Sie sind auf Seite 1von 8

The Textual Tradition of Quintilian 10.1.46 f.

Author(s): M. Winterbottom
Source: The Classical Quarterly, New Series, Vol. 12, No. 1 (May, 1962), pp. 169-175
Published by: Cambridge University Press on behalf of The Classical Association
Stable URL: http://www.jstor.org/stable/638040 .
Accessed: 22/06/2014 03:18

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Cambridge University Press and The Classical Association are collaborating with JSTOR to digitize, preserve
and extend access to The Classical Quarterly.

http://www.jstor.org

This content downloaded from 62.122.76.45 on Sun, 22 Jun 2014 03:18:00 AM


All use subject to JSTOR Terms and Conditions
THE TEXTUAL TRADITION OF QUINTILIAN
Io.I.46 f.

THIS article does not set out to cast doubts on the established textual basis of
the bulk of Quintilian's Institutio,or on the history of the work's fortunes in the
Middle Ages.' What I say about these things will be unoriginal and, I hope,
uncontroversial.My object, however, is to show that what is true of the bulk is
not true of Io. I. 46-131 ; and to fill in some details in the history of the tradi-
tion.
The Institutioas a whole is transmitted to us by way of two hyparchetypes.
One, which we may call f3,was at some stage mutilated so badly that perhaps
a third of the text disappeared, including, in Book io, the first 107 sections of
the first chapter. What remained can be reconstructedfrom a ninth-century
manuscript, the Bernensis (B), with the help, primarily, of the slightly younger
Nostradamensis(N). A transcriptof the Bernensis,the Bambergensis(Bg), was,
perhaps in the early tenth century, converted into a hybrid representativeof
the other, unmutilated hyparchetype (a) by receiving corrections (b) and
supplements (G), probably from one and the same source. The only other
witness we have to the second hyparchetypeis the ninth-centuryAmbrosianus
(A), which is very similar to, but not identical with, the manuscript that
generated b and G. Unfortunately A, though originally complete, had by the
fifteenth century lost a good deal of Books and the result is that the
celebrated sections on Greek and Latin literature
10-12-:
beginning at io. I. 46 and
running on to the end of that chapter are found complete in only one of these
primary manuscripts of the Institutio-namely G. The f-stream mutili contain
merely io. I. Io07-31.
It is a singular fact that though G generated at least four other complete
Quintilians before 1400, neither these nor G itself nor A were generally known
to the articulate reading public before the Renaissance. Scholars like Petrarch
had merely mutili from the P-stream: they could only pray for a complete
text, and the prayer was not answered until Poggio's discovery at St. Gall.
All the same, it is quite certain that early humanists like Salutati knew io. I.
46 f.; and it is generally agreed that they possessed it in the form of unab-
breviated excerpts, that circulated either separately or tacked on to the end of
mutilated Quintilians. What has never been established is exactly what we
can say about the relation of these excerpts to the tradition of the bulk of
Quintilian; and that is what this article will try to show.

G-VERSION AND S-VERSION


The text of Io. I. 46-107 found in G and its descendants (which I shall call
the G-version) has, among many other errors, a number of characteristic
omissions-e.g. 55 nulla-adfectus:56 quid?Euphorionem-Vergilius. Contrasting
sharply with this version is the text found in the excerpts I have mentioned:
I Forthissee esp. P. Lehmann,Philologuslxxxix (1934), 349 f. : P. S. Boskoff,Speculum
xxvii
(1952), 71 f.

This content downloaded from 62.122.76.45 on Sun, 22 Jun 2014 03:18:00 AM


All use subject to JSTOR Terms and Conditions
170 M. WINTERBOTTOM
this I shall call the S-version (after the now lost fifteenth-century Codex
Argentoratensis,which, under the name S, was used by Halm, and which, as
his apparatus clearly shows, contained this sort of text); it is marked by, for
example, the omission of dialogisve-illeat Io7. Mistakes common to both ver-
sions include 104 remutifor Cremuti.
Clearly, any critical edition of the passagemust take account of both versions.
Thus Halm used on the one hand G, and on the other various fifteenth-century
manuscripts influenced in one way or another by the S-version.Radermacher,
the only later scholar' to edit the passage on a new critical basis, used a slightly
different set of Renaissance manuscripts to balance G, and added what he
called E, the text of Io. I. 46-1o7 found separately at the end of Stephen of
Rouen's epitome of Quintilian in Paris. lat. 14146, the twelfth-centuryPraten-
sis. Since E, too, gives the S-version, Radermacher triumphantly asserted that
this proved what he held on other (no less disputable) grounds, that at least
some fifteenth-centuryQuintilians are descended from earlier sources now lost
to us. ('Quid multa?' wrote Radermacher Vol. ii. praef. IV, oddly ignoring
his own use of E: 'Codices recentesnisi possideremus,decimus Quintiliani liber
ille celeberrimus etiam plures lacunas haberet quam nunc habet.') I shall
show later that this is true even of io. i. 46 f. only in a limited and uninteresting
way; and I hope to show elsewhere that of the great bulk of Quintilian's text
it is not true at all.

THE S-VERSION: Six EARLY REPRESENTATIVES

Our first concern must be to establishwhere the S-versioncan be found in its


purest form. It appears (to my knowledge) in six extant pre-Renaissance
manuscripts, which fall readily into three pairs.
(i) E (already mentioned), and what I call D, Paris. lat. 7719, where, fol-
lowing a P-stream mutilated but unepitomized text of the Institutio,we have,
transcribed separately, io. I. 46-107, together with (as in E) 12. IO.
IO-15.
It is known that D was copied from, and E based on, the lost Beccensis,zand so
we can with some certainty say of the Beccensis that it too had these two
extracts bound up with it. Collation of the excerpts confirmsthat D and E are
gemelli. Peculiar errorsof DE include 49 innuit(for nuntiat); 73 praecipuare (for
praeclare);82 omission of comoediae.
E (but not D) omits docereat 78 (D actually
has dicere)and veterum at 97 (D has verum).D (but not E) omits diligenterat 69
and oratorat 8o.
(2) The same excerpts appear at the end of two (perhaps twelfth-century)
manuscripts at Leyden, p (Voss. lat. q. 77) and a (Voss. lat. f. 80). The main
text in both these manuscripts does not,as in the Beccensis, finish at Io. 3- 32,
but at 12. 10. 43, as in most P-stream manuscripts; and collation of the excerpts
shows that p and a are independent of, though related to, the Bec excerpts.
Peculiar errorsof pa include the omission offormandamensat 59 and of Cratinus-
queat 66. Errorsshared with DE include the omission of in bucolicisat 56, esse
Peterson's edition of Book Ten, despite and there is no formal apparatus.
his elaborate manuscript discussion and 2 See esp. Peterson, Introduction, p. lxi.
critical notes, is eclectic rather than critical:

This content downloaded from 62.122.76.45 on Sun, 22 Jun 2014 03:18:00 AM


All use subject to JSTOR Terms and Conditions
THE TEXTUAL TRADITION OF QUINTILIAN io. I. 46f. 171
at 57, and tenebrasat 72. pa are gemelli. p (but not a) omits longeat 73. a (but
not p) omits longemagisat 7o. I suggest then the following stemma:

[Becc]

D E

(3) Finally we have two slightly longer sets of excerpts, 10. I. 46-I 31 and I2.
10. o0-15, contained among other rhetorical material (not the same in both
books) in Paris. lat. 7696 (which I call X) and 7231 (which I call Y). X is from
Fleury: neither X nor Y is later than the twelfth century. X and Y avoid all
the errors I have cited above from DEpa, whether jointly or separately; but
they have some (not many) errors of their own where GS gives the correct
reading: e.g. 61 credit(for credidit); 96 longofor longe.At 84 X omits, Y includes
valuerunt(cf. also I04 X excitamus:Y, correctly, excutimus). At 78 Y, but not X,
omits si; further,Y has at 46 deditexemplum et ortum,an inversion avoided by X;
and there are other similar cases. X and Y then, too, are gemelli, and we can
add them to the stemma (it is not, I think, quite out of the question that both
X and Y are independent transcriptsof y, but practically speaking this would
make little difference to an apparatus).'

x Y [Becc]
p 0-

D E

THE S-VERSION:ITS SOURCE

The Beccensis, as we have seen, in all probability contained


Io. I. 46-1o7
and 12. I0. 10-15 separately transcribed at the end of a mutilated a-stream
manuscript of the Institutio.The same can be said of the common ancestor of
the Vossiani; and so, therefore, of 8. And collation of the main bulk of D, p,
and a shows that they, like the excerptsappended to them, go back to one single
Some of these conclusions were arrived about the Vossiani, and he did not realize
at by C. Fierville in his edition of Book One the superior quality of XY. Nor did he draw
(Paris, 189o): see especially his stemma on any wider conclusions.
p. lxxxvi. But he was badly misinformed

This content downloaded from 62.122.76.45 on Sun, 22 Jun 2014 03:18:00 AM


All use subject to JSTOR Terms and Conditions
172 M. WINTERBOTTOM
lost manuscript (a copy, very likely, of the hyparchetype/: this account differs
from the one given by Max Niedermann in the introduction to his edition of
the grammatical chapters of the first book of Quintilian; but it accounts as
well as his for the evidence he adduced and better than his for evidence that he
did not cite). We must now ask about 8 why it contained these particular added
excerpts; and where it got them from (8 itself may have been copied, excerpts
and all, from some other lost manuscript; but to say this is only to take the
problem a furtherstage back). 8, like all the mutili, omitted the passage 9. 3. 2-
10. I. 107; anyone therefore who had access to io. I. 46-10o7 had a clear motive
for adding the passage to supply part of this lacuna. But 8 contained (forp and a
have it, though the Beccensis did not) 12. 10 up to ? 43: why then did the
supplementer of 8 copy out the unwanted 12. 10. o0-15 after the sections from
the tenth book? The answer must surely lie in the nature of his source. If he
had been drawing on a manuscript of Quintilian that contained, in addition
to the normal mutilus text, io. I. 46-o107, he would have copied out that
passage but notalso the passage from 12. His source then must have been some-
thing that juxtaposed Io. I. 46 f. with 12. 10. 10-I5. But io. I. 107 is by no
means a natural stopping place; it is midway through an elaborate (and for the
Middle Ages important) discussion of Cicero. It is not clear why any manu-
script source should have contained together with six sections from I2 merely
a truncated fragment of the discussionof Greek and Latin literature. Is it not
much more likely that the source juxtaposed the complete passage 0o. I. 46-131
with the sections from I2: the two together forming a handy epitome of
Quintilian's judgements on literature? And when we observe that this is pre-
cisely what X and Y do, and when we remember the relationshipsestablished
by collation, we can only conclude that y was a set of excerpts of Quintilian
containing all that X and Y now contain; and that whoever supplemented
8 was using y or a descendant of y. He correctly observed that io. i. io8-
31 was not needed in 8: he failed to notice that equally 12. 10. 10-15 was
not wanted either; but because it was there in his source he copied it out
again.
y, then, was a set of excerpts of Quintilian. From what manuscript was it
taken ? Here we have the guidance of the two passages, io. I. 108-131 (found
in XY) and I2. 10. o0-15 (found in XYDEpu), where the excerpts overlap
with the oldest #-stream manuscripts,B and N, from which, for most practical
purposes, the hyparchetype of this stream is to be reconstructed. Now the
excerpts can be classedwith the ~-streamagainst the a-stream (here represented
only by the corrector of the Bambergensis, b): see, for example, 12. 10. 14
alienigenam etrum BN: alienigena metrumXYDEpu: alienigenamet parum b, cor-
rectly; io. I. og9extulitom. BNXY: hab. b. But there are some mistakes of
BN that the excerpts avoid: e.g. 10. I. 121 explicandobXY: plicando BN; 12. 10o.
Io velis bXYDEpr: velit BN; I2. 10. 13 ingenii affluentia bXYDEpr: ingenia
fluentiaBN. y, then, cannot be identified with the common parent of BN: and
so we have in y something that takes us back beyond the mutilated hyparche-
type-back, perhaps, to a time when the a-stream was not mutilated at all, or
much less mutilated (it would be very odd if, when the original excerptor was
at work, the lacuna ceased just at io. I. 46, at the start of so interesting and
important a passage). y, then, is a primary witness to the passages it covers;
it can be reconstructed for 10. I. 46-Io7 and 12. 10. IO-I5 from XY3: for IO0.I.
Io8-31 XY are the only witnesses. Thus the S-version has the same sort of

This content downloaded from 62.122.76.45 on Sun, 22 Jun 2014 03:18:00 AM


All use subject to JSTOR Terms and Conditions
THE TEXTUAL TRADITION OF QUINTILIAN 10. 1. 46f. 173
high authority against the G-version that in other parts of the Institutiois given
to the f-stream against the a-stream.

ARCHETYPE

G-b

Independent excerpts 10.1.46-131/12.10.10-15


"

( Excerpts 10.1. 46-107: 12.10.10-15


used for supplementation purposes
B N

THE RENAISSANCE: EXCERPTA

Even in the fifteenth century we find Io. I. 46-107 sporadically leading a


separate life alongside the complete texts of Quintilian now available. These
late excerpts, as we should expect, give a pure S-version. There is, for example,
the incomplete excerpt Io. I. 46-91 (up to doctius at the end of a folio) in
Bodley's MS. Rawlinson 48, which is dated as late as 1439. More interesting
still, in Vaticanus lat. 1807 we have io. I. 46-107, and, obviously to fill the
page, 12. 10. io-i I, ending in mid-sentence at Calvi. Closely related to this is
the Naples codex V.B.34, where the same material reappears, here ending in
mid-page as well as mid-sentence (suggesting very strongly that this is a copy
of the Vatican fragment). To the Naples extract has been added in a contem-
porary hand a note in terms that suggest its writer had no access to a complete
Quintilian but only to a mutilus: 'huic profecto coniungendus est textus ille
qui nostris codicibus prius habebatur.' In view of the fragment from 12 there
can be little doubt that the latter two excerpts descend from excerpts of the
DEpa variety. And collation, while confirming this, shows that the Bodley
excerpt is closely related to the other two.' These fifteenth-century excerpts
have great affinity with DEpcr when those manuscripts agree in error; some-
times strikingly with pa against the rest (e.g. 51 clarissimapa c. 15 exc.: durissima
GXYDE correctly); sometimes, no less strikingly, with DE against the rest
(e.g. 54 superatumDE c. 15 exc.: superari GXYpc correctly). Whatever their
exact relation to the stemma (and they may well be based on fourteenth-
century conflation of excerpts of the DE and the pa variety) they are clearly
technically eliminable, except as a possible source of conjecture; this enables
us to keep out of the apparatus such blots on Radermacher's as 46 tum brevitate
tumrcopia,.52 lenitasque,and 56 frustra transibimus,all readings that appear first
in these excerpts and in fifteenth-century complete Quintilians (see below)
that are clearly indebted to them; and that, if they go farther back, reflect
merely an eliminable relation of excerpts we already possess.
' It
may be added that a was almost cer- supplement what appears to be a copy of the
tainly known in the fifteenth century: Nostradamensis (or perhaps of the Sal-
Johannes Poulain copied it, I think, into his mantinus).
manuscript that is now Paris. lat. 7721, to

This content downloaded from 62.122.76.45 on Sun, 22 Jun 2014 03:18:00 AM


All use subject to JSTOR Terms and Conditions
174 M. WINTERBOTTOM

THE RENAISSANCE:COMPLETETEXTS

Examination of 0o. I. 46 f.' in complete fifteenth-centurytexts of Quintilian


does nothing to upset the conclusion that the Renaissance has only conjecture
to offer here. There are three main types of text:
(I) G-versionspure and simple, found in manuscriptsdescended from T (and
so from G).
(2) S-versionspure and simple, found for instance in the lost Argentoratensis
and in manuscriptsrelated to Guelferbytanus.In these manuscripts,whatever
the source of the rest of the text, the version given of Io. I. 46 f. is closely related
to that found in the fifteenth-centuryexcerpts discussed above. This is easily
explicable on the hypothesis that at some stage the excerpt from Book Ten was
added to a fuller text of Quintilian at the appropriate point: either a mutilus
text, or, after Poggio's discovery of T, a complete text. Such an addition would
be the natural thing to do (even after the discoveryof T, because T's G-version
is hopelessly corrupt).
(3) Contaminated texts. These of course predominate, but we can dis-
tinguish two by no means all-inclusive types:
(a) Texts where an original G-versionhas been overlaid by c. 15 corrections
from an S-version. These can be diagnosed on the assumption that
correctorstend to fill gaps and alter words but not, on the whole, unless
they are particularly conscientious, correct inversions. Among texts of
this kind, we may place P and the numerous manuscripts descended
from it.
(b) Texts descended from F, where the G-version (descended from G itself
via H) was, perhaps before the c. 15, corrected from an S-text. Among
such texts are L, the Lassbergensisused by Halm.
There is thus a very good reasonwhy c. 15 manuscriptsoften have versionsof
Io. I. 46 f. superior to that of the earliest available, G. They are making use of
a tradition of excerpts outside, though intimately connected with, the general
textual tradition of Quintilian. But we do not need their assistance, because
pre-c. 15 excerpts, nearer the source than they, are still extant. No doubt the
actual excerpts that generated the c. 15 versionsdirectly are not now available;
but we know all we need to know about them. Far more important, we can now
see clearly that, even if in this limited sense some c. 15 manuscriptshave access
to sources lost to us in Io. I. 46 f., this proves absolutely nothing about the
general question of the relationship of c. 15 Quintilians to earlier complete or
mutilated texts.2

i The S-version of 12. 10. Io-15 was not texts. In fact, however, the limits can be
influential in the fifteenth-century (or any narrowed still further, to those that I have
other) texts: naturally, because even the suggested--Io. I. 46-107. Outside these,
mutilated texts had included the passage (in Kroll cites only two passages, 9. 4. 144 and
a less corrupt form) in the body of the book. 10. I. 42: in both of these, Renaissance con-
2
Radermacher, in fact, could have saved jecture is clearly at work. Kroll calls his
himself much trouble by pondering on a dis- demonstration 'eine Warnung fUr alle
cussion by W. Kroll in Satura Berolinensis Stammbaumfanatiker'. But in fact a stemma
(1924), pp. 61 f., where it is suggested that is clearly profitable in sorting out the com-
only between 9. 4. 135 and Io. i. io8 is there plexities of this part of the tradition.
any sign of an independent tradition in c. 15

This content downloaded from 62.122.76.45 on Sun, 22 Jun 2014 03:18:00 AM


All use subject to JSTOR Terms and Conditions
THE TEXTUAL TRADITION OF QUINTILIAN io. I. 46f. 175

SOME CONCLUSIONS
The resultsarrived at above can be used materially to simplify the apparatus
of Io. I. 46- 07; in particular they would enable an editor to avoid the main
fault in Radermacher's, by allowing him (and the reader) to distinguish
between variants and remote sub-variants.A considerablenumber of readings
quoted by Radermacher can be formally eliminated: and c. 15 manuscripts
can be ignored completely, except for one or two conjectures (e.g. 47 nonus).
Apart from this, the text can be based on G, X, and Y, with DEpa as a control
to establish peculiar errors of XY (and such peculiar errors are notably few:
for example, in Io. I. 46-59 there are none; DEpa in the same sections share
no less than thirteen, besides many more private to each of them).
Thus, in the apparatus to 46 alone: hic and caperecan be given the symbol y
alone. E's omission of rebus,its superavitand its poeticaeare all eliminable errors.
The order brevitatetumcopia,not in itself preferable, has the support neither of
G nor of y and need not be mentioned.
I add some points of interest from other sections:
50 The order PriamirogantisAchillemis supported by GXY and so should
be read in the text.
52 medio,besides being correct, is supported by Gy.
55 Read paremcrediditsupported by GXY (XY credit).
59 XYpa read exi** with lacunae of varying length (exitumD: exemplum E;
two bad conjectures) and so all but anticipate Philander's correction

70 oratoriis,given no support at all by Radermacher, is read by Y (X


oratioris).
79 compararat, P2's c. 15 correction, appears, as a correction, in X.
8o mediocriis supported only by DE, and besides being wrong is thus
eliminable.
88 hocis supported by Gy: G's quathen can safely be corrected to quo,the
reading of all the excerpts except E.
90 senectutematuruit, given no supportat all by Radermacher,is read by XY.
94 non labor G: nisi laborXYpa: mihilaborDE. y clearly read nisi labor,and
this is very probably right.
99 XY correctly read aeli stilonis.
102 clarivir ingenii,attributed by Radermacher to 'ed. vett.' is read by XY.
126 XY agree with b in omitting (surely correctly?) the awkward in before
quibus.
ChristChurch, Oxford M. WINTERBOTTOM

This content downloaded from 62.122.76.45 on Sun, 22 Jun 2014 03:18:00 AM


All use subject to JSTOR Terms and Conditions

Das könnte Ihnen auch gefallen