Immortality and the Nature of the Soul in the "Phaedrus"

Author(s): Richard Bett

Source: Phronesis, Vol. 31, No. 1 (1986), pp. 1-26
Published by: BRILL
Stable URL: http://www.jstor.org/stable/4182241 .
Accessed: 22/11/2014 11:06
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.

BRILL is collaborating with JSTOR to digitize, preserve and extend access to Phronesis.

http://www.jstor.org

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

Immortalityand the Nature of the Soul in the

Phaedrus
RICHARD BErr

"Firstwe mustinquirewhatkindof soul the discussionis about".So begins

Hermeias'commentaryon the argumentfor immortalityin the Phaedrus
(245c5-246a2).'It is a piece of advicethat has not, I think,been sufficiently
heeded by modern scholars. There has been some discussionof various
textualproblemsandpointsof translation,andthishas, of course, involved
some scrutinyof the argument'sstructureand presuppositions;but I am
not aware of any really detailed, step by step analysisof the argumentotherthanHermeias'own.2As a result,the preciserelationbetweenPlato's
views on immortalityand the natureof soul in the Phaedrusand in other
dialogueshas not received as close attentionas it could. In particular,the
questionof whatconceptionof soul Platois operatingwithin this argument
has tended to be dealt with too superficially- by examinationof single,
I

HermeiaeAlexandriniin Platonis PhaedrumScholia, ed. P. Couvreur(Paris, 1901)ad.

loc.

The most detailed discussions I am aware of are in R. Hackforth, Plato's Phaedrus,

(Cambridge, 1952), pp. 64-8, and T.M. Robinson, "The Argument for Immortalityin
Plato's Phaedrus",Essays in Ancient GreekPhilosophy, ed. J.P. Anton and G.L. Kustas
(Albany, 1971). But both of these, while they do cover the main steps of the whole
argument, deal with most of them extremely briefly; in addition, Robinson's analysis
seems to me mistaken in several places. Among other writersand commentatorson the
Phaedrus, W.J. Verdenius ("Notes on Plato's Phaedrus", Mnemosyne, Series 4, 8
(1955), 265-89) and G.J. de Vries (A Commentaryon Plato's Phaedrus, Amsterdam,
1969) offer only isolated remarks, mostly on passages where the text is in doubt. I.M.
Crombie (An Examinationof Plato's Doctrines, London, 1962) in a long and exhaustive
chapter on Plato's views on the soul (Vol. I, ch. 7, "The Philosophyof Mind"), devotes
only a page and a half (325-7) to the argument,mostly at a very abstractlevel; andMartha
Nussbaum, in a discussionof the Phaedruswhich covers almost every major topic in the
dialogue ("'This Story Isn't True'; Poetry, Goodness and Understanding in Plato's
Phaedrus", Moravcsik and Temko (eds.), Plato on Beauty, Wisdom and the Arts,
Totowa, N.J., 1982), dismisses it in literally a sentence (pp. 106-7).

Phronesis 1986. Vol. XXXIII (AcceptedJuly 1985)

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

troublesome phrases, rather than by assessment of the argumentas a

whole.
My purpose,then, is to analysethe Phaedrus'argumentfor immortality
as minutelyas I can. This argumentis of considerableintrinsicinterest,and
I shallbeginby treatingit largelyas an isolatedpiece of logic, examiningthe
natureand cogency of the variouspremisesand inferences.However, the
argumentmust also, of course, be seen in its context. By examiningit in
detail, I hope also to clarify, to some degree, the place of the Phaedrus
among Plato's works - or at least, among those works that deal with
questionsaboutthe soul and immortality.In addition,I hope to arriveat a
better understandingof the argument'splace in the Phaedrusitself. For
another neglected issue, I believe, is that of the relation between the
argumentfor immortalityand the mythof the charioteersthatimmediately
follows it. Typically, the two sections have been discussedtoo much in
isolation from one another;yet clearly the natureof the soul is a central
topic for both of them.
That Plato intends the argumentas a rigorous proof, and hence as
admittingof the detailedexaminationI intendto give it, is not, I think, in
doubt. In this respectit contrastssharplywith the myth that follows, and
Plato purposely signals the contrast in two ways.3 First, there is the warning

at 246a6 (immediatelyprecedingthe myth) that what follows is a description of the soul only T EOLXEV,
not o[ov {GTL;the proof has no such
qualificationsattachedto it. Secondly, the styles of the two passagesare
very different. The proof is presentedin extremelyspare, choppyprose;
Platohere seems to be aimingfor maximumclarityandlogicalperspicuity.4
3 This contrast should not be taken to imply that the proof is more importantthan the
myth, or that the myth is not "real philosophy". In this connection, it is interestingthat
Plato uses the word &x6bettg (245c1,4; cf. &no6eLXtwoV,b7) to refer not simply to the
proof of immortality,but to the entire ensuing discussion;the "demonstration"is of the
fact that the madness of the lover is divinely inspired, and so encompasses the myth as
well as the proof. In Plato's view, then, mythand proof are equallyvalid waysof showing
things- though of course, each may be appropriatein different circumstances.
4 Some (e.g. de Vries, op.cit., following Denniston) have likened the style of the proof
(in additionto its thought- but that is anothermatter)to that of certainPresocratics.The
similaritymay be there, and may be intentional;but this does not, I take it, detract from
the plausibility of my suggestion here - there may be more than one reason for his
adopting the style that he does. Raphael Demos (in "Plato's Doctrine of the Soul as a
Self-Moving Motion", JHP 6 (1968), 133-45) goes so far as to suggest, largely on the
groundsof this stylisticdifference, that the argumentis a laterinsertion.This seems to me
quite unnecessary;it is not at all unusualfor Plato to change his style radicallybetween
adjacent passages.

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

When he embarks on the myth, his language loosens up, and seems
positively florid compared with the rigor and economy of the previous
passage. In fact, of all the argumentsfor immortalityin Plato, this one is
muchthe closest to whatwe wouldconsidera formalproof. So in dissecting
it, we are not approachingit in a spiritanydifferentfromthatin whichPlato
composed it.
I
The basic shape of the argumentis as follows:
1) Soulis thatwhich is its own source of motion.
2) Thatwhichis its ownsourceof motionis immortal.
Therefore3) Soulis immortal.

Plato presentsit, though,in the reverseorder.The conclusion3) is statedat

the beginning(245cS);the argumentfor2) occupieshimfromthereuntile2,
and 2) is stated at e2-3; 1) is then arguedfor between245e3and 246al; and
finally, at 246al-2, the conclusionof the whole argumentis restated.The
argumentfor 1) is relativelystraightforward,andwe shalldeal withit later;
2), however, is establishedin a muchmore complexfashion, by meansof
two independentsub-arguments.The premisesof the first sub-argument
(which I shall label A) are as follows:
Ai) Thatwhichis its own sourceof motionis always in motion.
Aii) That which is always in motion is immortal.

The premisesof the second sub-argumentB are:

Bi) that which is its own source of motion is a source of motion for everythingelse
that moves.
Bii) that which is a source of motion for everything else is ungenerated and
imperishable.

A is also presentedin the reverseorder,but B occursin the orderin whichI

havejust exhibitedit.5I shallnow analysethe stepsone by one, in the order
in which Plato gives them, beginningwith sub-argumentA.
I

This analysisof the argument'sbroadoutline agrees most closely with that of Hackforth
(op. cit.). The only point at which I would take issue with him is that he regards the
second of the two argumentsfor 2) as subordinateto the first;it seems to me that the two
are parallel and equally important. Hermeias's reconstructionis also roughly in agreement with my own (see p. 104.4-12 for his introductorystatement of the premises). He
thinks that the argumentas a whole divides into two; in his view, the first sub-argument
runs (using my symbolism) 1), Ai), Aii), therefore 3), and the second (roughly) 1), Bi),
Bii), therefore 3). (This is not quite right, since he furthersubdivides my Bii), and his

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

We begin with a statement of Aii) - TO yaQ a'ELXLVTOV aOtvacVov

(245cS). No argumentis offered for this premise;it seems to be simply
taken for granted.6Presumablythe underlyingthoughtis that something
which is always in motion must always exist, and so must be immortal.
Provided"alwaysin motion" means "in motion for eternity"(on which
more below), this seems hardto quarrelwith. It is only when we move to
Ai) that the real argumentationbegins.
The effect of Ai) is to identify that which is always in motion with that

whichmoves itself;given Aii), this establishesthe main premise2). Plato

now states Ai) at 245c7-8 - "only that which moves itself . . . never ceases

moving"- prefacingit with what amountsto a statementof its converse"that which . . . is moved by something else, since it admits of a cessation
of movement, admitsof a cessationof life" (c5-7).7 At the same time, he

offers a reason for the assertion;this is containedin the terse and cryptic
aTr6O(c7-8).
phrased-E o"uxaunoXkiLnov
view of the logic at that point is slightlydifferentfrom mine; on this, see furthernote 14.)
The main difference, then, is that he does not include Plato's statement of my main
premise 2); but this difference is not important, except from a strictly formal point of
view. The analysisof Robinson (op. cit.) is rathermore distantfrom mine. Partlyfor this
reason, it would be a somewhatarduoustask to criticiseit directly;I prefersimplyto offer
my own rival interpretation,and hope that it prevailson its own merits.
and
6 There has been much division of opinion over the alternativereadings&ELXt(v-Tov
of
basis
the
on
decided
be
cannot
matter
the
me
that
to
seems
It
in 245cS.
aU1roxLvrlrov
their relative appropriatenessto the argumentas a whole; either readingwould yield a
closely-knit logical progression. It is true that the readinga1'Tox'vToV makes for some
repetition; but this would be quite tolerable - it by no means suffices (as J.B. Skemp
thinks - see The Theoryof Motion in Plato's LaterDialogues, Cambridge, 1942, p. 3, n.
2) to rule that readingout. Hackforthdefends &ELXCViTov on the groundsthat it gives us
a statement of an ?vbo?ov - that is, a premise which could be expected to find general
acceptance;but againstthis, Ackrill (Mind62 (1953), p. 278) seems rightto point out that
an argument does not need to begin with a statement of some premise that is
uncontroversial. Logical considerations aside, however, the evidence for &ElXiVoTOV
seems to me entirely superior. The mss. agree on &etx(v-9TOV; aUToxLvTjTov is found

only in one Oxyrrhynchus papyrus (1017). In addition, Hermeias clearly had

the passage for the
&tLXLvItOV in front of him, and so did Cicero when he translated
SomniumScipionis (a section which he also quotes in TusculanDisputations,1, 54); the
Latin is "quod semper movetur, aeternumest". For a full discussionof the issue, which
supportsthis general position, see F. Decleva Caizzi, 'AEIKINHTON o AYTOKINHTON?', Acme 23 (1970), 91-97.
7Thus c5-7 (T0 b'&Xo xtvoi3v, etc.) seems to me to look forward to the following
sentence,

rather than back to T6 y'Q

&ElLX1VqTOV

&06vaTov,

despite the evident

QVa
parallelism between the pairs &eLx(v1qTov/&0vaTov and 7tafOav XLV 'EJOW
4X95. Burnet's punctuation suggests the opposite. (Translations,here and elsewhere,

are my own.)

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

Whatis meant by alE oUx &3toXkE'Lov

a'vTo? Literally,the phrasemust
mean something like "in as much as it does not abandonitself"; and the
naturalway to understandthis is as saying that for a self-moverto cease
movingwould be for it to abandonits own nature. It is, then, an essential
propertyof a self-movingthing (to speak in a later, but surely an appropriate, terminology)that it be movingitself at any given time;self-moving
things are necessarilyin constantself-motion. Two questions now arise.
First,why does Platothinkthatthis is so? And second, if we acceptthatit is
so, does this sufficeto show thatself-movingthingsare eternallyin motion?
On the firstpoint, Platooffersus no furtherenlightenment;butperhapsthe
underlyingidea is somethinglike the following.Supposethat a self-mover
were to cease to move itself. Then it could not start moving itself again
unlessgalvanisedinto motion eitherby itself or by somethingelse. But if it
were galvanised into motion by itself, this would have to occur by the
agency of some part of itself that was alreadyin motion - in which case,
contraryto hypothesis,the thingwouldnot reallyhaveceasedto move itself
after all. But if by somethingelse, it could no longercount as a self-mover.
Thus somethingwhich trulymoves itself must move itself continually.8
Supposethatwe acceptthisconclusion(whetheror not Platoarrivedat it
in the way I just proposed).The otherquestionis whetherthis is equivalent
to sayingthatsomethingwhichmovesitselfdoes so eternally;andto thisthe
answer is clearly "no". As we saw, a straightforwardunderstandingof
premiseAii) requiresthat we take &ELXiLVTOV in c5 as meaning"eternally
in motion";so if the argumentis to work, oThnoTE XT'1yEL
XLVOV?CVOV in c8
must presumablyhave the same sense. However, it is not in this sense that
Plato has shown us that self-movers "never cease moving". If being in
constant self-motionis an essential propertyof a self-mover,then a selfmover cannotcease to be in motionandstill be a self-mover;but this is not
If this were the line of thought that was motivatingPlato - and it is my best conjecture
on the subject - then it would in one respect anticipateAristotle's ideas on self-motion.
For on this conjecture, Plato is led very naturallyinto thinkingof self-movers as consisting of (at least) two parts, the active and the passivepart, so to speak. Aristotle's analysis
of the concept of self-motion (Physics Bk. VIII, ch. 5) makes central use of just such a
division;a thing which moves itself must, he thinks, consist of a partwhich is moved and a
part which causes this motion. Aristotle argues convincinglythat this latter part cannot
itself be in motion, which leads him to the concept of the unmoved mover; it turns out,
then, that, contraryto Plato, the ultimatecause of motion is not a self-mover. However, a
picturein which Aristotle developed beyond Plato, havingbegun by holdingideas similar
to his, is presumablyone we are bound to adopt in any case; and the development would
perhaps be a little smoother if Plato's views were of the type I suggest. But of course, all
this is pure speculation.

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

to say thatit will necessarilykeep movingforever(andso existingfor ever).

There is, then, a gap in the argument.At the same time, however, it is
quite understandablethat Plato should not have been aware of it. The
troubleis that "X never ceases to be F" (of which"thatwhichmoves itself
neverceases moving"is an instance)can meaneither"It is impossiblethat
there shouldbe an X whichis not F" (i.e., F is an essentialpropertyof Xs)
or "Xs are eternallyF"; the second entails that Xs are immortal,but the
first does not. It is only the first that Plato is really entitled to in this
instance;for as I understandit, &TEotUx&doXntXEov
(aiavi6makes a point
about an essential property, not about the eternal possession of any
property.But it is not surprisingthat the ambiguityshould have escaped
him.
It is interestingthat preciselythe same illegitimatemove is made in the
finalargumentfor immortalityin the Phaedo- an argumentwithwhichthis
one is often compared;indeed, I take this parallelas confirmingevidence
for my readingof what is going on here in the Phaedrus.9In the Phaedo
Plato moves from "The soul always comes bringinglife to whatever it
occupies"(105d3-4),and its corollary(dlO-11)"Thesoul mayneveradmit
the oppositeof thatwhichit alwaysbringswithit" (i.e. death, as he goes on
to say) to "Thesoul is immortal"(e6). Again, it is in one sense uncontroversial, given the commonsenseGreek view of the soul, that "the soul always
bringslife". Thatis, anythingwhichhasa soul is therebynecessarilyalive;it
is an essentialpropertyof soulsthattheyare not presentin organismswhich
have died. But this is not to say thatsoulsexist eternally.Nothingcancease
to bringlife and still be a soul; this follows from the essentialpropertyof
soulswe havejust noticed,andin thissense we mayagreethatthe soul does
not "admitdeath".It does not followthatthe soul mustcontinueto possess
this life-givingcapacitythroughouttime, and so be immortal.As in the
9 That there is an error common to the two passages is noted in passing by Demos (op.
cit., p. 135). The standardview of the relation between them seems to be that they are
argumentsof essentially the same kind, but that the argumentin the Phaedrus is of a
more empirical nature; see Hackforth, op. cit., p. 68 and Robinson, op. cit., p. 347. 1
agree about the similarity,but I cannot see that the Phaedrusargumentis more empirical. It is true that the Phaedo argumentmakes use of the terminologyof Forms;but as far
as I can see, this is not really essential to it. The crucial point is simply that the soul
necessarilybringslife. In the Phaedrus,the crucialpoint is that the soul is necessarilyin
constantself-motion. Each argumentrests, then, on a very basicfeatureof the soul, and I
do not see that one of these featuresis any more empirical,or observable, thanthe other.
It is also true that the Phaedrusargumentowes more to Presocraticthought(on whichsee
also Skemp, op. cit., pp. 3-10); but to equate "Presocratic"with "empirical"(as Robinson, at least, appears to do) seems to me highly inadvisable.

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

Phaedrus,the difficultystemsfromPlato'suse of the words&drandoVU'OTE

- wordswhose primarysenses have to do with temporalduration- to talk
aboutwhatwe wouldcall essentialproperties.10It is hardto knowhow else
he could haveexpressedhimself,giventhe vocabularyavailableto him;but
the resultingambiguitywas a treacherousone."I
So much, then, for the argumentgiven in the phrase&TEOVX 0koXELetoV
EaTO6.Whilethe argumentis not one we can accept, it is significantfor the
pictureof the soul that it obliges Plato to hold. He mustembracea view of
the soul as not simplya self-movingthing, but as necessarilyin unceasing
self-motion. We will return to this point later; for the moment, let us
continuewith the analysisof the argument.Sub-argumentA for the main
premise 2) is now complete, and Plato moves on to sub-argumentB; the
discussionhere is somewhatmore lengthy.
Premise Bi) is stated in the same sentence we have been looking at for
some time; "only that which moves itself .

. is a source and first principle

of motion for the other things that move" (c7-9).'2Bi) is not arguedfor
For a similaruse of 6e( and o'vnoxe,considerone of the ways in which Plato commonly
contrastsForms and particulars.Formsare frequentlysaid to be "always"the same; each
Form "always" possesses the property which makes it the specific Form that it is.
Particulars are sometimes one way, sometimes the opposite way; but the Form of
Beauty, for example, is "never" other than beautiful. Examples of this usage are Rep.
479a2-3, A,. 74b7-c3, and Symp. 211a3,bl-2.
11Curiously enough, Plato appears to recognise the error in the Phaedo immediately
after he has committed it. (Curiously, because the interlocutorsare made to regardthe
argument we have been discussing as entirely conclusive - 105e9.) He continues by
pointing out that it needs to be shown that the soul is imperishable. He appears to be
saying that to establish that the soul "does not admit death" - i.e., that nothing is both
dead and a soul - is not to establishthat it never ceases to exist; and this is essentially the
point that I have just been making. (See in particularthe paragraph106b1-c7.)However,
his way of closing this gap is rather lame; he simply says that of course that which is
6d6varogis also imperishable(106d2ff.). If "the soul is e0dvatos" means only that the
soul does not admit death, in a sense which is compatiblewith the soul's ceasing to exist as he appears to have just conceded - then this is plainly begging the question.
12 Indeed, this sentence is dense to the point of being very confusing; it contains a
premise for sub-argumentA - as well as a reason for it - and a premise for sub-argument
B; moreover, the same words g6vov bh lb abTr6xLvo!Bv
serve as subject in the statement
of both. I was originally tempted to think that what I am now calling Bi) was a second
supportingconsideration, parallel with &rEoOx &okoEinov taUT6, in favor of Ai). But
while this would reduce the numberof differentthingsgoing on in the sentence, it cannot
be the rightway to read it; for no mention is made, here or anywhere,of the firstprinciple
of motion being always in motion - as would be needed if the assertionin question (that
that which moves itself is a first principleof motion) were to constitute supportfor Ai).
Instead, this assertion leads directly into claims about an d4eXs freedom from yiveoLt
and WoQ&- claimsleadingto what I now call Bii); the idea of constantmotion, whichis at
the center of sub-argumentA, is left behind.
10

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

be
directly. Instead, Plato immediatelystates one half of Bii) -a'X
245e2.
&ykvrltov(dl) - and the argumentfor Bii) takes him from there to
(Bii) is stated in full at d7-8; "it is not possible that this [i.e., that which
moves itself]shouldeitherperishor come to be".) However,the firstpoint
he presentsin favor of Bii) - namely, that everythingwhich comes to be
must do so from a first principlewhich does not itself come to be from
anythingelse (dl-2, repeatedat d6)- is also a pointin favorof Bi) (provided
we understandyeveovLas a species of xLvroLg - I shall return to this
later). Indeed, as if to signalthat he has been confirmingBi) at the same
time as he has been arguingfor Bii), Plato restates Bi) at d6-7; "thus
that which moves itself by itself is a first principleof motion".
The argumentfor Bii) is as follows. (I havejust mentionedthe firststage
of it, but I shallrepeatthis for clarity'ssake.) A firstprincipleis ungenerated (ay'v-qTov),since everythingthatcomes to be comes to be out of a first
principle, and it (i.e., the first principle) does not come to be out of
anythingelse (dl-2); if a firstprincipledidcome to be out of somethingelse,
then "it would no longerbe out of a firstprinciplethat all thatcomes to be
comesto be" (d2-3).1 Thislastpointis somewhatobscure,andthe text may
be faulty. But the general idea, that a first principleis necessarilynot
generated from something else, seems clear enough; and it does follow
plausiblyenoughthata firstprincipleis ungenerated,assumingthatnothing
can generate itself. Hence, Plato continues, it must also be imperishable
(d3-4);for (in view of what has just been said) if it were to perish,nothing
else couldbringit backinto being, norcouldanythingelse come to be out of
it (d4-6). One mightstill ask why this showsthat it could not perish. Plato
gives his answerat d8-e2, in conjunctionwith his explicitstatementof the
premiseBii) withwhichwe are now concerned.Essentially,the pointis that
if it were to perish (so bringingto an end all possibilityof yEvFotg), the
universewould collapse into immobility.4 However, it is not stated why
13 Reading *t &?XQlg
as the subject of yLyvoLo; in
T8 yLyv6Rtevov
in d3, and taking JC&V
6?x1i
this I follow Verdenius and de Vries (opp. cit. ad loc.). Buttmann'sconjecture tTL
would make for much better sense (and for a closer fit with Cicero's translation), if we
as equivalent to dl; but this is surely impossible.
could understandytyVOLTo
"4The balance of opinion now seems in favor of the mss. yEvrotvin el (also found in
Hermeias - see p. 117.24), and against Bumet's reading yfv d; Ev (following
Philoponus). It is generally agreed that yfvEotv can be taken as equivalent to tca
ytyv6pEva, and this gives good sense.
Hermeias'view of the logic at this point is puzzling.He says that once it has been shown
that a first principle is ungenerated, that by itself suffices to show its imperishability
(p. 117.17). In his view, therefore, the considerationsoffered in d4-e2 form a supplementary, and strictly speaking superfluous, argument dg &biGvactov(p. 117.17-25; cf.

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

this, in turn, could not happen.'5

This concludes sub-argumentB for the main premise 2); immediately
after, at e2-3, 2) is explicitlystated. Whatare we to make of sub-argument
B? There are two points about it which I find of interest. First, the
argumentrequiresthat we think of yEVEOLgas a species of XIVaCLg. For
Plato introduces the notion of an aQxi XLV9(aWo;, and then immediately
goes on to say (by wayof arguingfor Bii)) that"everythingthatcomes to be
comes to be" out of an a'QxTI.
For the argumentto work, it mustobviously
be the same kind of aQxqunderdiscussionthroughout;and it follows that
cannotsimplymean "(loco)motion",but mustreferto any kindof
XLVYiOL
changewhatever.However, there is nothingvery startlingin this. We find
just such a broad notion of xivqcL; explicit in the Laws (again in the
context of a discussionof soul as self-mover);at 893b6-894c8ten kindsof
"motion" (xLvivtL, also interchangeablyRETcaokX)are distinguished,
two of which are yEVEFLg and pOoQa(see, e.g., 894bll).16 Similarly,in
Physics I11.1 Aristotle gives his definition of motion (xivrJoL;) as the

fulfilmentof a potentialityqua potentiality,and immediatelygoes on to

distinguishvariousspeciesof motion, amongwhichare yEVEOL;and aoQd
(201alO-16). And indeed, a strong connection between

yEvrOL;

and

occurs as early as Parmenides:"Thatwhich is" in Parmenidesis

explicitlysaid to be both aye'- ov and&xLvTTov,and these two properties
appear to be seen as closely connected - see especially DK B8.26-8.
Parmenides does not appear to treat y~VEGL; as actually a species of
xLvrloGL;
but in his use of these notions, he certainlyseems to foreshadow

xivroL;

103,11.3-6, where he says that the argumentas a whole consists of two syllogismsplus one
further argument ?i5 dEbivvaTov).
I fail to understandthis; it seems to me that d4-e2 is
integral to the main argument, and builds very directly upon the demonstration just
above that an &pxQi
is &yv- og. Robinson (op. cit.) claims agreementwith Hermeias on
this issue, but seems to me also to misconstrue Hermeias' view as to the extent of the
argument ErigM6vvaTov,locating it from d8-e2 only.
15 The best comment on this point may be
Hackforth's (op. cit., pp. 66-7) who simply
says "Thatwas a possibilitynever contemplatedby any Greek thinker".As he points out,
the possiblity in question is that there should be no yvEaLg whatever, not that some
particular cosmos should perish. The Greeks had no trouble conceiving this latter
possibility; Plato himself suggests in the Timaeus(41a) that it is possible (though in fact,
since the creator is good, it will not happen to this cosmos), and many Presocraticand
Hellenistic schools held that it actually occurred.
16 The Theaetetusis also relevant
here; in the context of the "secretdoctrine"associated
with Protagoras, Heraclitus and others, the words nactvta yLyvETaL and nacvTa
XLVELttaL
appear to be intended as equivalent (compare 152d8 with 183a5).

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

those who do. There are, then, respectableparallels for the idea that
"comingto be" is a kind of x(vrt'Cg.'7
On the other hand, this does not meanwe shouldsimplygrantPlato the
idea and move on. It may be understandablethat he shouldhave regarded
y?vws1; as a species of xUvroLV;but I think that questions can still be
raised about the role this assimilationplays in the argument.Recall that
Plato has just claimed that a self-moveris a first principleof motion for
everythingelse that moves. We have now discoveredthat "motion" includes "comingto be"; but presumably(thoughthis is not explicitlystated
here) it includesother thingsas well - at least some of the other kindsof
"motion"mentionedin the Lawsand in Aristotle'sPhysics.But if this is so
- if there are severaldifferentkindsof x(V1OL; - one mightwonderwhythe
same thing shouldhave to be a first principlefor all of them. If we accept
that there are such thingsas self-movers,we can also acceptthat these are
responsiblefor the "motion"of everythingelse. But it is not clearwhy any
one self-movershouldhave to be responsiblefor everyone of the different
species of xCvrioM.For example, why should not one thing (or kind of
thing) be a first principleof yCvFsog,and some other thing (or kind of
thing) a first principle of locomotion? Plato's argument requires that
anythingwhich is a self-moveris a first principlespecificallyof y'vECtL;
but it is not obviousto me whythisshouldbe takenfor granted.Needlessto
say, Plato does not respondto this query;and in view of the parallelsI just
adducedfromParmenides,Aristotleandelsewherein Plato,thisshouldno
doubt not surpriseus. Among other things, these parallelsdemonstratea
pervasive tendency in Greek thought to see all the processes subsumed
under the heading of xivrjo5L,in the broad sense we have examined, as
being very intimately related. (In Parmenidesand Plato, at least, this
tendency results from the centralcontrastin both their philosophiesbetween the worldof changeand the worldof the changeless,the worldof To
ov; someone for whomthis distinctionwasprimarywouldnaturallytend to
grouptogether all species of change, as being on a par.)
Here, then, is one assumption,or set of assumptions,in this partof the
argument.I move now to the secondof my two observations.Not only does
the argumentrequire that "motion"includes "comingto be", and that
anythingwhich is a first principleof "motion" is a first principleof all

17

Hermeiasseems quite unruffledby this idea; in discussingthis partof the argument,he

simply remarks in passing iy y4p y tvEaL x(vT)o;t

tOrtLV

(p. 116.13), as if this is

entirely commonplace.
10

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

species of "motion";'8it also appears,at least at firstsight, to requirethat

thereis justone principleof motionin the entireuniverse.Forthe argument
for the indestructibilityof the principleof motion rests, as we saw, on the
assertionthat all yFVetLg wouldcease with its destruction.But if therewas
morethanone principleof motion, it is not clearwhythe destructionof any
individualprinciplewouldresultin the collapseof the universe;so longas at
least one principle was in operation at any given time, it seems (given
Plato'sother assumptions)as if yFVEOLt could continue. Apparently,then,
Plato is simplytakingfor grantedthat there is not more than one principle
of motion.
But there is somethingvery peculiarabout this. Recall that Plato is just
aboutto identifythatwhichmovesitselfwithsoul (thisis step 1) of the main
argument,accordingto my analysis).Since sub-argumentB has identified
that which moves itself as the principleof motion, we are led to infer, by
simpletransitivity,that soul is the principleof motion. But now, if the logic
of sub-argumentB requiresthat there be just one principleof motion, it
appearsto follow that there mustbe just one soul. And this, besidesbeing
plainly contraryto any commonsenseview of the soul, seems flatly to
contradict the myth which immediately follows the proof, in which a
pluralityof individualsouls is discussed.Somethinghas gone wronghere;
andsince, on thisview of the matter,the inconsistencyseems so blatant,we
should probablyconclude that the fault lies not in Plato, but in our interpretationof him.
One wayto eliminatethis apparentcontradictionmightbe to suggestthat
Plato is implicitlyoperatingwhith some notion of a World Soul. That is,
perhapshe is thinkingof individualsouls as being all ultimatelyaspectsof
some larger,unitarysoul of cosmicproportions.In this way the pluralityof
individualsoulswouldbe only superficial,andwouldbe consistentwiththe
principleof motion in the universe being in a deeper sense one, as the
argument for imperishabilityrequires. Besides making Plato's train of
thought coherent, this proposal seems to have at least some historical
plausibility. First, a doctrine of World Soul certainly was adopted by
various later philosophicalschools - by the Stoics and, perhaps more
significantly,by the Neo-Platonists- and it is hard not to see the present
passageas in some way prefiguringthese later ideas. Second, the Timaeus
makes large use of the conceptof a WorldSoul (thoughit looks here as if
individualsouls exist in theirown right,in additionto the WorldSoul, not
18 This is perhaps too strong. Strictly, all that the argument requires is that anything
which is a first principle of any species of "motion" is also a first principle of yE'VEGLg.

11

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

as aspectsof it'9).Despite these parallels,however,I do not thinkthisis the

rightway to understandthis partof the Phaedrus.It would not be surprising, in view of the parallels,if Platowere operatingherewitha conception
of WorldSoul; but the parallelsdo not provideany positiveevidence that
he is doing so. And the fact is that there is absolutelyno hint of any such
conceptionin the Phaedrusitself. We findno suggestion,eitherin the proof
of immortalityor in the succeedingmyth, that all our souls are ultimately
aspects of the same thing, or that our ultimategoal, in strivingto escape
from the cycle of rebirth, is reabsorptionin some larger unity. On the
contrary,I would say, it is the individualityof our souls, the differences
between them, that is emphasisedin the myth.20
I concludethatwe are not entitledto readthe hypothesisof a WorldSoul
into the Phaedrus;andthe problemremainsas to how we are to makesense
of Plato's apparent assumptionthat there is just one self-mover. But
perhapsthere is anotherwayto understandthe matter.Possiblyhe is using
as a mass term, analogousto "water"or "electricity".2'In this
"VPuX"
case, it wouldnot be thattherewasone soul- the WorldSoul;rather,there
would be a single kind of stuff (just as water is a single kind of stuff), of

t9 The creation of individualsouls (or rather, of those souls and parts of souls that are
immortal) is described at 41d4ff. The creator is said to fashion these souls out of the
remainderof the elements previouslyused to fashion the WorldSoul, mixingthem in the
same bowl, and in the same manner (except with some dilution), as was used for the
World Soul. Obviously, it is hard to know how, if at all, this is to be cashed out literally.
But it is at least implied, I take it, that individualsouls are not simplyaspectsof the World
Soul; they enjoy a separate existence.
It is sometimes said that a WorldSoul is also to be found in the Laws, but this is much
less clear; as far as I know, a WorldSoul is never explicitlyreferredto in that dialogue. I
shall briefly touch upon this point again near the end.
20Think, for example, of the distinctionbetween gods' souls and humansouls (246a-b);
between the twelve companies of souls, each led by a different god (246d6-247a4);and
between the ten types of human lives, correspondingto various conditions of human
souls at birth (248c2-e3). While none of these distinctionsis formallyinconsistent with
the notion of a World Soul, they do indicate that Plato is choosing to focus on the
differences between souls, not on their fundamentalunity; and this, I think, makes it
implausible to suppose that he has a World Soul in mind - given that there is nothing
whatever in the text to suggest it.
21 In what follows, I am expanding on a
suggestion by Tony Long. This is perhaps an
appropriate place also to thank him more generally for much helpful criticism and
encouragementas this paper was developing- especially on the mattersI am now about
to introduce.

12

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

which individual souls (like individual pools of water) consist.22Now,

anythingwhich is true of water as such (for example, that its chemical
compositionis H20), or of electricityas such, is also true of any individual
pool of water,or of anyindividualelectricalcurrent;andsimilarly,if soul is
being conceived of along the same lines, anythingwhich is true of soul as
suchwill also be trueof individualsouls. But the propertywhich,by the end
of the argument,we discoverto be true of soul as such is preciselyimmortality. Thus it will follow, on this understanding,that any given individual
soul is immortal- a conclusionwhich Plato obviously wants to uphold.
Moreover,if this is right,we can respectboth the logic of sub-argumentB
and the pluralityof individualsouls. Sub-argumentB requires, in some
sense, the singularityof soul. But thisdoes not meanthatwe need to ascribe
to Platothe view that, out of all the particulars in the universe,only one is a
soul. If "soul" is a single kind of stuff, the argumentwill work perfectly
well. For if this stuff is responsiblefor all y'4veOLg in the universe, the
destructionof this stuff would indeed resultin the universe'scollapse;and
My use of the term "stuff"is perhapsa little unusual,and I should define it as precisely
as I can. To begin with what is uncontroversial,all stuffs are designated by mass terms,
but not all mass terms are the names of stuffs. For abstractnouns, such as "hunger"or
"wisdom", qualify as mass nouns (they do not occur in the plural, nor can they be
qualified by numerical adjectives); yet it would be bizarre to call hunger or wisdom a
stuff. Now, this might suggest that stuffs are by definition material;and in this case I
would not, of course, want to suggest that Plato thinks of soul as a stuff. But however it
may be in ordinaryusage, I intend to use the word "stuff"in such a way that there can be
said to be non-materialstuffs. In my usage, the referentof any mass noun whichis not an
abstractnoun will count as a stuff. (Here I am assumingsome intuitivesense of what is an
abstract noun; I admit that it is very hard to give a foolproof definition. On this and
several other complications in the notion of a stuff, see V.C. Chappell, "Stuff and
Things", PAS 71 (1970-1) 61-76 - an article from which I have profited in writingthis.)
Thus electricity, though not in any obvious sense material,is a stuff; and so, too, is soul if it is true that Plato is using "pu " as a massterm. For this reason, electricityis perhaps
the most useful analogy to keep in mind in what follows.
Another parallel which may be helpful is the concept of "spiritualsubstance"in early
modern philosophy; a famous passage which makes use of this concept is the chapteron
identity in Locke's Essay (Bk. II, ch. XXVII, "Of Identity and Diversity"). Here Locke
raises the possibility that some contemporaryperson's soul might be composed of the
same spiritual substance as that which made up the soul of Socrates (though this
possibility is, according to him, irrelevant to questions about identity) - just as two
material objects existing in non-overlappingperiods of time might be composed of the
same materialsubstance. In my usage, we might say that spiritualsubstance, like Plato's
soul, is a kind of immaterialstuff of which souls are composed (though I do not want to
take the parallel any further than that); while we may find such a picture highly alien,
Locke's account (even if it is not offered in an entirely serious spirit) does, I think,
illustrate that the concept of an immaterialstuff is a perfectly coherent one.

22

13

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

from this it does indeed follow, on Plato'sassumptions,that the stuff will

necessarilynot perish.But all thisis quitecompatiblewithsayingthatthere
are many individualsouls - just as water (or electricity)is a single kind of
stuff, yet there are many pools of water (and many electricalcurrents).
Equally,the singularityof the stuff "soul"is quitecompatiblewiththe idea
thatsoulshavedifferentiableparts,andwiththe idea thattherearequalitative differencesbetweensouls (both of whichwe subsequentlylearnin the
mythof the charioteersandtheirhorses). Forwater,electricityandthe like
are similarlydifferentiable;water is composed of hydrogenand oxygen,
and differentelectricalcurrentscome in differentvoltages.
This understandingof soul seems, then, to fit the context somewhat
betterthanthe previousidea of a WorldSoul. In addition,Plato'susageof
the term "9vx" itself seems more consonantwith the notion of soul as a
stuff (in the sense I have explained). Throughoutthe argument,"pVX"
occursin the singularwithoutarticle. If he had in minda WorldSoul, one
mighthaveexpectedhimto referto it usingthe definitearticle;buthe never
speaksof "Aivx", onlyof "Vu "323Grammatically,thatis, he does seem
to be using the word as a mass term;24and this is exactly what we would
expect if he is thinkingof soul as a stuff. While the evidence is all too
23 The one possible exception is 246b6, at the beginningof the myth, where some mss.
have ni&oaA Vuxi, also i Vuxh n&aoa.As far as I can gather from LSJ and from Greek
grammars,both these phraseswould have to mean "The whole soul". And since it is clear
from the context that the reference is not to individual souls (for no individual soul,
whether in its entiretyor not, "takes care of all that is soulless"), it would have to be to a
WorldSoul. If so, however, it would be the sole explicit mention of the WorldSoul in the
whole dialogue; and its intrusion here, after a complicated proof of the immortalityof
soul which makes no explicit use of such a notion, would, I think, be very surprising.My
claim regardingthe usage of Vux should not, therefore, be upset by these ms. variants;
we should follow Hermeiasand Simplicius,and read VurX2&aa. (All moderneditors of
whom I am aware do indeed adopt this reading.) Given the prevalence of doctrinesof a
World Soul in later antiquity(including, as I said, Neoplatonic doctrines), one might in
any case argue that VuyX3tdoa, which does not entail any notion of a World Soul, is the
lectio difficilior.
24 This is not necessarilytrue of
Vuxh 3rtoa in 245c5;taken in isolation, the phrasecould
equally be translated "all soul" or "every soul", and commentators have sometimes
exercised themselves over which sense Plato intends. Plainly this cannot be decided by
inspection of the phrase itself. The question must be whether one or the other sense is
required by the logic of the argument; and this question we are now in a position to
tackle. Given the presuppositionsof sub-argumentB that I have unearthed, the collective sense "all soul" seems clearly more relevant. As we saw, the principle of motion
there under discussionis thought of as in some sense single and whole (whether or not I
am right about the exact conception); and the distributivesense "every soul" would run
counter to this. On the other hand, the argumentis presumablyintended to show that

14

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

nebulousto admitof certainty,I thinkwe are entitledat least to a tentative

conclusion- namely,that the underlyingpictureof soul withwhichPlatois
operatingis indeed this latterone, and not the one whichinvolvesa World
Soul.25

I have dwelt at some length on the presuppositionsof sub-argumentB,

both because of their interest for our broaderunderstandingof Plato (a
matterto whichI shallreturnshortly),andbecauseof the greatdifficultyof
elucidatingthem. The remainderof the argumentcan be dealt with much
more briefly. Our analysishad reached245e2. The followingsentence (e24) containsa statementof both the mainpremises2) and 1) (in that order).
To recall, premise 1) states that soul is that which is its own source of
motion; and Plato now offers us a considerationin favor of this assertion.
The argumentis simplythat any body with an externalsourceof motion is
"soulless" (&ivXov), whereas one with an internal source of motion is
"ensouled"(EtVuXov)(e4-6). On its own, this seems somewhatflimsy;it is
not at all clear that it justifiesthe conclusionthat "soul"is a self-moverin
the very strong sense used above. We may certainlymake a distinction
between bodiesthat need to be pushedor pulledby somethingelse in order
to move, and bodies that do not; and these two classesdo seem to correspond with the classesof soullessand ensouledbodies respectively.But it is
surelyquitepossible,for all thatPlatohas said, thatbodiesof the latterclass
should have received some initial propulsioninto motion from outside,
even if they need no furtherpropulsionafter that. And this, of course,
would disqualifythem as self-movers;for as we saw, Plato is thinkingof a
each individualsoul is immortal;so that the sense "every soul", while less suited to the
tenor of the argument itself (and to the usage of "Vux" elsewhere in the argument),
cannot be decisively ruled out. Hackforthmay be right to conclude (op. cit., p. 64) that
"the distinction between collective and distributivesenses is not here before his mind";
see his discussion for several other useful insights.
25 I should perhaps briefly allude at this point to Hermeias' views on how Plato is
conceiving of soul - since it was with his adviceon this subject that I began. While he cites
Posidonius as having thought that Plato had in mind the World Soul (apparently as
distinctfrom individual souls), he himself rejects the idea. Relying heavily on 0ecag mE
xai &vOQ(w;ivTSin 245c3, he says that Plato must be thinking of the rational soul since this is the kind of soul possessed by gods and humans, but not by other ensouled
beings (p. 102.15ff.). This seems to me acceptable as far as it goes; but as the last few
pages should have made clear, I do not think it goes far enough. Plato's logic raises
deeper problems, which I have tried to address;neither Hermeias nor anyone else seems
to have faced these problems. Robinson (op. cit.) follows Hermeias, but then makes
matters worse by suggesting that Plato is perhaps thinking not of the rational soul in
toto, but of the rationalpart of the soul. This seems to me obviously wrong in view of the
myth, where the immortal soul is clearly protrayed as tripartite;more on this in section II.

15

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

self-moveras afirstprincipleof motion.26His argumentis farfromshowing

of everything
thatsoul is responsiblefor the motion(includingthe ycvEoLg)
else that moves. His firstpremise- and with it, a key point in the striking
conceptionof soul the argumentfor immortalityreveals- is moreassumed
thanarguedfor. This concludesmy detailedexaminationof the argument;
for the remaininglines 245e6-246a2merelycontain a restatementof premise 1), followed by 3), the conclusionof the entire argument.
II
Let us now shift our approachsomewhat.As I said, Plato intendsthat the
argumentshouldconvinceus by the forceof its logic;we are not to take on
trust what he says, but to give our assent if and only if we are rationally
compelled.This is nothingunusualfor Plato- thoughthe densityand rigor
of this particularpassageis unusual;he constantlyexhortsus, throughthe
mouthof Socrates,to "followthe argument"whereit leads us, and not to
base our views on anythingelse. So far, I have been tryingto approachthe
argumentin precisely this spirit, attemptingboth to understandexactly
what Plato is saying, and to determinehow muchof it we can reasonably
accept. For the restof the paper,I shalladopta moreexclusivelyhistorical
attitude. As we have seen, the argumentcontainsvariousassumptionsor
presuppositionswhich we might very well question;havingengaged with
him in the fashionwhichhe himselfapparentlyinvites,we findthat he fails
to convinceus of hisconclusions.Thereare no doubtmanyreasonswhythis
is not surprising,and to discuss them would take us far afield into more
generalissues in philosophyand in intellectualhistory.However, I prefer
to stick more closely to the argumentitself, and to see how it can help to
increaseour understandingof Plato'sthought.The assumptionsor presuppositionsthat I just mentionedare of greatinterestin themselves.For as I
havesuggested,they reveala certainunderlyingconceptionof the natureof
soul; and it is instructiveto comparethis conceptionwith other, generally
more explicit, conceptionsof soul, both in other dialoguesand elsewhere
in the Phaedrus.This is what I now intendto do.
The underlyingconceptionof soul has severalfeatures, and we should
brieflyreviewthem. The most obviousfeature, and the one scholarshave
noticedmost often, is that soul, andsoul alone, movesitself;indeed, this is
simplymy premise 1). But Plato's view is strongerthan this alone would
imply. Recall the feature of soul which we uncoveredin examiningthe
26 Here it is instructiveto compare Aristotle's apparentvacillationon the issue of what it
is for something to be a self-mover. On this, see D.J. Furley, "Self Movers",Aristotleon
Mind and the Senses, ed. G.E. R. Lloyd and G.E.L. Owen, Cambridge, 1978.

16

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

phrase &TE oIUx &oXeikitov EaT6O, Plato's support for premise Ai) namely, that soul, the self-mover, is in unceasing self-motion, and
necessarilyso. Further,we discovered,in connectionwithpremiseBi), that
soul has a cosmicsignificance;for as self-mover,soul moves not only itself
but everythingelse - where"moves"is not restrictedto whatwe wouldcall
".motion",but includes (probablyamong other things) generation. And
finally, there was the puzzlingpoint raisedby premiseBii), that "soul"is
beingconceivedof as in some sense unitary.Whilethe issueis veryobscure,
we concluded that it was best to understand"soul" as a unitary (nonmaterial)stuff, of which our souls consist. How does a conceptionof soul
containingthese featuresrelateto other conceptionsthat we find in Plato?
I shall begin by summarisinga few familiarpoints from the Phaedoand
the Republic.In these two dialogues,a strictdivisionis madebetweentwo
realms, the realmof the changeableand the realmof the changeless.The
formeris the realmof sensibleobjects, the latterof intelligibleobjects;and
in the Republic,at least, it is the latterclass of objects which, just because
they are changeless, alone fully warrantthe title of "things that are"
(v&Ta).Now, in the Phaedothe soul is arguedto be akinto the Forms- that
is, to the objects granted to belong to the non-composite, changeless,
intelligibleclass; and from this it is deduced that the soul, too, is noncompositeand changeless,and thereforeimmortal(78b4ff.). Or at least, it
is claimed, the soul in its essentialnatureis non-compositeand changeless.
However,becauseof its associationwiththe body, it is bound,in thislife, to
partakeof the changeableto some extent - thoughthis taint is something
we should strive to minimise.
In the Republic,the pictureof the soul is more complicated;for in Book
IV we are told that the soul consistsof three parts, and that our task is to
achievethe appropriateharmonybetweenthem. Now, the two lowerparts
of the soul, at least, appearto be very muchsubjectto change;27and so it
This is not, as far as I know, explicitly stated. But there is much emphasis, in the
description of the inferior states and inferior souls in Bks. VIII and IX, on the
progressivelygreater instabilityand variabilityof the various kinds of life. The less the
influence exerted by the rational part of the soul, and the greater the power of the two
lower parts (particularly iR1LORLa, which is throughout seen as standing in starker
opposition with the rational), the more the person, or the state, is subjectto change. Also
relevantis the discussion, again in Bk. IX (580d-588a),of the differenttypes of pleasures
associatedwith each partof the soul. Here the pleasuresof EhI&RLeVa
are identified as the
pleasuresof the body, and as such intrinsicallychangeable;by contrast, the pleasuresof
the rational part come from its communion with that which truly is - that which is
changeless - and so are themselves not subject to change. Plato has some trouble fitting
the OVUtOeLEbUinto this discussion, but he does place it in the same category as the
bALOlURilltLXoV(586c7-d2).
27

17

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

mightseem that the Republic'sview of the soul was a fundamentaldeparture from that of the Phaedo. However, a strong case can be made for
saying that the two dialoguesare reallyconsistentwith one another. For
one thing, the ideal state for the soul in the Republic(thoughmost people
are incapableof attainingit) is still rationalcontemplation;and this is a
state where the lower two parts of the soul, and change itself, appearto
have no place. But more importantly,the argumentfor immortalityin
Book X seems to implythat, as in the Phaedo, the soul in its truenatureis
non-compositeand changeless.As we see it, it is "composedout of many
elementsandnot enjoyingthe finestcomposition"(611b5-6),andis "fullof
variabilityandunlikenessanddifference"(b2-3). However,thisis due to its
associationwith the body;if we were to see the soul as it reallyis, free from
such association,we wouldprobablyfind that it wasjust the opposite. It is
hard not to read this as suggestingthat in its true nature, the soul is not
tripartite- that the divisionarguedfor in Book IV pertainsonly to the soul
as embodied.18To be sure, the tone is tentative; but the view being
expressedseems clear enough.
viewof the
Let us returnto the Phaedrus.How does the Phaedo/Republic
soul compare with the underlying conception we discovered in the
Phaedrus'argumentfor immortality?The first feature in that underlying
conceptionwas that soul was that whichmoves itself; and since "motion"
here includesall kindsof change,this appearsto standin simplecontradiction with the view I havejust been describing.Now, if we restrictourselves
to the bare assertion "soul is that which moves itself' (which was my
premise1)), it is not, in fact, clearthat the two viewsare formallyinconsistent. For the Phaedo and the Republicallow that the soul does undergo
changewhen in the body;andindeed, the evidencefor premise1) concerns
the distinctionbetweenensouledandsoullessbodies.The issueis not, then,
as simpleas it mightseem. However, as we saw, the underlyingconception
in the Phaedrusis not just that souls do at timesmove themselves,but that
soul is necessarilyalwaysin motion. In otherwords,soul mustbe in motion
both in and out of the body; in the Republic'sterminology,the soul "in its
truenature"is in unceasingmotion.And thisconceptiondoes undoubtedly
stand in oppositionto that of the Phaedo and the Republic.Instead, it is
aligned with views which we can confidentlyplace at the end of Plato's
career. I am thinking in particularof the Laws, where the definition
28 The disclaimerat the beginningof the Bk. IV argument,to the effect that the present
methods are inadequate for a precise understanding(435c9-dS), may be intended to
accommodate this point.

18

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

(Xoyog)of soul is saidto be "thatmotionwhichis capableof movingitself'

(896al-2); except that soul is here seen not as the thingwhichmoves itself,
but as itself a speciesof motion,29this is identicalwith what the Phaedrus
refersto as the oia'Cv TE xaa X6yovof soul (246e3). But anotherimportant
passageis Sophist248a4-249d4.Here Plato argues,againstthe view of the
Republicandelsewhere,that"thatwhichis" is not restrictedto the changeless, but encompassesthe changeable as well; in fact, it is plausible to
suppose that the putativetarget of the argument,some people called the
"Friendsof Forms",is really his own earlierself. Most interestingly,the
key point in the argumentis that if the Friendsof Formswere right, then
"life, soul and thought" (248e6-7) could not belong to the realm of that
whichis. Obviouslyit is assumednot only that soul is a "thingwhichis" which, I take it, Plato assumedat every stage 0- but also that soul belongs
to the class of changingthings.This argument,then, is highlycongenialto
the underlyingconceptionof soul in the Phaedrus.For there, as we saw,
soul is in constant self-motion;and by the PhaedolRepubliccriterionfor
Being, this would entail that soul was not a "thing which is". So far,
therefore, the matter appears relatively straightforward;the Phaedrus
seems to squarewith Plato'slaterthought,not with what is usuallyseen as
his middle period.
Let us nowturnfromthe Phaedrus'argumentfor immortalityto the myth
whichfollows. At first sight, this seems to runagainstthe conclusionI just
reached.For the mythincludesseveralelementswhichare clearlyreminiscent of the Phaedoandthe Republic.First,the imageof the charioteerand
two horses, one of which is obedient to the charioteerand the other not,
evidentlyrecallsthe tripartitesoul in the Republic;at 247c7-8,indeed, the
charioteeris explicitlyidentified as vouv. Again, the supremevision the
soul can experienceis the "vision"of the Forms;andas in the Republic,the
Formsconstitutethe realmof "thatwhichis" (247d3), and are the entities
of whichtrue knowledgeis possible- as contrastedwiththe sensiblerealm,
the realmof yCvVEGL,of whichone can have only opnion. Moreover,it is
throughrecollectionof the Formsbeheldbeforebirththatunderstandingin
thislife is saidto be possible(249b-c);thisdoctrinegoes backto the Phaedo
and to the Meno. Finally,of course, as in the Phaedo, the goal of life is to
escapefromthe cycle of rebirthas soon as possible,andthe meansis to live
as a philosopher.
29 This difference is noted by Demos, op. cit., p. 136.
30 Except in the Timaeus.For the Timaeus'very peculiarview of the compositionof soul,
see further note 36.

19

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

We cannot deny, then, that there are manycontinuitiesbetween these

At the same time, though, there are
other dialogues and the Phaedrus.31
several crucialdifferences,whichmake the continuitiesmuchless significant thanthey mayseem. One pointwhichhasoften been noticedis thatin
the Phaedrus,unlike the Republic(if my own and others' readingof it is
correct), all three partsof the soul are immortal;even the gods' souls have
three parts, though in their case the differentparts are not in conflict.32
Again, the end-pointof the soul's progressis not changelessand eternal
contemplationof the Forms, but an eternal traversingof the heavens,
punctuatedby contemplationof the Formsat intervals.Moreover,if we
take the detailsof the mythseriously- as I thinkwe should- the lowertwo
partsof the soul (representedby the horses) play an indispensablerole in
this eternal traversing,and in transportingthe charioteer,reason, to the
31 My wording here and elsewhere assumes the standardview that the Phaedrusis later
than the Phaedo and the Republic; this seems to me justified by precisely the kinds of
comparisonswith other dialogues to which this entire half of the paper is devoted.
32 Pace Guthrie, who thinks that the soul is tripartiteonly during the cycle of rebirth.
(See "Plato's Views on the Nature of the Soul", reprintedin Vol. 11of Plato, ed. G.
Vlastos, New York, 1971; and Vol. IV, pp. 421-5 of Guthrie's A History of Greek
Philosophy, Cambridge,1975.) Because the gods' souls are free frominternalconflict, he
thinks that they must be unitary; and the same goes, he believes, for those originally
humansouls who have escaped the cycle. In this way he is able to hold that the Phaedrus
is fundamentallyconsistentwith the Republic.That the gods' souls consistof a charioteer
and two horses he puts down as just partof the machineryof the myth. This seems to me
quite untenable. It is true that Plato switches to the mythicalmode here because, as he
says, it is beyond humanpowers to give a fully accurateliteraldescriptionof the soul; the
picture conveyed, therefore, is only tentative and only partial. But this does not mean
that we are free to discard any element of the myth that we choose. He composed the
mythas he did because he felt that this was the closest approximationto the truthof which
he was capable. If he had meant to suggest that the gods' souls did not consist of three
parts, he could very easily have composed that part of the myth differently. Nor,
incidentally,does he ever suggest that the black horse of the humansoul becomes white
on the soul's escape from the cycle; rather,it becomes progressivelymoreobedient to the
charioteer. Guthrie's interpretationthus involves a furtherdeparturefrom what Plato
actually tells us.
Another author who seeks to show that Plato's views on the composition of the soul
'
remain essentially the same is Robert W. Hall, "4u as Differentiated Unity in the
Philosophyof Plato", Phronesis8 (1963) 63-82. As the title suggests, Hall arguesthat in
the Republic,the Phaedrusand every other dialogue that deals with the subject, the soul
is viewed as a differentiatedunity. This seems to me almost triviallytrue, but obviously
inadequate to show that Plato's views did not change. For the question now simply
becomes "into what aspects is this unity differentiated?"And to this question, as Hall
seems quite ready to admit, different dialogues return different (and incompatible)
answers.

20

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

pointwhereit can contemplatethe Forms;reasonon its own appearsto be

relativelyinert. The ideal state is no longer one where the lower, changeable parts of the soul are transcended.On the contrary,they are just as
importantas reasonitself to the soul'sfulfillingof its final destiny;andthis
finaldestinyitself consistsnot of freedomfromall change,but of constant,
albeit regular,motion. Finally,while the PhaedolRepublicconceptionof a
schism between Being and Becoming is maintained,the criterionfor the
divisioncan no longer be quite the same. For as I said earlier, I presume
that at everystage Platowouldhave held thatthe soul is a "thingwhichis";
but if so, changelessnesscan no longerbe a necessaryconditionfor Beingfor as we have seen, the soul, even in its true nature,is no longerchangeless. Plato does not explicitlytell us thatthe soul is an ov; indeed, he seems
to restrictthe title to the Forms.However,he does say thatthe soul "feasts
on" and "is nourishedby" To ov; andthe metaphorof nourishmentmust, I
take it, imply that the nature of the soul is not basicallyalien from that
whichnourishesit - in other words, that the soul, too, is an ov.
On this last point, it maybe that Platodid not manageto take accountof
all the implicationsof what he was saying. But even so, there is at least a
shift away from the Phaedo and the Republicand towards the Sophist,
where it is arguedthat we must not restrictthe title of To ov to changeless
entities. As for the other issues I just mentioned,the Phaedrusmythseems
to be definitelyalignedwith the view of soul propoundedin the Laws, and
againstthose earlierdialogues. As we saw, the Laws defines soul as selfmotion;and the souls in the Phaedrusmyth, insteadof being ideallystatic,
are in constant, self-propelledmotion. Moreover, the myth assigns an
essential role in this motion to those partsof the soul other than reason.
Again, in the Laws, the followinglist of motionsof the soul is given:"wish,
reflection, foresight, counsel, judgement, true or false, pleasure, pain,
hope, fear, hate, love" (897al-3, A.E. Taylor's translation).The Laws
does not makeexplicituse of the notionof a tripartitesoul. However,if one
approachesthis list withthe tripartitesoul in mind,andaskswhichitemson
the list can be assignedto the rationalpart, the answerseems clearlyto be
"not all of them": some, at least, would have to belong to the other two
parts.
So it turnsout, I think,thatthe Phaedrusmythis in centralrespectscloser
to Plato'slaterviews than to the Phaedoand the Republic;the similarities
with the latter dialoguesseem to me less significantthan the differences.
We can now see, too, that the myth is, after all, in agreementwith the
argument for immortality which immediately precedes it. Given the
similaritiesbetween the myth and the earlierdialogues,it looked as if the
21

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

myth and the proof were at odds with one another.In fact, however,both
view the soul as beingin eternalmotion;andthissets thembothapartfrom
the earlier dialogues, where the soul is in its true naturechangeless,and
alongsidecertainlater dialogues.33There is, then, a deep connectionbetween the proofandthe myth- a connectionwhichhasrarelybeen noticed.
I have so far said nothing about the place of the Timaeus.There is, of
course, much in the Timaeusabout the soul; and since it is usuallyconsidered a late dialogue, one might expect that it, too, would be naturally
grouped with the Phaedrus,Sophist and Laws. Now, the dating of the
Timaeushas been the subjectof muchheateddebate, andI cannotattempt
to deal with this thorny question here. However, it is strikingthat, with
regardto the matterswe have looked at so far, the Timaeusis in several
respects closer to the Phaedo and the Republicthan to the later group.
First, the Timaeus,like those two dialogues, makes a strict divisionbetween the realmsof Being and Becoming, with changelessness(including
motionlessness) repeatedly cited as a feature of the former realm and
changeas a featureof the latter.34This clearlyseparatesthe Timaeusfrom
the Sophist and the Laws, and separates it to some extent from the
Phaedrus- though, as we saw, the Phaedrusis not entirelystraightforward
about the relation between change and Being. Second, the Timaeus
employsthe notionof a tripartitesoul; however,here Platois quiteexplicit
thatonly the rationalpartis immortal,whichplacesit withthe Republicand
againstthe Phaedrus.Again, at 69c8ff. there occursa list of "terribleand
necessaryaffections"associatedwiththe mortalpartsof the soul;these are
pleasure,pain, rashness,fear, anger,hope, sensationandlove. The immortal part has none of these, except in so far as it is pollutedby the mortal
parts;in fact, it is to minimisethispollutionthatthe mortalpartsarehoused
in separatepartsof the body.35Butthislist hasmuchin commonwiththe list
of "motions"of the soul in the Laws, whichI quoteda littlewhileago; and
those "motions"are certainlynot regardedas merelytemporary,or as a
pollutionof any kind. Once more, the Timaeusappearsto fit less with the
For a fuller discussion of ways in which the Phaedrusmyth departs from the Phaedo
and the Republicwith regardto the natureof the soul, see Nussbaum,op. cit., section II
("Moral Psychology"), pp. 92-107.
34 See, for example, 27d5ff., 29b5ff., 35al-3, 48e5ff., 51e6ff.
3S It is curious, however, that at 42a-b most of the items on this list are apparently
assignedto the immortalpartof the soul also. But Plato does say that this is what happens
to it when it is "implantedof necessity in bodies" (a3-4), and that the goal should be to
overcome these states as much as possible. Perhaps, then, this is a reference to the
pollution by the mortal parts that we find in the other passage; if so, however, it is a
confusing one, since the mortal parts have not yet been introduced.
33

22

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

later dialoguesthan with the earlier ones. Finally, the Timaeusdoes not
definesoul as thatwhichmovesitself, or suggestthatit is responsiblefor the
motion of everythingelse. It is true that the World Soul does undergo
certainharmoniouscosmicrevolutions(35alff.); but these are initiatedby
its creator,not by itself, andthey arenot whatcausesmotionin otherthings
- though they do appearto play some kind of regulatingfunction.36
Whatwe seem to have found so far is that there are roughlytwo sets of
ideas about the soul in Plato - one of which appearsto be earlierand one
later - and that the Phaedrus, both the proof and the myth, displays
essentially the later set. Now, it may look as if the introductionof the
Timaeusdoes not importantlychangethis picture.For my remarksin the
last paragraphmightsuggestthat we couldretainthe twofolddivision,and
simplygroupthe Timaeuswith the earlierset. There are some oddities, to
be sure; but the Timaeus'view of the soul, it mightbe said, is basicallythe
same as that of the Republicand the Phaedo. However, the situationis not
so simple. Recall that the underlyingconceptionof soul that I detected in
the Phaedrus'argumentfor immortalityhad severalfeatures.Two of these
featureswe have not yet discussed;one wasthatsoul playsa vitalrole in the
operationin the cosmos, and the otherwas that soul is in some sense one probablya singlekindof stuff. These featurescomplicatethe picturein two
ways. First,they alert us to the fact that what I have been callingthe later
conceptionis actuallya lot less unified than I have implied. And second,
they inclineus to groupthe Timaeuswiththe laterset of dialogues,andnot,
as we were doingjust now, with the earlierset. I shallbrieflyexpandupon
these points.
In the Phaedrus'proof, Plato regardssoul as having a cosmic significance;for it turnsout that soul is responsiblefor the motion, includingthe
yEVEOL;,of everything that is not soul. This agrees with the myth; at 246b6

we are told that "allsoul takescare of all that is soulless".Now, this idea is
not presentin the Phaedo or the Republic.In both dialoguesthere is the
suggestionthat the universeis orderedfor the best, and this may suggest
some beneficentorderingforce. But Plato does not tell us about any such
orderingforce, and he certainlydoes not suggestthat it is soul. Again, the
Phaedrusmarks a departurefrom those dialogues; and in this respect,
again, it is in agreementwith the Laws. As the "motionwhichmoves itself
by itself', soul is saidin the Lawsto "'directeverythingin heaven, earthand
36Given the Timaeus'criteriafor Being, the fact that the WorldSoul is in motion should
mean that it is not in the full sense a "thing which is". This is indeed the case; for the
mixtureout of whichthe WorldSoul is formed (as well as other souls) is a mixtureof T6 ov
and T6 yLyv6otcvov(35al-3). As far as I know, this idea is perfectly unique in Plato.

23

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

sea" (896e8-9). But the other dialoguewhichmakessoul a cosmicforce is

the Timaeus.As we saw recently, the view is not the same as that of the
Phaedrusand the Laws. It is not suggestedthat the WorldSoul initiates
motion in other things, nor that it is self-moving.In the frameworkof the
Timaeus,the originof motioncan be accountedfor in other ways;there is
the Craftsman,who creates and sets in motion the WorldSoul itself, and
there is also the disorderlyprecosmicmotion in the Receptacleof Becoming. Instead,the WorldSoul is said to be the "mistressand ruler"(34c5)
of its body, the cosmos; and its function is apparentlyin some way to
preserve order. A further difference (if I was right earlier) is that the
Phaedrusdoes not employthe notionof a WorldSoul. (The positionof the
Laws in this respect is not clear to me, for reasonswhich I shall go into
shortly.) But despite these differences,the Timaeusseems in at least one
way closer to the later dialoguesthan the earlierones.
The final feature of the conception of soul we extracted from the
Phaedrus' proof - albeit tentatively - was that soul is a single kind of stuff

(in the sense explained- see note 22); how does this comparewith other
dialogues?The Phaedoand the Republiccontainno hint of sucha conception, though they do not obviously exclude it. The one other dialogue
wheresoul definitelyis regardedas a stuffin my sense, and a singleone, is
the Timaeus.As we have noted, the Craftsmanmixesa certaincompound
substance, and out of this substanceall souls are formed. Now, the substanceput togetherin the mixingbowl clearlyqualifiesas a stuff;this stuff
could very well be designatedby the massterm"soul"(thoughPlato does
not appear to do so). Individualsouls (both the World Soul and human
souls) consist, then, of portionsof the stuff "soul";and this is preciselythe
pictureI foundto be suggestedby the Phaedrus.One mightsaythatthisis a
fartoo literalreadingof the Timaeus'mythicalapparatus.But I believethat
one should alwaystake seriouslythe details of Platonicmyths. Obviously
some features,suchas the mixingbowl, will be factoredout whenwe tryto
penetrateto the core of doctrinePlato is tryingto present.But if Plato had
not meant to convey the impressionthat there is a kind of stuff of which
souls are composed, he could have writtena differentmyth that did not
have these implications;he was not forced to write it as he did. (On this
point, see also notes 3 and 32.)
Here, then, is anotherapparentpoint of contactbetween the Phaedrus
and the Timaeus.37Let us see, finally, whetherwe can relate the Laws to
I have throughoutbeen cautious about attributingthe notion of soul as a stuff to the
Phaedrus;and it will be rememberedthat the alternative,which I considered but found

3'

24

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

these two. Plainly,the evidencefor any notionof soul as a stuffin the Laws
will be indirect;but there is one point that seems to me suggestive.This is
that Plato is strikinglyevasive about the number of souls that run the
cosmos. He is clear that this is done by one or more good souls; but he
refuses to commit himself as to how many. (This is why it is hard to tell
whetheror not he has in mind a WorldSoul.) Mostlyhe uses the singular
4uxiwithoutarticle,as in the Phaedrus'proof;but he speaksalso of "soul
or souls"(899b5,cf. 898c7-8),andhe neverdecideswhetherthe singularor
the plural is more appropriate.He seems, then, to be either uncertain
about, or uninterestedin, the individuationof souls;yet this appearsnot to
detractin anywayfromhis confidencein the doctrineof soul as activatorof
the cosmos. Now, this does not, of course, show conclusivelythat he is
thinkingof soul as a stuff. On the other hand, if he was conceivingof soul
(or simplyof good soul) as a uniformstuff distributedabout the universe,
this insoucianceabout the numberof good souls would be entirelyunderstandable.If thereis a certainkindof stuffwhose natureit is to activatethe
cosmos, then it is indeedof minorimportance- supposingthatthe motions
of the cosmos are whatwe are seekingto understand- how portionsof this
stuff are to be individuated.However, this is as far as we can pursuethe
question. It is possiblethat the Laws as well as the Timaeusis in line with
the Phaedruson this matter;but the evidenceis too slenderfor us to decide
withanyconfidence.Moreover,even if all thisis correct- andI amfarfrom
insistingon it - it still leavesout of accountthe Laws'idiosyncraticnotionof
evil soul. The one thingPlato is clearaboutin the Laws, with regardto the
numberof souls, is thatthereareat leasttwo. The soul or soulsthatactivate
the cosmos are good; however, there must also be one or more evil souls
(896e4-6), to accountfor the presence of evil in the world. Of course, it
mightstill be thatthere was a single stuff "soul"(whose essence was selfmotion) encompassingboth good and evil soul. But it is obviouslyimpossible to tell whetheror not this is Plato's conception;the remarkson evil
soul - indeed, on soul in general - in the Laws are just too sketchy.

Whatgeneralconclusions,if any, can we drawfromthis investigation?It

is sometimessuggestedthatthe Phaedrusis a "transitional"dialogue.Now,
we have examinedonly a tiny fractionof the dialogue;and it may well be
that, concerningtopics whichwe did not touch upon, this verdictis a fair
less likely, was that the Phaedruspresupposesa notion of WorldSoul. It is worthpointing
out that even if I was wrong in my choice between these alternatives- even if we should
read the Phaedrus as implying a World Soul - the closest parallel would still be the
Timaeus.

25

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions

one. But with regardto the natureof the soul, I believe, it needs at least to
be substantiallyqualified.For to say that the dialogueis transitionalis to
implythatthereis an earlierset of doctrinesanda laterset of doctrines,and
that the Phaedrusis between the two. This is not entirelydevoid of truth,
but it is misleadingin two ways. First, as we have seen, in as muchas one
can speak of two groupsof dialogues,an earlierand a later, the Phaedrus
seems in most respects(again, of those that have been the subjecthere) to
belongsolidlywith the latergroup;the transition,for the mostpart,occurs
before the Phaedrus,not duringit. But second, thereare not just two views
of the soul in Plato. The Phaedoandthe Republicmaywithsomejusticebe
considered as expressinga single view - though even this is perhapsan
exaggeration;but amongthe laterdialogues,thereis far less homogeneity.
In particular,the position of the Timaeus,on this as on so many other
issues, is baffling. It is dangerous,then, to speak of "transitions"in this
context. We cancertainlyfindvarioustrendsin Plato'swritingson the soul;
andwe can tryto describethe positionof the Phaedrusamongthesetrends.
But we shouldnot thinkin termsof stablebodiesof doctrinebetweenwhich
the Phaedrusmight be consideredas intermediate.This conclusionis no
doubtnothingsurprising;people haveoften told us not to forcePlatointo a
straitjacket. Much more interesting - or so I hope

are the detailed

considerationswhichhave led us there.

Universityof California

26

This content downloaded from 200.3.149.179 on Sat, 22 Nov 2014 11:06:30 AM

All use subject to JSTOR Terms and Conditions