Two Conceptions of Vacuum

7/29/2019 Two Conceptions of Vacuum

http://slidepdf.com/reader/full/two-conceptions-of-vacuum 1/20

Two Conceptions of Vacuum

Author(s): David SedleyReviewed work(s):Source: Phronesis, Vol. 27, No. 2 (1982), pp. 175-193Published by: BRILL

Stable URL: http://www.jstor.org/stable/4182149 .

Accessed: 26/11/2012 06:01

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 [email protected].

.

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

http://www.jstor.org

This content downloaded by the authorized user from 192.168.82.216 on Mon, 26 Nov 2012 06:01:43 AMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=bap

http://www.jstor.org/stable/4182149?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp





http://www.jstor.org/stable/4182149?origin=JSTOR-pdf

http://www.jstor.org/action/showPublisher?publisherCode=bap



Two Conceptionsof Vacuum

DAVID SEDLEY

1. The question

Writerson ancientphilosophyhave alwaysbeen impressedby the early

atomists'conceptualbreakthroughn introducingvacuum,or void, as an

intelligiblenotion. But discussionof the concept itselfhas been rareand,

on the whole, disappointing.It is as if the doctrine were so obviouslycorrectand sensiblethatno one hadpausedto askwhether t really s quite

so straightforward.Mycontention s that it raisesconsiderable onceptualdifficultiesof itsown, andthatthesegradually merged,andweretackled,

over a long period. I shall be concerned primarilywith the theoretical

aspectsof void, and not with its cosmologicalrole or with the empirical

arguments orand against tsexistence.

Void,To ?v6v, is literally theempty'.Does thatmean emptyspace?So it

isregularlyassumed.In the index to CyrilBailey'sThe GreekAtomistsand

Epicurus,or example, the entryunder'void'simplyreads'see space'.Butthereis, as a matterof Greek usage,at least one otherthingthat Torevov

could mean,and that is 'emptiness'.Viewedas emptiness,void wouldnot

be a kind of space or place. Rather, a portion of void would be the

emptiness n such and such a place.And as an occupierof place,it mightevenbe consideredcapableof locomotion.

There should be nothing intuitivelyabhorrentabout the idea of some-

thingwitha purelynegativecharacterisation ccupyingplacesand moving- a gap in the traffic,for example.When you carryyourthermos laskto

work,you woulddo well to think of thevacuum n it as movingfromplace

toplacewithit. Ifyouinsist nsteadthatthe vacuum nit is empty pace and

therefore ncapableof moving in space, you may have to concludethat

throughoutyourjourney the vacuum in your flaskis being replaced n a

constantstream.These considerationsare not meant to tell decisivelyin favourof the

space-occupiernterpretation f void, but to showthat that interpretationis not too implausible to be entertained.Whetheror not a particular

thinkeractuallyadoptsit, consciouslyor unconsciously,willdependpartly

on other features of his system. For example, if he considersvoid an

element,capableof being partof a compoundbodyand of movingaroundwith it, he is more likely to think of it as a negativesubstanceoccupying

175







space;whereas f he introduces t only in order o provideroom for bodies

to move,he may be satisfiedwiththeconceptionof it as empty space. Mycontention will be that the formerview is characteristic f early Greekthought, but that a transition o the latter can be discerned n the fourthandthirdcenturies,B.C.

2. The orerunners f atomism

It is generallyagreed that the notion of absolutevacuum was first put onthemap by theearlyGreekatomists,LeucippusandDemocritus.But their

work grew out of a backgroundof theoreticaldebate, which we cannotafford to overlook f we areto understand heirconceptual ramework.The earliest fumbling attempt to come to terms with void must be

attributed othe Pythagoreans f the6th centuryB.C., nwhose cosmologythe world is said to inhale void from the infinitesurroundingbreath.'Itwould be pointless n the presentcontext to probe the precisemeaningofthisstartlingdoctrine,and I mention t merelyto pointout how easily voidwas assimilated o an insubstantial-seeming ccupantof space, air. Evenwithout this explicit evidence, it would be certainthat at least one early

conception of void was as something ike thin air, for Anaxagoras,prob-ably writingbeforethe emergenceof atomism, oundit an adequaterefu-tation of void (as understoodby his contemporaries)o demonstrate hecorporeality f air.2Absolutevacuum s somethingwhich fallsrightoutsideordinaryhumanexperience,and we shouldnot besurprisedf it wasatfirstinsufficientlydistinguished rom the least substantial tufffamiliarto thesenses. Reflection on the English idiom 'vanish into thin air' may helpconfirmthis.

I shall not suggestthat Leucippusand Democritusare themselves o besaddledwithacomparablyprimitivenotion of vacuum.Their erminology,as we shall see, puts its entirely negativecharacterbeyond doubt. Butinsofar as their theory is likely to have evolved as a modificationandrefinementof earlierassumptionsabout vacuum,we should not be toosurprised f we find them retainingat least some of their inheritedcon-ceptual framework.I shall maintainthat the status of void as a space-occupier s an exampleof this.

It is widely recognisedthat the Eleatics Parmenidesand Melissushadposed a challengeto the logicalcoherenceof void, and that the atomists'achievement must be understood at least partly as a response to that

challenge. Here too, then, we might hope to throwlight on the atomisticconception of void, by asking precisely what it was that they were

176







attempting o rescuefrom the Eleaticelenchus.Unfortunately here s too

littleagreementon the interpretation f Parmenides'poem to make thisaverypromisingproposal. I thereforeconfine myself to outliningone pos-sibleinterpretation, ne which seesa verydirect Parmenidean ncestry orthe atomisticdualismof body and void.

Thejob of cosmologywas to reducethe worldto the simplestpossiblesetof constituentprinciples.BeforeParmenides, osmologists ended to countmonisma virtue: f all phenomenacould be reduced o manifestations f asingle primal substance,the extreme of conceptual economy had beenreached.Parmenides,whateverwe take to be his own philosophicalpur-

pose, discreditedmonismas a cosmologicaltheoryby suggesting hat nodifferentiations f anykind werepossible n a worldso conceived,becausebeing,quabeing, is entirelyhomogeneous.In the secondhalf of his poemhe went on to demonstrate that the minimum condition for therehabilitation f cosmologywas anunacceptableone, the introductionof asecondelement,homogeneous n itself but distinctfromthe first element.At least one reader,Aristotle,3got the impressionthat there was somecorrespondencebetweenthese two elementsand what Parmenideshad intheWayof Truthcalledrespectivelybeing'(roreov) nd'not-being' 'or ijiov). That suppositionwould, if

nothing else, make good sense of Par-menides'rejection of cosmology as conceptuallymuddled,because not-being, he observed, does not exist.The atomistsdecidedthat they couldworkwith a dualisticscheme such as Parmenideshad hypothesisedandrejected,by holding that not-beingwas not, afterall, a self-refuting on-cept. Thus the atomistic universe consists, with enviable simplicity, ofbeingandnot-being.Ofthese, being,or 'theexistent', sequatedwithbody,not-being,or 'thenon-existent',with void.

If there is any truth in this derivationof atomistic dualismfrom thedualismdismissivelysketchedby Parmenides n his Way of Seeming, itssignificancefor presentpurposes is as follows. Parmenides' chemewas,quite properly,understoodas a dualism of two elements. Hence anyoneequatingvoid with one of theseelementsshould be predisposed o treat tas a movableentitycapableof occupyingwhateverplacesareoccupiedbythecompoundobjectsof which it is part.Indeed,evenif the Parmenideanancestryshould be doubted, thereis sufficient ndependentevidencethatthe earlyatomiststreatednot only body but also void as an element(thewordthey would probablyhave actuallyused is 'principle',&pxiq).4hevery clear andnumerousreports o thateffect mightbe suspectedby some

of forcingatomism nto an ill-fittingdoxographical traitjacket.But sinceEpicurus s in one reportdistinguished rom his atomistpredecessors n

177







thisvery point,5 here seemsgood reasonto trustthem.

Perhapsthe strongestchallenge to my interpretation f the atomists'vacuumwill be basedon Melissus.Did Melissusnot arguethatwhatexists

is immobilebecauseif it movedit wouldmove ntovoid,andvoid does not

exist?Surely then, it will be argued,the Melissanchallenge which the

atomistswere answeringhad already dentifiedvoid as emptyspace.

In my view, this, the standardreadingof Melissus30 B 7.7-10,is mis-

taken.6A translationwillhelp to explainwhy.

(7) ov& XEVE6V CGTLVoV6Ev- TO yap XEV?0V o68Ev kaTLV OV'x &v oVv Etq TO yE ujqv. Oi68

xivrLTav v'OXWp7lOaL yap ovx 'xeXL Ua8tL, &XXairTxiv koi;V. El EV yap XEVEOVjV,

iMT avtS 70bXEV6VXEVOOE pi E6VTOS UxEX?LbXtiM,noxpAUIeL. (8) VvxvovE

xvi&paLov oix av E?vj.TOyap &poa6v o'vx&vvTOv iTXEwV E'LVaLLOiwVrT3L tVXV@o,&XX'

rb o &pLov yE XZVE?MTOV yLVETaLTOrrxvo. (9) xpLoLV & Tovv XPI 1OL1UMOL

TOV1TXEi(XOToV I 'nX&.ELs?V OVVX.pet Tl * ELcb&XVroat,vrXfVw- Et 8E isirreXwPtd

FLAVrEL8iXeTtl, 'rrXi,V. (10) &V61yXnTOiVVVTX\EV EIVOL, t X?VOVRhWaTLV.E 'roivuv1XEWV JTLV,, XLV.

. .L

(a) And nothingof it is empty. For what is empty is nothing.Well, what is nothing

could not verywell exist.7(b) Nor does it move. For it cannot give way at any point, but is full. For if there

were such a thingas emptyit would give way intowhat wasempty; butsince thereis

not such thing as empty, it has no pointat which to give way. (Dense and rarecould

not exist. For whatis rare cannotbe as full aswhat is dense, but whatis rarealreadythereby becomes emptier than what is dense. And that is the criterion for dis-

tinguishingbetweenwhat is full and what is not full. Hence if something gives way

or accommodates, it is not full, but if it neithergives way nor accommodates, t is

ful.)(c) [summary]So (a) it must be full, if thereis no such thing as empty;and so (b) if it

is full, it does not move.

The standard nterpretationacesthreeobjections.First, t mustsomehow

construe 'rroXwpetvand hense also XwpeLv,hichby a regular inguistic

convention will retain the sense of the precedingcompoundform) as a

simpleverbof locomotion:hencethe ratherpuzzled-soundingranslation,'it has nowhere o retreat'n 7. Second, t must gnore8, on rareanddense,

as an irrelevant intrusion. Third, it cannot easily explain why the

immobilityof what exists arisesfrom ts beingfull.There is no need to imagine an externalvoid into which what exists

would transport tself. Why should Melissusconcernhimselfabout any-

thingexternal,givenhisdemonstrationhatwhatexistshasnoboundaries?

Whathe is denyingis an internaladmixtureof void, whichwould make

what exists rareor spongyand thusenableit to 'giveway'((MoXwpeiv)t

some point. For according o 8-9 the 'empty' s characterised y rareness,thesignsof which areabilitytogivewayandabsorbency.His claim,then,is

178







that the non-existenceof void makes what exists internally mmobile -

morelike a stackof bricks hana sponge.Void is envisagedas thatwhichmustbe mixedin with a substance o makeit less thantotallydense.Andthat, as far as I can see, is at least as compatiblewith the 'emptiness'interpretation s with 'emptyspace'.

3. The early atomists

Aristotle n his discussionof void in PhysicsIVseveraltimes(e.g.213b31)asserts hat the proponentsof void regard t as unoccupiedplace;and both

he in the fragments of his work On Democritus(fr. 208 Rose) andSimplicius n his commentaryon the Physics8attributethe same view toDemocritus by name. But on closer inspection this evidence crumbles.Simplicius' attributionis derived from Aristotle's,and is therefore notindependentevidence.And Aristotle's ssertions itselfof dubiousvalue.Itsuits him to treat void as place, because he has alreadydefined place insucha way as to deprive t of independentexistence,and he now seizes theopportunityto tar void with the same brush (especially214a16-22).Butelsewhere(215a11) he is just as happy to treatvoid instead as a kind ofnegative substance (pa ov TL), or privation,where this will afford him afurtherground of refutation.9

I thereforeconcludethat Aristotle'sevidence on this question s of littlehistoricalvalue,'0 and I prefer to settle the matterby relying on our onesecure item of evidence, the atomists'own nomenclature or void. Threeinterchangeablepairs of terms were adopted for the twin principles ofatomism,bodyandvoid.These are 'the existentandthe non-existent'T ovxai TO [LOv), thethingand the nothing' 'orBEvxai Tbotr#v), and 'thefulland the empty' To ITXipEsxxt T xev6Ov).he first twopairs of terms 'thethingand the nothing',and 'the existentand the non-existent' can only

be understoodas designationsof that which occupiesspace. To call emptyspace itself 'nothing'or'non-existent'would indeedbe to deny that there ssuch a thing. But to call the occupantof a pocket of space 'nothing'or'non-existent'would be to asserton the contrary hatthere s such a thingasempty space; and that is what the atomists wanted to assert. And thesymmetrical ooking pairingof To rrXpesand Tox?vOvsuggests that if To

x?v6v is empty space then Tb rXiipes s filled space; but ToXipes is in fact

identified with atoms, and if an atom is a filled space it will becomeextremelyhard to see how it can move. So there s everyreason o expectTo

xev6v oo to designatenot empty space but the negativesubstancewhichoccupies empty space.

179







It may be felt that in thatcase the atomistswere unwiseto hypostatise

this mysterious negative occupant of space. Epicurusseems to havethoughtso, as we will see in the nextsection.Butforthe timebeing I shall

stick to showing the enormous advantagewhich it conferred on their

system.By makingbothof theirprimary lementsoccupantsof spacethey

wereable to maketheminto formalcontradictories.t seemsclearenough

that the threepairs of termswerechosenwith this objectin mind.Everypoint in space is either occupiedor unoccupied.If it is occupied,it is

occupiedby something,Biv;if it is unoccupied, t isoccupiedby nothing,

p.q8v. If it is occupied by something,that thing is existent,and can be

called 'fullness'(rorArpes); if it is occupiedby nothing,its occupantisnon-existent,and can be called'emptiness'ror mvOv).o byrelyingon the

plausibleandwidespreadassumption hatto existis to occupya place,the

atomistscould comfortably it back and allowany existingthing,i.e. any

occupant of place, to sort itself out into one or other of their pair of

contradictoryelements or into a complex of the two. The Law of the

ExcludedMiddlewouldensurethat nothingescapedclassificationunder

the scheme. By selectingas elementsa pair of formalcontradictorieshey

came up with anontologicalschemeof theutmosteleganceandeconomy.

All thatwouldhave beenjeopardized f only one of theelementshadbeen

an occupantof placewhilethe otherhadbeen a speciesof placeitself.

Itmaybe objected hat theschemeas Ipropose o interprett is indanger

of hypostatisingplaceas a thirdkindof existingthing.I do notmyselffeel

that thedanger s, historically peaking,a realone.The interpretation oes

not requirethat any theorisingaboutthe natureor statusof placeshould

have been done. I have triedto show that, given theirconceptualback-

ground,the view of vacuumas a quasi-substance nd place-occupierwas

simplythe naturalassumption or the atomiststo make.Besides,even if

they had faced up to the challenge,theywouldprobablyhave thoughtit

ill-conceived.The body-void dualismwas their answerto the question

'What s therein the universe?', nd anycandidatefor thisrolewouldbe

assumed o existin a place.Tomakethesamedemandof placeitselfwould

be to initiate an infinite regress as theircontemporaryZeno of Elea

mischievously riedtodo.11Questionsaboutthe statusof place,asalsothat

of time,arehigher-ordermetaphysicalquestionswhichshouldbe thought

no morepertinent o the cosmological ystemof theatomists hanto those

of AnaxagorasandEmpedocles.Of course it is well knownthat Leucippusand Democritusboughtthis

schemeat a price;the priceof declaring hatthe non-existent xists(to pilov etvac), or, on an alternative version, that the nothing (o IUi?v) exists.12

180







These paradoxical logans are the legacy of the Eleatics,whoseargument

againstvoid was preciselythat,being the nothingand the non-existent, tdid not exist.Buthow did theatomists hinktheycouldgetawaywithsucha bare-faced self-contradiction?How can they have supposed that byannouncingthat the non-existentexists they were doing anythingmorethan reassert he Eleaticrefutationof void?

One type of solutiontakes it as a purelyhumorousad hominemmove.Theslogan is mostcommonlycitedin the form The existentno moreexiststhan the non-existent',but Democritus s alsoreported ohaveexpressed tas 'Thething ('r6Uv) no more existsthanthenothing TrRqUiv)'.hislatter

formulationcould be thoughtto havea rhetorical orce: 'YoudenythatToW#9?vexists,but actuallyitscontradictory,T'oUv, sounds if anythingmoreproblematic'.On thisinterpretation E'vwouldbe a humorouscoinagelike'couth'or'ept',invented to embarrass nopponent.But even if theword slikely to have been an unfamiliarone to Democritus'readers,13it is soregularlyisted alongwith his otherdesignationsof body andvoid thatit ishardtodoubt that itwas one of hisowntechnical erms.He andLeucippuswere,arguably, hefirstphilosophers o developa technical erminology. tis tempting to connect this fact in turn withone of his fourarguments orthenon-natural haracter f language, hevcwvvpovrgument,14whichrestson the observation that there are some things in the world for whichlanguagehas failed to supply a name:To'8Ev ould be meantas anexampleof a missing name.

Often a 'no more'(ov [a&XXov)lausebehaves as a self-evidentpremissfor some argument:no more (i.e. 'there s no more reasonfor it to be thecase that. . . ') p than q; but p: thereforeq. This is, for example, how theatomists argued that since there is a world here there must be worldselsewhere,by askingwhat morereason herewas for the one to be truethanthe other.15But 'the existent no more existsthan the non-existent'couldhardlyfunctionin that way. On the face of it, there is farmore reason forthe existent to exist than the non-existent.The paradoxmust, therefore,havereceivedsome theoreticaldefence.

JonathanBarneshas recentlyproposed the much morepromisingsug-gestion that we understand 'The non-existent exists' in terms of theFregeandistinctionbetween' x exists' n the sense of'x is a real thing',and'x exists' in the sense of 'there is x'. By 'the non-existent'the atomistsintendedthe firstof these twosenses, .e. 'thatwhich isunreal'.By sayingofthis that it 'exists'they meant'There s thatwhich is unreal' a perfectly

intelligible claim, which a non-platonistmight, for example, make withreference o numbers.

181







I have two worriesabout this.First,whyshould theatomistswant tocall

void 'unreal'?Barnes akes this conclusion o follow from the premisses I)that to be real is to be a space-fillerand(2) that void is not a space-filler.hope I have alreadyshown reasonto doubtthe second of these premisses.Void is a space-filler. At least, it occupies some parts of space just aseffectivelyasbody occupiesothers.Moreover,f theatomistshadregardedvoid as unreal t would be hardto understandwhy they gave it the statusofan element alongside body, or what Democntus meant by his celebrateddeclaration, Byconvention there is colour,by conventionsweet,by con-vention bitter, but in reality ( ) atoms and void'17 odd language for

someone who considersvoid unreal.There is a second objection.Now thedistinctionwhich Barnesattributesto Democritus s moreor less the one whichthe Stoics werelaterto make.Void for them is not one of rarovTor, eal existent things, because onlybodies'exist'.Voidonly 'subsists'(q,C 'TxL), which meansroughly thereis such a thingas' void. Butit would be hopelesslyanachronistic o expectLeucippusor Democritusto have expressedany such distinction notmerely,that is, to have isolated the existential ense of the verb'tobe' butalso to have found two distinctshadesof meaningwithin it. Barnes, o be

fair, admits this,and suggestsonly that the atomistswere'feelingtowardssuch an insight'.But that would hardly be enough.In declaring hat thenon-existentexists they were blatantly nvitingridicule,unlessthey had adefenceprepared. f thisparticular nsight s one whichtheywereno morethan feeling towards,they needed some other insightthere and then tooffer to theiropponents.

Democritusdid, it is true,recognise hatsome wordshavemore thanonemeaning. But I know of no evidence that he, or anyone else until a muchlater date, put this insight to work on the solution of philosophicalpro-blems. What comparable device can we attribute to him withoutanachronism? can only think of one: the move of explaining hatsome-thing is true in one respect but not in another. This can be illustrated from

Leucippus'contemporaryEmpedocles,who thought he could evade theEleatic ban on change by makingall cosmicchange strictlycyclical(B 26,8-12):

Thusinsofar as (jL pv) one has learntto come intobeing out of many, and when onedisintegratesmany are formed, to that extent theyare becoming and have no stablelife. But insofar as (f&8e) these things nevercease from their continual change, tothat extent they are for ever unmoved in a circle.

I see no reason why Democritus should not have answered the Eleatic banon void by a similar tactic, defending the paradox that the non-existent

182







exists by explainingthat in one respectvoid existsbut that in anotherit

doesnot. Inwhat respects?Let us assumeonce again the equationof 'exist'with 'occupya place'.The atomistscould then say that when a place is

occupiedby nothing,insofaras the occupantis nothingit does not exist,butinsofaras it occupiesa placeit does exist.

4. Epicurus

ForEpicurus' onceptionof void theprincipal extis Ep.Hdt. 39-40:-

1. &OX Riv xoL TO 'rT&v artL <awILaTaT Xal Xev6v>. c4.a'rT iV y&p AS FUTLV yi'aIp

aLUo0Ls i-'TLTdXvTwvotprvpEL,xaO'iv &VayxCLOV To &8&Xov T XoyUY[L() TEX)aipE06aL$oGlTEp lTpOELVTOV.O1TOOS EEAa aV, 'OVXEVOV xiX1 x(pcl XCXL VfXqYIq)ULV6VvopL&oJ.v,Ovx

av Exe T&oaroTX bsov 'qvoVOE 0L' Ou XlVel'rO,xaotxrep qmiVe'raL XLVo0ULEVa.ITaptbe

5. raDia i'vo'V '8'bt'qVi"vaL MVxTOfLVTe nEpLXi1iTrCosVi?rEvaX6ycs TOLSTEPAX1'TOLS

Wsxa0' bXocs foausXaAf3sv6geva, xeviRh 's Tr&ovwv ovRTuT'xaTaaoex6mr

1. < a6RaraxaixEvov> Gassendi: <0au L,aTaxai orros> Usener 3. '71TEpf'rpOELITOV

To .rrp6O0EV'Li1 iv ov (&BB TD) Z3 f ) xEv6v XT\.B' P Co FZ1:WosMep'ITpOELITOVO

7rp6O0EEl Rh iV bV XEV'OVrX. B': 'Wa'TrepnpOEvtoV.T6070 8e eC 1 6v, bV xevov XTX.

emendavit Usener: IlTx?Ep -npoEtv?ovTrb -fpOEV. Et <be> Rh 'v 8 XEvOvwrX. aiii. 5.ovW' sener ovre

Moreover, the totality of things is <bodies and void>. That bodies exist is uni-

versallyattested by sensationitself, in accordancewithwhich it is necessary o judge

by reason that which is non-evident, as I said above; and if place, which we call

'void', 'room' and 'intangible substance', did not exist, bodies would not have

anywhere to be or to move through as they are observed to move. Beyond these

nothing can even be thought of, either by imaginationor by analogy with what is

imagined,as things grasped n termsof completesubstancesand not as what we call

accidentsor propertiesof these.

There are two crucial emendationsto the text, and I must start with a

defenceof them.Fortunatelywe have a check on this textin a veryclosely

parallelpassageof Lucretius,1.419-428.The openingsentenceis hopelessas it stands, and the supplement < uatwar xai XeV`v> is the obvious one to

adopt, both because the same formulais attestedfor Epicurus n other

sources18 nd becauseit corresponds xactlyto Lucretius' orpora untet

inane at the equivalent point in his version (1.420). The second

emendation, in 3, is more controversial.Usener's brilliantconjecture,

which I have adopted, has been virtually gnoredby editorsthis century.Most have preferredto follow Gassendi in retaining 'U?EJ?'rppoevov TO

183







rp6a00Evs the end of one sentence, and in addinga connective n thenext

sentence,to make ed<8e> jLilv 8 x?vOv---. Usener ends the firstsentenceat ffpodfTfov,and emends To ffpo6ev to ro6'os ?, retaining ovwith the best

mss. in place of B.Several considerations avourthis.First,Tbopo60ev isdisturbingly redundant with ?rpoetlrov,especially where the reference is

only to the previousparagraph; o an emendationwhich gets rid of it iswelcome.Second,Usener'stextrequiresonly two lettersto be changed ntheoriginalreadingof the best ms., the Borbonicus the readingof B antecorrectionemis frommy ownautopsyof thecodex - Usenerdidnot realisequitehowstrongly t supportedhim).The alternative mendation equires

four letters in B's original readingto be changed. Third, the resultantreading corresponds much more closely to the equivalent clause inLucretius 1.426-7), Thenagain, if there werenot placeandroom, whichwe call void'; the alternativereadingleaves nothing in Epicurus' ext tocorrespond o Lucretius'ocus.

Why have all these advantagesbeen overlooked?Perhaps t is becauseUsener's extseems to make voidandplacestraight ynonyms,and editorshavehesitated to introducesuch an obvioushowler nto Epicurus'extbyway of an emendation.The answer o this is twofold.

First,with or without the emendationthe very sameconflationof voidwith place is implicit at the end of the same sentence:'(Withoutvoid)bodies would not have anywhere to be or to move through,as they areobserved to move'. Void provides stable location as well as passage.Giussaniand Baileytry to mitigate heconfusionbyobserving hatstrictlyspeaking hereareno stablelocations n Epicureanphysics,becauseatomsarein perpetualmotion.19But thiswill not do. The laws of atomicmotioncannot be assumedat thisearlystagein Epicurus' rgument,because heythemselves ollowupon the proof of void. Besides,the last three wordsofthe sentence make it clear that Epicurushas not atomsbut phenomenalbodies in mind;and these can be said to have stable location.Moreover,the sametreatmentof void asplacerecursatEp. Hdt. 42 andseveral imesin Lucretius.

Second, s thisconflationof placewith void a howler?Thosewho believeso mightbe temptedto argue that 'roIros,ccupiedspace,can hardlyberegardedas a secondconstituentof the universeon a parwith the bodieswhichoccupy it, whereas the void intervalsbetweenthose bodiesclearlymightbe. It is for the controversial oid intervals hatEpicurus houldbearguing,and that is a quite separate task from that of arguingfor the

existence of occupied place, which no opponent would have denied.Althoughthese difficulties are real ones, I want to argue in whatfollows

184







that there is no simple oversightor confusion on Epicurus'part, but a

doctrine evolved to cope with conceptual difficulties first raised byAristotleand stillkeptaliveby sceptics n the Hellenisticera.The story, as I reconstruct t, startswith Aristotle'scriticismsof the

notion of void. I have alreadysuggestedthat Leucippusand Democritusconsideredvoid an occupantof placerather hana speciesof place itself,but that Aristotle or reasonsof hisown tendedto attribute he latterviewto them. This assumptionallows him to raisea fundamentalconceptualdifficulty PhysicsIV,213alS5-19):-

Those who speak of void set it up like a sort of place and vessel.They thinkthat it is

fullwheneveritcontains the masswhich it is fittedto receive,butvoid whenever t isdeprived of it - as if void and plenum and place are all the same thing yet theiressence is different.

Thisdescription s meantto introduce henotion of void fordiscussion,notas anargumentagainst t. Butit is clearlya loadeddescription,and ithelpssoften us up for the formalrefutationof void which follows at 216a26-b16:-

Also considered in itself the alleged void would appear to be a trulyemptynotion.Forjust as, if someone puts a cube in water, waterequal in volume to the cube willbe displaced, the same happens in air, only it is imperceptible o the senses.Indeed,

it is always the case with every body capable of displacement that it must, unlesscompressed, be displaced in its natural direction of displacement, always eitherdownwards, f its movement is downwards ike that of earth, or upwards, f it is fire,or in both directions,whateversort of thing be insertedin it. But in the void this isimpossible,since it is not a body, but the cube must be penetratedby an extensionequal to that which previously existed in the void - just as if the water, or air, werenot displaced by the wooden cube but completelypenetrated t.

Butthe cube has a magnitudeequal to the void it occupies, and this, althoughit ishot or cold or heavy or light, is different in respect of being from all its attributes,even if not separablefrom them: I mean the volume of the wooden cube. So if itcould actuallybe separatedfrom all the other attributes,and be neither heavy nor

light, it will occupy an equalvoid and be in the same place as the portion of placeorvoid equal to itself. What then will be the differencebetween the body of the cubeand the equal void or place? And if there will be two such things in the same place,why not any numberof things?This is one absurdand impossibleconsequence.Andagain, it is clear thatthe cubewill even when it is displaced have this thing whichallotherbodies have too, so that if it is no differentfrom theplace why should we makea place for bodies over and above each one's volume (if the volume is somethingwithout attributes)? tcontributesnothing to suppose such a distinctequal extensionbelonging to it.

The crucialquestion raised here is what happens to void when a body

enters it? The first stage of the argument s the importantpart for ourpurposes.Thevoidcannotbe displacedby thebody,because t is notitself

185







a body, and we are expectedto agreethat the only sort of thingwhich a

body can displaceis anotherbody. Thereforethe void mustremainandbecomecompletelycoextensivewiththe body.Aristotledoesnotconsiderthe possibility that on being filled the void perishes,perhapsassuming,withsomejustification, hatthe atomistswouldnot haveboughttheirwayout of trouble at the priceof makingone of their two primaryelementsperishable.

The absurdity of a void coextensive with a body might have beenthoughta sufficientrefutationof thevoidtheory.Aristotle's wnapproach,however, s to supposethatthe voidwill survive n theguiseof the body's

place. But this conception of place as the intervalor extensionfilledby abody is one he has alreadyrebutted 21lb14-29),and the secondstageofthepresentargument s really,I think,justa reinforcement f thatrebuttal.If thisextensionwhichthebodycomesto fill issomethingdistinct rom hebody's own volume, then we will be landed with two indistinguishablethingsin the same place; and if two, thenwhy not any number?Aristotlecan only maintain this line of argumentby ignoring his own earlierdemonstration hat strictlyspeaking a placecannot itself be in a place.Idoubt f it impressedhisreadersverymuch,andIfindeveryreason o think

that it wasthe firststageof theargument akenon its own thatcausedmostperplexity o thechampionsof void.To see what sort of dilemma(or,moreaccurately, rilemma)Aristotle's

argumentposed for Epicurus, t will be helpful to look at it in a laterHellenistic ormulation,as presentedby SextusEmpiricusM 10.20-3).Ofcourse,thisactualformulationpostdatesEpicurus,and is probablyaimedin the first instanceat the Stoicconceptof place. But it is of interestbothbecause t sets out fairlystarkly hechoicewhichEpicurushadfaced,andalso because it could well representa continuingtraditionof criticisminspired

bythe

Physics IV argument(the immediatelyfollowing argu-ments,M 10.24-36,can be seen to drawheavilyon the material n PhysicsIV):

For if there exists any body-housing place, it is either body or void. Now, thebody-housing place is not body; for if every body must be in a place, and place isbody,theplacewill be ina place,and thatplaceina thirdplace,the third n a fourth,and so on ad infinitum.Therefore the body-housing place is not a body. But if thebody-housing place is void, then when thebodyarrivesat it either thisvoid remains,or it is displaced,or it is destroyed. And if when the body arrivesat it it remains, twill be simultaneously void and full, void insofar as it remains, full insofar as ithouses the body. But it is unthinkable to call

the same thing both voidand full.

Therefore the void does not remain when the body arrivesat it. But if the void isdisplaced, the void will be body; for that which is displaced from here to thereis

186







body. But the void is not body. Thereforeit is not displacedeither,when the bodyarrivesat it. Besides, if it is displaced when the body arrivesat it, it will no longerbehousing the body, and that too is absurd.It remainsthereforeto say that the void isdestroyed,which is again impossible.For if it is destroyed t undergoeschange andmovement; and if it is destroyedit is generable. But the generableand destructiblethingwhich undergoeschange and movement is body. Therefore the void does notperish either. Thus if place is neitherbody, as we have proved,norvoid, as we havepointed out, no place can exist.

With these argumentsin mind, let us consider what possible responsesEpicurus ouldoffer to Aristotle's onundrum,whathappenstovoidwhenit

is approachedby a body? Plato had proposedthe law that anythingwhich is approachedby its own opposite musteitherwithdrawor perish(Phaedo102dff.).Andon anotheroccasionEpicuruswashappyto agree.Ideduce this fromone of Lucretius'argumentsagainstEmpedocles' our-elementtheory(1.760-2),where it is observedthatsome of theseelementsaremutuallyopposed;thereforewhentheyapproacheachothertheymusteitherperishordodgeeachother.But n thecaseofvoidneitheralternativewas verypalatable.Take the firstalternative, hat void withdrawson theapproachof body.SextusechoesAristotle ncommenting hatonly a bodycan be displaced.Epicuruswouldhave to agreewiththis, sincehe rightlyholdsthatvoid, unlikebody,cannotact or be actedon (Ep.Hdt.67).Thesecond alternative,that it perishes, looks no more promising.Sextus'argument s thatperishing nvolveschangeand movement,of whichonlybody is capable.ButI wouldnot expectthatto convinceEpicurus, f onlybecause he holds that, strictlyspeaking,even body does not perish,butonlyundergoesredistribution.The reasonwhyhe cannotcontemplate hepossibilityof void's perishing s that his entiresystemis founded on thefamiliarprinciple that the universewill be unstable and unintelligibleunless itsultimateconstituentsarepermanentones;andtheseconstituentsarebody andvoid.

If neither alternative is acceptable,the only available move short ofabandoningvoidaltogether s to allowthatvoiddoesafterall remainwhena body entersit. But the only way in whichit could coexistwith a bodywould be bybecomingthatbody'splace.HenceEpicurushasnochoicebutto follow Aristotle's ead in conflatingvoid with place. Democritus'con-ceptionof void asa negativesubstancewouldno longerdo,sincethatcouldin nowaybe imaginedasexisting nthesameplaceasbody.HowEpicurushandledthis conflation is admirablyexplainedby SextusEmpiricus n a

passage(M 10.2)which,extraordinarily,asbeen almosttotallyneglectedby modernscholarswhendiscussing he Epicurean heoryof void:-

187







Therefore one must grasp that, according to Epicurus,of 'intangible substance',as

he calls it, some is named 'void',some 'place', and some 'room',the namesvaryingaccording to the different ways of looking at it, since the same substancewhenempty of all body is called 'void', when occupied by a body is named 'place',andwhen bodies roam through t becomes'room'. But generically t is called 'intangible

substance' n Epicurus'school, since it lacks resistant ouch.

Epicurus nvents the technicalexpression intangible ubstance' orspacein its broadest sense, whetheroccupied or unoccupied.He then explainsthe familiarwords 'void', 'place'and 'room'as being merelythe termsbywhichwe refer o it in specificcontexts: void'when it is unoccupied,place'

when it is occupied, and 'room'when bodies move through t (this lastdefinition is backed up by an etymologicalassociationof -xpa, 'room',withx(pCLv,go', which I heretranslate roam').All three termsnamethesame thing, intangibleextension.When bodies pass into and out of this,itremains unaffected in all but name. As Aetius puts it (1.20.2 = fr.271

Usener),

Epicurussays that the differencebetween void, place and roomis one of name.

Itmay be askedwhy he shouldcall place, i.e. occupiedspace,anintangiblesubstance.Presumably he justificationwould be that it is not the spaceitself which is tangible,but the body occupying t - when the bodymovesout of it the space itself will not offer it any resistance.At any rate,a verysimilar view was held, accordingto Simplicius, by the majorityof thePlatonistsand Stratoof Lampsacus fr.60 Wehrli).These people spokeof

space as in its own nature a void coextensive with the cosmos, but inpracticealwaysfilled with body.Epicurus' intangible ubstance'mayhavea strongclaim to be the first clear recognitionof geometrical paceas athree-dimensionalxtensionwhich persistswhetherornotit isoccupiedbybody. But if anyone arrived at the same notion before him,21t is these

Platonistsof whom Simpliciusspeaks, workingno doubt from Plato'snotoriouslyproblematicdepictionof space in the Timaeus.Epicurus'irstphilosophical training was in Platonism, and it may be that we would

understand heoriginsof his theoryof spacebetter f we knew moreabouttheworkof men likeXenocrates.

To returnto Ep. Hdt. 40, I understandEpicurus'wording"'place",which we call "void", "room" and "intangible substance" , as an

annoucement hat he will use its variousnames indifferently,probably n

orderto emphasizethat the differencebetweenthemis one of context,not

of essence. True to his word, he does elsewherefluctuatein his usagebetween 'void', 'place' and 'intangiblesubstance'withoutapparentdis-

188







tinction,although'void' is certainlyhis favourite. t mayseem odd that he

puts 'place'first,rather hanthe masterconcept intangible ubstance'.Theobject is no doubt strategic:place is the least controversialof the fourterms.

The rangeof alternativenames also serves to stressthat at least some

parts of space are occupied and at least some unoccupied.And this isconfirmedby his proofsof itsexistence: a) it is needed to provide ocation,i.e. there must be occupiedspace;(b) it is neededto allowmotion, .e. theremust be unoccupiedspaceinto whichthingscan move.

I thereforeconclude that Ep. Hdt. 40 is not muddled,but representsa

coherentdoctrineevolved in responseto Aristotle's riticismof the notionof vacuum.As a result of this shift of position, Epicurus has jeopardised the

symmetrical ntithesisof bodyand voidwhich had been sucha meritof theearlyatomists' system. There is now none of Democritus'pairs of con-tradictory erms which Epicuruscan use. 'Thingand nothing',and 'theexistent and the non-existent'are unsuitable,not only because they are

semantically puzzling but also because they are, as we saw, names foroccupantsof space, not for spaceitself. As for'the full and theempty',nowthat 'the empty'meansemptyspace 'the full'would inevitably mply'fullspace',which is not a suitabledefinitionof an atom. (WhenEpicurususes

-nXipqst is as a predicatenotof bodyper se but of those portionsof bodywhich contain no void gaps (Ep. Hdt. 41, 42).) He does, however,haveanotherpairof contradictorieswhich characterisebody and void respec-tively, and these are 'tangible'and 'intangible'.He never uses them inDemocriteanfashion as the names of body and void, but he appeals tothem to show thatbody andvoid themselvesare true contradictories.Therelevantargument s preservedby Lucretius 1.430-439):-

Praetereanil est quod possisdicere ab omni 430corporeseiunctumsecretumqueesse ab inani,quod quasi tertia sit numeronatura reperta.nam quodcumqueerit, esse aliquid debebit id ipsumaugminevel grandi vel parvodenique, dum sit.cui si tactus erit quamvis levis exiguusque, 435

corporisaugebit numerum summamque sequetur;sin intactileerit, nulla de parte quod ullamremprohiberequeat per se transiremeantem,scilicet hoc id erit, vacuum quod inane vocamus.

Beyond these there is nothingwhich you can call distinct from allbody and separatefromvoid, to play the role of a third discoveredsubstance. For whatever will existwill have to be in itself somethingwith extension,whether arge or small, so long as

189







it exists. If it has tangibility,however light and faint, it will extend the measureof a

body and be added to its sum. Whereas if it is intangible, and unable to prevent

anythingfrom moving through t at any point, it will undoubtedlybe the emptiness

which we call void.

Most editorstranspose ines434 and 435, butI have retained hems.orderandtakenaugmine 434) to mean 'extension'.22 his gives Lucretius vitalpremiss: 'Whatever will exist will have to be in itself somethingwithextension,whether large or small, so long as it exists'. Presumably hismeans 'three-dimensional xtension'.The premiss ooks like one usedbyZeno of Elea (29 B 2 D.-K.), that somethingwithout magnitude ould not

exist because when added to something else it could not increase thatthing'ssize. With this premissestablishedLucretius an safelyproceed o adilemma: either this extendedthing is tangibleor it is intangible.If it istangible t will, when added to a quantityof body,increase t. Theimplicitconclusion s thatit must thenitself be body,and thisfollowsprovided hatwe assume a further premiss,that if a quantityof body is increasedbyaddition that which has been added to it is itself body. The argumentcontinues: f on the otherhandthe extendedthing n question s intangible,

being unable to resist moving bodies it will allow them to pass straightthrough precisely he essentialfunctionof void. It will be seen now thatthe premissthatonly spatiallyextended thingsexist is a necessaryone -

otherwise all sorts of impostorsmight get included under the heading'intangible', ven PlatonicForms.

So the antithesisof 'tangible'and 'intangible'comes to the rescue andseems to make Epicurus' body and void' a neatly symmetricalpair of

contradictories,as they had been for Leucippusand Democritus.Butunderneath hedefinitionalsymmetry herenowlurksa strongontological

asymmetry.Body and space are in some sensejoint constituentsof the

world,yetmany partsof spaceare completelyoccupiedby body.This snotin itself an absurdity: he Stoic principles,matterand God, alsocoincidespatially,as do body and mind in the view of many philosophers.Evenso,I thinkEpicuruswas perfectlywell aware hatvoid as he conceived t was ofa very different order of being frombody. He resistedthe temptation ofollow Leucippusand Democritus n callingit an element,and used thatname for atoms alone (Ep. Pyth. 86).23 He never makesthe mistakeof

regardinga compound body as made out of atoms and void in combin-

ation.The void of the earlyatomists,being a substancehousedin space,could presumablybe an elementof a compoundbody and move around

with it; but once Epicurushad identified void with place, it becamestationaryand no longer availableas an elementof movablecompound

190







bodies. Compound bodies consist of atoms variouslyspaced out. Space

provides he locationof theseatoms,theintervalsbetweenthem,androomforthemto move;but it is not itselfpartof the compounds.

Whythendoes Epicuruspairbody andvoid in the formula thetotality

of things s bodiesand void'?He does not mean by thisthattheuniverse s

compoundedout of them in the way that a house is compoundedout of

bricksand mortar.Rather,he means thatthey are the only two ordersof

beingthatare required oaccount or theuniverse.All othercandidates or

the title 'existent', ncludingtime,eventsandproperties, an be accounted

for as attributes of body, incapable of separate existence (Lucretius

1.449-82).Space alone cannot. And that is because it exists even wherebody does not.24

Christ'sCollege,Cambridge

1 Aristotle, Physics IV 213b22-7. In defence of an early date for this doctrine,see C. H.

Kahn, 'Pythagorean philosophy before Plato', in A. P. D. Mourelatos(ed.), The Preso-

cratics 1974).2 59 A 68 Diels-Kranz. No doubt the argumentwas, like 59 B 17,phrasedas a criticismof

currentusage in general,notjust of a specific doctrine.3

Metaphysics 986b27-987a2.4 According to Eudemus, fr. 31 Wehrli,CJrOLXELOVn the sense 'element'was firstused byPlato.5 Aetius, Plac. 1.3.14-18,p. 285 Diels, DoxographiGraeci 1879). Cf. frs. 173, 184, 188-9,

190, 192, 194, 262, 264, 487 in S. R. Luria, Demokrit Leningrad1970).6 Cf. G. Reale, Melisso, testimonianze frammenti(1970), 176-92, or a surveyof current

interpretations,and add J. Barnes,The PresocraticPhilosophers 1979) I 217-9. Some of

theweaknesses which I pointout in the standard ype of interpretationhave alreadybeen

discerned by Reale, Barnes,P. Albertelli (GliEleati, testimonianze frammenti 1939)),J.

Loenen(Parmenides,Melissus, Gorgias 1959)), and others.7 The real subject of the first sentence is 'it' (whatexists),as comparisonwith theclosing

sentence shows. The first oMiv must therefore be adverbial. But mytranslation is

designed to bring out the word play of the first two clauses. oux ... ovv ... yE is the

negation of YoiBv,o that the ovv has no inferential force:see J. D. Denniston, The GreekParticles(ed. 2, 1954), 422-5, and Loenen, op.cit., 163-4.8 E.g. Simplicius In Ar. Phys. 394,25 ff., 397,2 ff. (= fr. 250 Luria),571,22 ff. (= fr. 254

Luria).Aristotle'sinfluence may also explain Eudemus'tendencyto treat earlyviews of

void in the same way: cf. Eudemusfr. 75 Wehrlion Democritus (= fr. 251 Luria),andfr.

65 on Archytas(= 47 A 24 Diels-Kranz);and note 21 below.9 Cf. Met. 1,985b4 ff. (= fr. 173 Luria), where the atomists'void is not 'place'but an

element, characterisedas Fav6v;De caelo 302a1ff., wherevoid is thought of as occupying

a place; and ib. 309b 17ff., where one of the possibilitiesconsidered s that voidwould be

capable of locomotion.10 The likelihood that Aristotle'sdepictionof void as placeis unhistorical s well notedby

F. Solmsen in Aristotle'sSystem of the Physical World 1960) 140-2.It is certainlynot my

191







purpose to question Aristotle'sintegrityin the matter.For one thing the view of voidwhich I attribute to the atomists is more likely to have been an assumption than aworked-out heory for whichchapterand verse could be cited. For another, hey no doubt

also had occasion to speak of empty places, and for all I can show to the contrarytheymay have misleadinglyused r6xEv6vof these as well. My point is that the cardinalnotionof void, as featured n their dualism of atoms and void, cannot be of emptyspace. Met. I,

985b4 ff. (see previous note) suggests that Aristotle may have appreciated his.11 29 A 24 Diels-Kranz.12 E.g. frs.78, 173, 194, 261 Luria.13 See Loenen, op. cit., 79, and A. C. Moorhouse,'LiEN n classicalGreek', CQ n.s. 12(1962), 235-8. The form is rare,but is found as early as Alcaeus, 320. L.-P., and still inTheodotion'sOldTestament ranslation,ap. Philop.De opif ii 1.59.12.Forattributions f

it to Democritus as a regularterm, see frs. 78, 172, 185, 188 Luria.14 Fr. 563 Luria15 Fr. 1Luria;cf. frs. 2-8, and Barnes,op. cit., II, 251-7.16 Barnes, op. cit., 11 100-3. I have learnt a great deal from this marvellous book, andwould not want my disagreementon this issue to obscure that fact.17 Fr. 55 Luria.18 Usener's <alxm-Ta xai T0'r6os> has had an undeservedly bad press. The same formula

occurs at fr. 76 Usener and Nat. 34.14.7-9 Arrighetti, and is suggested by a4uaTa pEV ...

8rros e in what follows (on the readingsee below). But Gassendi's<o ta.ra XvL xevOv>

also has good parallels n frs.74-5 Usener, in addition to Lucretius' upport.19 C. Giussani, Studi Lucreziani 1896), 25; C. Bailey, Lucretius,de rerumnatura ibrisex

(1947), II653; cf. Brad Inwood, 'The origin of Epicurus'concept of void', CP 76 (1981),273-85, p. 280-1. The Democriteanthesis that the only 'real'truthsare truthsabout atomsand void, never about phenomenal objects, is regularly assumed to be shared byEpicurus.I argue against the assumption n 'Epicurus'Refutationof Determinism' n G.

Pugliese Carratelli ed.), XuvtrpIS Naples, 1983).20 This text has been consistently overlooked because Usener included it only in theaddendato his Epicurus 1887): fr. 271, pp. 350-1. It is now discussedby Inwood(art.cit.281-2), who dismisses it as Stoic-contaminated.This, however, could only be establishedif it were shown to be at variancewith Epicurus' heory, and my own view is thaton the

contrary t makesexcellent sense of it. I see no conflict between saying thatvoid, placeand room differ accidentally,or accordingto different ways of looking at the same thing

(bfrtj3oX&sa very Epicureanterm: cf. Ep. Hdt. 70 for just this usage),and saying thatthey areall namesfor the same thing, as at Ep. Hdt. 40 and fr. 271 Usener (citedbelow):one might,forexample, describethe relationshipof the MorningStar to the EveningStarin both ways without inconsistency. And Inwood's other objections are, I believe,adequatelymet by what I say in the main body of this paper. Only one item of evidenceneeds explaining away. At 1.503-6 Lucretiussuggests that body and space cannot be

coextensive. This is technically incorrect,on the account which I adopt. But note thatLucretiusquickly corrects this to empty space (1.507-9). Inwood's interpretation,onwhich Epicureanvoid comes out looking rather ike Democritean void as I have inter-preted it, conflicts with at least three items of evidence (art. cit., 280-1) as well as with

Epicurus'principlethatvoid cannot be acted upon (Ep. Hdt. 67).

21 The same notion of container space may already be implicit in Archytas47 A 24Diels-Kranz,as reportedby Eudemus,although t is hard tojudge how fartheconception

192







of 'r6nos here may reflect Eudemus' Aristotelian understandingof void (see note 8

above).22 In this I am following Brieger's1894 Teubner edition, although he emends aliquid(433) to aliquo, perhaps rightly.23 Cf. Aetius, cited note 5 above.24 Versions of this paper were read to meetings at London, Princeton, Stanford and

Chicago.I benefited from the discussionson all fouroccasions,andoweparticular hanksto Myles Burnyeat, John Cooper, Rick McKim, Henry Mendell, Brad Inwood, IanMueller and ElizabethAsmis for their comments and suggestions.Above all, it is to Brad

Inwood's paper (cited note 19above) and to ourextended correspondenceabout it, thatI

owe the impetus to write on this subject. The reader can be safely referredto it for awider-rangingdiscussion and fuller bibliography than I have provided here. It is a far

richer paper than my foregoing criticisms may suggest, and its demonstration thatEpicurusreadPhysicsIVclosely (pp. 282-4) is so powerful that I have felt able to take thisforgranted n myown paper. Finally, for the leisure to write this final versionof the paperI am deeply indebted to the Humanities Council of PrincetonUniversityfor the awardofa visiting fellowship in the Fall Semester 1981-2,and to the Institute or Advanced Study,Princeton, or membershipduring the second term of that year.

193

Documents

Two Conceptions of Vacuum