ARISTOTLEONPOETS:

ACRITICALEVALUATIONOFRICHARDJANKOSEDITIONOFTHEFRAGMENTS

MalcolmHeath

Abstract
ThispaperprovidesacriticalexaminationofRichardJankoseditionofthefragmentsofAristotleOnPoets
(Janko2011).Section1discussespassagespreservedinlateancienttextswhichJankoassignstoOnPoets.
Section2identifiesproblemsintheevidencepreservedinthePhilodemuspapyri.Section3assessesindi
rectevidenceforAristotlestheoryofkatharsis,andconsiderstwocontestedpointsinAristotlesdiscussion
ofkatharsisinPolitics8:themeaningofmousik,andthenatureofhisresponsetothechallengeposedby
PlatoinRepublic10.

RichardJankoseditionofPhilodemusOnPoemsbook11combinedexceptionalphilo
logical expertiseand ingenuity withaseeminglysuperhumanpatienceandattention
todetail.Indeed,hiseditorialachievementfaroutshoneinintellectualstaturetheme
diocrity of Philodemus and his opponents. He has now edited the remains of books
34.2These,thoughlessextensivethanthoseofbook1,holdoutaprospectofmorere
wardingcontent:inbook4(atreasure,asJankocallsit:viii)Philodemusapparently
begins an engagement with Aristotles lost dialogue On Poets which continues into
book5.Forthisreason,JankosupplementshiseditionofPhilodemuswithanedition
ofthefragmentsofOnPoets,supportedbyasubstantialintroductionandcommentary.
Anearlierversionofthiscollectionoffragments,withoutaGreektextoranyelaborate
scholarlyapparatus,wasappendedmanyyearsagotoJankostranslationofAristotles
Poetics.3Itdidnotthenreceiveclose scrutinyfromreviewers,andthereisariskthat
thisnewversion,similarlyappendedtoothermaterialbywhichitmaybeovershad
owed, will not be reviewed in detail. An edition which more than quadruples the
number of fragments (400) certainly merits careful scrutiny. Yet few scholars com
mand the full range of expertise needed for a comprehensive assessment of material
drawn from late ancient sources as well as from Philodemus. Since I am neither a
papyrologistnoranexpertonPhilodemus,myowncontributionmustdrawprimarily
onmyknowledgeofAristotleandoflateancientintellectualculture.Accordingly,the
firstsectionofthisreviewdiscussesinsomedetailtheidentificationandinterpretation
ofsomeofthepassagespreservedinlateancienttextswhichJankoassignstoOnPoets;
thetreatmentofevidencepreservedinthePhilodemuspapyriinSection2willneces
sarily be briefer. Section 3 will give separate consideration to a variety of sources of
evidenceforAristotlestheoryofkatharsis,including(mostimportantly)evidencepro
videdbyAristotlehimself.
Inwhatfollows,IfollowJankosnumerationtoidentifytextswhichhehasprinted
asfragmentsofOnPoets.Buttoavoidprejudgingthequestionofwhethertheidentifi
cation of a given text as a fragment of On Poets is correct, I shall use the prefix J
Janko2000.
Janko2011.Inwhatfollows,referencestothisbookwillbegivenbypagenumberalone.
3 Janko1987.
1
2

STUDIAHUMANIORATARTUENSIAvol.14.A.1(2013)
ISSN14066203http://sht.ut.ee

(instead of the standard abbreviation F).4 Similarly, since the interpretation of frag
mentsissometimesanissue,Iprovidemyowntranslations,givingJankosrendering
forcomparisonwheredivergencesarepotentiallysignificant.
1.Theevidenceoflateantiquity
Janko separates testimonia from fragments. The latter include, for example, excerpts
fromPolitics8,thoughthesearenotfragmentsofOnPoetsinanynormalsense.Janko
explainsthatheclassestextsthatattesttotheexistenceofOnPoetsastestimonia,while
his fragments include texts which report or provide other indirect evidence for the
contentofOnPoets,eveniftheydonottransmitAristotlesownwords(403).5Combin
ingthetwokindsofevidenceisadefensiblepolicy,thoughusingthetermfragments
(ratherthan,forexample,sources)6forbothispotentiallymisleading.Itis,atanyrate,
importanttomaintainaconceptual,ifnotaterminological,distinction,sincedifferent
kindsofsourceplacedifferentconstraintsoneditorialpractice.Itislegitimatetotryto
restore the probable wording of the original text by emendation if one is editing a
fragmentinthestrictsense,butnotifoneiseditingalaterauthorsparaphrase.Janko
isnotconsistentlyalerttothisdistinction.7Consider,forexample,J36c,fromJohnthe
DeaconscommentaryonpseudoHermogenesOnMethod(150.79Rabe):8
, .
Arion of Methymna introduced the first tragic drama, as Solon taught in his work with the
titleElegies.

WhenJankoemendstheexpression... to ... (Arion

of Methymna introduced the first form of tragedy), what has he emended?
SubstitutingthemoreAristotelianexpressionwouldbejustifiedinareconstructionof
Aristotlesoriginaltext.ButsuchareconstructionpresupposesthatJohnhasgivenusa
fragmentinthestrictsense(transmittingAristotleswords,thoughinacorruptform),
ratherthaninJankosextendedsense(transmittingAristotlesthoughtinwhatmaybe
afreeparaphrase).Sincewehavenoreasontobelievethatthispassageisafragmentin
thestrictsense,eliminatingfromthetextofJohntheDeaconaperiphrasisfortragedy
attestedelsewhereinlateancientscholarship9isillegitimate.
AnevenmorefundamentalquestioniswhetherwehavereasontobelievethatJ36c
isanAristotelianfragmentinanysense.Itisnotanamedcitation. Thesourceofthe
information is explicitly identified in Johns text as Solon, not Aristotle. Perhaps the
AconcordanceofJankosnumerationwiththoseofRose1886,Gigon1987andLaurenti1987canbe
foundinAppendix2below.
5 Jankodistinguishesdoubtfulfragmentswithanasterisk;heincludesonespuriousfragment,marked
withadagger,butalsoprovidesalistofotherexclusions(4037).
6 Seee.g.Fortenbaughetal.1992:49.
7 This point was made long ago by reviewers of Janko 1984: Barnes 1985: 1045; Schenkeveld 1986:
214;Fortenbaugh1987:1579;Bremer1988:168.
8 Rabe1908apublishedextractsfromVat.gr.2228.Jankodesignatesthisascod.unicus,whichisnot
strictlycorrect:Rabe1908b.
9 sch.AristophanesRan.95( );sch.Lucian21.41(... );sch.Lycophron7.4Scheer( ).The
fact that this reading commits John to what is, to our eyes, a very implausible claim is not in itself
evidencethathedidnotwriteit.
4

statement about Arion attributed to Solon was transmitted by Aristotle: but do we

have any reason to believe that? Aristotle held that tragedy arose out of dithyramb
(Poet.4,1449a911),andProclusreports(J36a)thatAristotleclaimedthatArionorigi
nated dithyramb. But there is a difference between originating dithyramb and intro
ducingtragedy.SoProclusreportcannotsafelyberegardedasavariantofJohnsre
port that Arion introduced tragedy, and gives no positive support to the conjecture
thatSolonsclaimthathedidsowastransmittedbyAristotle.Nordoesthatconjecture
gainanysupportfromthefactJohnhasmentionedAristotleonlyafewlinesearlier.
Hereisthesentenceinaslightlyextendedcontext(150.110Rabe):10
(*J33a.3) , ,
, .(J35)
, . .
. (J36c)
, . .
(*J33a.3)Oncecomedyhadbeendiscoveredinthisway,toavoidtherebeing uninterrupted
merrimenttheydiscoveredtragedy,introducingfrowninganddowncasteyesbymeansofit.
(J35) Both were invented in Athens, as Aristotle says. For it is there that humans first came
into existence. That is why one should attribute to this city the finest and most serious of
studies. (J36c) Arion of Methymna introduced the first tragic drama, as Solon taught in his
work with the title Elegies. But Strato of Lampsacus says that drama was first produced at
Athens,whenThespiscomposedit.

DoesAristotlesauthorityextendbeyondthesentenceinwhichheisnamed?Notifmy
translation is correct: Aristotle is unlikely to have said that humans originated in
Athens,sinceinhiseternalworldthequestionofhumanoriginsdoesnotarise.11But
Jankostranslationofthecrucialphraseisdifferent:itisinAthensthathumanbeings
have become preeminent. This, however, neglects the specific tradition that lies be
hindJohnstext.TheAtticoriginofhumansgoesbacktoPlato(Mnx.237d8b);Aelius
Aristidestook it up inthe Panathenaicus(25,33),andinthe samework healsomen
tionedtheAthenianoriginofweaponry(43).ThesetwoideaswerecombinedbyThe
mistius,whosaysthattheclaimthatweaponryfirstappearedinAtticaisreasonable,
given the Attic origin of mankind (Or. 27, 337a). This passage in Themistius comes
immediately before a statement about the origins of tragedy (J34). It is therefore not
likelytobeacoincidencethatJohnssupportingargumentfortheAthenianinvention
oftragedyreproducesthestructureofThemistiussupportingargumentfortheAthe
nianinventionofweaponry.ItfollowsthatJohnscitationofSolonsstatementabout
Arion(J36c)isseparatedfromthestatementattributedtoAristotlebyname(J35)bya
statementthatJohn(or,rather,hislateancientsource)hasmodelledonsomethingsaid
by Themistius in the context of, but not about, the origin of drama. This intervening

See Appendix 1 below for a more extended presentation of the passage in John from which
*J33a+J35+J36careextracted(149.6150.10Rabe),alongsidethepartialparallelinthecommentaryonOn
MethodattributedtoGregoryofCorinth(RG7.1328.89.1Walz).Thetwocommentatorssharedacom
mon late antique source, from which they independently selected and adapted material (Rabe 1908a:
1302).Notethat*J33b(fromGregory),whichabbreviatesthesourceof*J33a.23,includesareference
tofestivalswhichisadisplacedsummaryofthesourceof*J33a.1.
11 SeeDC1.3,270b1625;Met.12.8,1074a38b14;Mete.1.3,339b1630;1.14.Discussionin(e.g.)Sedley
2007:11920;Cambiano2002;Verlinsky2006.
10

statementisnotAristotelian.ThereisthereforenoreasontoregardJ36c,oranypartof
J35beyondthenamedcitation,asafragmentofAristotle.12
Themistius demands further attention. He is prominent in Jankos collection of
fragmentsofOnPoets,furnishingsubstantialadditionstothefragmentstobefoundin
Rose or Gigon. Janko maintains that Themistius was exceptional, for someone of his
latedate,inhavinghaddirectaccesstoAristotleOnPoets:inthefourthcenturyADthe
De poetis became the preserve of learned philosophers, since so far as we know only
IamblichusandThemistiusreadit(392).IshallhavemoretosayaboutIamblichusin
3. For the present, it is enough to note that Janko offers no evidence for his direct
knowledgeofOnPoets:hesimplyrefersustoJ55,inwhichIamblichusdoesnotname
Aristotle,letaloneOnPoets,andwhereJankoscommentaryspeaksmorecautiouslyof
dependenceonAristotle,directorindirect(520).InthecaseofThemistius,thebelief
thathehaddirectknowledgeofOnPoetsisapparentlygroundedon,andinturnpro
videssupportto,thebeliefthatheisthesourceofotherwiseunattestedfragmentsof
Aristotle. This reasoning need not be viciously circular: the more reason we have to
believe that Themistius mediates fragments of On Poets that are not attested in any
earliertext,themoreplausibleitisthathehaddirectaccesstoOnPoets;andthat,in
turn, will lend support to the identification of fragments in Themistius which might
otherwise have seemed doubtful. But the risk of circularity is there, and we need to
satisfyourselvesthatthecaseforidentifyingfragmentsofOnPoetsinThemistiusisa
strongone.
We may start with J34 (Themistius Or. 27.337ab). This follows the passage men
tionedabove,onwhich thesecond part of J35ismodelled, andcontainsThemistius
statementabouttheoriginoftragedy.AccordingtoThemistius,tragedy,thoughper
fectedinAthens,wasinventedinSicyon,justascomedyoriginatedinSicilyandwas
enhancedinAthens.AfirstpointtoobserveisthatJankoshypothesiscreatesanap
parentcontradictioninAristotle.IfJ34,J35andJ36careall,asJankoclaims,fragments
of On Poets, then Aristotle said that tragedy was invented in Sicyon, invented in
Athens,andfirstintroducedbyArion.Jankoproposesacomplicatedwaytoreconcile
thisseeminglyextravagantvarietyofclaimsabouttheoriginsoftragedy(36571).But
thereconciliationisunnecessary:thefactthatAristotlesnameappearsinonlyoneof
these three fragments (indeed, only in the first sentence of J35, according to which
Aristotle said that tragedy was invented in Athens) suggests that the problem is an
artefactofJankosmistakenattributions.
ItmightbeobjectedthatJ34,atleast,ismarkedasAristotelianbyitsreferencetothe
Sicilian comic dramatists Epicharmus and Phormus. Janko comments: the presence
of the ungrammatical phrase at Poetics 1.5, 1449b6, which has
often beendeemedaninterpolationfroma parallelpassageintheDepoetis,strongly
supports the assignation of J34 to the De poetis (366). But what exactly is it that
ThesubsequentreferencetoStrato(thetransmittedtexthasDraco:JankoadoptsPatzersconjecture,
plausibly)suggeststhepossibilitythatitwasStratowhomentionedtheSolonfragmentanddissented
from it. But there are no strong grounds for asserting this: Johns source is likely to have combined
snippetsfrommanydifferentsources.Thispossibility,andthesourcesknowledgeoflateantiqueliter
ature, are illustrated by the fact that the named reference to Aristotle (J35), which is followed by the
adaptationofThemistius,isalsoprecededbyanapparentechoofChoricius(*J33a.3
~ Choricius1.93: :theparallelis,sofarasIhavebeenableto
determine,unique).
12

providesthissupport?Thefactthatthephraseisnotanauthenticpartofthetextofthe
Poetics(i.e.isinterpolated)isnotinitselfevidencethatitisanauthenticpartofthetext
ofOnPoets.NoristhefactthatthephraseinPoeticshasoftenbeendeemedaninter
polationfromOnPoetsevidencethatthephraseisafragmentofOnPoets:thecrucial
questioniswhetherithasrightlybeensodeemedinotherwords,weneedevidence
thatthephraseisafragmentofOnPoets.Combininganunsupportedconjectureabout
thesourceofJ34withanunsupportedconjectureaboutthesourceoftheinterpolation
in Poet. 5, 1449b6 would yield only the flimsiest of selfsupporting structures. Let us
thereforeconsiderthematterinalargercontext.Itisbeyondreasonabledoubtthatthe
phraseinthePoeticsisaninterpolation.13Fromthis,itdoesnotfollowthatthepairing
ofEpicharmusandPhormismusthavereachedThemistiusfromsomeotherAristote
lianwork.TheyarealsopairedinSuda2766and609,whichsupplyadditionalin
formationaboutthem;thesourceisidentifiedin2766asLycon.Soitispossiblethat
ThemistiustookthenameseitherfromLyconorfromsomelinkinadoxographictra
dition leading from Lycon to the Suda. The interpolation in the Poetics might derive
from the same tradition. Alternatively, since the interpolation in the Poetics is appar
entlynotanearlyone,14itmightderivefromThemistiushimself.Theexistenceofsuch
alternativepossibilitiesmeansthatwehavenoreasontosupposethatEpicharmusand
PhormismadeajointappearanceinanyAristoteliantext.ThereforeJankosargument
foridentifyingJ34asafragmentofOnPoetsfails.
Thereisanother,muchlongerextractfromThemistiusthatJankoincludesamong
thefragmentsofOnPoets(Or.26,316a9a=J38+J43a).Thecontextisadefenceofinno
vation.Themistiusfirstmentions(J38)variousexamplesofhumanprogress:statuary,
shipbuilding,diet15andhousing,paintingandmusic,andthedevelopmentoftragedy
fromitspurelychoralbeginningsthroughtheinventionsofThespisandAeschylusto
SophoclesandEuripides.Aristotlessupportisinvokedbynameinthecaseoftragedy.
Themistius then proceeds (J43a) to trace at much greater length the development of
philosophy from ThalesandthePresocraticsthrough SocratesandPlatoto Aristotle.
SinceAristotlesOnPoetsisunlikelytohaveincludedahistoryofphilosophyculmi
natinginAristotlehimself,JankoprudentlytruncatesthepurportedAristotelianfrag
mentbeforeThemistiusreachesAristotlescontributiontophilosophy.Butwhyshould
weattributeanyofthishistoryofphilosophytoAristotle?Itcanhardlybeclaimedthat
Themistius,whoJankorightlydescribesasalearnedphilosopher(392),wouldhave
been incapable of composing such an account himself. But there is evidence that he
did,at least,havesome helpinconceivingtheidea: DiogenesLaertius hasasimilar,
thoughbriefer,comparisonofthehistoriesoftragedyandphilosophy(3.56=J42).This
parallelmakesitprobablethatsuchacomparisonwaspresentinasourcecommonto
Diogenes and Themistius. In Diogenes the history of philosophy does not extend
beyondPlato,andJankoacknowledges(505)thatThemistius(or,lesseconomically,an
SeeTarninTarnandGutas2012:245.Jankohimselfisnoncommittal(365n.2;cf.1987:80,where
the phrase is retained), which makes his appeal to the support of the (in his view, possibly mistaken)
viewthatthephraseisinterpolatedinPoet.5,1449b6evenstranger.
14 The names were not in the Greek ancestor of the SyriacArabic tradition: see Gutas in Tarn and
Gutas2012:3346.
15 Themistius here mentions balanophagy, which also appears in *J33a.1, from John the Deacon (see
n.10above).ThatmaygivefurthersupporttotheinferencethatthecommonsourceofJohnandGreg
orymadeuseofThemistius.ThereisnoreferencetobalanophagyinextantAristotle.
13

intermediatesource)haselaboratedthematerialtakenoverfromthecommonsource
atleasttotheextentofextendingthehistoryofphilosophy.Butitwouldnotbesafeto
assumethatthisistheonlyrespectinwhichThemistiusmodifiedthecommonsource.
EvenifweknewthatAristotlewasthecommonsourceofthepassagesinDiogenes
(J42) and Themistius (J38+J43a), therefore, we would not know how far Themistius
was adapting and expanding, rather than simply paraphrasing, Aristotle. But we do
notknowthatAristotlewasthecommonsourceofDiogenesandThemistius.Wedo
notevenhavereasontosupposethatthecommonsourcemadeanyextensiveuseof
Aristotle.ThemistiusnamesAristotleasanauthorityonlyinsupportoftheclaimthat
tragedy developed by a series of incremental innovations. This reference to Aristotle
could have been introduced by the common source into a body of material that was
notderivedfromAristotle;orThemistiushimselfcouldhaveintroduceditintoabody
ofmaterialderivedfromthecommonsource.16Indeed,thereferencetoAristotle,sofar
from authenticating the Aristotelian provenance of other material in the context,
countsagainstit:ifallormostofJ38+J43awasderivedfromAristotle,whyshouldhe
benamedjustatthisonepoint?SincethefullextentofThemistiusadaptationofthe
common source is not known, we must recognise the possibility that he was able to
supplyfromhisownwidereadingtheexamplesthatprecedethehistoryoftragedyin
J38,orwhichfollowitinJ43a.Theideathatanaccomplishedandscholarlyoratorwho
had occasion to compose a defence of innovation must have taken it over slavishly
fromasinglesourceisnotrealistic.
Jankodoes,however,haveafurtherargumentforconnectingThemistiushistoryof
philosophywithAristotle.TheendofthetributetoPlatorunsasfollows(J43a.6):
,
, ,
.
Inaddition,hemadeverymanyotherinnovations,andintroduced(blendingastyle[idea]of
discourse [logos] from poetry and bare prose) people asking questions and answering them
andtellingstories,bywhomallwehumansareenthralledandraisedabovetheearth.

There is an apparent parallel in Diogenes, who reports Aristotle as saying that the
style (idea) of Platos discourses (logoi) is between poetry and prose ( ,J43b=D.L.
3.37). Since Diogenes makes this point in a completely different context, the parallel
does not license us to assume that the point about Platos style was made in the
commonsourcewhichwehaveinferredforThemistiusOr.26,316a9d(J38+J43a)and
Diogenes3.56(J42).Here,too,wemustrecognisethatThemistiuscouldhimselfhave
made this addition to material that he had adapted from the common source. More
over, this case provides no evidence to support Jankos claim that Themistius had
directaccesstoOnPoets,sincetheparallelinDiogenesisevidencethatbyThemistius
timethepointwaspreservedinindirecttradition.
More importantly, there are issues of interpretation. Jankos text and translation
differ from those given above in two important respects: he translates logos as dia
16 Laurenti1987includesthisashisF4c(i.e.onlythelast5ofthe23linesofJankosJ38,andnoneofthe
59linesofhisJ43a).IfthiscitationofAristotlewasinsertedbyThemistius,Iamnotconfidentthathe
wouldhavescrupulouslycheckedhismemoryofwhatAristotlesaidagainsttheoriginaltext.Jankode
fendsThemistiusreportofAristotleagainstvarioussuspicionsthathavebeenraisedagainstit(3701):
nevertheless,reportsarenotalwaysreliable(seen.19below).

logue and prints a supplement <>: he... introduced a form of dialogue,

blending it from poetry and bare prose, <imitating> people asking questions and
answeringthemandtellingstories(443).17Thesetwodecisionsnudgethepassagein
the Aristotelian direction that he thinks it should take. But there are two problems.
First,thesupplementisnotadequatelymotivatedunlesswehaveindependentreason
tobelievethatThemistiusismakinganAristotelianpoint.Secondly,thewordwhich
Themistiususeselsewherewhenhewishestoreferspecificallytodialogueisnotlogos,
butdialogos.Inthispassage,Themistiusdoesacknowledgethedialogicalcharacterof
Platosdiscourse;buthedoesbymeansofanallusiontoastandarddefinitionofdia
logue in terms of question and answer (D.L. 3.48; Alb. Intr.1), combined with a re
minderthatPlatoalsoincludesnarrativeelements(suchasthemyths).18Asweshall
see,however,Themistiusconcerninthispassageisnotwiththedialogueformassuch.
InhisdiscussionofThemistius(J43a)Jankomaintainsthattheparaphraseofthis
sameideainDiogenes(J43b)demonstratesthatThemistiusdependsontheDepoetis
(508).Whatexactlyisthisidea?Jankoisconfidentthattheideainbothpassagesrelates
totheframeworkofgenericclassificationwhichAristotleconstructsinPoetics1,with
Socraticdialoguesasaformofmimesisinprose(Poet.1,1447b913).Butwhateveritis
thatThemistiusisreferringto,heexplicitlyregardsitasoneofPlatosmanyinnova
tions.AmultiplyattestedfragmentexplicitlyattributedtoOnPoetsrecordsthatAris
totleawardedpriorityinthecompositionofdialogues,nottoPlato,buttoAlexamenus
ofTeos(J44a,c,e).19SoifThemistiusconcernhereiswiththemimeticdialogueform
assuch,theassignmentofprioritytoPlatocontradictsAristotlesattestedviewsonthe
originsofmimeticdialogueinprose.ButitisquestionablewhetherthatiswhatThe
mistiusisconcernedwith.First,inanauthorofthefourthcenturyAD,itishardnotto
seestylisticanalysisinthephraseblendinganidea,whichechoesthetechnicaltermi
nology of contemporary rhetorical theory.20 Stylistic comment would not be out of
place:thelegitimacyofthepoeticelementsinPlatosprosestylehadbeenamatterof
debateforcenturies.21Sinceideaisusedelsewhereinconnectionwiththeinventionof

Foreaseofcomparison,Ihavegivenanadaptedformofmyowntranslation.InJankosownwords:
he...introducedaformofdialoguebymixingitfrompoetryandprose,<representing>peopleasking
questionsandansweringthemandnarrating.
18 Ithinkthismorelikelythanareferencetothedramatic/diegematicclassificationofdialogues(Plut.
QC711bc;D.L.3.50).
19 Elsewhere Janko highlights Aristotles award of priority to Alexamenus, claiming that it bringsto
light one of two contradictions in Platos argument (that is, in his provocative condemnation of
poetryas):thathecannothavethecreditforinventingtheSocratic...dialogue(319).Ifwehad
toreconstructPlatosargumentfromthispassageofJanko,wewouldhavetoinferthatPlatoclaimedto
haveinventedtheSocraticdialogue,andthatthisclaimplayedanimportantroleinhiscondemnation
of poetry. Otherwise, how would Aristotles denial that Plato invented Socratic dialogue have any
bearingonPlatosargument?(Eventhen,strictlyspeaking,denyingPlatosprioritywouldbringtolight
afalsepremiseinPlatosargument,notacontradiction.)Similarly,whenJankospeaksofPlatosrejection
of poetry in general (320), we would have to infer that this rejection embraced all kinds of poetry
which is false: Plato explicitly licenses hymns to the gods and encomia of good men (Rep. 10, 607a).
When assessing evidence forlost texts, anddrawing inferences from that evidence, weneed totakea
realisticviewofthereliabilityofsecondhandreports.
20 SeeHermog.Id.215.911,305.6Rabe;LonginusF49.1089PatillonBrisson=212.46SpengelHammer.
21 See e.g. Dionysius of Halicarnassus Letter to Pompeius, where he defends his earlier criticisms of
Plato;CaeciliuscriticismsofPlatoevokeasharpresponseinLonginusOnSublimity.Forasurveysee
Walsdorff1927.
17

thedialogueformassuch(Athen.11,505c),thisargumentisnotinitselfdecisive.But,
secondly, Themistius interest is in the effect that Plato has on his readers: all we
humansareenthralledandraisedabovetheearth( ).22Itwouldbeimplausibletoclaimthattheuseofprose
mimesisassuch(whichwasnot,inanycase,distinctivelyPlatonic)wasthesourceof
thatelevatingeffect:whatenthralsandelevatesisPlatosdistinctivestyle.
ThemistiusinterestinJ43aisthereforeintheinnovativeandelevatingadmixtureof
poeticelementsinPlatosprosestyle,ratherthaningenericclassification.Wheredoes
that leave Diogenes (J43b)? He, unlike Themistius, does refer explicitly to Aristotle.
Given that this is a genuine fragment of Aristotle, it is an easy assumption that the
pointbeingmadeinitistheoneaboutgenericclassificationfamiliarfromPoetics1.Yet
betweenpoetryandproseisnotaparticularlygoodformulationofAristotlesargu
mentinPoetics1,thatimitationsinverseandimitationsinprosearesubsetsofasingle,
asyetnamelessart.23Moreover,AristotleinPoetics1doesnottalkaboutPlatoinpar
ticular, but about Socratic dialogues in general: by contrast, the Aristotelian observa
tion which Diogenes reports is specifically about Plato. We know that Aristotle had
viewsontheuseofpoeticelementsinprosestyle(Rhet.3.1,1404a2040),andIseeno
reason why On Poets should not have included discussion of stylistic differences
betweenpoetryandprose.SoJ43bisagenuinefragmentofAristotle,andJankomay
wellberightinattributingittoOnPoets;butbyhastilyassimilatingittoafamiliarpas
sage from Poetics, he has misidentified its original context. If that construction of the
facts is correct, we need not assume that Aristotle took so favourable a view of the
poeticelementsinPlatosprosestyleasThemistiusdoes:heinsistsonthedistinction
betweenproseandpoeticdiction(Rhet.3.1,1404a389).24Thathypothesisis,ofcourse,
speculative;but,inaccordancewiththeneedforcautionwhendealingwithfragments,
ithasthemeritofnotassertingitselfwithunwarrantedconfidence.
There is one further piece of evidence that might be seen as supporting the argu
ment for Themistius dependence on Aristotles generic classification in J43a. Janko
pointsoutthatpsilometriaoccurselsewhereonlyinthePoetics,andcomments(508):
ThemistiususageprovesthatbyAristotlemeantlanguagethatisbareofverse,
i.e.prose...HenceatPoet.1.2,1448a11, cannotmean,as
everyone has thought, prose and unaccompanied verse, but rather prose literature that
lacksverseform,withepexegetic.

Thisargumentisseriouslyflawed.Themistiususageproveswhathemeantwhenhe
usedtheword:whatAristotlemeantshouldbejudged,notfromsomeoneelsesuseof
the word, but from the context in which Aristotle himself used it. I find Jankos
reinterpretation of that context entirely unconvincing. The structure of Poetics 2,
1448a718isbasedonafourpartsequence:

.(i)

ForthemetaphorcompareLong.Subl.7.2,36.1,36.3.
At one point Janko says that Aristotle argues that nonverse mimesis is poetry (217). If this were
true,thenbetweenwouldbeevenlessappropriate.
24 Dionysius of Halicarnassus includes Aristotle among the critics of Platos style (ad Pomp. 1.16,
226.1015UsenerRadermacher).AristotleissharplycriticalofPlatosemptyverbiageandpoeticmeta
phorsatMet.1.9,991a202(=13.5,1079b246;cf.An.Po.2.13,97b379).TheassessmentofPlatoatPol.
2.6,1265a1013ismoresubtlyambivalent:Halliwell2006providesanacuteanalysis.
22
23

 ,(ii) [] ,
, ,
(iii)
, [] . (iv)

.
It is clear that each of the kinds of imitation mentioned above will also exhibit these differ
ences,andwillbedistinguishedbytheimitationofdistinctobjectsinthisway.Forthesedis
similaritiesarepossible(i)bothindanceandinmusicforpipeorlyre,(ii)andalsoinconnec
tion with language and unaccompanied verse: for example, Homer imitates better people;
Cleophon people similar to us; Hegemon of Thasos, who invented parodies, or Nicochares,
theauthoroftheDeiliad,worsepeople;(iii)similarlyalsoinconnectionwithdithyrambsand
nomes one could imitate as Timotheus and Philoxenus did the Cyclopes; (iv) and the very
samedifferencedistinguishestragedyandcomedyfromeachother:forthelatteraimstoimi
tatepeopleworsethanourcontemporaries,theformerbetter.

This sequence recapitulates the arts mentioned in 1, 1447a1316,25 dividing them ac

cordingtothepermutationsofrhythm,languageandmelodyestablishedinchapter1:
(i)=rhythm,withorwithoutmelody(~1447a238);(ii)=rhythmandlogos(~1447a28
b13); (iii) = rhythm, language and melody used in combination throughout
(~1447b246);and(iv)=rhythm,languageandmelodydistributedinseparateparts
(~1447b268).Jankopositsanunnoticedparenthesis (508) in1448a911,embracing
... (Forthesedissimilarities...unaccompaniedverse).26
But this parenthesis destroys the internal logic of the passage, and in doing so also
disrupts its cohesion with chapter 1. I therefore conclude that Themistius used the
word in a different sense from Aristotle. That is not to say that he misunderstood
Aristotlesusage.ThemistiuswasundernoobligationtoconformtoAristotlesusage
whencraftinghisownprose.Hecouldachieveapleasinglyeruditeeffectbydeploy
inganitemofvocabularywhichhehadmetinthePoetics,andcoulddosowithlittle
risk of confusing his readers, since he was using it in a contextually transparent
sense.27Themistiuswasasophisticatedliteraryartist,pursuinghisownagenda,andit
does him an injustice to treat his text as a passive conduit for the evidence for
Aristotleslostworksthatwewouldsomuchliketohave.
2.TheevidenceofPhilodemus
LetusturnnowtothefragmentsfromPhilodemus.Theconditionofthepapyrifrom
which traces of Philodemus text have to be recovered poses special difficulties. The
reconstruction of the text itself from these incomplete and indistinct remains is also
fraughtwithproblems.IamnottheonlyreaderofthefirstvolumeofJankosedition
tohavebeenstruckbythepeculiarityofsomeoftheGreekattributedtoPhilodemus;
thetranslations,too,areoftenonesthatIwouldhavestruggledtogetoutoftheGreek.
2, 1448a7 (each of the kinds of imitation mentioned above) ~
1,1447a21 (theartsIhavementioned).
26 In1448a11, signalstheprogressionfrompurelyinstrumentalmusictoimitationinlan
guage,whileindicatesthatwearenotyetconcernedwithlanguagewithmelodicaccompa
niment.ThusI,likeJanko,readtheepexegetically:butthemeaningislogoi,i.e.unaccompaniedverse
(notproseliteraturethatlacksverseform).
27 SynesiuswassufficientlyimpressedtoreproducetheeffectinDion18,ifBernaysplausibleemenda
tion of to is correct (1869: 116). Less plausibly, Garzya 19723 defends inSynesius,emendingThemistiustomatch.
25

ThatdoesnotprovethatJankosreconstructioniswrong.IfweknewAristotlesMeta
physics or Plotinus Enneads only from papyri excavated at Herculaneum, our recon
structions would be wrong if they were not linguistically challenging.28 On the other
hand,itisdoubtfulwhetherwewouldhavesucceededinreconstructingthepeculiari
tiesofAristotlesorPlotinuslosttextscorrectlyhadwebeenreliantonHerculaneum
papyri.
Theseconsiderationsprovidegroundsforapproachingthereconstructedevidence
ofPhilodemuswithcaution,butnotforignoringit.AtOnPoems4,col.106.89(=J3)
the restoration [ ] [ is compelling. In principle, this formula
would be consistent with a collective reference to Aristotle and his followers; but
readingit(inaccordancewithafamiliaridiom)asaperiphrasticsingularreferenceto
Aristotleis,asJankoargues(2201),moreconsistentwiththemannerofthefollowing
polemic. Janko also makes a good case against the Poetics as the Aristotelian source
(33062);theinferencethatthesourceisOnPoetsisthereforeattractive.Ontheother
hand, we know that Philodemus made use of doxographic aids when compiling On
Poems;29sothepossibilitythatheisengagingwithOnPoetsatsecondhandneedstobe
considered,30 as do his own polemical habits. There are therefore still questions that
needtobeposedaboutthevalueofthisevidenceandaboutitsresponsibleuse.
First, let us consider an example of the work of restoration that needs to be per
formedontheevidenceofPhilodemustextinordertoproducesomethingthatlooks
likeAristotle.Philodemushasjustintroducedthetopicofcharacters;thisishowJanko
reconstructsthetextthatfollows(OnPoems4,col.109.1721):
[ ] [ ] [][. ] []
[ ...]
Andtheonewhosaidtheactionseitherwellorworsewaswithoutdeliberativecapacity.If
itisofthosebetterthanus...(followedbyc.48missingwords)31

FromthishereconstructsafragmentofAristotle(J8):
< > < > <
>
<The poet imitates> the actions either well or worse <of agents either> better than us <or
worse>.32

Longinus,anativespeakerofGreekandaphilosophicallyeducatedcontemporaryofPlotinus,com
plained to Porphyry that copies of some ofPlotinus works that had been sent to himwere full of er
rorsyetPorphyryassuresusthattheywereaccuratecopies(Porph.VP19.213,20.59).
29 See Janko 2289. Janko 391 acknowledges that Philodemus knowledge of other Aristotelian dia
loguesmayhavebeendirectorindirect.
30 TherearesomeitemsofvocabularynototherwiseattestedforAristotle:(elsewhereonly
in Plut. QC 9.14, 744e), (only in [Hipp.] Oct. 86.6 Grensemann before Plut. Mor. 360a,
582d).
31 Janko277translates:Thepersonwhoclaimedthat(thepoetrepresents)theactionseitherwellor
badlywasincautious.Ifontheonehandtheactionbelongstothosewhoarebetterthanweare...
Thephraseinparenthesisisaclarification,notasupplement:thatis,itdoesnottranslatewordswhich
JankobelieveswerepresentinPhilodemustext(seenextnote).
32 Janko417translates:(...thatthepoetrepresents)eitherwellorbadlytheactionsof(characters)who
are(either)betterthanweare(orworse).InAristotelianfragments,Jankoplaceswordswhichhebe
lieveswereomittedbyPhilodemusinroundbrackets(undefinedinhissiglabutexplainedon330;cf.
403).Iwillfollowthestandardconvention,anduseangledbracketstomarkeditorialsupplementstoa
textastransmitted.
28

10

DidAristotlewritethosewordsinOnPoets?Possibly.IsthetextofPhilodemussuffi
cient evidence that he did? No: it provides Janko with a peg for a conjectural recon
structionthatismadepossiblebyhisknowledgeofthePoetics.Therearethreemeth
odologicalcorollaries.First,wherewedonothaveindependentknowledgeofthekind
ofthingthatanauthorislikelytohavewritten,Philodemusmaynotprovideadequate
evidenceforwhattheauthoractuallywrote.Secondly,totheextentthatsuchconjec
turalreconstructionsdependonourpriorknowledgetheycannotsecurelyaddtoour
knowledge of Aristotles views. Thirdly, convergence between On Poets as recon
structedbyJankoandthePoeticsisnotevidencethatAristotlesviewsremainedcon
stant:itisaninevitableresultofJankosmethod,sinceconsistencywiththePoeticsisa
criterion for his identification and reconstruction of fragments of On Poets. Consider
J12:facedwithastatementthatseemstoconflictwithwhatAristotlesaysinthePoetics,
Janko suggests that it belongs to a speaker in the dialogue who opposes Aristotles
view,unlessthephilosopherhimselfarguedinutramquepartem(337).So,whenJanko
later concludes that Aristotle does not seem to have changed his views of poetry
betweenOnPoetsandthePoetics(389),weshouldreflectthatevidenceforachangein
Aristotlesviewscouldnotbebroughttolightbyaprocedurewhicheliminatesappar
entinconsistencies.
Itis,nevertheless,truethatOnPoetswasadialogue.Itisthereforeprobablethatnot
everythingsaidinOnPoetswasastatementofAristotlesownviews.Weneednot,for
example,feelunderanyobligationtoconcludethatAristotlehimselfwasofthecon
sideredopinionthatHomersfatherwasoneofthespiritsthatdancewiththeMuses
(J65a).AsJankosaysofsuchanecdotes,sincetheywerewithinadialogue,Aristotle
did not have to vouch for them himself (317). But this does not apply only to anec
dotes:oneoftheformalcharacteristicsofAristoteliandialoguesacceptedbyJankois
that everything was discussed from opposing points of view (324, my emphasis).33
This,too,hasmethodologicalconsequences.Ontheonehand,argumentsagainstthe
attributionofafragmenttoOnPoetsongroundsofinconsistencywithotherevidence
forAristotlesviewsarenotconclusive.Forexample,ifwehadpositivereasontobe
lieve that Jankos J34, J35 and J36c were all fragments of Aristotles On Poets, that
worksdialogueformwouldprovideapossibleexplanationoftheapparentlydiscrep
antopinionsabouttheoriginsoftragedywhichtheycontain.Jankosharmonisingex
planationoftheapparentdiscrepanciesbetweenthesefragments(36571)wouldstill
besuperfluous.34Ontheotherhand,thedialogueformgivesusanadditionalreason
for being cautious in reconstructing Aristotles views from citations of On Poets, in
Philodemusorelsewhere.
ThecaseofJ12providesagoodillustrationoftheimportanceandthedifficultyof
decidingwhetherornotaviewexpressedwithinthedialogueisAristotlesownview.
I am not convinced that Janko is right to say that the fragment is inconsistent with
what Aristotle says in the Poetics. The statement (extracted from Philodemus) that
some have tried to thoroughly humanize () tragedy is not in conflict
with Aristotles idea that tragedy represents characters superior to ordinary human
beings(Janko337:cf.Poet.2,1448a118),anymorethanthestatementthatsomepoets
CitingLaurenti1987:I.5573.
Thatisnot,ofcourse,enoughtoconfirmJankosspeculativeattributions:westillhavenopositive
reasontobelievethattheseareallfragmentsofOnPoets.

33
34

11

distortthecontinuityoftheirplots(Poet.9,1451b352a1)isinconflictwiththeideathat
tragedyisanimitationofacompleteaction(Poet.6,1449b245;7,1450b2134).Anor
mativetheory(forexample,thatstealingiswrong)isnotcontradictedbyareportof
deviations from the norm (for example, that some people steal). Either way, this
example has important implications for Philodemus reliability as a source. If I am
right,Philodemuschargeofinconsistency(OnPoems4,col.111.10112.3)dependson
treating a practice reported by Aristotle as a practice endorsed by Aristotle. If Janko is
right, Philodemus charge of inconsistency depends on treating a view expressed in
AristotleasaviewexpressedbyAristotle.Onbothreadings,Philodemusargumentis
worthless.Thoughsuchcrudemisrepresentationwouldnotbeoutofkeepingwithhis
style of polemic,35 we should acknowledge that the confusion might not be Philode
musfault:ifhewasnotworkingdirectlyfromOnPoets,herantheriskofbeingmis
ledbyhispredecessorsattemptstosummariseAristotlesviews,aswellasbyhisown
polemical zeal. Wherever the fault lies, there are limits to the confidence that can be
placed in Philodemus reports of Aristotle; and even when we think we can make
senseofhisrebuttals,thereareseverelimitstotheconfidencewithwhichwecanre
coverfromthemwhatAristotlemeantbythewordstowhichPhilodemusisrespond
ing.
3.Theevidenceforkatharsis
PotentiallythemostimportantofthefragmentsfromPhilodemusarethosethatrelate
tokatharsis,aconceptwhichreadersofAristotleandofGreektragedycontinuetofind
irresistiblyfascinatingandfrustratinglyimpenetrable.Jankosaysthatthefragmentsof
Philodemusconfirmtherecentconsensus,derivedaboveallfromPol.8,thatAristote
lian catharsis was a process in which the emotional effect of watching a mimesis
habituatesustoapproachmorecloselytothemeanintermsofourreactionstoreallife
events,andhencehelpsustoattainthevirtuesofcharacter(374).Idonotthinkthat
thatsuchaconsensusexists(exceptinthetautologicalsensethatthereisaconsensus
amongthosewhoacceptthatinterpretation).Buttotheextentthatsuchaconsensus
does exist, Philodemus does not confirm it. As we shall see, Janko himself identifies
one aspect of the view of katharsis which he finds in Philodemus as a surprise. But
therearealsosurprisingandproblematicelementswhichJankoseemsnottohaveno
ticed.
Therearethree relevantfragments,allfromOn Poems book 5.One,whichmerely
mentionstragickatharsisofpity(J50] [... ]),addsnothingtoour
knowledge. Another (J46) has more substantial content, indicating that katharsis is
beneficialwithregardstovirtue:
[] [], , [] , .
Poetryisusefulforvirtue,purifying[kathairousa],aswesaid,thepart.

ThatisthesomewhatcryptictextofPhilodemus.InthefragmentofAristotleasJanko
reconstructs it, Philodemus text is supplemented with what he surmises (after Nar
delli)arethewordsofAristotlewhichPhilodemusomitted:

JankosometimesdignifiesPhilodemuspointscoringdistortionswiththetitlereductioadabsurdum
(e.g.274n.6;297n.8).Thatisatechnicalterm,andshouldbeusedwithmoreprecision.

35

12

 , , , <> <
>.
Poetryisusefulforvirtue,purifying[kathairousa],aswesaid,the<nonrational>part<ofthe
soul>.36

ThesurprisecomesinJ47:
[ ] , [] [], [
] [] [] [], [
].
For there is vice even in the best souls: (i) folly in the wisest, and (ii) selfindulgence in the
most temperate; similarly there are also (iii) fears in brave souls, and (iv) jealousies in
magnanimousones.

Inviewofwhatfollows,thesupplementintheinitiallacuna(Forthereisviceevenin
thebest...)isplausible.Buttwoproblemsareworthnoting.First,wisdomisanintel
lectualvirtue,notavirtueofcharacter.Theextensionofkatharsistointellectualvirtue
isnotpartofanyrecentconsensus:itis,inJankoswords,arealsurprise(513).Vari
ousquestionsspringtomind.Cananintellectualdefectberepairedbythemechanism
of emotional habituation, or must its katharsis involve some different process? If so,
whatprocess?Doesanintellectualvirtuehavethesamestructureasavirtueofchar
acter?Inotherwords,iswisdomamean?Ifitis,betweenwhatexcessanddefectdoes
wisdomlie?37CanthekatharsisofanintellectualdefectbereconciledwithJankostext
of J46? Janko does not explain how katharsis of an intellectual deficiency can be
achievedbyoperatingonthenonrationalpartofthesoul.AsecondprobleminJ47is
thatthecatalogueofvicesinthebestsoulsisstrikinglyanomalous.In(i)anintellectual
virtue (sophia, theoretical wisdom) is paired with a deviation from a different intellec
tualvirtue.38In(ii),avirtueofcharacterispairedwithadeviationfromthesamechar
acter virtue. In (iii), a virtue of character is paired with one of that same virtues
affectiveprerequisites.39In(iv)avirtueofcharacterispairedwithanaffectivedisposi
tionwhich,thoughespeciallyincongruentwiththatvirtue,isalwaysdeviant.40Onthe

See n.32 above. Jankos translation: The (art of) poetry is something useful with a view to virtue,
purifying,aswesaid,the(irrational)part(ofthesoul).
37 Jankosanswerseemstobethatfollyisanexcess:follyisjuxtaposedwithcowardice,intemperance,
andirritabilityinasimilarlistofexcessesat7.7,1149a56(514).Buthehasmisreadthetext:Aristotle
saysthatanyofthesethingsinanextremeformiseitherbestialorpathological.
38 AsJankoputsit,thegreatestscientificorartisticgeniusmaystillbeadunceattakingpracticaldeci
sions (513). One would expect practical folly () to be contrasted with practical wisdom
(:EE7.2,1236a5;8.1,1246b57,18;Rhet.1.11,1371a13;3.9,1410a78).Itis,infact,possiblethat
Aristotledidthinkofpracticalwisdom(phronsis)asameanatsomepointinhiscareer(considered,but
not confidently affirmed, in Pearson 2007; see also, briefly, Reeve 2012: 129). This possibility is of no
helptoJanko,however,sinceheidentifiessophiaastheoreticalortechnicalwisdom,inaccordancewith
Aristotlesnormalusage(forthelatterpossibilityseeNE6.7,1141a912).InPoet.18,1456a223,where
Sisyphusisdescribedassophosbutwicked,thewordmayrefertothemorallyneutralcapacityforprac
ticalreasoningwhichAristotleelsewheretermscleverness(:NE6.12,1144a239).Clevernessis
necessary,butnotsufficient,forpracticalwisdom.ButPhilodemusspecifiesthebestsouls:andasoul
thathasclevernessbutnotpracticalwisdomwouldnotqualify.
39 Theremustbefearsinthecourageoussoul,sincecourageisameanwithregardtofearandconfi
dence(NE3.6,1115a67;cf.3.7,1115b720,248).Sointhiscase,theevidenceforviceinthebestsoulsis
thattheyhavesomethingthattheymusthaveiftheyaretobecourageous(andnotreckless).
40 Forthedevianceofphthonos seeNE2.6,1107a814;2.7,1108a35b5;cf.EE2.3,1221a389;Top.2.2,
109b378;Rhet.2.11,1388a345.ItreceivesnospecialmentioninAristotlesdiscussionofmegalopsukhia
36

13

faceofit,thisseemsdesperatelyincoherent.Iamnotsorashastoassertdogmatically
that these problems are insoluble, or that it is impossible to extract a coherent and
Aristotelianaccountofkatharsisfromthefragmentsthatgiverisetothem;butIhave
notyetsucceededindoingso.NorhasJanko:heshowsnosignofhavingnoticedthat
theproblemsexist,andmakesnoattempttoaddressthem.41
OnereasonwhyJ47isasurpriseisthatitisentirelyisolated.Noneoftheothertexts
thathavebeenproposedassourcesofindirectevidenceforAristotlestheoryofpoetic
katharsiscontainsanyhintofitspromotingintellectualvirtue:theyallspeakofaneffect
ontheemotions.Weought,then,atleasttoconsiderthepossibilitythatPhilodemus
evidenceonthispointiserroneous.Alternatively,ifPhilodemusevidenceisright,it
follows that the rest of the indirect evidence for Aristotles theory is positively mis
leadinginitsincompleteness.Bothhornsofthisdilemmahavediscouragingimplica
tionsfortheprojectofusingindirectevidencetorecoverAristotlestheory.Butthereis
athirdpossibility,stillmorediscouraging:foronecoulddoubtthereliabilityofPhilo
demus evidence without being convinced that the rest of the indirect evidence is
reliable.Theonlyresponsiblewaytoproceedistoconsiderineachcasewhetherornot
therearesufficientgroundsfortreatingagivenpassageasindirectevidenceforAris
totlesthinkingonkatharsis.
Sincetherewereothertheoriesofkatharsis,thefactthatapassageisconcernedwith
tragicandcomickatharsisisnotsufficientonitsowntowarrantourtreatingitasevi
denceforAristotlestheory.Here,atleast,JankoandIagree.Heonceincludedapas
sage from Olympiodorus commentary on Platos Gorgias (33.3, 172.623 Westerink)
among the testimonia to Aristotles theory, but has since changed his mind: the pas
sageisomittedfromthiseditiononthegroundsthatitprobablyderivesfromasimi
lar theory of Theophrastus (406).42 In a series of passages in his commentary on the
First Alcibiades, Olympiodorus distinguishes various conceptions of katharsis, one of
which is associated with Aristotle by name (6.1012), though also later described as
Peripatetic or Stoic (54.1555.14) and Aristotelian (146.24).43 These passages do
appearinJankoscollection,butonlyasaspuriousfragment(J75).Hisconclusionthat
OlympiodorusdoesnotprovideuswithreliableevidenceforAristotlestheoryis,in
myjudgement,correct.Weagree,therefore,thatreferencetokatharsisisnotsufficient
towarrantapassagebeingtreatedasevidenceforAristotlestheory.Thatistrueeven
whenAristotleisnamedbythesourceoftheputativefragment;afortioriitwillbetrue
whenheisnotnamed.

(NE 4.3; EE 3.5); however, it is associated with mikropsukhia (Rhet. 2.10, 1387b345) and servility (Pol.
4.11,1295b213).
41 Nordoeshegiveanyadequateexplanationwhenheextendskatharsisfromthecorrectionofmoral
andintellectualerrorsintheaudiencetothecorrectionoftheerrorsofpoetsandofpeopleinthepoems
(5145,onJ48andJ49).Ihavefailedtomakeanysenseofthisatall.
42 IncludedinJanko1987:60(cf.1984:1478);retractedinJanko2009.
43 PeripateticdoesnotreferspecificallytoAristotle,andAristotelianneednot(seee.g.Olympiod.in
Cat. 48.37 Busse; in Mete. 27.223 Stve; Proclus Tim. 1.262.5 Kroll; Philop. in DA 597.21, 28, 39 Hay
duck). The element of this theory which Olympiodorus specifically attributes to Aristotle (6.911) is
combatingthumoswithepithumiaandviceversa,whichisnotfoundinanyextantworkofAristotle(the
parallelas54.1755.1isprinted,withoutindependentattestation,asChrysippusSVF3.489);thisideais
explainedashealingbyopposites(6.11;cf.6.9;54.1718),whichcanbefoundinAristotle,thoughnotin
connectionwithkatharsis(NE2.3,1104b1618;EE2.1,1220a347).

14

Iamblichus,too,wasfamiliarwithatleastoneaccountofkatharsisotherthanAris
totles: in On the Mysteries 3.9 he mentions a theory that aulos music can have thera
peuticeffectsonemotionsandonbodilystates;Iamblichusdenialthatthisshouldbe
described as katharsis indicates that it had been so described. Some have treated this
passageasanAristotelianfragment(F81cRose;F6cLaurenti);Jankoisrightnottodo
soindeed,hedoesnotevenincludeitinhislistofomissions.44What,then,istheevi
dence for Aristotles anonymous presence in On the Mysteries 1.11 (J55, mentioned
brieflyin1above)?JankoclaimsthatalthoughIamblichusdoesnotnameAristotle,
hisconceptsandwordingaresosimilarthatdependenceonAristotle,whetherdirect
orindirect,isgenerallyaccepted(51920),anddrawsattentiontothetypicallyAris
totelianantithesisbetweendunamisandenergeia(520;cf.1987:186).Thecogencyofthis
evidenceiseasilytested.Ifthepresenceofthisantithesisistocountasevidencethat
thepassagederivesfromAristotle,thentheantithesisshouldbeabsentfrompassages
inIamblichusthatarenotderivedfromAristotle.InOntheMysteriesalone(andwith
outincludingtechnicalusesofthecognateverbs)thereareapproximately180and80
occurrencesofdunamisandenergeiarespectively,andtheyarejuxtaposedascloselyas
theyareinthepassageunderdiscussioninatleast15otherpassages.Thesetermsare
therefore not evidence that the passage is derived from Aristotle: they are part of
Iamblichus own technical vocabulary,45 and he was perfectly capable of deploying
themifheneededtosummariseatheoryofkatharsisinhisownwords.
Moreover,onedetailinthispassagecountsagainstitshavingpreservedAristotles
own wording: the potentials of human passions are to be brought to activation to a
degree that is proportional ( ). Aristotle uses summetria and its
cognatesfreely.Heroutinelyusesitinconnectionwithphysicalhealth;andhedoesso
inethicalcontextswhereheisusingphysicalhealthasananalogyforvirtue(NE2.2,
1104a1419; 10.3, 1173a238). Yet he never applies this terminology to emotions
thoughtheideathatemotionsshouldbeproportionatetotheirobjectis,ofcourse,fun
damentaltohisethics.Idonotthinkhisavoidanceofsummetriainthiscontextiscoin
cidental.Hisregularformofexpression,thatemotionsshouldbefeltastheyought,in
responsetothethingstheyought,etc(e.g.NE4.5,1126b56),isclearer,moreflexiblein
itscapacitytoexpressallthevariousdimensionsinwhichemotionsmaybeappropri
ate,andalso(perhaps)lessliabletobemisunderstoodasmeaningnotverymuch.46
ThecombinationofpsychologicalandphysicalhealingpointstoTheophrastus(F726Fortenbaugh);
the phrase is consistent with Theophrastus (Sens. 35, 509.12 Diels; CP 2.3.8, 6.5.4,
6.5.6),thoughnotexclusivetohim(Plut.Mor.383b;432d,438c;496a;918de;Simpl.inDA19.34Hay
duck).
45 See Steel 1978: 5361; Shaw 1995: 34 n.28, 723, 789 and index ss.vv. Iamblichus is not alone: for
PlotinusseeSleemanandPollet1980:26876,37483.
46 The Magna Moralia (which I take not to be by Aristotle) speaks of summetria between reason and
emotion(2.7,1206b911).ThetermisapplieddirectlytoemotionsbylaterPeripatetics(Asp.inNE1.5,
42.22,44.1819Heylbut;Alex.Aphr.inTop.99.1011,19Wallies;[Alex.Aphr.]Eth.Probl.146.23,149.31
150.3,etcBruns)andPlatonists(e.g.Plut.Mor.443cd;Max.Tyr.Or.27,3c;Plot.6.8.5.2830);Olympi
odorususesitintheaccountsofPeripateticStoicorAristoteliankatharsisthatwehavealreadyseenare
notAristotles(inAlc.54.22,146.4).ComparingProclusinRemp.1.82.57KrollwithRep.2,377e8ewill
provide oneexample of the tendency to introduce the term summetria into reports of texts in which it
didnotappear.Theuseoftheterminconnectionwithkatharsiswasnodoubtparticularlyencouraged
by Pl. Soph. 227c8d. As Janko notes, the Tractatus Coislinianus also has this usage in its definitions of
tragedy (tragedy... wants to have a summetria of fear) and comedy (summetria wants to be of fear in
tragedies, and of the laughable in comedies). The status of the Tractatus as evidence for Aristotle is,
44

15

OlympiodorusisnottheonlylatePlatonisttomentionAristotleinconnectionwith
katharsis:Proclusdoesso,too,thoughhedoesnotmentionAristotlealone.Inhiscom
mentary on the Republic (1.49.1720 Kroll = J56) Proclus speaks of the objection to
Platosrejectionoftragedyandcomedybasedonthethesisthatamoderatesatisfaction
oftheemotionswillmakethemeducable.Hecontinues:
,
, ... .
This,then,whichprovidedAristotlewithampleopportunityforcriticism,andthecombatants
onbehalfofthesekindsofpoetryfortheirargumentsagainstPlato,weshallsolve...moreor
lessasfollows.

That is my translation; others have understood the Greek in the same, or a similar,
way.47ButJankotranslatesthepassagedifferently(459):
It was this that gave Aristotle, and the defenders of these (kinds of) poetry in his dialogue
againstPlato,mostofthegroundsfortheiraccusation(againsthim).Weshallrefutethis(ob
jection)...asfollows.

ThistranslationiscrucialforJankosclaimthatProclus...makesclearthatheisciting
theDepoetis...,andprovesthatAristotleandthecharactersinthedialogueresponded
explicitlytoPlatoschallenge(520).48Butitisnottenable.AsinapassagefromThe
mistius discussed earlier, Janko offers dialogue as a translation of logosor, more
preciselyandevenlessplausibly,oflogoi.YetOnPoetswashardlyadialogueagainst
Plato, though it may well have included arguments (logoi) against Plato. More im
portantly,itishardtoseeanyreasonwhyProclusshouldhavethoughtitworthgiving
separate mentions to Aristotle and other characters in the dialogue if they all shared
Aristotles critical stance towards Platos views on poetry.49 (That is not like distin
guishing,forexample,theviewsofSocratesandCalliclesinGorgias.)Itismucheasier
to see why Proclus would record, with disapproval, the fact that others (who would
includeTheophrastus)hadjumpedonAristotlesantiPlatonicbandwagon.SincePro
cluspresentshimselfasrespondingtoatraditionofresponsestoPlatothat,thoughit
originatedwithAristotle,wasnotlimitedtohim,wehavenorighttobeconfidentthat
hissummarycarefullydistinguishesthespecificsofAristotlestheoryfromlatervaria
tions. That is, for our purposes, inconvenient. But Proclus, like Themistius, was not
however,itselfdisputed,andthisuseofsummetriaisoneofthereasonsfortreatingitwithcaution(as
alreadynotedbyBarnes1985:106).
47 1857:165(=1880:47):DiesenPunktnun,welcherdemAristotelesvielenAnlasszuVorwrfenund
denVerfechternjenerPoesienzuEntgegnungengegenPlatongegebenhat...(Itwasthisthatprovided
most of the grounds both toAristotle for accusation and to the defenders of these kinds of poetry for
their arguments against Plato: Janko 459 n.1). Bernays version is accepted by Vahlen 1874: 296.
Festugire 1970: I 67 understands the passage in the same way: Ce deuxime problme donc, qui a
fourni Aristote une ample occasion dattaque, et aux dfenseurs de ces genres potiques prtexte
leurs crits contre Platon....Compare (with a different, and in my view less satisfactory, rendering of
)Laurenti1987:I.219:Questoproblema,dunque,cheoffreunampiaocca
sionedicriticaeadAristoteleeaquantidifendonotaligeneridipoesiacontrogliscrittidiPlatone...;
Flashar,inFlashar,DubielzigandBreitenberger2006:79:DashatnunauchdemAristoteleseinennicht
geringenAnlassfrseineKritikgegebenundallen,diefrdieseDichtungeneingetretensindgegendie
WortePlatons.
48 Cf.Janko1987:186(providesaclearstatementthatheiscitingtheOnPoets,andthatAristotleand
the characters in the dialogue are responding to Platos challenge). Janko is here following Rostagni
1955:I.2878.
49 AsalreadynotedbyLaurenti1987:I.2634,inresponsetoRostagni.

16

writingtoserveasconvenientrepositoryofevidenceforAristotle;wemustrespecthis
agenda.
We have, then, no alternative: we must fall back on the direct evidence for
Aristotlestheoryofmusicalandpoetickatharsis(Poet.6,1449b278;Pol.8.61341a214;
8.7, 1341b322a18). Readers of the Poetics are acutely aware that its brief allusion to
katharsisisnotsupplementedbyanyexplanation.Mostscholarsturntothediscussion
ofmusicalkatharsisinPolitics8.7(J54)inthehopeoffindingcluestotheexplanation
thatismissingfromtheextanttextofthePoetics;yetinthePoliticsAristotlerefersusto
aworkonpoeticsforacleareraccount(1341b3840).50Hisadmissionthatthediscus
sioninPolitics8isanoutline,lackinginexplicitness,isdiscouraging.Nevertheless,the
discussioninthePoliticshassignificantadvantagesoverallindirectsources.Weknow
that Aristotle is expounding his own theory of katharsis; we can be confident that he
did not misunderstand or misrepresent his theory; and the crossreference gives us
someassurancethatthetheoryofmusicalkatharsishassomerelevantconnectionwith
thetheoryofpoetickatharsisthatismissingfromthePoetics.
HowclosethatconnectionismaydependonthescopeofAristotlesdiscussionof
music in Politics 8. If mousik is used in a narrow sense, approximating to music in
modernusage,thenthisdiscussionofmusicalkatharsismightstandatsomedistance
fromthelostdiscussionofpoetickatharsis.Janko,however,maintainsthatAristotles
usageinPolitics8isbroad:notonlymeanswhatwenowcallmusic,butin
cludes the words, and therefore comprises poetry and song as well (375, cf. 516:
comprises song and poetry also). Elsewhere he has claimed that everyone
usedmousikinthisbroadsensedowntoPhilodemusDemusica.51Ifthismeansthat
thewordwasalwaysavailableforuseinthebroadsense,itisunobjectionable,butin
sufficienttodeterminethesenseofthewordinanygivencontext.Butifitmeansthat
thewordwasonlyusedinthebroadsense,andneverreferredtosongand/orinstru
mental music as distinct from nonmelodic poetry, it is false: in Poet. 26, 1462a1416,
mousikissomethingwhichtragedyhasandepicdoesnot;andthatcanonlybesong
asdistinctfromnonmelodicpoetry.ThesenseinwhichAristotleusestheterminPoli
tics8isthereforeanopenquestion.
Janko believes that the question can be settled on internal evidence: Aristotle de
fines as working , via songcomposition and
rhythms (Pol. 8.7, 1341b24) (375 n.1). This argument turns, obviously, on the claim
thataresongs,nottunes(517).Thetranslationisdoublyanomalous.First,
Olympuswereapparentlysolopiecesforthewithoutwords(Janko517):
such purely instrumental compositions are not songs in normal English usage.
Secondly, Janko classifies epics and tragedies as : these, too are not songs in
normalEnglishusage.Moreimportantly,epicsandtragediesarenotinAristotles
usage,either.Tragediesuseincertainparts(Poet.1,1447b248;cf.6,1449b2831)
JankodoesnotinterpretthisasacrossreferencetoOnPoets:hetakesitasareferencetopassagesin
the lost part of the Poetics, of which (more controversially) he believes the Tractatus Coislinianus pre
serves an epitome (518, citing Janko 1984: 13651, 2112: see n.46 above). J53 and J54 are therefore
fragments of On Poets only in an exceptionally attenuated sense: they are indirect evidence for what
AristotlesaidinthePoeticsaboutatopicthatwasalsodiscussedinOnPoets.
51 Janko2009:270:ThisissupportedbythefactthatTheophrastushadatheoryofcatharsisthatap
pliedto;bythistermhemeantbothmusicandpoetry,asdideveryonedowntoPhilodemusDe
musica.
50

17

and therefore include . It does not follow that they are . is not co
extensive with the composition of tragedy, and is far from being its most important
component(6,1449b3150b20);anditisprimarilythroughtheimitationofactionsthat
tragedy evokes pity and fear and thus effects katharsis (6, 1449b278, with e.g. 9,
1452a13;14,14531b114),notthroughthesongs.
JankomayobjectthatthereferencetopityandfearinPol.8.7,1342a7(J54.2)proves
thatsongsrelatingtoactionincludestragedy,whichrepresentsactionsthatarouse
theseemotionsinparticular(519).Butthatdoesnothelp.Toshowthatatragedyisa
song relating to action, it is not enough to show that it relates to action or that it
arousescertainemotions:onemustalsoshowthatitisasong.Nordoesthefactthat
tragedy and epic represent character and action combine with the fact that rhythms
and are associated with character and action (J53.3 = Pol. 8.5, 1340a1423) to
confirmthataresongs,nottunes(517).Toshowthatarenottunes,it
is not enough to show that and songs can represent character and action: one
mustalsoshowthattunescannotdosoforiftheycan,theassociationofwith
character and action is consistent with being tunes. In what follows, Aristotle
treatsandrhythmsalikeasrepresentativeofcharacter(J53.45=Pol.8.5,1340a38
b10).ThispairingofrhythmsandoccursfourtimesinPolitics8.Translatingthe
phraseasrhythmsandsongsawkwardlyyokessongswithonestructuralcomponent
ofsongs.Itmakesbettersenseifthephrasedesignatesthetwodimensionsinwhich
songsandinstrumentalcompositionsarepatterned,i.e.rhythmandmelody.Itisthen
equivalenttothephraseandrhythms,whichoccursfourtimesinPolitics8.52
This interpretation is more consonant than Jankos with the dominant emphases of
Aristotles discussion of musical education, which seeks to determine which
instrumentschildrenshouldlearntoplayatschool,andwhichharmoniaiandrhythms
areeducationallyappropriate.
Ifmusicisnotthesameaspoetry,thereisanelementofuncertaintyabouttheiden
tity of musical and poetic katharsis. Yet the crossreference indicates some significant
connection. The question, then, is whether the account of musical katharsis that
emergesfromthediscussioninthePoliticscanbetransferredtopoetry.Ifthetransferis
possible(perhapswithsomedegreeofmodification),thenwewouldhaveanaccount
ofpoetickatharsisthatcouldbeacceptedasAristotleswithsomedegreeofconfidence.
The problem, obviously, is that we must first extract an account of musical katharsis
from the Politics. The history of scholarship gives no grounds for optimism; but nor,
perhaps,needwedespair.Inadditiontothegeneraladvantagesofdirectoverindirect
evidencementionedabove,thedirectevidenceinthisinstanceprovidesuswithsome
resources. To attempt a full interpretation would exceed the remit of the present
paper.53 But it is worth considering what can be concluded about the nature of
AristotlesresponsetoPlatoscritiqueofdramaandHomericepic.
InRepublic10,Platoexcludesthesekindsofpoetrybecausetheystimulateintense
emotion.Hemaintains(i)thattheyareharmful,becausetheymakeuspronetoinap
52 Rhythmandmelos:8.5,1340a13,19;8.6,1341a1,14.Rhythmandharmonia:8.5,1340a40b10;1340b17
18; 8.7, 1341b19, 2021. Cf. Poet. 6, 1449b29, where I agree with Janko (332 n.3) that the epexegetic
isnottobedeleted;so,too,TarninTarnandGutas2012:245
53 Ihaveinthepasttriedtomaintainanagnosticpositiononthenatureofmusicalandtragickatharsis
(mostrecentlyinHeath2013:934).Mycurrentresearchprojectwillforceme,induecourse,tocommit
myself.SeefurtherHeath(inpreparation).

18

propriate emotion in real life (605c6d); and (ii) that an argument in their defence
wouldneedtoshowthattheyarebeneficial(607d69).Ifwesetasidethelibertarian
response that harmfulness does not warrant legislative control (a view that Aristotle
couldnotaccept),anyrebuttalwillneedtomaintainthatsuchpoetryisnotharmful
i.e. to deny (i). But two different strategies then become available: the first would
maintainthatsuchpoetryisnotonlyharmless butalso beneficial;thesecondwould
maintainthatharmlessnessissufficientdefence,denying(ii)aswellas(i).Whichstrat
egydoesAristotleadopt?
Hereisakeypassage(Pol.8.7,1342a418):
(a4) , ,
, , (a7)
, ,
, (a11)

, ,
.(a15)
(a16)
<> .
(a4)Fortheaffectthatoccursinsomesoulsstronglyexistsinall,butindifferentdegrees:e.g.
pityandfear.Andalsoenthusiasm.(a7)Forsomearepossessedbythisdisturbance,too,and
weseethat,undertheinfluenceofthesacredmelodies,whenevertheyemploythemelodies
thatexciteextremefrenzyinthesoul,theyarerestored,asifhavingundergonemedicaltreat
ment [iatreia] and katharsis. (a11) The very same thing, necessarily, is experienced by those
pronetopityortofearor,ingeneral,toanyaffect,andbyotherstotheextentthateachissus
ceptibletosuchthings;andforalltherecomesaboutacertainkatharsisandalleviation[kou
phizesthai] with pleasure. (a15) In the same way, kathartic melodies, too, provide harmless
pleasure to human beings. (a16) This is why the use of such harmoniai and such melodies
shouldbeallowedtocompetitorstakingpartinpublicmusicalperformances.

Aristotleisheresettingoutanargument.Thatargumentispromptedby,andaimsto
justify,thestatementthatimmediately precedes:practicalandenthusiasticharmoniai,
thoughnotappropriateforchildrenseducation,areappropriateforlisteningtoothers
performing (a34). The arguments success is marked by an explicit statement of the
conclusion reached (a1618); careful attention to the arguments structure should re
vealwhatAristotleseesassufficientgroundsforreachingthatconclusion.Theargu
ment proceeds directly from asserting that kathartic melodies provide harmless
pleasure(a1516)totheconclusionwhichlicensestheuseofsuchmelodiesinpublic
performance(a1618).TheimplicationisthatAristotleregardsharmlessnessassuffi
cient defencein other words, that he is following the second of the two strategies
mentionedabove.
DenyingthatsuchmusicneedstobebeneficialdoesnotcommitAristotletodenying
thatsuchmusiceverisbeneficial;anditisclearthatAristotlebelievesthatmusiccan
bebeneficial.Theeffectofthesacredmelodiesonthosewhoarepathologicallyprone
to enthusiasm is therapeutic ( , a711), which can hardly fail to be a
benefit. Why, then, does Aristotle shift to the weaker claim that kathartic music is
harmless immediately before drawing his conclusion? Here, too, attention to the
structureoftheargumentisimportant.Astheargumentprogressesitsscopeextends
beyondthosepathologicallyprone toenthusiasmtoincludepeopleacrossthewhole
range of emotional susceptibility; in the course of that progression, Aristotle moves
fromasitweremedicaltreatment andkatharsis(a1011)tokatharsisandalleviation

19

()withpleasure(a1415)toharmlesspleasure(a1516).Thatmakessense
ifhethinksthatthebenefitinthepathologicalcasescannotbegeneralisedtoallcases:
whatcanbegeneralisedisthecombinationofharmlessnessandpleasure;andthat,in
Aristotles view, is sufficient. Another passage in Politics suggests that Aristotle took
the same view in the case of poetry: he permits adults to attend performances of
comedy,notbecausetheywillbenefitfromit,butbecausetheireducationimmunises
themagainsttheharmthatcomedycandotoyoungpeople(7.17,1336b203).
ThoughIhavespokenofAristotlerespondingtoPlatoscritique,thatmaynotbe
the most helpful way to describe what is happening in the passage on musical
katharsis.CompareAristotlescommentontherepresentationofgodsinpoetry(Poet.
25, 1460b351a1), which is a response to Plato only in the sense that it manifests
Aristotles disagreement with Republic 2; it is not a response to Plato in the sense of
arguing that he was wrong. Similarly, the discussion of katharsis manifests Aristotles
disagreementwithanargumentaboutmusicanalogoustotheargumentaboutpoetry
inRepublic10,butitdoesnotengagedirectlywiththatargument:whenitassertsthat
suchmusicisharmless,itimplicitlyrefusesthedemandthatadefenceshoulddemon
stratebenefits,butdoessowithoutargument.AristotleisnotwillingtoallowPlatoto
settherulesforthedebate(whyshouldhe?),butinsteadpresentsconclusionswhich
hehasreachedfrompremisesbasedonhisown(notPlatos)understandingofethics
andhumanpsychology.
Thosepremisesarepresupposed,andnotarticulatedinthepassageitself.Theyare
not drawn to the readers attention. Readers in whose thoughts Platos critique is
prominentwillunderstandablyfinditnaturaltointerpretAristotleontheassumption
that Platos challenge was being met on the terms that Plato had laid down: that is,
theywillassumethathemusthavebeentryingtoshowthattheemotionalstimulusis
beneficial.54Modernreaderswhomakethatfalseassumptionwillfindconfirmationof
theirunderstandingofAristotleintheindirectevidenceprovidedbyancientreaders
whomadethesamefalseassumption.Theresultingconsensuswillbeaconsensusin
error.
4.Conclusion
Workingwithfragmentsisahazardousenterprise.Excessivecautioncondemnsusto
remaininignorance,whichisabadthing;butlackofrestraintmaybetrayusintobe
lieving falsehoods, which is worse. The history of scholarship is littered with the
corpsesofsourcecriticalhypotheseswhich,afterbecoming thebasisforfarreaching
conjectural constructsandbeingwidelyacceptedassecurely established, succumbed
tothescepticalscrutinythatexposedtheirinadequatefoundations.Withhindsightitis
easy to see in such cases how overconfident acceptance of insecure conjectures put
researchontoafalsetrail.Yetinadisciplineinwhichcertaintyishardtocomeby,total
abstinence from conjecture is necessarily stultifying. The balance between excessive
54 Jankosmisreading(520:Ford...deniesthatAristotlerespondedtoPlatosattackonpoetry)ofFord
2004: 334 (My reading of Politics 8 will disappoint those who want to see Aristotle taking up Platos
challengeintheRepublictoshowinproseisnotonlyasourceofpleasurebutalsoofbenefittopolitical
communitiesandhumanlife(607d)...TorespondthattragedyiseducationalisnottodefeatPlatobut
tocapitulatetohim)illustratesthepowerofthatassumption:ittransformsFordsdenialthatAristotle
respondedonPlatosowntermsintoadenialthatherespondedatall.

20

cautionandoverconfidenceishardtostrike.Atboththeindividualandthecollective
levels,theprogressofresearchdependsonadialecticbetweenboldnessandcaution.55
TodisapproveonprincipleofJankosboldendeavourwouldthereforebeabsurd.56
IfthereisevidencethatwillhelpustounderstandAristotle,weneedtoknowaboutit;
andwewillneverfinditifnooneispreparedtolookforit.But,asJankosays,itiscru
cialthatweshouldbepreparedtoformulateandtestsuchhypotheses(362,myem
phasis).Inthispaper,IhaveputsomeofhypothesesadvancedinJankoseditionofthe
fragments of On Poets to the test,57 and have found them compromised by multiple
failuresofmethodologyandjudgement.Butthereisalsoadeeperfailure.Jankodoes
notrecognisethefragilityofhisconjectures,andfailstoalertusersofhisworktothe
need for caution. Indeed, his hypotheses are consistently delivered with the utmost
confidence and a tendency to overstatement: confirms, proves and certainly are
favouredwords.Thiseditionisthereforelikelytogiveitsusers,especiallyuserswho
arenotexpertsonAristotleorthelatertextswhichareadducedasindirectevidence,a
dangerouslymisleadingimpressionofthesolidityofitsconclusions.58

MalcolmHeath
ProfessorofGreekLanguageandLiterature
UniversityofLeeds
Email:M.F.Heath@leeds.ac.uk

Heath2002:107.
Janko362:Withtheoddsstackedagainstthosewhowouldargueforauthenticityinsuchcases,itis
nottobewonderedatifscholarsprefernottotaketherisk,andevendisapproveonprincipleofthose
whodo.ButJankodoesnotidentifyanyscholarwhohasexpresseddisapprovaloftheendeavouron
principle,andIcanthinkofnone.
57 Otherexamplescouldbeadded.Icannotresistgivingpassingmentiontothepages(3278)inwhich
Janko, reluctant to believe that Aristotle could have endorsed an allegorical interpretation of Homer
(F175 Rose), hypothesises that the Homeric Questions were in dialogue form (those who share Jankos
doubtsaboutF175woulddobettertoconsiderthepossibilitythatoursourcesrecordasolutionmen
tioned but not endorsed by Aristotle), and then founds a conjecture about the speakers in this hypo
thetical dialogue on the work of that supremely inventive fantasist, Ptolemy son of Hephaestion, also
knownasPtolemyChennus(onwhomseeCameron2004:13463).
58 ResearchforthispaperwasundertakenduringmytenureofaLeverhulmeMajorResearchFellow
ship(F10099B).TheLeverhulmeTrustssupportisgratefullyacknowledged.
55
56

21

Appendix1
Seen.10aboveforanexplanationofthisAppendix.

JohntheDeacon149.6150.10Rabe

,
.
.



,

,

,

.

,


.

*J33a.1

.


.

*J33a.2

*J33a.3

J35

,


,

...

.
,

,
,

.
,
.

GregoryofCorinth,RG 7.1328.89.1
Walz
,
,
.



,

,

,

.

*J33b

,
,

.

22

J36c

:

.

,

.


.

23


Appendix2:Concordance
*=fragmentdoubtfullyattributedtoOnPoets(Janko2011),ordoubtfullyattributed
toAristotle(Rose).J32ismarkedasdoubtfulinJankostext,butnotinhiscom
mentaryorconcordance.InJanko1987asterisksmarkfragmentsconjecturallyat
tributedtoOnPoets:sincethisisadifferentcriterion,Ihaveomittedthemhereto
avoidconfusion.
=spuriousfragment(Janko2011).
italicised fragment numbers = fragments which other editors (Rose, Gigon) do not
assigntoOnPoets.IhavesuppliedsomenumbersfromGigonsFragmenteohne
BuchangabeomittedfromJankosconcordance.
Understatus,IhaveindicatedwhichtextsIwouldincludeifIwerecompilingacol
lectionofsources;sinceIamnotcompilingsuchacollection,thejudgementsshould
beregardedasveryprovisional.
P=atextthatIwouldincludeasasourceforOnPoets(insomecaseshesitantly).
A=atextthatIwouldincludeasasourceforAristotle,butcouldnotassigntoOn
Poetswithanyconfidence.
InsomecasesIwouldincludeonlyalimitedpartofthetextprintedbyJanko(seethe
discussions of J35 and J38 above). Inclusion does not imply endorsement: texts in
which Aristotle is explicitly named have been included even if I believe that their
evidence is unreliable (e.g. J75); this includes cases where a named reference to
Aristotleiscomplicatedbymentionofothers(e.g.J56).

Janko
2011
*1*2
310
*11
1230
31a
31b
31c

*32
*33a
*33b
34
35
36a
36b
36c
36d
*37a
*37bc
*37d
38

1987
3
3

3
11

12
16

14b
14a
13

15a

*39*40
41

Rose3

*676

*677a

*677b

Gigon

937

919

Laurenti

4a

4b

4c

Source
Philodemus
Philodemus
ThemistiusOr.34.11
Philodemus
sch.Cicero
Didymus
GL6.107.1012Keil
anon.POxy.2389
sch.DionysiusThrax
JohnDiaconus
GregoryofCorinth
Themistius27(337ab)
JohnDiaconus
ProclusChrestomathy
sch.Plato
JohnDiaconus
Suda(s.v.Arion)
Pausanias(Atticist)
Zenobius(paroemiast)
anon.PBerol.9571
Themistius26(316ad)

Ar.Byz.(?),POxy.3219

24

Status

P
P?

A(part)
P?

P?
(part)

42
43a
43b
44a
44b
44c
44d
44e
4552b
*52c
5354
55

56
57
58
59
*60*61
6263
*64
65a
65b
65c
65d
65e
66a
66b
66c
*67a
*67b
*68
*69
*70
*71
*72a
*72b
73
*74
75

15b

2
1a

1b

1c
4

A
B

C
E
7
5

73
72b

72a

81b
81c
81a

74

862
15

14

893

921

16

5
3b

3a

6b
6c
6a
T1
7

D.L.3.56
Themistius26(316d9a)
D.L.3.37
Athenaeus
Eustathius
Ar.Byz.(?),POxy.3219
Heracleodorus
D.L.3.48
Philodemus
JohnPhiloponus
Aristotle(Pol.)
IamblichusMyst.1.11
IamblichusMyst.3.9
ProclusinRemp.
Aristotle(Poet.)
Macrobius
Aristocles

6
9

10

Di
Dii

76a
76e
76b
76c
76d
75a
75b

7b

66
71a
71b
70

77

20.1
20.4
20.5
20.2
20.3
21.1
21,2

27

865
18
19
17

22

9a

9b

2c
2a
2b
1

10

Philodemus
[Plut.]DeHom.
anon.Vit.Hom.
Gellius
[Plut.]DeHom.
anon.Vit.Hom.
D.L.2.46
D.L.8.489
Tzetzes
Aelian
Eustathius
Cicero
anon.POxy.2506
anon.POxy.2734
D.L.8.63
D.L.8.512
D.L.8.74
D.L.8.57
Plutarch
OlympiodorusinAlc.
OlympiodorusinGorg.
BekkerAnecdota1.101

P
P

Aetal.
P
P

P?

P
P
P
P
P
P

A
A
A
A
A
A
P
A
Aetal.

REFERENCES

Barnes,J.(1985)Editorsnotes.Phronesis30,99110.
Bernays,J.(1869) Die Heraklitischen Briefe. Ein Beitrag zur philosophischen und religions
geschichtlichenLitteratur.Berlin:Hertz.
Bernays,J.(1857) Grundzge der verlorenen Abhandlung des Aristoteles ber Wirkung der
Tragdie. Breslau: Trewendt. Reprinted in Zwei Abhandlungen ber die aristotelische
TheoriedesDrama(1880).Berlin:Hertz.
Bremer,J.(1988)ReviewofJanko1984.Mnemosyne41,16670.

25

Cambiano,G.(2002)CatastrofinaturaliestoriaumanainPlatoneeAristotele.RSI
114,694714.
Cameron,A.(2004)GreekMythographyintheRomanWorld.NewYork:OxfordUniver
sityPress.
Festugire,A.J.(1970)Proclus.CommentairesurlaRpublique.Paris:Vrin.
Flashar,H.;Dubielzig,U.;Breitenberger,B.(2006) Aristoteles. Fragmente zu Philosophie,
Rhetorik,Poetik,Dichtung.Darmstadt:WissenschaftlicheBuchgesellschaft.
Ford,A.(2004) Catharsis: the power of music in Aristotles Politics. Murray,P.;
Wilson,P.(eds.),MusicandtheMuses.Oxford:OxfordUniversityPress,30936.
Fortenbaugh,W.W.(1987)ReviewofJanko1984.CP82,15664.
Fortenbaugh,W.W. et al. (1992) Theophrastus of Eresus. Sources for his Life, Writings,
ThoughtandInfluence.PartI.Leiden:Brill.
Garzya,A.(19723) Varia philologica. Athina 7374, 3658; reprinted in Storia e
interpretazioneditestibizantini:saggiericerche(London1974),XX(ix).
Gigon,O.(1987)AristotelisOperaIII:Fragmenta.Berlin:deGruyter.
Halliwell,S.(2006) An Aristotelian perspective on Platos dialogues. Herrmann,
F.G.(ed.),NewEssaysonPlato.LanguageandThoughtinFourthCenturyGreekPhilo
sophy.Swansea:ClassicalPressofWales,189211.
Heath,M.(2002)InterpretingClassicalTexts.London:Duckworth.
Heath,M.(2013)AncientPhilosophicalPoetics.Cambridge:CambridgeUniversityPress.
Heath,M.(inpreparation)PoeticalAnimals.Aristotle,AnthropologyandPoetry
Janko,R.(1984)AristotleonComedy.TowardsaReconstructionofPoeticsII.London:Duck
worth.
Janko,R.(1987) Aristotle, Poetics. With the Tractatus Coislinianus, a Hypothetical Recon
structionofPoeticsII,theFragmentsoftheOnPoets.Indianapolis:Hackett.
Janko,R.(2000)Philodemus.OnPoemsBookI.Oxford:OxfordUniversityPress.
Janko,R.(2009) A new comic fragment (Aristophanes?) on the effect of tragedy.
CQ59,2701.
Janko,R.(2011)Philodemus.OnPoemsBooksThreeandFour.WiththefragmentsofAris
totleOnPoets.Oxford:OxfordUniversityPress.
Laurenti,R.(1987)Aristotele.Iframmentideidialoghi.Naples:Loffredo.
Pearson,G.(2007)PhronesisasameanintheEudemianEthics.OSAP32,27395.
Rabe,H.(1908a) Aus RhetorenHandschriften: 5. Des Diakonen und Logotheten Jo
hannesKommentarzuHermogenes .RhM63,12751.
Rabe,H.(1908b)AusRhetorenHandschriften:6.WeitereTextquellefrJohannesDia
conus.RhM63,51217.
Reeve,C.D.C.(2012)Action,Contemplation,andHappiness.AnEssayonAristotle.Cam
bridgeMA:HarvardUniversityPress.
Rose,V.(1886)Aristotelisquiferebanturlibrorumfragmenta.Leipzig:Teubner.
Rostagni,A.(1955)ScrittiMinori.Turin:BottegadErasmo.
Schenkeveld,D.(1986)ReviewofJanko1984.Gnomon58,2127.
Sedley,D.(2007) Creationism and its Critics in Antiquity. Berkeley: University of Cali
forniaPress.
Shaw,G.(1995) Theurgy and the Soul. The Neoplatonism of Iamblichus. University Park
PA:PennsylvaniaStateUniversityPress.
Sleeman,J.H.;Pollet,G.(1980)LexiconPlotinianum.Leiden:Brill.

26

Steel,C.(1978) The Changing Self. A study on the Soul in Later Neoplatonism: Iamblichus,
DamasciusandPriscianus.Brussels:PalaisderAcademin.
Tarn,L.;Gutas,D.(2012)AristotlesPoetics.EditioMaioroftheGreekTextwithHistorical
IntroductionsandPhilologicalCommentaries.Leiden:Brill.
Vahlen,J.(1874) Wo stand die verlorene Abhandlung des Aristoteles ber Wirkung
derTragdie?SWAW77,2938.
Verlinsky,A.L.(2006)ThefloodinAristotlesMeteorologica(I.14).AntikeNatur
wissenschaftundihreRezeption16,5168.
Walsdorff,F.(1927)DieantikenUrteileberPlatonsStil.Bonn:Scheur.

27