 10, 1910). Proc. London Math. Soc. (2) ix. (1911), pp. 6780. Notes on Bessel Functions. Quarterly Journal, xlii. (1911), pp. 216224. The products of Bessel functions. Quarterly Journal, xwn. (1912), pp. 78100. The pressure of radiation on a cylindrical obstacle (Dec. 14, 1911). Proc. London Math. Soc. (2) XI. (1913), pp. 104126. The Lateral Vibrations of Bars of Variable Section (May 11, 1917). Proc. Royal Soc. xciii. A (1917), pp. 506519. 329. Generalisation of a theorem due to Sonine. Quarterly Journal, xlviii. (1920),pp. 321 A Problem in the Theory of Heat Conduction (Sept. 29, 1921). Proc. Royal Soc. c. A (1922), pp. 226240.
NICOLAS,
norm. sup.
(2) xi. (1882),
J.
espece).
Ann.
sci.
de VEcole
3-90.
30, 1898).
NIELSEN, N.t Sur le produit de deux fonctions cylindriques (July pp. 228242.
Udviklinger efter Cylinderfunktioner (Aug. 28, 1898): Nyt Tidsskrift, ix. B (1898), pp. 7383. Sur le developpenient de zero en fonctions cylindriques (March 25, 1899). Math. Ann. Lll. (1899), pp. 582587. Flertydige Udviklinger efter Cylinderfunktioner. Nyt Tidsskrift, x. B (1899), pp. 73 81. Note sur les developpements schloemilchiens en sene de fonctions cylindriques (Nov. 15, 1899). Oversigt K. Danske Videnskabernes Selskabs, 1899, pp. 661665. Note supplemental relative aux developpements schloemilchiens en sene de fonctions cylindriques (June 23, 1900). Oversigt K. Danske Videnskabernes Selskabs, 1900, pp. 55 60. Sur une classe de polynfimes qui se presentent dans la theorie des fonctions cylindriques (Feb. 26, 1900; Jan. 8, 1901). Ann. diMat. (3) v. (1901), pp. 1731 (3) vi. (1901), pp. 331340. Evaluation nouvelle des integrales ind^finies et des series infinies contenant une fonction cylindrique (May 23, 1900). Ann. di Mat. (3) vi. (1901), pp. 43115. Sur une classe de series infinies analogues a celles de Schlomilch selon les fonctions cylindriques (Aug. 17, 1900). Ann. di Mat. (3) vi. (1901), pp. 301329.
des fonctions cylindriques dues a MM. C. Neumann et de VEcole norm. sup. (3) xvm. (1901), pp. 3975. Note sur la convergence d'une sene neumannienne de fonctions cylindriques (Feb. 10, 1901). Math. Ann. lv. (1902), pp. 493496. Recherches sur une classe de series infinies analogue a celle de M. W. Kapteyn 146. (April 13, 1901). Oversigt K. Danske Videnskabernes Selskabs, 1901, pp. 127 Sur les series de factorielles (Jan. 20, 1902). Comptes 7fenrfiw,cxxxiv.(1902),pp.l57 160. Theorie nouvelle des series asymptotiques obtenues pour les fonctions cylindriques et pour les fonctions analogues (March 16, 1902). Oversigt K. Danske Videnskabernes Selskabs,
Recherche sur
les
series,
W. Kapteyn. Ann.
sci.
BIBLIOGRAPHY
driquoa Nouvelles Annates de Math.
777
liquations differentieltes lineaires obtenues pour le produit de deux fonctions cylin(4) n. (1902), pp. 396410. Handbuch der Theorie der Cylinderfunktionen (Leipzig, 1904). Note sur lea senes de fonctions bernoulliennes (Dec. 14, 1903). Math. Ann. lix. (1904),
pp. 103109. Sur une integrate definie (June 5, 1904). Math. Ann. lix. (1904), pp. 89102. Sur quelques proprtetes nouvelles des fonctions cyiindriques (April 1, 1906). Atti della R. Accad. dei Lincei, (5) xv. (1906), pp. 490497. Recherches sur quelques generalisations d'une identity integrate d'Abel (Sept. 17, 1906). K. Danslce Videnskabernes Selslcabs Skrifter, (7) v. (1910), pp. 137. Sur les series des fonctions cyiindriques. Journ. fiir Math, cxxxii. (1907), pp. 127 146. Sur quelques proprietes fondamentates des fonctions spheriques (Dec. 20, 1906). Ann. di Mat. (3) xiv. (1908), pp. 6990. ftber dieVerallgemeinerung einiger von F.und C.Neumann gegebenen nach Kugel- und Zylinderfunktionen fortschreitenden Reihenentwickelungen (Jan. 26, 1908). Leipziger
3361.
tJber das Produkt zweier Zylinderfunktionen. Monatshefte fiir Math, und Phys. xix. (1908), pp. 164170. tJber den Legendre-Besselschen Kettenbruch (July 4, 1908). Miinchener Sitzungsbenchte, xxxvm. (1908), pp. 8588.
NIEMOLLER,
F.
Ueber Schwingungen einer Saite deren Spannung eine stetige Function der Zeit ist. Zeitschrift fUr Math. xxv. (1880), pp. 4448. Fonneln zur numerischen Berechnung des allgemeinen Integrals der Bessel'schen Differentialgleichungen. Zeitschrift fur Math. xxv. (1880), pp. 6571.
OLBRICHT, R. Studien iiber die Kugel- und Cylinderfunctionen (June Caes. Leop. {Halle), 1888, pp. 148.
2,
1886).
OLTRAMARE,
Note sur
l'integrale
G.
""
/:
^y^d*.
I.
p.
182
(6)
ONO, A. On the First Root of Bessel Functions of Fractional Order (April 2, 1921). Phil. Mag. xml (1921), pp. 10201021. ORR, W. McF. On Divergent (or Semiconvergent) Hypergeometric Series (May 16, 1898; April 3, On
Trans. Camb. Phil. Soc. xvn. (1899), pp. 171199, 283290. the product Jm (x)Jn (x) (May 15, 1899). Proc. Camb. Phil. Soc. x. (1900), pp.
14, 1908). Proc. R. Irish
1899).
93100.
Extensions of Fourier's and the Bessel-Fourier Theorems (Dec. Acad, xxvii. A (1910), pp. 205248.
Arlciv for
OTTI, H.
Eigenschaften Bessel'scher Funktionen
[1899], pp.
6196.
PAOLI,
Sull' integrazione dell'
Sopra gl' integrali definiti (Oct. 8, 1827). Mem. di Mat. Sci. (Modena), xx. (1828), pp. 161182.
equazione
Mem. di Mat.
183
778
les series et sur l'integration complete d'une Equation aux differences partielles lineaires du second ordre a coefficieiis constans (Le 16 germinal, an 7 April 5, 1799). Mem. presented a I'Inst. par divers savans, I. (1805), pp. 639 648.
Memoire sur
PEARSON,
On
(1880), pp.
K.
127131.
of the Random Walk. Nature, lxxii. (1905), pp. 294, 342. mathematical theory of random migration. Drapers Company Research Memoirs, Biometric Series, m. (1906).
The Problem
PEIRCE,
See under Willson.
B. 0.
J.
9,
PERES,
Sur
les fonctions
1915).
Comptus Rendus,
PERRON,
(Dec.
7,
0.
504.
PETZVAL,
J.
I.
(Vienna, 1851).
PICARD, C. E. Application de la theorie des complexes lineaires a l'etude des surfaces et des courbes gauches. Ann. sci. de VEcole norm. sup. (2) vi. (1877), pp. 329 366.
PINCHERLE,
S.
Sopra alcuni sviluppi in serie per funzioni analitiche (Jan. 12, 1882). Bologna Memorie, (4) in. (1881), pp. 151180. Alcuni teoremi sopra gli sviluppi in serie per funzioni analitiche (March 23, 1882). Rend, del R. 1st. Lombardo, (2) xv. (1882), pp. 224225. Delia trasformazione di Laplace e di alcune sue applicazioni (Feb. 27, 1887). Bologna Memorie, (4) vm. (1887), pp. 125143.
20, 1822).
Mem.
della R.
Accad
delle Sci. di Torino, xxvi. (1821), pp. 519538. Recherches analytiques sur la decouverte de la loi de pesanteur des planetes vers le Soleil et sur la theorie de leur mouvemcnt elliptique (June 20, 1847). Mem. della R. Accad. delle Sci. di Torino, (2) x. (1849), pp. 249332.
POCHHAMMER,
L.
Ueber die lineare Differentialgleichung zweiter Ordnung mit linearen Coefficienten. Math. Ann. xxxvi. (1890), pp. 84 96. Ueber einige besondere Falle der linearen Differentialgleichung zweiter Ordnung mit linearen Coefficienten (Sept. 1890). Math. Ann. xxxvin. (1891), pp. 225 246. Ueber eine binomische lineare Diflerentialgleichung nter Ordnung (Sept. 1890). Math. Ann. xxxvm. (1891), pp. 247262. Ueber die Differentialgleichung der allgemeineren .F-Reihe (Jan. 1891). Math. Ann. xxxvm. (1891), pp. 586-597. Ueber eine specielle lineare Differentialgleichung 2 ter Ordnung mit linearen Coefficienten (May, 1891). Math. Ann. xli. (1893), pp. 174178. Ueber die Differentialgleichungen aer Reihen ${p, <r x) und $(p, a, r x) (June, 1891). Math. Ann. xli. (1893), pp. 197218.
POISSON,
Sur une nouvelle maniere d'exprimer
S.
D.
ix. (cahier 16)
Memoire sur les integrates denies. Journal de VEcole Polytechnique, (1813), pp. 215246.
elliptique.
les coordonnees des planetes dans le mouvement Connaissance des Terns, 1825 [1822], pp. 379 385. Memoire sur l'Integration des equations lineaires aux differences partielles. Journal de VEcole Polytechnique, Xli. (cahier 19) (1823), pp. 215248.
BIBLIOGRAPHY
les
779
Suite du Memoire sur lea integrates denies et sur la sommation des series insere" dans precedens volumes de ce Journal. Journal de VHcole Polytechnique, xii. (cahier 19)
404509. Memoire sur le calcul des variations (Nov. 10, 1831). Mem. de VAcad. R. des Sci. xn. (1833), pp. 223331. Sur le developpement des coordonnees d'une planete dans son mouvement elliptique et la fonction perturbatrice de ce mouvement. Connaissance des Terns, 1836 [1833], pp. 3-31.
(1823), pp.
La
PORTER, M. B. Note on the roots of Bessel's Functions. Bulletin American Math. Soc. iv. (1898), pp. 274275. On the Roots of the Hypergeometric and Bessel's Functions (June 3, 1897). American Journal of Math. xx. (1898), pp. 193214. On the roots of functions connected by a linear recurrent relation of the second order (Feb. 1901). Annals of Math. (2) m. (1902), pp. 5570.
PUISEUX,
Sur
V.
la convergence des series qui se presentent dans la theorie du mouvement elliptique des planetes. Journal de Math. xiv. (1849), pp. 33 39.* Seconde note sur la convergence des series du mouvement elliptique. Journal de Math.
242246.
PURSER,
On
F.
the application of Bessel's functions to the elastic equilibrium of a homogeneous isotropic cylinder (Nov. 11, 1901). Trans. R. Irish Acad. xxxn. (1902), pp. 31 60. Some applications of Bessel's functions to Physics (May 14, 1906). Proc. R. Irish Acad. xxvi. (1907), pp. 2566.
RAFFY,
Une
lecon sur l'equation de Riccati.
L.
(4) n. (1902), pp.
529
545.
RAMANUJAN,
RAWSON,
R. On Cognate Riccatian Equations (1876). Messenger, vn. (1878), pp. 69 72. Note on a transformation of Riccati 's equation. Messenger, xii. (1883), pp. 34
310.
36.
RAYLEIGH
On
(J.
W. STRUTT), LORD.
the Vibrations of a Gas contained within a Rigid Spherical Envelope (March 14, 1872). Proc. London Math. Soc. iv. (1873), pp. 93103. Investigation of the Disturbance produced by a Spherical Obstacle on the Waves of Sound (Nov. 14, 1872). Proc. London Math. Soc. iv. (1873), pp. 253 283. Notes on Bessel's Functions. Phil. Mag. (4) xliv. (1872), pp. 328344. Note on the Numerical Calculation of the Roots of Fluctuating Functions (June 11, 1874). Proc. London Math. Soc. v. (1874), pp. 112194. [Scientific Papers, I. (1899),
pp. 190195.]
The Theory of Sound (2 vols. London, 1877, 1878; 2nd edition, 1894, 1896). the Relation between the Functions of Laplace and Bessel (Jan. 10, 1878). Proc. London Math. Soc. IX. (1878), pp. 6164. [Scientific Papers, I. (1899), pp. 338 341.] On Images formed without Reflection or Refraction. Phil. Mag. (5) xi. (1881), pp. 214
On
218.
[Scientific Papers,
I.
(1899), pp.
513
517.]
the Electromagnetic Theory of Light. Phil. Mag. (5) xii. (1881), pp. 81101. [Scientific Papers, i. (1899), pp. 518536.] On the Vibrations of a Cylindrical Vessel containing Liquid. Phil. Mag. (5) xv. (1883), 389. [Scientific Papers, n. (1900), pp. 208211.] pp. 385 On Point-, Line-, and Plane Sources of Sound (June 14, 1888). Proc. London Math. Soc. xix. (1889), pp. 504507. [Scientific Papers, in. (1902), pp. 4446.1 On the Theory of Optical Images, with special reference to the Microscope. Phil. Mag. 260.] (5) xlii. (1896), pp. 167195. [Scientific Papers, iv. (1904), pp. 235 On the Passage of Waves through Apertures in Plane Screens, and Allied Problems. Phil. Mag. (5) sun, (1897), pp. 259272. [Scientific Papers, iv. (1904), pp. 283296.] On the Bending of Waves round a Spherical Obstacle (May 23, 1903). Proc. Roy. Soc. lxxii. (1903), pp. 4041. [Scientific Papers, v. (1912), pp. 112114.] On the Acoustic Shadow of a Sphere (Jan. 21, 1904). Phil. Trans, of the Royal Soc. ccra. A (1904), pp. 87110. [Scientific Papers, v. (1912), pp. 149161.]
On
780
the Open Organ Pipe Problem in two Dimensions. Phil, Mag. (6) vm. (1904), [Scientific Papers, v. (1912), pp. 206211.] On the Experimental Determination of the Ratio of the Electrical Units. Phil. Mag. (6) XII. (1906), pp. 97108. [Scientific Papers, v. (1912), pp. 330340.] On the Light dispersed from Fine Lines ruled upon Reflecting Surfaces. Phil. Mag. (6) xiv. (1907), pp. 350369. [Scientific Papers, v. (1912), pp. 410418.] The Problem of the Whispering Gallery. Phil. Mag. (6) xx. (1910), pp. 10011004. [Scientific Papers, v. (1912), pp. 617620.] Note on Bessel's Functions as applied to the Vibrations of a Circular Membrane. Phil. Mag. (6) xxi. (1911), pp. 5358. [Scientific Papers, VI. (1920), pp. 15.] On a Physical Interpretation of Schlomilch's Theorem in Bessel's Functions. Phil. Mag. (6) xxi. (1911), pp. 567571. [Scientific Papers, vi. (1920), pp. 2225.] Problems in the Conduction of Heat. Phil. Mag. (6) xxn. (1911), pp. 381 396. [Scientific Papers, vi. (1920), pp. 5164.] On the propagation of Waves through a stratified Medium (Jan. 11, 1912). Proc. Royal Soc. lxxxvi. A (1912), pp. 207226. [Scientific Papers, vi. (1920), pp. 111120.] Electrical Vibrations on a thin Anchor Ring (June 27, 1912). Proc. Royal Soc. lxxxvii. (1912), pp. 193202. [Scientific Papers, vi. (1920), pp. 111120.] Further Applications of Bessel's Functions of High Order to the Whispering Gallery and Allied Problems. Phil. Mag. (6) xxvu. (1914), pp. 100109. [Scientific Papers, vi.
pp. 481487.
On
(1920), pp.
211219.]
Further remarks on the Stability of Viscous Fluid Motion. Phil. Mag. (6) xxvin. (1914), pp. 609619. [Scientific Papers, vi. (1920), pp. 266275.] On the Stability of the simple Shearing Motion of a Viscous Incompressible Fluid. Phil. Mag. (6) xxx. (1915), pp. 329338. [Scientific Papers, vi. (1920), pp. 341349.] On Legendre's Function* Pn (0), when n is great and 6 has any value (April 27, 1916). Proc. Royal Soc. xcn. A (1916), pp. 433 437. [Scientific Papers, vi. (1920), pp. 393397.] On Convection Currents in a horizontal Layer of Fluid, when the higher Temperature is on the under side. Phil. Mag. (6) xxxn. (1916), pp. 529 546. [Scientific Papers, vi. * (1920), pp. 432446.]
On the Problem of Random Vibrations and of Random Flights in one, two, or three dimensions. Phil. Mag. (6) xxxvu. (1919), pp. 321 347. [Scientific Papers, vi. (1920), pp. 604626.] REINECK, A. Die Verwandtschaft zwischen Kugelfunktionen und Besselschen Funktionen. Dissertation, Bern (Halle, 1907). (72 pp.) [Jahrbuch iiber die Fortschritte der Math. 1907, p. 495.]
RICCATI,
COUNT
J. F.
Animadversationes in aequationes differentiales secundi gradus. Actorum Eruditorum quae Lipsiae publicantur Supplementa, vm. (1724), pp. 66 73. Com. Jacobi Riccati Appendix ad Animadversationes in aequationes differentiales secundi gradus, editus in Actis Eruditorum quae Lipsiae publicantur Tomo vm. Supplementorum, Sectione II. p. 66. Acta Eruditorum publicata Lipsiae, 1723, pp. 502 510.
RIEMANN,
G. F. B.
Zur Theorie der Nobili'schen Farbenringe (March 28, 1855). Ann. der Physik und Chemie, (2) xcy. (1855), pp. 130139. Partielle Diferentialgleichungen und deren Anwendung auf physikalische Fragen. Von Bernhard Riemann. FiiY den Druck bearbeitet und herausgegebenen von Karl Hattendorf (Brunswick, 1876).
ROHRS,
Soc. v. (1874), pp.
J.
H.
(May
14, 1874).
Proc.
London Math.
125139.
RUDSKI,
P.
ill.
Ueber eine Klasse transcendenter Gleichungen. Prace Mat. Fit. [Jahrbuch iiber die Fortschritte der Math. 1892, pp. 107 108.] Note sur la situation des racines des fonctions transcendantes 1891). Mem. de la Soc. R. des Set. de Liege, (2) xvm. (1895), no. 3.
(1892), pp.
69
81.
10,
JH +t(#)=0
(May
RUSSELL,
The
Effective Resistance
A.
and Inductance of a Concentric Main, and Methods of Computing the Ber and Bei and Allied Functions (Jan. 22, 1909). Phil. Mag. (6) xvn. (1909), pp. 524552.
* This is the function
(cos 0).
BIBLIOGRAPHY
RUTGERS,
Over die bepaalde integraal
/
781
J.
G.
(2)
e-**z- dz.
vl
(1905),
pp. 368373. Over eene reeks met Besselsche Functies. Nieuw Archie/ voor Wiskunde, (2) VII. (1907), pp. 8890. Over reeksen van Besselsche functies en daarmede samenhangende bepaalde integraleu, waarin Besselsche functies voorkomen. Nieuw Archie/ voor Wiskunde, (2) vn. (1907), pp. 164181. Sur les functions cylindriques de premiere espece. Nieuw Archie/ voor Wiskunde, (2) vii. (1907), pp. 385405. Over eenige toepassingen der Fourier'sche ontwikkeling van een willekeurige functie naar Bessel'sche functies. Nieuw Archie/ voor Wiskunde, vm. (1909), pp. 375380.
RYBCZYNSKI, W. VON.
die Ausbreitung der Wellen der drahtlosen Telegraphie auf der Erdkugel (March 5, 1913). Ann. der Physik und Chemie, (4) XLI. (1913), pp. 191208.
Uber
On
pp. 45
Math. Journal,
v. (1914),
47.
SCHAFHEITLIN,
SAVIDGE, H. G. Tables of the ber and bei and ker and kei Functions with further formulae for their Computation (Nov. 12, 1909). Phil. Mag. (6) xix. (1910), pp. 4958.
P.*
die Darstellung der hypergeometrische Reihe durch ein bestimmtes Integral (Feb. 1887). Math. Ann. xxx. (1887), pp. 157178. Ueber eine Integraldarstellung der hypergeometrischen Reihen (Jan. 1888). Math. Ann. xxxi. (1888), p. 156. Ueber die Gattss-sche und Bessel-aehe Differentialgleichuug und eine neue Integralform der letzteren (Feb. 16, 1894). Journal /ur Math. cxiv. (1895), pp. 3144.
Ueber
Die Nullstellen der Besselschen Functionen. Journal /Ur Math. cxxn. (1900), pp. 299
321.
Uber
der Math,
III.
die Nullstellen der Besselschen Funktionen zweiter Art (Jan. 10, 1901). Archiv und Phys. (3) i. (1901), pp. 133137. Uber den Verlauf der Besselschen Funktionen (May 18, 1904). Berliner Sitzungsberichte,
(1904), pp.
8385.
Die Lage der Nullstellen der Besselschen Funktionen zweiter Art (June 27, 1906). Berliner Sitzungsberichte, v. (1906), pp. 82 93. Uber den Verlauf der Besselschen Funktionen zweiter Art. Jahresbericht der Deutschen Math. Vereinigung, xvi. (1907), pp. 272 279. Die Theorie der Besselschen Funktionen. Von Paul Schafheitlin (Leipzig und Berlin, 1908; Math. Phys. Schri/ten/ilr Ingenieure und Studierende, 4). Beziehungen zwischen dem Integrallogarithmus und den Besselschen Funktionen (Feb.
Berliner Sitzungsberichte, vm. (1909), pp. 62 67. Die semikonvergenten Reihen fur die Besselschen Funktionen (Sept. 129. bericht der Deutschen Math. Vereinigung, xix. (1910), pp. 120
24, 1909).
1,
1909).
Jahres-
SCHEIBNEll, W.
the asymptotic values of the coefficients in the development of any power of the radius vector according to the mean anomaly (March, 1856). Goidds Astronomical Journal, IV. (1856), pp. 177182. [Math. Ann. xvn. (1880), pp. 531544.] Ueber die asymptotische Werthe der Coefficienten in den nach der mittleren Anomalie vorgenommeuen Entwickelungen (May 31, 1856). Leipziger Berichte, vm. (1856), pp. 40 64. [Math. Ann. xvn. (1880), pp. 545 560.
On
SCHERK,
Uber
pp.
H. F.
Journal /ur Math.
x.
(1833),
92 97.
*
782
SCHLAFLI, L. Sulle relazioni tra diversi integrali definiti che giovano ad esprimere la soluzione generale della equazione di Riccati. Ann. di Mat. (2) I. (1868), pp. 232 242.
Einige Bemerkungen zu Herrn Neumann's Untersuchungen iiber die Bessel'schen Functionen (May 4, 1870). Math, Ann. in. (1871), pp. 134149. [With note by C. N.
p. 149.]
Sopra un teorema di Jacobi recato a forma piu generale ed applicata alia funzione cilindrica (Aug. 1871). Ann. di Mat. (2) v. (1873), pp. 199205. SulP uso delle linee lungo le quali il valore assoluto di una funzione e constante (Oct. 4, 1872). Ann.di Mat. (2) VI. (1875), pp. 120.
Ueber die Convergenz der Entwicklung einer arbitraren Function f(x) nach den Bessel'schen Functionen Ja (^x), J* (fax), Ja (j8j.tr), ..., wo ft, 32, fts, ... die positiven Wurzeln der Gleichung Ja (ff) vorstellen (Jan. 17, 1876). Math. Ann. x. (1876), pp. 137142.
SCHLOMILCH,
Note sur
0. X.
la variation des constantes arbitraires d'une integrate dennie. Math. xxxm. (1846), pp. 268280. Analytische Studien, n. (Leipzig, 1848).
Journal fur
(1857), pp.
137
SCHONHOLZER,
Bessel'schen Funktionen, n. (Bern, 1900).]
J. J.
Ueber die Auswerthung bestimmte Integrate mit Hiilfe von Veranderungen des Integrationsweges (Dissertation, Bern, 1877). [Graf and Gubler, Einleitung in die Theorie der
SCHWARZSCHILD, K. Die Beugung und Polarisation des Lichts durch einen Spalt. Math. Ann. lv. (1902), pp. 177247.
SCHWERD,
SEARLE,
On
xxxix. (1908), pp. 5166.
J.
F.
M.
H.
C.
Quarterly Journal,
SEGAR, H. W.
On
the roots of certain continuants. Messenger, xxn. (1893), pp. 171
181.
SERRET,
J.
A.
Note sur quelques formulas de calcul integral. Journal de Math. vm. (1843), pp. 1 27. Memoire sur ^integration d'une equation differentielle a l'aide des differentielles a indices quelconques (Sept. 4, 1843). Comptes Bendus, xvn. (1843), pp. 458475; Journal de Math. ix. (1844), pp. 193216.
SHARPE,
On
H.
J.
the Reflection of Sound at the Surface of a Paraboloid (Nov. 1874). Quarterly Journal, xv. (1877), pp. 1 8. On a diflerential equation. Messenger, x. (1881), pp. 174 185 XI. (1882), pp. 41 44. On a transcendental differential equation. Messenger, xi. (1882), pp. 56 63. On a differential equation. Messenger, xill. (1884), pp. 6679. Note on Legendre's coefficients. Quarterly Journal, xxiv. (1890), pp. 383 386. On The Reflection of Sound at a Paraboloid (June 20, 1899). Proc. Camb. Phil. Soc. x.
(1900), pp.
101136.
SHEPPARD, W. F. expressions of a function of a single variable in terms of Bessel's functions. Quarterly Journal, xxin. (1889), pp. 223 260.
On some
SIACCI, F. Sulla integrazione di una equazione differenziale e sulla equazione di Riccati (April 13, 1901). Napoli Rendiconto, (3) vn. (1901), pp. 139143.
Bibliography
sibirani,
Sopra
1*
783
f.
equazione di Riccati. Riv. fis. mat. xix. (1909), pp. 216220. [Jahrbuch iiber die Fortschritte der Math. 1909, p. 369.]
SIEMON,
P.
Ueber die Integrale einer nicht homogenen Differentialgleichung zweiter Ordnung. Programme Luisenschule, Berlin, 1890, 22 pp. [Jahrbuch fiber die Fortschritte der Math. 1890, pp. 340342.] SMITH, B. A * Table of Bessel's Functions F and Ft Messenger, xxvi. (1897), pp. 98101. Arched Dams. Proc. American Soc. of Civil Engineers, xlvi. (1920), pp. 375 425.
.
SMITH, CLARA
E.
theorem of Abel and its application to the development of a function in terms of Bessel's functions (June 8, 1906). Trans. American Math. Soc. vni. (1907), pp. 92106.
SMITH,
Quelques relations integrates entre Mat. (2) xil. (1905), pp. 365373.
O. A.
SOMMERFELD, A. J. W.| Die willkiirlichen Functionen in der mathematischen Physik. Dissertation, Konigsberg, 1891 (75 pp.). [Jahrbuch fiber die Fortschritte der Math. 1891, pp. 519523.] Zur analytischen Theorie der Warmeleitung. Math. Ann. xvv. (1894), pp. 263277. Mathematische Theorie der Diffraction (Summer, 1895): Math. Ann. xlvii. (1896), pp.
317374. Uber die Ausbreitung der Wellen der drahtlosen Telegraphic (Jan. 15, 1909). Ann. der Physik und Chemie, (4) xxviii. (1909), pp. 665736. Die Greensche Funktion der Schwingungsgleichung. Jahresbericht der Devtscher Math.
Vereinigung, xxi. (1912), pp. 309
353.
SONINE, N.
J.
the resolution of functions into infinite series (Jan. 17/29, 1870). Mathematical Collection published by the Moscow Math. Soc. [MaTenaTHiecKift C6opHHKt], y. (1870),
On
pp.
271302, 323382.
.
Recherches sur les fonctions cylindriques et le developpement des fonctions continues en series (Aug. 1879). Math. Ann. xvl (1880), pp. 180. Sur les fonctions cylindriques (Oct. 24, 1887). Math. Ann. xxx. (1887), pp. 582583. [Note on a paper by Schafheitlin.] Sur les fonctions cylindriques. (Extrait d'une Lettre adressee a M. Niels Nielsen, a Copenhague, May 6, 1904.) Math. Ann. lix. (1904), pp. 529552.
Integral der Differentialgleichung
n. (1857),
pp. 165170.
(1858), pp.
Entwickelung von e***^x in unendliche Reihen. 244246. Darstellung des unendlichen Kettenbruches
2a?
, +l +
2x+3 +
2#+52* + 7 + ...
in geschlossener Form. Archiv der Math, und Phys. xxx. (1858), pp. 331 Ueber die Integration der Differentialgleichung
334.
d*y
dxn
(1859).
'
87.
STEARN,
On some
(1880), pp.
H.
Quarterly Journal, xvn.
90104.
* See also under Michell.
784
Intensitats-verhaltnisse der Beugungserscheinung durch eine kreisformige Offnung (Jui)e 19, 1893). Math, und Naturwiss. Berichte aus Ungarn, XL (1892 93) [1894], pp
362373.
STEINTHAL,
On
the Solution of the Equation
.
A. E.
,
cPu
dar
'
du ax
, '
Quarterly Journal,
xvm.
(1882), pp.
330
345.
A.
Messenger, xxxiii. (1904),
STEPHENSON,
pp.
On Expansion
549.
Mag.
547
STERN, M.
Ueber
die
A.
365.
STIELTJES,
in. (1886), pp.
T. J.
sci.
201258.
On
1141.]
On
(May
the Discontinuity of Arbitrary Constants which appear in Divergent Developments 11, 1857). Trans. Camb. Phil. Soc. x. (1864), pp. 106128. [Math, and Phys. Papers,
77109.] Supplement to a paper on the Discontinuity of Arbitrary Constants which appear in Divergent Developments (May 25, 1868). Trans. Camb. Phil. Soc. xi. (1871), pp. 412 425.
Papers, iv. (1904), pp. 283 298.] On the Communication of Vibration from a Vibrating Body to a surrounding Gas* (June 18, 1868). Phil. Trans, of the Royal Soc. clviii. (1868) [1869], pp. 447 463. [Math, and Phys. Papers, iv. (1904), pp. 299 324.] Smith's Prize Examination Papers, Fet>. 1853 and Jan. 29, 1867. [Math, and Phys. Papers, V. (1905), pp. 319, 347.] Note on the Determination of Arbitrary Constants which appear as Multipliers of Senii-convergent Series (June 3, 1889). Proc. Camb. Phil. Soc. vi. (1889), pp. 3652 366. [Math, and Phys. Papers, v. (1905), pp. 221225.] On the Discontinuity of Arbitrary Constants that appear as Multipliers of Semi-convergent Series t (April 23, 1902). Acta Math. xxvi. (1902), pp. 393397. [Math, and Phys. Papers, v. (1905), pp. 283287.]
[Math,
and Phys.
STRUTT,
See Rayleigh (Lord).
J.
W.
H.
STRUVE,
Ueber den Einfluss der Diffraction an Fernrbhren auf Lichtscheiben (May 25, 1882). Mem. de I'Acad. Imp. des Sci. de St. Petersbourg, (7) xxx. (1882), no. 8. Beitrag zur Theorie der Diffraction an Fernrohren (Aug. 1882). Ann. der Physilc und
Chemie, (3) xvn. (1882), pp. 1008
1016.
STURM,
Sur les Equations differentielles Math. l. (1836), pp. 106186.
lineaires
J. C.
F
(Sept. 28, 1833). Journal de
du second ordre
* An abstract will be found in Proc. Royal Soc. xvi. (1868), pp. 470471. f This, the last published paper which Stokes wrote, contains a summary of his researches composed for the Abel centenary volume.
BIBLIOGRAPHY
SUCHAR,
Sur
les
785
P. J.
equations differentielles lineaires reciproques du seco. d ordre (Nov. 18, 1903). Bull, de la Soc. Math, de France, xxxn. (1904), pp. 103116.
SVANBERG,
De
pp.
A. F.
Soc. Sci.
Nova Acta R.
T.
Upsala,
x.
(1832),
231288.
TAKEUCHI,
Gn
the integral
J
17, 1920).
THEISINGER,
Bestimmtc
Integrate.
L.
THOMAE,
J.
Die Brauchbarkeit der Bessel-Fourierechen Reihe. Berichte der Math. Seminar zu Jena, 191213, pp. 810. [Int. Catalogue of Set. Lit. xin. (1913), p. 100.]
Note on
""f*
llt dx.
Recent Researches in Electricity and Magnetism (Oxford, 1893). Nature, xlix. (1894), pp. 367369.]
[Review by A. Gray,
THOMSON, W.
See Kelvin (Lord).
TODHUNTER,
TURRIERE,
Une
application
I.
Elementary Treatise on Laplace's Functions, Lamd's Functions and BesseVs Functions (London, 1875).
E.
(4) ix. (1909), pp.
geom&rique de
la serie considered
UNFERDINGER, F. C a tiber die beiden Integrate fe*n * ? (njc-cosx)dx (April 16, 1868).
berichte, lvii. (1868), pp.
Wiener Sitzungs-
611
620.
VALEWINK,
Over asymptotische ontwikkelingen
die Fortschritte der Math. 1905, p. 328.]
G. C. A.
{Dissertation,
Haarlem, pp.
138).
[Jahrbuch
iiber
VAN VLECK,
On
pp. 75
E. B.
,
VERDET,
I.
(Paris, 1869).
VESSIOT,
Sur quelques equations
des Sci. de Toulouse, ix. (1895), no.
6.
E.
differentieltes ordinaires
du second
ordre.
Annates de la Fac.
VOLTERRA, V. Sopra alcune questioni di inversione di integrali definiti. Ann. di Mat. pp. 139178.
VORONOI,
G.
(2)
xxv. (1897),
Sur une fonction transcendante et ses applications a la sommation de quelques senes. Ann. sci. de VtkoU norm. sup. (3) xxi. (1904), pp. 207268, 459534. Sur le developpemeut, a l'aide des fonctions cylindriques, des sorames doubles
2 2 If (pm +2qmn + rn )
oh pmt + Zqmn+rn*
est
une forme
786
i. Art. Bern Mittheilungen, 1894, pp. 204266. Uber die Darstellung einiger bestimmten Integrate durch Bessel'sche Funktionen (June 12, 1895; Aug. 5, 1895). Bern Mittheilungen, 1895, pp. 115 119; 1896, pp. 53 60.
G. T. Some formulae for transforming the origin of reference of Bessel's functions. Messenger, xxv. (1896), pp. 76 80. WALKER, G. W. The scattering of electromagnetic waves by a sphere. Quarterly Jburnal, xxxi. (1900), 49. pp. 36
WALKER,
WALKER,
J.
WALLENBURG,
G.
fiir
Math. cxxi.
Die Differentialgleichungen deren allgemeines Integral eine lineare gobrochene Function Journal fiir Math. cxxi. (1900), pp. 210 217. ist. Sur l'equation diff&rentielle de Riccati du second ordre (Dec. 14, 1903). Comptes Rendus,
pp 1033 1035.
WANGERIN, A. Cylinderfunktionen oder Bessel'sche Funktionen. Encyclopddie der Math. Wiss. Bd. n. Teil 1 (Leipzig, 1904 757. [Encyclopedic des Sci. Math. Tome II. vol. 5 16), pp. 742 (Paris et Leipzig, 1914, pp. 209 229), translation by A. Lambert.]
On
pp.
5055.
Bessel Functions and
Kapteyn Series (April 26, 1916). Proc. London Math. Soc. (2) 150174. Bessel Functions of Equal Order and Argument (June 17, 1916). Phil. Mag. (6) xxxn. (1916), pp. 232237. Simple types of Kapteyn Series. Messenger, xlvi. (1917), pp. 150157. Bessel functions of equal order and argument (Nov. 13, 1916). Proc. Camb. Phil. Soc. xix. (1918), pp. 4248. The limits of applicability of the Principle of Stationary Phase (Nov. 22, 1916). Proc. Camb. Phil. Soc. xix. (1918), pp. 4955. Bessel functions of large order (June 14, 1917). Proc. Camb. Phil. Soc. xix. (1918),
xvi. (1917), pp.
pp.
96 110.
The Zeros
of Bessel Functions (Aug. 17, 1917).
Proc.
Phil.
(1918), pp.
190206.
Bessel Functions of Equal Order and Argument. 370. The Diffraction of Electric Waves by the Earth
Mag.
(May
29, 1918).
(1919), pp.
8399.
xcv.
of Electric Waves round the Earth (Jan. 13, 1919). Proc. Royal Soc. pp. 546563. On Nielsen's functional equations. Messenger, xlviii. (1919), pp. 49 53. The zeros of Lommel's polynomials (May 15, 1919). Proc. London Math. Soc. (2) xix. (1921), pp. 266272.
The Transmission
A (1919),
WEBB,
The expansion of an
(1904), pp.
H. A.
Messenger,
xxxm.
Phil.
55-58.
On
the Convergence of Infinite Series of Analytic Functions (Nov. 10, 1904). Soc. cciv. (1904), pp. 481- 497.
WEBER,
222237.
H.
Journal
fiir
partiellen Difierentialgleichung
~2 + x^ +1^11 =
Math. Ann.
I.
(1869), pp.
13d.
BIBLIOGRAPHY
787
Ueber die Besselxvhcu Functionen und ihre Anwendung auf die Theorie der elektrischen Strbme (May, 1872). Journal fUr Math. lxxv. (1873), pp. 75105. Ueber die stationaren Stromungen der Electricitat in Cylindern (Sept. 1872). Journal
fiir
Ueber erne Darstellung willkiirlicher Functionen durch Bessel'sche Functionen (Oct. 1872). Math. Ann. vi. (1873), pp. 146161. Zur Theorie der Bessersohen Functionen (July, 1890). Math. Ann. xxxvn. (1890), pp. 404416.
WEBER, H. F. Die wahrc Theorie der Fresnel'aehen Interferenz-Erscheinuugen. Zurich Vierteljahrsschrift, xxiv. (1879), pp. 33 76; Ann. der Physik und Chemio, (3j vm. (1879), pp. 407 444. WEBSTER, A. G. Application of a definite integral involving Bessel's functions to the self-inductance of solenoids (Dec. 29, 1905). Bulletin American Math. Soc. xiv. (1907), pp. 16.
WENDT, CACILIE.
Eine Vcrallgemeinerung des Additionstheoremes der Bessel'schen Functionen erster Art. Monatshe/te fiir Math, und Phys. xi. (1900), pp. 125131.
WEYL,
Singulare Integralgleichungen (April, 1908).
(1909), pp.
273324.
Oes.
WEYR,
(Prag), (6)
Wiss.
1.
WHEWELL,
Of the
Intrinsic Equation of
W.
a Curve and its Application (Feb. 12, 1849). Trans. Camb. Phil. Soc. vijr. (1849), pp. 659671. Second Memoir on the Intrinsic Equation of a Curve and its Application (April 15, 1850). Trans. Camb. Phil. Soc. IX. (1856), pp. 150156.
.WHIPPLE, F. J. W. Diffraction by a wedge and kindred problems (Nov. 8, 1915). (2) xvi. (1917), pp. 94111.
The
Proc.
London Math.
Soc.
WHITE, F. P. Diffraction of Plane Electromagnetic Waves by a Perfectly Reflecting Sphere (June 9, 1921). Proc. Royal Soc. C. (1922), pp. 505525.
WHITEHEAD,
On
On
the functions ber x, bei x, ker x, kei x.
C. S.
342.
WHITTAKER, E. T. the General Solution of Laplace's Equation and the Equation of Wave Motions and on an undulatory explanation of Gravity. Monthly Notices of the R. A. S. lxii. (1902),
pp.
617620.
On
Proc.
the partial differential equations of mathematical physics. Math. Ann. lvii. (1903),
pp. 333
355.
On a new Connexion
London Math.
Soc.
of Bessel Functions with Legendre Functions (Nov. xxxv. (1903), pp. 198206.
13, 1902).
WIGERT,
S.
1913).
WILLIAMSON,
On
(1856), pp.
B.
5, 1856).
Phil.
Mag.
364371.
(4) xi
WILLSON, R. W. PEIRCE, B. O. forty roots of the Bessel equation (x)=*0 with the corresponding values of J\{x). Bulletin American Math. Soc. in. (1897), pp. 153 155.
Table of the
first
AND
WILTON, J. R. continued fraction solution of the linear differential equation of the second order. Quarterly Journal, xlvi. (1915), pp. 318334,
788
15.
WORMS DE ROMILLY,
Note sur
l'integration
P.
de
l'e'quation
+ e! + r-a *J ax* x ax
Journal de Math.
(3) iv. (1878), pp.
177186.
On
1911).
On
pp.
Quarterly Journal, xliii. (1911), pp. 161 177. series of Bessel functions (Dec. 6, 1917). Proc.
6,
London Math.
163200.
ZELINSKIJ,
On
I.
I.
the integration of Riccati's equation (Jan. 27, 1890). Proc. Phys. Math. Section, Naturalist^ Soc. Imp. Univ. Kazan, vm. (1890), pp. 337 342.
INDEX OF SYMBOLS
[The numbers refer
to the
pages on which
Aw
pFq (i, 2,
ay
Pi
>
P*>
>
J,
(*),
308
An
584
...,/>,;*),
100
3.(*),61
j,
A n>y (0,283
#,(,*); 598 #,,(*), 529 #,,(*). 571
a,
P F
(a;),
k
(a),
639 646
1>l'\ >^)>37i
507
jm, 576
j**j,',j.",
j,,,
A*.P\
485
479
/<*)*, 526
if,
559
78 [see also
p.
Bn 6
,
/(*)*, 536
^n CO,
65]
(*), 247
n;M> , (0,293
#,(*), 597
#,,(*),
529
bei (x), 81
ber
(#),
(a;),
bei (#), 81
Ktli (z), 46
iT(20n 288
,
K(s), 78
ker
(e),
kei
(*),
81
ber
(0,276
,(0 284 5rm ,,(z),303
(*), 71
I- r
554 C(*), 658 C/,, (a), 320 Cn "(*), 50 r<), 82 (*[?), 129 c, 508
C,
I, l r ,
(*),329 111
H, 577
fr, (*),#.(#), 73
M, 6
505
H,(*),328
herj, hei(*), 81
her(s), hei(.z), 81
if (a;), 661
iW,
N-n m
,i,
Dn
598 (*), 74
,
h\
13
JV,
0,197
324 661
,
h(z)> 77
J",(5), 10,
#,6
16
14,
E
e,
n (z),
56
Jn (z),Q,
./-(*),
19
0- n (0,276
< (0, 569
16
^,, 10
111
P(cos<p), 129
790
Pn (r;
(a,
o),
420
2U(jr),343
7/n
*,
22
(*,a),320
k<*),56
B, 112
?;,(*,*),
582
@, 27; 291
0,112
*, 19; 112
Ajn,
fi,
k^,7.
Q 227
Q"(*),156
Q(*),46 Q(x,v), 205
i2,
26 38
577 38
;
537
488
537
<r r
w 502
,
606
537 538
HI Tn (s),64
5, *r,
<M)>49
339
% 514
i/r,
42 552
*(*), 56
Yn (z), 64 F(s),64
F (*),
F<">(*),
67
^(4
70
^ (*),
(*),
59
Y(*),58
Y-(*), 59
Y(s), 63
y*. y/, y*",
485
S^(z),S^(z), 263
a)*"*
* (s), 204
Sn (s),353
*,
67
**,*(*).
34 5
,20
51
'
//
^, 33 Tn (s),340
V39
e,6
n, 1,103
LIST OF
[The numbers refer
to the pages.
AUTHOES QUOTED
References are not given to entries in the bibliography,
Bourget,
J., 6,
K 60
R,
65, 142, 214, 247, 319, 502, 505,
659, 660
Airey, J.
Brassinne, E., 91
W.
S., 65,
Bromwich, T.
664
Alexander,
P.. 579 Anding, E., 313, 657 Anger, C. T., 21, 22, 308, 309, 310, 311, 312 Anisimov* V. A., 92 Appell, P. E., 146, 371 Autonne, L., 94
601 Bruns, H., 327 Bryan, G. H., 127, 480 Burkhardt, H. F. K. I/., 236, 280 Burnside, W. S., 305
Cailler,
C,
536, 537
Cantor, G. F. L.
268, 572
P.,
637
367, 383, 387, 402, 409 Basset, A. B., 76, 77, 78, 80, 172, 173, 180,
Carslaw, H.
509, 583
S.,
385, 388, 395, 425, 426, 454 Bateman, H., 130, 131, 367, 370, 372, 373,
376, 379, 380, 389, 406, 417, 437, 456, 530, 533, 535
Catalan, E.
C,
Cauchy, (Baron)
L., 7,
Bauer, G., 50, 128, 368, 370 Beltrami, E., 51, 358, 361, 374, 386, 389, 390, 391, 500, 579, 621 Bernoulli, Daniel (1700-1782), 2, 3, 4, 9, 85,
86, 87, 88, 111, 123, 478,
319, 324, 449, 545, 554, 557, 579 Cayley, A., 88, 90, 96, 102, 103, 109, 188,
502
H. W., 91 S., 621 Chessin, A. S., 135, 175, 325, 346, 382 Chree, C, 6, 597 Christoffel, E. B., 154 Chrystal, G., 102, 288, 295
Challis,
Chapman,
576
1, 2, 3,
Bernoulli,
James (1654-1705),
88, 90,
92
Bernoulli, John (1667-1748), 1, 2, 3 Bernoulli, Nicholas, (1687-1759), 2 Bernoulli, Nicholas, (1695-1726), 2
Bessel, F. W.,
1, 3, 4, 5, 9,
Cinelli, M.,
633
Clebsch, R. F.
Clifford,
A,
359, 363
W.
K., 90, 91
21, 24, 25, 38, 84, 140, 148, 153, 160, 295,
Crawford,
L.,
27
Bocher, M., 46, 57, 64, 376, 494, 495, 517 Bohmer, P. E., 142
Boole, G., 27, 47, 110,627
302
Curzon, H. E.
J.,
395
792
D'Alembert, J. le Rond, 3 Dandelin, G. P., 503 Darboux, J. G, 233 Darwin, C. G., 437 Debye, P., 225, 235, 237, 240, 241, 247, 249, 250, 251, 255, 262, 263, 268, 513, 516 De la Vallee Poussin, Ch. J., 53, 160, 189 De Morgan, A., 188, 190 Dendy, A., 107 Dini, U., 10, 577, 578, 597, 600, 616, 651 Dinnik, A., 579, 659, 660
Dirichlet, P. G. Lejeune, 157, 230, 406, 581,
365,
383,
396,
430, 438, 439, 480, 508, 517, 522, 524, 525, 579
662
B23
Dixon, A. C, 35, 480, 481, 482 Donkin, W. F., 109
Dougall,
J., 65,
Gilbert, L. P., 545, 548, 549 Giuliani, G., 155, 156, 324, 326, 327 Glaisher, J. W. L., 89, 96, 102, 103, 108, 109, 140, 171, 173, 183, 664
Gordan, P. A., 55
Goursat, Edouard
J. B.,
120
411
P. D. G., 183, 455,
503
470
108
Graf, J. H., 32, 64, 75, 145, 153, 160, 165, 175, 197, 215, 227, 286, 287, 290, 295, 296, 299, 301, 302, 303, 341, 344, 345, 359, 360, 362, 398, 498, 502, 583
Emde,
Gray, A., 64, 65, 78, 194, 206, 454, 480, 655 656, 657, 658, 660 Green, G, 124
Greenhill, Sir A. George, 91, 96
145, 290
Gubler, E., 32, 64, 145, 160, 165, 177, 197, 215, 227, 286, 287, 301, 329, 341, 351, 398,
408, 410, 426, 498, 502, 583
Falkenhagen,
Fejer, L.,
Hadamard,
J.,
Fields, J.
G, 110
Filon, L. N.
G,
629
Hague,
Hall,
B.,
656
Ford,
W.
B., 578,
605
A,
15
Hamilton, Sir William Rowan, 12, 195, 655 Hankel, H, 10, 38, 57, 58, 61, 62, 63, 65, 73,
75, 76, 77, 160, 163, 164, 165, 167, 175,
9, 10, 13,
195, 196, 203, 206, 208, 211,' 384, 386, 390, 393, 395, 424, 427, 428, 429, 430, 434, 453, 454, 456, 457, 458, 459, 462, 464, 465, 471, 488, 513, 514, 577, 579, 581, 582, 633 Hansen, P. A., 14, 20, 30, 31, 37, 152, 154, 155, 158, 195, 292, 406, 655,
Frobenius, F.
Frullani,
G,
57
656
G,
14, 19
Hanumanta Rao,
Hardy, G. H.,
C. V., 437
Gallop, E.
G,
8, 111, 180, 183, 188, 189, 200, 309, 320, 321, 322, 324, 373, 382, 386, 395, 406, 421, 422, 437, 441, 442, 463, 464, 542, 546, 547, 573, 575, 579, 606, 615, 621
J., 88,
Gauss,
C F.
Hargreave, C.
170, 171
LIST OF
Hargreaves, R., 538 Harnack, A., 577 Hams, R. A., 15
AUTHORS QUOTED
Lacroix, S. F., 27
793
Lagrange,
551
Lamb,
Lambert,
Lame\ G,
Landau,
96,
485 159
6, 7, 8, 9, 53,
E. G. H., 197
Laplace, P. S. de,
Heine, H. E.,
450
Largeteau, C. L., 501 Laurent, Paul Mathieu Hermann, 157
Wera
Myller-, 99
Hopf,
L., 178,
J.,
406
Lebesgue, Henri, 457 Lebesgue, Victor AmeMee, 110, 123 Lefort, F., 6 Legendre, A. M., 52, 90, 183, 204, 485, 557 l'Hospital, G. F. A. (Marquis de St Mesme),
Horn,
225, 526
134
Leibniz, G. W.,
1, 2,
Hurwitz, A., 9, 297, 302, 303, 304, 305, 306, 307, 483, 484
Ignatowsky, W. von, 365 Isherwood, J. G., 657, 658, 664
Jackson, Dunham, 343 Jackson, Frank Hilton, 43, 44 Jackson, William Hartas, 177
Jacobi, C. G.
J., 6, 8, 14,
Le
Paige,
C, 96
Lerch, M., 382, 393, 433, 434, 617 Lindner, P., 484
Lindstedt, A., 545, 661
Liouville, J., 27, 28, 87, 111, 112, 116, 117,
Lipschitz, R. O.
572
Jamet, E. V., 94 Johnson, W. W., 92 Jolliffe, A. E., 528, 529 Julius, V. A., 65, 200
Kalahne, A., 505, 507, 660 Kapteyn, W., 35, 183, 200, 268, 279, 281,
282, 292, 351, 370, 373, 376, 380, 386, 404,
Lommel,
38, 43, 45, 46, 47, 49, 53, 59, 62, 65, 66,
71, 73, 76, 77, 96, 97, 98, 99, 106, 107, 132,
152, 154, 196, 200, 294, 295, 296, 297, 298, 299, 300, 303, 308, 315, 328, 341, 345, 348,
413, 498, 499, 531, 532, 533, 535, 536, 538, 551, 559, 560, 562, 565, 566, 568, 569,
350, 364, 374, 406, 478, 479, 482, 529, 531, 537, 538, 539, 540, 542, 543, 544, 545, 546, 548, 549, 550, 576, 654, 655, 656, 658, 660,
661, 664
Lorenz,
500
658
Kepinski,
Kepler,
S.,
99
552
367, 419, 420
J., 551,
225, 229, 233, 365, 377, 385, 386, 389, 395, 396, 412, 413, 439, 440, 482, 509, 511
Kluyver,
J.
C,
McMahon,
J., 64,
Kummer,
MacRobert, T. M., 197 Maggi, G. A., 13 Malmsten, C. J., 99, 173, 185, 187, 188, 196, 203
Manfredius, G., 92
794
March, H. W., 56, 225, 449 Marcolongo, R., 135 Marshall, W., 505, 506 Mathews, G. B., 64, 65, 78, 194, 206, 454, 480, 655, 656, 657, 658, 660
Maxwell,
J.
664
P^res, J., 44
49
E.,
Clerk, 125
Phragmen,
358
554
Plummer, H. C, 270, 552, 555 Pochhammer, L., 100, 101, 297, 346, 410 Pocklington, H. C, 537
Poincare\ J. Henri, 236
Poisson, S. D.,
6, 9, 10, 11, 12, 13, 24, 25, 38,
M.
G., 83,
497
Moore, C. N., 479, 578, 579, 597, 649 Morton, W. B., 65, 66
47, 49, 52, 67, 68, 69, 73, 95, 96, 160, 173,
Murphy,
501
Lebedeff)
Porter,
Myller-Lebedeff,
W.
(see
M.
517
Preece, C. T., 27
Neumann, Carl
31, 32, 33, 34, 36, 37, 46, 59, 60, 65, 66, 67,
68, 69, 70, 71; 73, 128, 143, 150, 151, 155,
271, 273, 274, 276, 277, 278, 280, 281, 284, 286, 290, 291, 292, 346, 358, 359, 361, 363, 365, 386, 418, 424, 440, 441, 453, 455, 456,
W.
525
Riemann, G.
150, 189, 226, 229, 231, 247, 248, 249, 260, 252, 262, 329, 332, 413, 415, 425, 426, 431, 440, 441, 446, 448, 505, 656
Nicolas,
J., 77,
84
Nielsen, N., 24, 44, 49, 64, 73, 74, 77, 82, 83, 132, 142, 145, 148, 149, 154, 169, 224, 297, 298, 299, 315, 350, 355, 357, 359, 392, 405,
455, 465, 522, 523, 525, 526, 527, 528, 571, 572, 574, 597, 622, 629, 636
Rutgers,
Rybczyriski,
W.
Savidge, H.
G,
82, 204,
658
Schafheitlin, P., 64, 137, 142, 168, 169, 207, 215, 373, 391, 392, 398, 401, 402, 406, 408, 421, 447, 477, 479, 482, 485, 487, 489, 490,
Oltramare, G.,
Orr, W.
73
146, 206, 224, 454,455,
McF, 145,
579
65, 67, 72, 79, 90, 91, 143, 145, 151, 160,
186
9, 21, 24, 68,
196, 215, 216, 228, 253, 274, 276, 278, 284, 105, 229, 358, 285, 286, 288, 289, 290, 341, 342, 344, 345,
Parseval, M. A.,
359, 384
LIST OF
AUTHORS QUOTED
Thomson, Sir Joseph John, 65, 173 Thomson, Sir William (see Kelvin)
Tisserand, F., 371
795
173,183,617,618,619,621,622,628,655,666 Schonholzer, J. J., 145 Schott, G. A., 551, 556, 572, 573
Todhunter,
I.,
Schwarz, K. H. A, 358, 643 Schwarzschild, K., 361 Schwerd, F. M., 477, 654 Searle, J. H. C, 199 Segar, H. W., 483 Serret, J. A, 171, 173, 188 Sharpe, H. J., 105, 157
Turriere, E., 15
Sheppard,
Siacci, F.,
W.
92
Siegel, C. L.,
485
Siemon, P., 328, 398 Smith, Bernard A., 655 Smith, Clara E., 621 Smith, Otto Andreas, 50 Sommerfeld, A. J. W., 56, 57, 178, 361, 389, 395, 406, 417, 464, 499 Sonine, N. J., 82, 83, 132, 137, 139, 143, 169,
170, 171, 175, 176, 177, 180, 279, 280, 281,
Wagner, C,
290, 353, 354, 362, 363, 367, 373, 374, 375, 376, 377, 378, 383, 386, 391, 394, 395, 398, 401, 411, 415, 417, 418, 431, 432, 433, 434, 439, 454
Spitzer, S., 68, 71, 153
13, 142 Walker, Gilbert Thomas, 360, 361 Walker, James, 328, 331, 333, 537, 544 Wallenburg, G., 94 Waring, E, 503 Watson, G. N., 11, 105, 125, 158, 226, 231,
575
Webb, H. A,
395, 396, 398, 402, 405, 406, 408, 421, 450, 451, 452, 453, 454, 455, 468, 469, 470, 495 Weber, Heinrich Friedrich, 308, 309, 310,311, 312, 315, 320 Weierstrass, C. T. W., 358 Wendt, Cacilie, 363 Weyl, H., 189, 454 Weyr, E, 93, 94 Whewell, W., 479 Whipple, F. J. W., 177, 313, 387 Whitehead, C. S., 81, 82, 132, 148, 203
M.
A.,
500
214, 464
Stirling,
James,
7, 8,
214
W.
(see
Rayleigh)
Whittaker, E.
Sturm,
521
Wilton,
J. R.,
154
Young, W. H.,
GENERAL INDEX
[The numbers refer
;
to the pages.]
Addition theorems, 358-372 (Chapter xi) for Bessel coefficients of order zero, 128, 359; for Bessel coefficients of order n, 29; for Bessel functions of the first kind (Gegenbauer's type), 362, 367; for Bessel functions of the first kind (Graf's type), 130, 143, 359 ; for Bessel functions or cylinder functions of any kind (Gegenbauer's type), 363 for Bessel functions or cylinder functions of any kind (Graf's type), 143, 361 for hemi-cylindrical functions, 354; for Lommel's functions of two variables, 543; for Schlafli's function fn (z), 344; for Schlafli's polynomial, 289; integrals derived from, 367 ; physical significance of, 128, 130, 361, 363, 366; special and degenerate forms of, 366, 368 Airy's integral, 188 expressed in terms of Bessel functions of order one-third, 192 generalised by Hardy, 320; Hardy's expressions for the generalised integral in terms of the functions of Bessel, Anger and Weber, 321 references to tables of, 659 Analytic theory of numbers associated with asymptotic expansions of Bessel functions, 200 Anger's function 3 v {z), 308; connexion with Weber's function, 310; differential equation satisfied by, 312; integrals expressed in terms of, 312; recurrence formulae for, 311; representation of Airy's integral (generalised) by, 321 with large argument, asymptotic expansion of, 313 with large argument and order, asymptotic expansion of, 316 Approximations to Bessel coefficients of order zero with large argument, 10, 12 to Bessel functions of large order (Carlini), 6, 7 (extensions due to Meissel), 226, 227, 232, 247, 521 (in transitional regions), 248; to functions of large numbers (Darboux), 233; (Laplace), 421 to Legendre functions of large degree, 65, 155, 157, 158 to remainders in asymptotic expansions, 213; to the sum of a series of positive terms, 8. See also Asymptotic expansions, Method of stationary phase and Method of steepest descents Arbitrary functions, expansions of, see Neumann series and Kapteyn Beries (for complex variables); Dini series, Fourier-Bessel series, Neumann series and Schlbmilch series (for real variables) Argument of a Bessel function defined, 40 Asymptotic expansions, approximations to remainders in, 213; conversion into convergent series, 204; for Bessel coefficients of order zero with large argument, 10, 12, 194; for Bessel functions of arbitrary order with large argument, 194-224 (Chapter vn) (functions of the first and second kinds), 199 (functions of the third kind), 196; (functions of the third kind by Barnes' methods), 220; (functions of the third kind by Schlafli's methods), 215; (functions with imaginary argument), 202; for Bessel functions with order and argument both large, 225-270 (Chapter vni); (order greater than argument), 241 (order less than argument), 244 (order nearly equal to argument), 245 (order not nearly equal to argument, both being complex), 262; for combinations of squares and products of Bessel functions of large argument, 221, 448; for Fresnel's integrals, 545 for functions of Anger and Weber (of arbitrary order with large argument), 313 (with order and argument both large), 316; for Lommel's functions, 351; for Lommel's functions of two variables, 549; for Struve's function (of arbitrary order with large argument), 332; (with order and argument both large), 333 for Thomson's functions, ber (z) and bei (z), 203; for Whittaker's function, 340; magnitude of remainders in, 206, 211, 213, 236, 314, 332, 352, 449; sign of remainders in, 206, 207, 209, 215, 315, 333, 449. See also Approximations
; ;
;
Basic numbers applied to Bessel functions, 43 Bateman's type of definite integral, 379, 382 Bei (z), Ber (z). See Thomson's functions
Bernoullian polynomials associated with Poisson's integral, 49 Bernoulli's (Daniel) solution of Biccati's equation, 85, 89, 123 Bessel coefficient of order zero, J (z), 3, 4; differential equation satisfied by, 4, 5 (general solution of), 5, 12, 59, 60; expressed as limit of a Legendre function, 65, 155, 157; oscillations of a uniform heavy chain and, 3, 4 Parseval's integral representing, 9 with large argument, asymptotic expansion of, 10, 12, 194; zeros of, 4, 5. See also Bessel coefficients, Bessel functions and Bessel's differential equation Bessel coefficients J (z), 5, 6, 13, 14-37 (Chapter n); addition theorem for, 29; Bessel's integral for, 19; expansion in power series of, 15; generating function of, 14, 22, 23; inequalities satisfied by, 16, 31, 268; notations for, 13, 14; order of, 14; (negative), 16; recurrence formulae for, 17; square of, 32; tables of (of orders and 1), 662, 666-697; (of order i), 664, 730-732; (with equal order and argument), 664, 746; tables of (references to), 654, 655, 656, 658. See also Bessel coefficient of order zero, Bessel'B differential equation and Bessel functions Bessel functions, 38-84 (Chapter m) argument of, defined, 40 differential equations of order higher than the second satisfied by, 106 expressed as limits of Lame functions, 159 expressed as limits of P-functions, 158 history of, 1-13 (Chapter i) (compiled by Maggi and by Wagner), 13 indefinite integrals containing, 132-138 ; order of, defined, 38, 58, 63, 67, 70 rank of, defined, 129 relations between the various kinds of, 74 representation of cylinder functions in terms of, 82; solutions of difference equations in terms of, 83, 355; solutions of Laplace's
; ; ; ;
;
;; ;
GENERAL INDEX
797
threeequation containing, 83, 124 ; solutions of the equation of wave motions containing, 125 ; and ten term relations connecting, 300 ; with negative argument, 75. See also the two preceding following entries, and Cylinder functions 367, 368 Bessel functions of the first kind, JV (*), 38 ; addition theorems for, 143, 359, 362, 363, cut in Barnes' type of integral representing, 190 ; Bessel's type of integral representing, 176 ; expansion of, plane to render uniform, 45 differential equation (Bessel's) satisfied by, 38 expressed ascending series, 40 ; expansion of, in descending series, see Asymptotic expansions limit of a hypergeometnc as a generalised hypergeometric function, 100, 101 ; expressed as the significance of), 155; function, 154 ; expressed as the limit of a Legendre function, 156; (physical generalisations expressed as the limit of a Lommel polynomial, 302 functional properties of, 45 infinite integrals containing, of, 43, 44, 308-357 ; inequalities satisfied by, 49, 255, 259, 270, 406 ; Chapter xm, passim ; of complex order, 46 of order n + *, 41 (expressed in finite terms), 52,
; ;
55
(notations for), 55, 80 ; Poisson's integral representing, 47, 48 ; (modifications of), 161, 163, of two, expressed as a 164, 169, 170 ; products of, see Products of Bessel functions ; quotient Lommel s continued fraction, 153, 154, 303 ; recurrence formulae for, 45, 294 ; relations with polynomial, 297; represented by integrals containing Legendre functions, 173, 174; symbolic formulae for, 170; tables of (of orders and 1), 662, 666-697 ; (of order n), 664, 730-732 ; (of order n+4), 664, 740-745; (of order $), 664, 714-729; (of order - *, method of computing), 664; to), (with equal order and argument), 664, 746 ; (zeros of), 664, 748-751 ; tables of (references with large argument, see 654, 655, 656, 658, 659, 660 ; Weierstrassian product representing, 497 Asymptotic expansions ; zeros of, see Zeros of Bessel functions
;
;
Hud, Y (z)
Yv (z)
;
(z) (after Hankel), 57, 63; n (z) (after Heine), 65 ; F<> (after Weber-Schlafli, the canonical form), 63; addition theorems
;
Bessel's type of integral representing, 177 component parts of, 71, 72, 340; continuity of (qua function of their order), 63; differential equation (Bessel's) satisfied by, descending series, 59, 63 expansion of, in ascending series, 59, 60, 61, 69, 72 expansion of, in the first kind, 5, see Asymptotic expansions ; expressed as an integral containing functions of 133, 382, 433 infinite integrals containing, 385, 387, 393, 394, 424, 425, 426, 428, 429, 430, 433; of), 169, 170; products of, Poisson's type of integral representing, 68, 73, 165; (modifications 149; (represented by infinite integrals), 221, 441,' 446 ; (asymptotic expansions of), 221, 448; recurrence formulae for, 66, 71 represented by integrals containing Legendre functions, 174 symbolic formulae for, 170 tables of (of orders and 1), 662, 666-697 (of order n), 664, 732735 ; (of order J), 664, 714-729; (of order -J, method of computing), 664; (with equal order and argument), 664, 747 (zeros of), 748-751 tables of, references to, 655, 656, 658 ; with large zeros of, set. Zeros of argument, see Asymptotic expansions ; with negative argument, 75 Bessel functions. See also Neumann's polynomial w (z), v (z), 73; Barnes' integrals representing, 192 v Bessel functions of the third kind, Bessel's type of integral representing, 178 ; Poisson's type of integral representing, 166 (modifications of), 168, 169, 170; represented by integrals containing Legendre functions, 174; symbolic formulae for, 170 ; tables of (of orders O'and 1), 662, 666-697 ; (of order ), 664, 714729 ; tables of (references to), 657 with large argument, asymptotic expansions of, 199, 210, 215 with large argument and order, asymptotic expansions of, 244, 245, 262 ; with negative
; ; ; ; ; ; ; ; ;
H
;
argument, 75
Bessel functions whose order and argument are equal, approximations to, 229, 231, 232, 259, 260, 448, 515 ; asymptotic expansions of, 245 ; integrals representing, 258 ; tables of, 658, 664, 746,
747
658
Bessel functions
stability of
whose order is (n + ) Bessel functions whose order is large, 225-270 (Chapter vm) asymptotic expansions of, 241, 244, 245, 262 Carlini's approximation to, 6, 7 ; (extended by Meissel), 226, 227 ; Horn's (elementary) approximation to, 225 ; Laplace's approximation to, 7, 8, 9 ; method of stationary phase applied to, 232 ; method of steepest descents applied to, 237 ; miscellaneous properties of, 252-261 tables of (reference to), 658 ; transitional formulae for, 248 ; zeros of, 513, 516, 517, 518. See
; ;
fraction. Of orders J (and Airy's integral), 190; (and the tables of, 664, 714-729 ; tables of (references to), 659 ; zeros of, orders |, tables of (references to), 659. Of orders J, j . tables of (references to), small fractional orders, tables of zeros of (references to), 502, 660. See also Bessel
whose order
is
a vertical pole), 96
also Bessel functions whose order and argument are equal Bessel functions whose order is =(n t J), 10, 52, 80 ; expressible in finite terms, 52 ; notations for, 55, 80 ; tables of, 664, 740-745 ; tables of (references to), 658, 659 Bessel functions with Imaginary argument, I{z), K (z), K v (z), 77, 78; differential equation integrals representing (of Bessel's type), 181 ; (of Poisson's* type), 79, 171, 172; satisfied by, 77 (proof of equivalence of various types), 185-188 monotonic property of, 446 ; of order (, + 1), 80 ; recurrence formulae, 79 tables of (of orders and 1), 663, 698-713 ; (of order J), 664, 714729 ; (of various integral orders), 664, 736, 737-739 ; tables of (references to), 657, 658; with large argument, asymptotic expansions of, 202 ; zeros of, 511 ; (computation of), 512 ; (references to),
660
Bessel's differential equation, 1, 19 ; (generalised), 38 ; for functions of order zero, 5, 12, 59, 60; for functions with imaginary argument, 77 ; fundamental system of solutions of, 42, 75 ; has no
798
algebraic integral, 117 ; soluble in finite terms when and only when the functions satisfying it are of order n + %, 52, 119; solution of, in ascending series, 39, 40, 57, 59-61 ; solution of, in descending series, see Asymptotic expansions ; symbolic solution of, 41 ; transformations of, 94, 97. See also Bessel coefficients and Bessel functions
Bessel's integral representing Bessel coefficients, 19, 21; generalisations
and extensions
of, see
Anger's function, Bourget's function, Brans' function and Weber's function ; modifications of, to represent Bessel functions of arbitrary order, 175, 176, 177, 178, 181 ; Theisinger's transformation of, 184 ; used in theory of diffraction, 177 ; used to obtain asymptotic expansions, 215. See
also Parseval's integral
Bounds, upper,
see Inequalities'
</,* (z),
Bourget's function
326
327
327
k),
approximation for Bessel functions of large order, 6, 7 324 ; recurrence formulae for, 325 OT Cayley's solution of Biccati's equation, 88 Chain, oscillations of a uniform heavy, 3, 4, 576 Cognate Riccati equations, 91
Cauchy's numbers i^_ n i
,
,
Complex variables, expansions of arbitrary functions of, see Kapteyn series and Neumann series Complex zeros of Bessel functions, 483 of Bessel functions with imaginary argument, 511 of Lommel's polynomials, 306
; ;
Composition of Bessel functions of the second kind of integral order, 340 Computation of zeros of Bessel functions by various methods (Oraeffe's), 500, 502; (Stokes'), 503; (Sturm's, for the smallest zero), 516. See also Zeros of Bessel functions Constant phase, Schlafli's method of, 216 Constants, discontinuity of arbitrary (Stokes' phenomenon), 201, 203, 238, 336 Continuants, connected with Schlafli's polynomial, 288 Continued fractions representing quotients of Bessel functions, 153 convergence of, 154, 303 Convergent series, Hadamard's conversion of asymptotic expansions into, 204 Crelier's integral for Schlafli's polynomial, 288. See also Neumann's integral for Neumann's polynomial
;
Cube of a Bessel function, expansion of, 149 Cat necessary for definition of Bessel functions, 45, 77 Cylinder (circular), motion of heat in, 9, 10, 576, 577 Cylinder functions, 9v {z), 4, 82, 480; addition theorems, 143, 361, 365; connexion with Bessel functions, 83 origin of the name, 83 rank of, 129; solutions of differential equations of order
;
higher than the second by, 106 three-term relations connecting, 300. See also Bessel functions and Hemi-cylindrical functions
;
;;
GENERAL INDEX
799
Dlni series, 577, 580, 596-605, 615-617 (Chapter xvra), 651-653 ; expansion of an arbitrary function of a real variable into, 580, 600 ; methods of theory of functions of complex variables applied of, to, 596, 602 ; Riemann-Lebesgue lemma, analogue of, 599 ; Riemann's theorem, analogue 649; summability of, 601, 615 ; uniformity of convergence of, 601, 604 ; uniqueness of, 616, 651
value at end of range, 602
Dirichlet's discontinuous factor, 406
Discontinuity of arbitrary constants (Stokes' phenomenon), 201, 203, 238, 336 Discontinuous factor (Dirichlet's), 406; (Weber's), 405 Discontinuous integrals, 398, 402, 406, 408, 411, 415, 421
Domain
K (Kapteyn's),
559
diagram
of,
270
Bessel functions, 183
Electric waves, 56, 226, 449 Electromagnetic radiation, 551, 556 Elementary transcendants, definition of, 111 order of, 111 ; solution of differential equations by, 112 Equal order and argument, Bessel functions with, 231, 232, 258, 260 ; tables of, 746, 747 ; tables of (references to), 658, 664 Buler's solution of Riccati's equation, 87 Exponential function, tables of, 698-713 tables referred to, 663, 664
; ;
Factors, discontinuous (Dirichlet's), 406 ; (Weber's), 405 ; Neumann's en of Bessel functions as products of Weierstrassian, 497 Fejer's theorem, analogue of, for Fourier-Bessel expansions, 610
(=1 or 2), 22
expression
Finite terms, Bessel functions of order Ht (n+i) expressed in, 52 ; Bessel functions of other orders not so expressible, 119 ; solutions of Riccati's equation in, 85, 86, 89 ; the solution of Riccati's equation in, not possible except in Daniel Bernoulli's cases and their limit, 123
Flights, problem of
random, 419
See also Fourier-Bessel series
4,
84
Fourier-Bessel integrals, see Multiple infinite integrals Fourier-Bessel series, 576-617 (Chapter xvm), 649-651 ; expansion of an arbitrary function of a real variable into, 576, 580 ; Fejer's theorem, analogue of, 610 ; Kneser-Sommerfeld expansion of a combination of Bessel functions into, 499 ; methods of theory of functions of complex variables applied to, 582, 607 ; order of magnitude of terms in (Sheppard's theorem), 595 ; RiemannLebesgue lemma, analogue of, 589 ; Riemann's theorem, analogue of, 649 ; summability of, 578, 606, 613 ; term-by-term differentiation of, 578, 605 ; uniformity of convergence of, 593, 594 (near origin), 615 ; uniformity of summability of, 612 ; uniqueness of, 616, 649 ; value at end of
range, 594, 603 Fractional differential coefficients, 107, 125 Fresnel's integrals, 544 ; asymptotic expansion of, 545 ; tables of, 744, 745 ; tables of maxima and minima of, 745 ; tables of (references to), 660, 661, 664 Functional equations denning cylinder functions, 82 ; generalised by Nielsen, 355 Functions of large numbers, approximations due to Darboux, 233 ; approximations due to Laplace, 8, 421 . See also Approximations, Asymptotic expansions, Method of stationary phase and Method of steepest descents Fundamental system of solutions of Bessel's differential equation, 42, 75, 78
Gallop's discontinuous infinite Integrals, 421 Gamma functions, representation of Bessel functions by integrals containing, 190, 192, 221 ; applications to determination of asymptotic expansions, 220, 223 ; applications to evaluation of infinite integrals, 383, 434, 436 Gamma functions, representation of Lommel's functions by integrals containing, 351 ; applications
to determination of asymptotic expansions, 352 Gegenhauer's addition theorem for Bessel functions, 362, 363, 367 Gegenbauer's discontinuous infinite integrals, 415, 418 Gegenbauer's function Cn " (z), 50, 129, 363, 365, 367, 368, 369, 378, 407 Gegenbauer's polynomial A nyV (t), 283 ; contour integrals containing, 284, 524 ; differential equation satisfied by, 283 equivalence with special" forms of Lommel's function, 351 ; recurrence formulae for, 283 Gegenhauer's polynomial B n>lkiV [t), 293, 525 Gegenbauer's- representation of Jv \z) by a double integral resembling Poisson's integral, 51
;
Gegenhauer's type of definite integral, 378 Generalised'hypergeometric functions, see Hypergeometric functions (generalised)
; ;;
800
Generalised Integrals (with implied exponential factor), 188, 441, 463, 464 Generating function of Bessel coefficients, 14, 22, 23 ; of Neumann's polynomials, 281, 282 Gilbert's integrals, 548, 549 Giuliani's function, see Bourget's funotion Graeffe's method of calculating zeros, 500, 502 Grafs addition theorem for Bessel functions, 359, 361 Group velocity, 229 Growth of zeros of Bessel functions, 485
Hankel's infinite integrals, 384, 386, 389, 390, 393, 395, 424, 427, 428, 434 Hansen's upper bound for Jn (x), 31 generalised, 406 Hardy's functions Ci n (a), Si n (a), Ein (a), (generalisations of Airy's integral), 320; expressed in terms of functions of Bessel, Anger and Weber, 321, 322 Hardy's integrals representing Lommel's functions of two variables, 546 Hardy's method of evaluating definite integrals, 382 Heat, conduction of, 9, 10, 450, 576, 577, 616
;
Hemi-cylindrical functions S w (z), defined, 353 353 addition theorem for, 354
;
Hypergeometric functions, limiting forms expressed as Bessel functions, 154 Hypergeometric functions (generalised), 90, 100 Bessel functions expressed in terms of, 100, 101 notations for, 100 relations between (Rummer's formulae), 101, 102 Sharpe's differential equation solved by, 105
; ; ;
gee Bessel functions with imaginary argument; Struve's functions with, 329, 332 Indefinite integrals containing Bessel functions under the integral sign, 132-138, 350, 581 ; tables of, 744, 745, 752 ; tables of (references to), 660, 661, 664 Inequalities satisfied by Bessel functions, 16, 31, 49, 255, 259, 268, 406; by Neumann's polynomial, 273, 282 ; by Struve's function, 328, 337, 417 ; by zeros of Bessel functions, 485, 489, 490, 492, 494, 515, 516, 521
under the integral sign, 383-449 (Chapter xm) ; discontinuous, 398, 402, 406, 408, 411, 415, 421 ; generalised, 441 ; methods of evaluating, described, 383 ; Bamanujan's type (integrals of Bessel functions with respect to their order), 449. See also under the names of various integrals, e.g. Lipschitz-Hankel infinite integral Infinity of the number of zeros of Bessel functions and cylinder functions, 4, 478, 481, 494, 495 Integrals, expressed in terms of Lommel's functions of two variables, 540 ; expressed in terms of the functions of Anger and Weber, 312 ; Fresnel's, 544, 545, 660, 661, 664, 744, 745 ; Gilbert's, 548, 549 values of, deduced from addition theorems, 367 ; with oscillatory integrands, 183 with the polynomials of Neumann and Gegenbauer under the integral sign, 277, 285. See also Definite integrals and Infinite integrals Interference, 229
Infinite Integrals containing Bessel functions
;
Interlacing of zeros of Bessel functions and of cylinder functions, 479, 480, 481 Irrationality of *-, 90, 485
Jacobl's transformation connecting sinnl with the (n - l)th differential coefficient of sin 2B_1 respect to cos d, 26 ; erroneously attributed to Bodrigues, 27 ; various proofs of, 27, 28
;
with
Kapteyn's domain K, 559 diagram of, 270 Kapteyn series, 6, 13, 551-575 (Chapter xvn) ; connexion with Kepler's problem, 551 ; expansions into, derived from Kepler's problem, 554, 555 expansion of an arbitrary analytic function into, 570; fundamental expansions into, 557, 559, 561, 564, 566, 568, 571 ; Kapteyn's domain K, of convergence of, 559 (diagram of), 270 nature of convergence outside and on the boundary of K, 574 ; second kind of, 572 Kapteyn's polynomial % (t), 568 ; expressed in terms of Neumann's polynomial, 569 Kapteyn's type of definite integral, 380 Kepler's problem, 6, 551, 554 ; Bessel's solution of, 13 ; Lagrange's solution of, 6 Kinds of Bessel functions, (first) 40 ; (second) 58, 63, 64, 65, 67 ; (third) 73 Kneser-Bonunerfeld expansion of a combination of Bessel functions as a Fourier-Bessel series, 499 Hummer's formulae connecting generalised hypergeometric functions, 101, 102
;
;
Lame functions, limiting forms expressed as Bessel functions, 159 Laplace's equation, general solution due to Parse val, 9 ; general solution due to Whittaker, 124 solutions involving Bessel functions, 83, 124 ; used to obtain addition theorems for Bessel functions,
127
GENERAL INDEX
Laplace's methods of approximating to functions of large numbers, 8, 421 Laplace's transformation, 280, 395 Large numbers, methods of approximation to functions of (Darboux), 233 ; (Laplace), 8, 421.
also
801
Sec
Large order, see Anger's function, Bessel functions whose order is large, Struve's function and Weber's function Lebesgue's lemma, see Riemann-Lebesgue lemma Legendre functions, Barnes' notation for, 156; integrals containing, 50, 173, 174, 339, 475; limits
(physical significance of), 155 of large degree, expressed as Bessel functions, 65, 155, 157 approximations to, 158 relation between two kinds of, 174 Whipple's transformation of, 387. See also Oegenbauer*s function Cn v (z) Lipschltz-Hankel infinite integral, 384 ; generalised, 389 Lommel's functions 8 Ml (z), p,i> (z), 345, 347 ; cases of expression in finite terms, 350 integrals special cases expressible by the polynomials representing, 346, 350 recurrence formulae, 348 of Gegenbauer, Neumann and Schlfifli, 350; special cases with /jlv an odd negative integer, 348 with large argument, asymptotic expansion of, 351 Lommel's functions of two variables, Uv (w, z), Vv (w, z), 537, 538 addition formulae for, 543 integrals representing, 540, 546 reciprocation formulae, 542 recurrence formulae, 539 special case of, 581, 752 ; tables of, 752 ; tables referred to, 660 with large argument, asymptotic expansions of, 549 Lommel's polynomial J?wt , (z), 294, 295 differential equation satisfied by, 297 ; Hurwitz' notation 9m, v (z), 303 limit of, expressed as a Bessel function, 302 of negative order, R^ m v (z), 299 recurrence formulae, 298 recurrence formulae in Hurwitz' notation, 303 ; relations with Bessel functions, 295, 297, 302 ; three-term relations connecting, 300, 301 zeros of, 304, 305, 306
of,
; ; ; ;
i;
Magnitudes of remainders in asymptotic expansions, 206, 211, 213, 236, 314, 332, 352, 449 MaTJitia of Bessel functions, 488; of FresnePs integrals, table of, 745 of integrals of Bessel func;
752 Mean anomaly, expansions of elements of an orbit in trigonometrical series of, 6, 13, 552, 554, 556 Mehler-Dirichlet integral representing Legendre functions, limiting form expressed as Poisson's integral, 157 Mehler-Sonine integrals representing Bessel functions, 169, 170 Meissel's approximations to Bessel functions of large order, 226, 227, 232, 247, 521 ; types of Kapteyn series, 557, 561, 564, 566 Membrane, vibrations of a circular, 5, 576, 618 ; vibrations of a sectorial, 510 Method of constant phase (Schlafli's), 216 Method of stationary phase, 225, 229 applied to Bessel functions, 231, 233
tions, table of,
;
Method of steepest descents, 235 applied to Bessel functions, 237, 241, 244, 245, 262 applied to functions of Anger and Weber, 316; applied to Struve's function, 333; connexion with Laplace's method of approximation, 421 Minima of Bessel functions, 488 of Fresnel's integrals, table of, 745 of integrals of Bessel func; ; ; ;
tions, 752
Monotonic properties of
Jv (vx)jjv (*),
257; of
Jv (v) and
,TV
'
(v),
260
of
(.r),
446
;
Multiple infinite Integrals, 450-476 (Chapter xrv) ; investigated by Neumann, 453, 470 (generalised by Hankel), 453, 456, 465; (generalised by Orr), 455; (modified by Weber), 468; RiemannLebesgue lemmas, analogues of, 457, 471 Weber's type of, 450
;
Neumann
series, 522-537 (Chapter xvi); expansion of an arbitrary analytic function into, 523; generalised, 525 ; (special series), 30, 31, 36, 69, 71, 151 Laurent's expansion, analogue of, 524; Pincherle's theorem on the singularities of, 526; special series, 18, 23, 25, 33, 34, 35, 128, 130, 138, 139, 140, 527, 581 ; Webb-Kapteyn (real variable) theory of, 533. See also Addition theorems
;
and Lommel's functions of two variables Neumann's factor eB (=1 or 2), 22 Neumann's Integral for <7m2 (z), 32 for Neumann's polynomial, 278, 280 Neumann's polynomial On (t), 271, 272, 273 connected with Kapteyn's polynomial, 569 connected with Neumann's polynomial O n [t), 292 connected with Schlafli's polynomial, 285, 286 contour integrals containing, 277 differential equation satisfied by, 276 expressed in terms of Lommel's
; ;
formerly called a Bessel function of the second kind, 67, 273 generalised by ; Gegenbauer, see Gegenbauer*s polynomial A n<v (t) generating function of, 281, 282 inequaliinfinite integrals containing, 433 Neumann's integral representing, ties satisfied by, 273, 282 278, 280; of negative order defined, 276; recurrence formulae for, 274 Neumann's polynomial On (t), 290, 291 ; expressed as integral containing Neumann's polynomial On (t), 292 Gegenbauer's generalisation of, see Oegenbauer's polynomial B n p, v (t) recurrence formula for, 292 Nicholson's infinite integrals, 431, 441
functions, 350
; ; ; ; ; ;
, ;
; ;
802
Nielsen-Hankel functions,
Null-functions, Lerch's theorem on integrals representing, 882 634, 636, 642, 647
Numbers, analytic theory of, associated with asymptotic expansions Numbers, Cauchy's, 324 recurrence formulae for, 325
; ;
200
Order of a Bessel function denned, 38, 58, 63, 67, 70 integrals with regard to, 449 Ordinary differential equations, see Differential equations Oscillation of solutions of linear differential equations, 518 Oscillations of membranes, 5, 510, 576, 618 of uniform heavy chains, 3, 4, 576 Oscillatory integrands, Du Bois Beymond's integrals with, expressed in terms of Bessel functions, 183
;
P-functions, limiting forms expressed as Bessel functions, 158 Parseval'8 integral representing J (z), 9, 21 ; modifications of, 21
Partial differential equations, see Differential equations Phase, method of stationary, general principles of 225, 229; applied to Bessel functions, 231, 233 Phase, Schlafli's method of constant, 216
,
Pincherle's
theorem on
Neumann
series,
526
Poisson's integral for Bessel coefficients, 12, 24, 25 ; for Bessel functions, 47, 48, 49 ; (generalised by Gegenbauer), 50 ; (symbolic form of), 50 ; for Bessel functions of imaginary argument, 80 ; for Bessel functions of the second kind, 68, 73 ; limit of the Mehler-Dirichlet integral for Legendre functions as, 157 ; transformation into contour integrals to represent Bessel functions of any order (of the first kind), 161, 163, 164 ; (of the second kind), 165 ; (of the third kind), 166, 167; (with imaginary argument), 171, 172; transformations of the contour integrals, 168, 169, 170. See also Parseval's integral and Struve's function
Polar coordinates, change of axes of, used to obtain transformations of integrals, 51, 374, 376, 378 used to express Bessel functions as limits of Legendre functions, 155
Probleme de moments of Stieltjes, 464 Products of Bessel functions, 30, 31, 32, 82, 146, 147, 148, 149; Bateman's expansion of, 130, 370 expansions of arbitrary functions into series of, 525, 572 integrals representing, 31 150, 221 438, 439, 440, 441, 445, 446, 448 ; series of, 30, 151, 152 with large argument, asymptotic expansions of, 221, 448 Products of Weierstrassian factors, Bessel functions expressed as, 497
; ,
Radius vector of an orbit, expansion as trigonometrical series of the mean anomaly, 6, 13,552,553,554 Ramanujan's integrals of Bessel functions with respect to their order, 449 Bamanujan's method of evaluating definite integrals, 382 Random flights, problem of, 419 Rank of Bessel functions and cylinder functions, 129 Real variables, expansions of arbitrary functions of, see Dini series, Fourier-Bessel series, Neumann series (Webb-Kapteyn theory), and 8chl6mllch series Reality of zeros of Bessel functions, 482, 483, 511 Reciprocation formulae for Lommel's functions of two variables, 542 Recurrence formulae for Anger's functions, 311 ; for Bessel coefficients, 17 for Bessel functions of the first kind, 45 ; for Bessel functions of the second kind, 66, 71 for Bessel functions of the third kind, 74 for Bessel functions with imaginary argument, 79 for Bourget's functions, 326 for Cauchy's numbers, 325 ; for cylinder functions, 82 for Gegenbauer's polynomials, 283 for
; ; ; ;
;
tions, 329 ; for Weber's functions, 311 ; for Whittaker's functions, 339. equations, Heml-cylindrlcal functions and Three-term relations Reduced functions, Cailler's, 536
of,
Stieltjes'
approximations
to,
213
of,
479
Riccati's differential equation, 1, 2, 85-94; connexion with Bessel's equation, 1, 90} equation cognate to, 91; limiting form of, 86; soluble cases .of (D. Bernoulli's), 85; soluble cases of, exhausted by D. Bernoulli's formula and its limit, 123 ; solutions by various mathematicians (D. Bernoulli), 2, 85, 89 ; (Cayley), 88 ; (Euler), 87 ; (Schlafli), 90 ; solved by means of infinite series by James Bernoulli, 1 ; transformations of, 86
equivalence
GENERAL INDEX
with the linear equation of the second order, 92 ; singularities bers (two, one or none) of quadratures, 3, 93 Riemann-Lebesgue lemma, analogues of the, 457, 471, 589, 599
of,
803
94
;
soluble by various
num-
Riemann's theorem on trigonometrical series, analogues for Schlomilch for series of Fouricr-Bcsscl and Dini, 649
Rodrigues' transformation, see Jacobl's transformation
series, 642,
647
analogues
Schafheitlln's discontinuous Infinite integral, 398, 402, 405, 406, 408, 411
71, 340, 343; addition theorems for, 344, 345; differential equations satisfied by, 342, 343; of negative order, 343 ; recurrence formulae for, 71, 342, 343
Tn (z)
and
Vn (z),
hypergeometric function, 90 polynomial .S' (t), 284, 286 addition theorem for, 289 connexion with Neumann's polynomial On (t), 285, 286; Crelier's integral representation of, 288; differential equation satisfied by, 285 expression by means of Bessel functions, 287; expression in terms of Lommel's function, 350 integrals evaluated in terms of, 350 recurrence formulae for, 285 Schlafli's solution of Riccati's equation, 90 Schlomilch series, 618-649 (Chapter xix) definition of, 621 ; definition of generalised, 623 expansion of an arbitrary function of a real variable into, 619, 623, 629 ; nature of convergence of, 637, 645 null-functions expressed by, 634 Riemann's theorem on trigonometrical series (analogue of), 642, 647; special cases of, 632; symbolic operators in the theory of, 626, 627; theory of functions of complex variables connected with, 623 uniqueness of, 643, 647
8chlafli's Schlafli's
; ; ; ; ; ; ; ; ; ;
Series containing Bessel functions, see Dini series, Fourier-Bessel series, series and Schlomilch series
sum of (greatest term method), 8 Sharpe's differential equation, 105 solution by generalised hypergeometric functions, 105 Sign of remainders in asymptotic expansions, 206, 207, 209, 215, 315, 333, 449 ; of Struve's function, 337, 417 Sine-Integral expressed as a series of squares of Bessel coefficients, 152 Singularities of functions defined by Neumann series (Pincherle's theorem), 526 of the generalised Riceati equation, 94 Smallest zeros of Bessel functions, 5, 500, 516 Sommerfeld's expansion, see Kneser-Sommerfeld expansion Sonine-Mehler integrals representing Bessel functions, 169, 170
Series of positive terms, approximation to the
; ;
generalised, 382
Bonlne's infinite integrals, 432 Spherical geometry used to obtain transformations of integrals, 51, 374, 376, 378 ; used to express Bessel functions as limits of Legendre functions, 155 Sound, Sharpe's differential equation in the theory of, 105 Squares of Bessel functions, see Products of Bessel functions Stability of a vertical pole associated with Bessel functions of order one-third, 96
Stationary phase, method of, 225, 229 ; applied to Bessel functions, 231, 233 Steepest descents, method of, 235 ; applied to Bessel functions, 237, 241, 244, 245, 262 ; applied to functions of Anger and Weber, 316 ; applied to Struve's function, 333 connexion with Laplace's method of approximation, 421 Stokes' method of computing zeros of Bessel functions and cylinder functions, 503, 505, 507
;
Stokes'
phenomenon
Struve's function H (z), 328 ; connexion with Weber's function, 336 ; differential equation satisfied by, 329 ; inequalities connected with, 328 ; infinite integrals containing, 392, 397, 417, 425, 436; integral representations of, 328, 330 ; occurrence in generalised Schlomilch series, 622, 623, 631, 645, 646, 647 ; of order (n -( $), 333 ; recurrence formulae for, 329 ; sign of, 337, 417 ; tables of, 663, 666-697; Theisinger's integral for, 338; with imaginary argument, 329, 332 ; with large
of,
332
Struve's infinite integrals, 396, 397, 421 Sturm's methods applied to determine the reality of zeros of Bessel functions, 483 ; of Lommel's polynomials, 304, 305, 306 ; applied to estimate the value of the smallest zero of Bessel functions and cylinder functions, 517, 518 Symbolic operators in expressions representing Bessel functions, 50, 170 ; in expressions representing solutions of various differential equations, 41, 51, 108; in the theory of Schlomilch
series,
627
804
Tables of Bessel coefficients (of orders and 1), 662, 666-697; (of order n), 664, 730-732 (with - n - ), equal 1 order and argument), 664, 746 of Bessel functions of the first kind (of orders n + J, and 664, 740-741 (of order i), 604, 714-729 of Bessel functions of the second kind (of orders (of order ^), 664, 714-729 ; (with equal order and (of order n), 664, 732-735 1), 662, 666-697 and 1), 662, 666-697 (of argument), 664, 747 ; of Bessel functions of the third kind (of orders "' "" ** " " order.Jj), 698-7i3;
.-,...
; '
"
integrals, and 1), 663, 666-697 tions (of orders order n and of order $, 664, 748-751
and functions
of integral
Tables (references to) of Airy's integral, 659 of Bessel coefficients and iunctions derivable from them, 654, 655, 656, 658 of Bessel functions (of orders n + , - n - h), 658, 659 (of orders I, 3), 659; (of orders J, ), 659; of Bessel functions of the second kind, 655, 656, 658; of Bessel functions of the third kind, 657 ; of Bessel functions with imaginary argument, 657, 658; of
; ; ;
;
Fresnel's integrals, 661 of integrals of Bessel functions and Struve's functions, 661 ; of LommePs functions of two variables, 660 ; of Thomson's functions ber r and bei x, etc., 658 ; of zeros of Bessel coefficients, functions and associated functions, 659, 660 Theisinger's integral representation of Bessel functions, 184 ; of Struve's and Weber's functions,
338
Thomson's
(Sir
;
sations, 81
William) functions, ber 2, beiz, 81 ; connexion with Bessel functions, 81 ; generalireferences to tables of, 658 ; squares and products of, 82, 148 ; with large argument,
asymptotic expansions of, 203 Three-term relations connecting Bessel functions, cylinder functions and Lommel's polynomials, 300, 301 Transcendants, elementary, definition of, 111 order of, 111 ; solutions of differential equations by, 112 Transitional regions associated with Bessel functions of large order, 248
;
Uniformity of convergence of Dini series, 601 of Fourier-Bessel series, 593, 594 of Kapteyn series, 575 ; of Schlomilch series, 632 Uniqueness of Fourier-Bessel and Dini series, 616, 649, 651 of Schlomilch series, 643, 647 Upper bounds, see Inequalities
; ;
;
Viscous
fluid,
motion
of, associated
Wave-motions, equation of, general solutions, 125 generalised to p dimensions, 128 ; used to obtain addition theorems for Bessel functions, 129 Waves, electric, 56, 226, 446 ; on water, and the method of stationary phase, 229 Weber's (H.) discontinuous factor, 405 Weber's (H.) infinite integrals, 391, 393, 395, 396; (discontinuous types of), 398, 402, 405, 406, 408, 411 Weber's (H. F.) function E (z), 308 connexion with Anger's function, 310 connexion with Struve's function, 336 ; differential equation satisfied by, 312 integrals expressed in terms of, 312 recurrence formulae for, 311 representation of Airy's integral (generalised) by, 321 tables of, see Struve's function Theisinger's integral for, 338 with large argument, asymptotic expansion of, 313 ; with large argument and order, asymptotic expansion of, 316 Weierstrassian products, expression for Bessel functions as, 497 Whipple's transformation of Legendre functions, 387 recurrence formulae for, differential equation satisfied by, 339 Whittaker's function v (z), 339 339 ; with large argument, asymptotic expansion of, 340
; ; ;
;
Wronskian determinant,
42, 76,
77
;
Zeros of Bessel functions, 477-521 (Chapter xv) computation of (various methods of), 142, 500, 502, 503, 516 inequalities connected with, limits of, rates of growth of, 485, 489, 490, 491, 494, 507, 513, 516, 518 infinity of, 4, 478 interlacing of, 479, 480, 481 ; non-coincidenee of (Bourget's hypothesis), 484 non-repetition of, 479 ; number of, in a strip of arbitrary width, 495 reality of, 482, 483; tables of, 664, 748-751 tables of (references to), 659 ; values of, 4, 5, 512, 516 with imaginary argument, 511 with unrestrictedly large order, 513, 516 Zeros of Lommel's polynomials (reality of), 304, 305, 306
;
;
Generalized
Hypergeometric
Functions
The theory
function
field of
mathematical statistics and the basic analogues of the Gauss functions have applications in the
is
also used in
number
theory.
treatment leads on from a discussion of the Gauss functions to the basic hypergeometric functions, the
Dr
Slater's
hypergeometric integrals, bilateral series and Appel series. This book was planned jointly with the late Professor
an extended revision of his Cambridge Mathematical Tract (1985) on the subject and Dr Slater has
W. N. Bailey
continued
it
as
incorporating in
the results of
many
of her
own
researches.