ON THE SECOND BOOK OF ARISTOTLE'S POETICS AND THE SOURCE OF THEOPHRASTUS' DEFI NI TI ON OF TRAGEDY SINCE the Renaissance any treatment of Aristotle's Poetics hasdiscussedandlamented thelossof a secondbook. Becausethisbook,asweshallsee,issupposedtohavecontainedatheoryof comedy,-itsloss, measured by thevalueof theAristotelian theory of tragedy,is incalculable. An attempt to investigate the facts on which thebelief inthislossisbased, andtodetermineitsreliability is,therefore,of fundamentalimportance.The belief rests on the observation that the Poetics, as it is nowconstituted,isincomplete, or rather fails to fulfill its apparent pro-gramme, being especially deficient in a symmetrical elaboration ofits initial divisions. Such incompleteness is usually defined as thelack of anentiresecondbook.Adirectstatementthatthereweretwobooksistobefoundinthelist of Aristotle's worksgivenby DiogenesLaertiusinhis lifeof thephiios~pher. ~ By this evidence, Bywater holds? " the fact is s&i-cientlyassured," although " we have no further direct testimony tothe existenceof a SecondBook." There is, however, the statementof the so-calledAnonymus Menagii, to be identified probably withThematerialsinthisarticleformedpartof adissertation, The Mediaeval Con- ception ofComedy andTragedy, submittedinpartialfuliilmentof therequirementsforthedegreeof Ph.D.at HarvardUniversityin 1916. Iamunderobligationsforsuggestionsandhelp, toProfessorsC. N.Jackson,W. H.Schofield,andespeciallytoProfessorE. K. Rand.aV. Rose,Aristotelisqui ferebanturlibrmum fragments(Teubner),Leipzig,1886, p. 6,1. 83. aIngramBywater, Arislotlem theArtofPoetry,Oxford, 1909,p. xx.Harvard Studies in Classical Philology 28 (1917)2A.Phi l i pMcMahon Hesychius,'wheretwobooksare ment i ~ned. ~Thisevidenceisalso acceptedbyvonChristasestablishingtheexistenceofthesecond book.3I n the third place, there is the additional testimony ofthe only other early indexofAristotle'sworks,inthe "fragmentsofaphilos- opherofPtolemy'sreign,"wherethereismentionoftwobooks " de arte poetica secundum disciplinam Pythagorae eiusque sectatorurn placita."This isexplained byWenrich:and accepted byRose:as beingaconfusion betweenAristotle'stwobooksofthe Poetics and a workonPythagoras,butneverthelesscorroboratingtheevidenceof the other two lists. Zellercitestheevidenceofallthreeliststoprovethatthe Poetics as we have it is only a fragment?The relation ofthe three is declared byRosetobeasfollows:Andronicusmadealistoftheworksof AristotleinhisbooksontheAristotelianphilosophy;anunknown philosopherofthe time ofPtolemy did the same, with certain changes, but using the same basis;Favorinus then derivedfrom Andronicusin hiscommentaries, uponwhichDiogenesand also Hesychius relied. Althoughthedirectassertionsoftheexistenceoftwobooksare comparativelyfew,manystatementsindicatethatthereweremore than one.Most important inthis class ofevidence is, ofcourse, that tobefoundintheotherworksofthephilosopher,especiallyinthe Rhetoric.I n all ofthese the definite article is used in the pluralwhen reference is made to the Poetics. I n the first bookofthe Rhetoric Aristotlerefers to the booksofthe Poeticsfora discussionoftheri di cul o~s. ~In thethirdbookofthe Rhetoric, againwithapparent reference to the ridiculous,atreatment E.Zeller,Aristotleatzdtke EarlierPeripatetics,trans.f r om Philosophyofthe Greeks by B.F. C. Costelloe andJ.H.Mui rhead,2vals., London,1897, i, p.48,n. 3.2Rose,Fragm.,pp.13, 75.3W.v o n Chri st , GeschichtedergriechisckenLitteratur,5t hed.,W.Schmi d,Muni ch, 1908, i ,p.702, n.4.V.Rose,AristotelesPseudepigraphus,Leipzig,1863, pp.79,80. J. G.Wenri ch, DeAzrctorumGraecorumzersionibus. . . commentariiscom- mentatio, Leipzig,1842, p.143. Rose,Arist.Pseud.,p.194. 8 Rose, op. cit.,pp.8 ff. Zeller, op. sit.,i, p.102,n.2.9Rhetoric,I ,11, 1371 B 33. 3 ALostBooko jthePoetics ofthat subject isomittedonthe groundthat it already exists inthe books ofthe P0etics.lIn the same part ofthe Rhetoric a cross-reference is givento the booksofthe Poetics -to chapter22, SO Jebbbelieve^.^ A little later inthis sectionanother reference is made to the booksof the Poetics -chapter 21according to Jebb.3A few paragraphs further on, another reference is made,for a treatmentofdiction,to the books ofthe Poetics,corresponding,likethetwoimmediatelypreceding,to chapters21and22.4In stillanother placewehaveareference, like thatinthefirst,tothebooksofthePoeticsforatreatmentofthe ri di cul o~s. ~ I nthePolitics,also,Aristotle,speakingoftheterm~hoapuis promises to discuss it more fully in the books on Poetics!The evidence presentedby all this testimonyiscertainly ofgreat weight. BesidestheindicationsofAristotle'sownworkswehavethe implicationsofsomeoftheearlycommentatorsonthe philosopher. Ammonius,probablyofthefifthcentury,inhisworkontheDe Interpretatione,referstothe Poeticsintheplural?Thiscommentis takenbyVahlentorefertochapter20:andiscitedbyZellerto support the traditionofa lost book. Boethius also treated the De Interpretatione,translating it once with acommentary in two books,and again, about507, witha fullercom- mentary in six books.1I n the second work he mentions"Aristoteles inlibris quos de poeticascripsit," l1which istakenby Zellerto mean that Boethius knewa Poetics in two booksJ2 Rhetoric, 3 ,I ,1404 A39. "Rhetoric,3,2,1404 B 5 ;R. C. Jebb,The RhetoricofAristotle, ed.J.E. Sandys,Cambri dge,1909. Rhetoric,3,2, 1404B26.Rhetoric,3,18,1419 B2. Rhetoric,3,2, 1404B37.6Politics,8, 1341 B39. ' -4. Brandi s,Scholia i n Aristotelem(Aristotelis Opera, i v ) , Berlin, 1836, p,g g A12;A.Busse,Ammoni iiAristotelisdeInterpretationeCommentarium, I ,Berlin,1897, P.13,1. J. Vahl en, AristotelisdeArt ePoeticaLiber,gded.,Lei pzi g, 1885,not eo nPoetics,20, 1456 B 2j( p.45) . 9 Zeller,op. cit.,i, p.102, n.2.lo J . E. Sandys,AHistoryofClassicalScholarship f romtheSi xt hCenturyB. C. to the EndoftheMi ddl e Ages,Cambri dge,1906, i ,p.253. l1Boet hi us,Commentarii i n l i brum Aristotelis IIEPIEPMHNEIAZ(Pars posterior secundam edi t i onem cont i nens),ed.C.Meiser, pp.6 ,11 ff.:i nlibris quos de poetica scripsitlocutionispartesesse syllabasvel et i am coniunctionest radi di t ,etc." l2Zeller,op.,p.102, n .2.4A.PhilipMcMahon The commentatorEustratius isappealedto by Bywateras show- ing " that there waseven inthe latest Aristotelian schools some faint traditionofanotherBook,"for,inhiswork on the Ethics,he speaks ofa first book ofthe Poetics, and this may be taken to mean that there wasalso another, a second book. Finally,the fragments ofan earlyanonymous commentatoronthe Rhetoric employthe plural when asserting that Aristotle discussed the ridiculousinthepoetic^.^The evidenceonthisscore is presentedby Vahlenin a series ofquotations that refer to Aristotle'streatment of the ridiculous.The words ofthe anonymous writer thus prove, accord- ing to Ritter, that Aristotle'swork On Poets, which was in three books, was differentfrom the Poetics intwo.' ThereareinAristotleandelsewheregroundsforbelievingthat certainmattersspokenofastreatedinthePoetics,butnotfound therenow, wereonceto bereadinasecond book.Thesereferences alsoindicatean orderoftheworks,accordingto Ritterand others, whothinkthat the compositionofthe Poetics is spokenofas a future work not only in Politics 8, 1341 B 39, but also in the De Interpretatiowe 41I 7A 5 .At any rate,inthe passages ofthe Rhetoric which wehave already cited,the philosopherapparentlyreferredtothe workas onealready done.ThecloserelationingeneralbetweentheRhetoricandthe Poetics is further established bythe largenumberofcorrespondences ofdifferentsorts noted by Vah1en.B Nevertheless, the explanation ofkatharsis to which the eighth book ofthe Politicslooks forwarddoesnotappeartobesutiticiently given in Poetics, 6,2,wherethe wordoccursonly inthe definition? Thereisnotreatmentoftheridiculoussuchasweshouldexpect from various statements in the Rhetoric (I,11, 1372 AI ; 3,I, 1404 A 39; 3,18,1419 B2).Thus,ofthereferencestothe Poeticsinthe 1Bywater, op.cit., p.xxi. 2Spengel,AristotelisArsRhetorics,Leipzig,1867,i, 159,15. Vahlen,09.cit., p. 77. F.Ritter,AristotelisPoetica,Cologne,1839,p. xi.6Op.cit., p. vi.6Vahlen,op.cit., pp.49ff., pp. 53ff. Poetics,6,1449B24.A Lost Book of the PoeticsRhetoric,one-halffind some counterpart,while there is nothing atallcor- responding to the other.'Ritter decides such inconsistencies compel us to conclude that either Aristotle does not mean our Poetics in these pas- sages ofthe Rhetoric, or else the Poetics as we have it (Poeticam nostram) "mancamad nos temporum hominumque iniuria pervenisse." InthePoeticsitselfthereisapledgethatisnotredeemedtothe satisfactionofreaders,whereAristotlesaysthat'(weshallspeak laterabout Comedy."It is naturalto suppose that thetreatment oftheridiculousasthebasisofcomedywouldhavebeenfoundin thepartofthePoeticswhichdiscussedcomedy,andthatifneither the promise ofthe Rhetoric nor that ofthe Poetics was kept, the reason would be the same:namely, the loss ofthe second book in which both were contained.The deficiency is explained on that basis by Gercke? following Rose ;6 and Bywater, inhis footnote on Poetics 6, 1449 B 21, supplementsAristotle'swordswiththephrase" inthelostSecond Bookofthe Poetics."Thus Bywaterisenabledto outlinethecon- tents ofthe second book as containing a further treatment ofkatharsis, andadiscussion ofcomedy,inwhichthe laughable wouldhavebeen analysed as corresponding to the pitifuland theterriblein tragedy? Theexposition ofsomeothermatterswouldprobablyhavefound a place in such a second book, according to most critics:in particular, certainaspectsofthesubjectofcomicdictiontowhichthereexist apparently two references; and, much more important, a defence ofthe drama againstthecensures ofPlato. In his commentaryonthe Categories, Simplicius quotes Aristotle on the subject ofsynonym^.^That this topic also stood in the second book E. M. Cope, TheRhetoric of Aristotle, ed. J. E. Sandys, Cambridge, 1877,i , p. 224; J. E. C. Welldon,TheRhetoric of Aristotle, London, 1886,p. 85.* Ritter, op. cit.,p. vii. Poetics,6 , 1449B 21.Gercke,"Aristoteles," Pauly-Wissowa,Real-Encyclopiidie,ii, I , col.1053 27.v. Rose, De Aristotelis Librwum Ordine et Auctoritate Commentatio, Berlin,1854,P.133.Bywater,Art of Poetry,p.149. Bywater,op.cit., p. xxiii. A. Brandis,Scholia,p. 43A13; Kalbfleisch,Simplicii i n Aristotelis CategoriasCommentam'um,Berlin, 1907,p. 36, 13.A.PhilipMcMahon isasuppositionto which bothBywater1 and Vahlen2 incline by plac- ingitamongthe fragmentsoncomedy immediately succeeding their texts ofthe extant Poetics. Theotherreferenceisapuzzlepresentedbyastatementinthe lexiconoftheanonymouscompiler,calledtheAntiatticist,published by Bekker, which may derive in part from O r ~ s , ~ whom Ritschl placed as earlyasthe secondcenturyofourera,whileReitzensteinassigns himto the fifth century.The Antiatticist, in defendingthe use ofthe wordKVVT~T~TOV, The portionofthecitesAristotleinthePoetics.' Poetics propertothisdefinitionwas,accordingto~ ~ i a t e r , 5 thelost secondbook. A defence ofthe drama against Platowould have been a feature of absorbing interestinthislostsecond book.In Aristotle'sPoetics, as Spingarn observes,'scholars ofthe Renaissancediscoveredasatisfac- toryvindicationoftheclaimsofpoetryagainstthePlatonicand mediaevalobjections.InPlatotheobjectionsweregroundedina metaphysicaltheory ofimitation,interpreted, some would hold,in a narrowspirit.Sinceheconceivedimitationasmerecopying,and since he held to the theoryofthe objectivityofideas, he allowed little scopefortherepresentativearts.8Ifanobjectinnatureisonlya comparativelyunrealcopy ofan eternal objectiverealityinGod,the reproductionofthat object inart istwiceremovedfrom reality, and if certainlyfalse probablydangerous also. After all, the matter is not entirely settled by the answer ofAristotle withrespecttotragedy.Plato'sobjectiontothedramaasexciting the passions without providinga means ofgoverning them is not fully 1Bywater, op. cit., p.93. 2Vahlen, op.cit., p.81. Sandys, 09.cit., i, p.325. Cf. Vahlen,op. cit., p.81;Antiatticistai nBekkerianecdotis,101, 3 2 rtvvr6-rarov:'Aprarori;X~snepinocl)rw?s.''766k nhvr ~vK V Y ~ ~ T ~ T OY . "6Bywater, op.cit., p. xxiii. 6The relations ofAristotle in the Poeticsto Plato have been welltreated, with results clearlydemonstratingthedependenceofAristotleonhismaster, in:Ch. Belger,DeAristoteleetiami n ArtePoeticacomponendaPlatonisdisci@lo,Diss., Berlin, 1872;Georg Finsler, Plalonunddie AristotelischePoetik, Leipzig, 1900. 7J.E.Spingarn, AHistoryofLiteraryCriticismi ntheRenaissance,nded., New York, 1908,pp.18, 19. B.Bosanquet, AHistoryofAesthetic,London, 1904, pp.47-55. ALostBookof thePoetics 7 refuted by Aristotle'stheoryofkatharsis,as it is usuallyinterpreted. I tmeetsthe Platonicobjectiononlysofarasit is probablethatthe possible vicarious aesthetic satisfaction withoutan inevitable impulse tomoralactionwillfindcompensationinthegeneralizingcharacter ofgenuinetragedy,whichisoneofitsinherentlymoralfunctions. Thatis,tragedygivesoccasionforaproperconceptionoflife,and maythus issue forth in properactions.But it is still possible for the passions to be aroused without finding a direct and immediate reaction appropriate to thecharacterofthe passionsarousedandatthesame timemorallyprofitable.Clearerintellectualperceptionisnotin- evitablysucceeded by improvedethical practice,andso Jamesurges us neverto allow ourselves an aesthetic excitation ofemotion without adeliberatelybeneficialconsequence in ouractions.' Whatevermay bethe correct theoryofthe effect ofcomedy on the spectators, the Platonic objections apply, it would appear, much more to comedythanto tragedy,and wereinfacturgedagainstitinthe first place ratherthan against tragedy. That a discussion ofall the objectionsraised by Plato, with definite refutationofthem,clearing away the difKculties justmentioned,was includedinthePoeticsisthejudgmentofmanyscholars.Although Vahlenopposedthetheory,urgedby Heitz,thatthistreatmentwas to befoundinthe lostending ofthe Politics, he supposed that it was containedinalostfinalchapterofthePoetics.Bernays,however, heldthat it wasinthe lamented second book, -areasonable conclu- sioninviewofthe otherargumentsurgedfortheexistenceofthat book.2 5 Victorius,3 thefirstgreateditorofthePoetics,was,asBywater notes:the firsttosaythat ourPoetics is onlypartofalargerwork. This opiniongrew steadily;Zellerstates it specifically as the loss of a second book, supporting his opinion with much ofthe evidence cited above.6Rose, in his work to determine the canon ofAristotle,finally W. James, Princi9lesofPsychology,NewYork,1890, ii, chapters 24and2 5. *Finsler, Platon,p.3. Vettori(P.)(Victorius), Commentationes in primurnlibrumAristotelisdeArle Poetarum,Florence,I560. Bywater, op. cit., p. xx.Zeller, op,cit., i, p.102, n.2 .8 A. Philip McMahotzlistedthePoetics ashavingconsistedoftwobooks,landSpengel also thoughtthattheremusthavebeentwobooks.2I nthissup- positionhe wasin accord withhis usualopponent Bernays,who even assertedthatinanothertreatisewestillpossesscertainfragments ofthesecondbook'streatmentof~o me d y. ~Mostrecentlyofall, Bywaterhas feltjustifiedindrawing upasketchofthecontents of thesecond book.4 The lossmaybepartlyexplainedbythe orthodoxtradition ofthe vicissitudesof Aristotle'smanuscripts.Sulla, according to this account, afterquellingtherevoltofApellicon,carriedoffhislibrary,which containedAristotle'sautographsthathadalreadylanguishedinthe caveat Skepsisforaconsiderableperiod.Sullaentrustedthese parchmentstoTyrannion,andthustheeditionofAndronicuswas prepared.6Iftheseincidentsare accurately reported,especially the story ofthe uniquemanuscriptsinthecaveat Skepsis, thesebooks rangreatrisksofdestructionorneglect.Asitis,thereexistsa totalgapinthe historyofall the Aristotelianwritings for afull cen- tury after Cicero.Thisiscertainlyduetotheentirelossofallthe commentaries ofthatperiod!andalike fate may easily beassigned to the second bookincase itsurvivedpreviousperils.Roseargues thatitwaslostat averyearlydate since itwasnotknownto the Arabs,Syrians, or other commentators, and must have perished before Andronicus, from whom he would date the presentstate ofthe text.' Bernays, however, believes that the second booksurviveduntil the fifthcentury,chieflybecausehefindsatthattimeinProclusacon- ceptionofkatharsis,correspondingto his owndeductions fromPlato andAristotleonthequestion,withwhichheoverthrewtheneo-classical doctrineinfavor ofamorepsychological s ~l ut i on. ~Hatz-1 Rose,De Arist. Libr.Ord.,p.241. 2H. Diintzer, " Di e Aristotelische Poetik und ihrVerhlltniss zu denBiichern I I e p i ~ O L ~ T L K ~ S , ' ' Zeitschrijtfur dieAlterthumswissenschaft (1842), pp.280,281.J . Bernays,DweiAbhandlungenuberdieAristotelische TheoriedesDrama,Berlin,1880. ( I . Grundziigeder verlorenenAbhandlung desAristotelesiiberWirkungderTragodie; 11.Erganzung zuAristoteles' Poetik.)Bywater,op. cit., p.xxiii.6 R. Shute,OntheHistory of theProcessbywhichtheAristotelian WritingsArrivedattheirPresent Form,Oxford,1888,pp.47,48.0 9 . cit.,p.66. 7 Rose,De Arist. Libr. Ord.,p.133. 8Bywater,op. cit., p.xxi. 9 ALostBookofthePoetics feld and Dufour go so far as to make the loss relatively recent,' because "c'estdanslesecondlivredecetouvragequelescornmentateurs alexandriis ont puisCla substancedeleursglosessurlespo8tesco- miques grecs et de leurs trait&' dela combdie.'" The existence ofasecond book,assumedto belost,cannot, inthe natureofthecasebeabsolutelydisproved.Bythelogicofsucha situationauniversalnegativecannotbeabsolutelyprovedevenof contemporaryfacts.Withtheincrease indistancefromthetimeof the facts considered the difficulty is immensely increased.No amount ofrational consideration applied to a complete collection ofthe avail- able facts surroundingthe pointatissue -facts relativelyfewafter solongatime -couldguaranteethetruthofitsconclusions.Pa-limpsests from some remote easternmonastery,or the accidental find ofsomearchaeologistinEgyptcouldeasilyoverthrowtheperfect theoryofan investigator.It istheobligation ofscholars, however, toerectsuchfrailstructures, basedonaconscientious surveyofall the evidence, with the humble reservation in every case, that the very paucityoftheevidence mustleavethestructurefrail.Thus,ina questionofthekindweareconsidering, itispossibletoweighthe valueoftheevidence andtojudgethetraditionwhichassertsthat therewasasecond bookofthe Poetics.Then,whileweare, bythe conditions ofthe problem prevented from making a categorical denial, wecan,I feelsure,assertthat sufficient reasoncannotbeshown to warrantthe beliefthatsuch abookeverexisted.Alltheconditions ofthe problem are morecompletelysatisfied, on the basis ofexisting evidence,bythehypothesisthattherewasnosecondbookofthe Poetics. To beginwith,thewhole traditiondepends toolargely onthe evi- dence oftheindices,thevalue ofwhich,underthescrutiny ofclose criticism, can be shown to be only limited. 1 A. Hatzfeld and M.Dufour, La PoLtipue d'rlristote, Lille, 1899,p. vii. 10A. Philip McMahonThatAndronicus did make a list may be accepted without hesita-tion ontheevidenceof Porphyryinhislifeof Pl ~t i nus . ~ Thatthesea i v a ~ ~ s were copied in turn by Favorinus, from whom DiogenesLaertiusobtainedhislist,isthetheoryof scholarsasdifferentasRoseand Bernays, whereas Shuteholds that these inferences can be dis-proved.3 Thereisagreatgulf between admittingthatAndronicus didmakealist,andthatthelistswehavereproducehim, especiallywhenthere is grave doubt about their intrinsic value, and whether theirauthorshipcannotwith safetybe assignedtoanotherancientscholar.Itishardongeneralprinciples tobelievethatthelack of orderandarrangementinDiogenes's listcouldhavebeentheresultof theother-wise admirablescholarship of Androni ~us. ~ The weight of evidenceshows rather that there is probably no relation between the indexcompiledbyAndronicus andtheonefurnishedby Di ogene~. ~I n the first place, although Diogenes cites Aristotle frequently,he doesnot follow his own list, but by implication appears to haveused thesamecanon as his predecessors and c~ntemporaries. ~ Howcould he refer to the third book of the Poetics,when hehad alreadysaidinhislist thatthereweretwo?7 Indeed,Diogenes's listcontainscomparatively fewworks, among his lengthyenumerations of titles,thatwecannow accept as genuinely Ari st ~t el i an. ~ Onpoetry aloneDiogenes ascribes five separate treatises to Ar i ~t ot l e, ~ and he seemselsewhere to have confused the Poetics and thedialogue On Poets.loInhim we seeclearly thebeginningsof theprocessbywhich,throughincluding forgeries, variant editions of the same work, editions ofseparateportionsof wholetreatisesbearinganothertitle,pupils' notes,enlargementsof latercommentators, andotheraccretions,David theArmenian found a thousand different works ascribed to Aristotle inthelibrariesof thePtolemies."Shut e, op.cit., p. 89.A.Ki rchhof f , Plotini Opera,Leipzig, 18j6,p. xxxi x. Shut e, op.cit., p. 86. Shut e, op.cit., p. 90. Shut e, op. cit., p. 80.Ri t t er, op.cit., p. x,n. Zeller, op.cit., i,p. 49. Shut e, op.cit., p.8. Rose, Fragm. pp. 3ff: 2, 83,118,119,136. 10 Rose, Fragm., p. 76; Diog.Laert., 8,57. Cf. Diog. Laert., 3,48 ( pp. 77, 78),and 2146 ( P. 79).Shut e, op.cit., p. 93.ALostBookofthePoeticsI I The compiler ofthe list given by Diogenes, because ofits inclusive- ness,was probablyan Alexandrine scholar.'In thisconclusion most critics, except Rose;agree, especially since Hermippusmay bedesig- nateditsauthor.8AclueisaffordedbyDiogeneshimselfwho, immediatelybeforegivingalistofthe works ofTheophrastus, cites FavorinusanddistinctlystatesthatthesourceofFavorinuswas Her mi pp~s. ~Since the origin ofhislistfor Aristotlemayhavebeen similar, it wouldbeeasytoexplain its character, whatevertheaddi- tions by Favorinus or Diogenes, as merely a librarian'slist ofthe titles borne by booksin a library. The works mentionedbyancientauthorsotherthanthe compilers ofthese lists, however, generally correspond to,whatwenowpossess, andCicero'sstatement ofthe range ofAristotle'sworks squares with our canon.5Dionysius ofHalicarnassususesvirtuallythesametext thatwenowhave,"andwhileGalen'scanonisidenticalwithours except fora few lost works,the roll ofthe missing does not include a second bookofthe Poetics?Thus, while from the timeofCicero on, thesuccessorsoftheeditorsTyrannionandAndronicusrefertoa uniform bodyofworksnearlyequivalenttoourcanon,oftheworks whichDiogenesmentions,hardlyany,exceptthedialogues,canbe identified inthe works wepossess? The index ofHesychius is plainly,as Rose points outlgonly acopy ofDiogenes, withthesuppression ofcertainrepeatedtitles,andthe addition ofsome more names, as incapable ofidentification as ofbelief. AltogetherHesychius managed to accumulatethirteendifferent titles which mighthavehadto do withpoetry.1 The index ofthe unknown philosopher in the time ofPtolemy is atonce dismissed by Bywater and most other recent critics.llIts devious Zeller, op. cit., i,p.51. 2Rose tries to maintainthe patently inconsistent position that the Aristotelian works and canon were always justas wehave them, and that wehave them all. aSandys, op. cit.,i, pp.122ff. Shute, op. cit., p. 92. Shute, op. cit., p. 51.Cf. Cicero, DeFin.,5, 4,9 ff. 6Sandys, op. cit., i, pp.279 ff.;Shute, op. cit., p. 67.' Op.cit.,p.77.8Op.cit., p. 86. Rose, Fragm., p.11,n.I. InOp.cit.,pp.11ff.Bywater, op. cit., p, xx. 12A.PhilipMcMahon history and the evident consequences offrequent mistranslation make itpracticallynegligiblewhenthesignificance oftheotherlistshas entirely disappeared. We may now approach the question from another point ofview.If various passages can beadduced to prove that therewere two books ofthe Poetics, a numbercan also be broughtforward in which one is assigned as the numberofbooks in that work.Ofcourse, it must be admittedinfairness, thatthe singular ofthedefinite articlehasnot the conclusive force possessed bythe use ofthe plural.It is, indeed, possible to referto the bookofthe Poetics,meaning the one whichI havein mind, withoutassertingthat there is only one book ; but the use ofthe pluralcarries withit the inevitable consequence that there was more than one book, and not less than two.Yet, ifthe occurrences ofthesingular alone are sdiciently numerous and ofvalueinthem- selves, theirsignificance cannotbe ignored. Zellercites Alexander Aphrodisiensisas usingtheexpression Cv TQ r c p iT O L ~ T L K ~ S whichhetakestoindicatethat Alexanderknewonly one b0ok.lZeller, however, apparentlydidnotexamine the passage to note the confused reading.In the Berlin edition it was noticed that Alexander'sreferenceinthissamepassagetotheRhetoricistobe found inthe Poetics i n~t ead. ~Following anemendation proposed by Vahlen, Wallies solved the difficulty by bracketing the reference to the Poetics, and allowing the mistaken reference to the Rhetoric to stand? We have here probablyonly acase where the original mistakeofthe authororanearlycopyistwascorrected byasucceedingscribe,to creep inlateralong withthemistake.Thispassage,then,does not prove that Alexander knew only one bookofthe Poetics;rather, that some scribe ofthe thirdcentury or later, knew only one. David the Armenian? probably in the fifth century,uses the singu- lar in a passage where he speaks ofother works by titles in the plural. Her mi a~, ~ another pupil ofSyrianus, was the father ofAmmonius, the Zeller, op. cit., i, p.102, n.z. 2IV,(Scholia i n Arist.), zgg B44. Wallies, AlexandriAphrodisiensisin AristotelisSophistichosElenchosCommen- tarium, Berlin,1898, pp. 33,26. 4Sandys, op.cit., i, p. 76. Vahlen, op.cit., p. 3.ALostBookofthePoetics pupilofProclus,andheusestheexpressionQv74 r epi ? T OL ~ T L K ~ ~ S >Thisisoneofthepiecesofevidence presentedbyZellertoshow a divergenceoft r adi t i ~n. ~ Herewehaveanapparentdserenceof opinionevenwithinafamilyofscholars,since Ammonius,theson, used the plural in spite ofhis father's singular.Hence Zeller'sobser- vationthatthemoreancientauthoritieswereacquaintedwithtwo books and the modern with only one loses some ofits force. WhenSimplicius, in his reference to synonyms,3 speaks ofthe book, it may be that he vaguely recollected that some matters ofdiction were taken up in the Poetics.Whatever the value ofhis citation ofAristotle in this connection, it is clear that he thought ofthe Poetics as one book. Disobedience to hisfathersecured forAmmoniusdisloyalty inhis ownpupilPhiloponus,forthelatteragreedwithhismaster'spro- genitor and usedthe ~i ngul ar. ~ AnArabic commentator, Alfarabi:inthe tenthcentury,alsoused the singular regarding the Poetics, but his allusion was mistaken.$ Eustratius,it willberemembered, wascited byBywaterto prove that whenthat author speaks ofa first book ofthe Poetics, he implies there was a second.Now, as I shall show later, neither the titles nor the numbers ofbooks have any definite meaning for us because ofthe absolutelyconflictingstatementswithregardtothem,sothatno inference canbemade from the title to the number or vice-versa, and thus the mention ofa first book in connection with a workcalled the Poetics by no means proves that there was more than one book inthe workwehave agreed tocall the Poetics.The only safe method is to see whether the content ofa given quotation corresponds to the nature ofthe works we have agreed to call the Poetics -a technical treatise,- or to the work Om Poets -a dialogue.Such examination ofthe com- 1F. Ast, Platonis Phaedrus(contains also the scholia ofHermias), Leipzig, 1810, p.111. *Zeller, op. cit.,i, p.102, n.2. Op.cit.,i, p.102,n.2.4Hayduck,IoannisPhiloponiinAristotelisDeAni maLibrosCommentaria, Berlin, 1897, p.269,l.28. 6Sandys, op. cit.,i, p. 395. 6Vahlen,op.cit.,p.3."AlfarabiusinterpreteSchmoeldersiodocum.philos. Arabum p,21, dedemonstrationeomnino fallacidisseritur in ipsizcslibro deartepo- etua." 14A.PhilipMcMahon pletepassage towhich Bywater refers showsclearly thatitpoints towhat we chooseto call Olz P0ets.l If the termfirst proves anything,itonlyshowsthattherewas more thanonebookineither thePoeticsorOlz Poets,andthebalance of evidenceinclinestowardsOnPoets.One of the citations, from an anonymous commentator on theRhetoric,isused toprove theexistenceof asecondbook, butanothercommentator, also anonymous, in paraphrasing his pasiage of theRhetoric, uses the singular of the poetic^.^ Both had the text of theRhetoric before them, but where our extant version has the plural,commentatorscouldstillvarywhen their workswere composed.In spite of the reasons brought forward to explain the loss of asecond book of thePoetics,thediiliculties in accountingfor this dis-aster lead more easily to theconclusion that it never existed. Theawkwardness of attempts to make the loss plausible become moreapparent when it is recollected that the tradition assumes definiteshapeonly since the time of Vettori. Thus Bywater saysJ3 " As forBook 11,one thingisquiteclear, thati t waswanting in thecommonarchetypeof 2,theMS. before the eighth centurySyriac translator,andAo, ouroldestGreekMS. Wecannotfix thedateof itsdisappear-ance; it is practically certain, however, that the loss must haveoccurred duringthepapyrusperiodof thetext,whenBookI1wasstillon a separate roll, so as to be easily detached from Book I,whichwas on another roll."Bywater,however, also acknowledges that there is no evidence toshow that later grammarians had any information about the secondbook or thetheory of comedy supposed tobe contained in it,whilethereis,on theotherhand, evidence toshowthatitwas unknown tothem. The history of the existing book in classical times is indeedobscure to thepoint of ignorance, andalthoughBywater thinks thatmuch of the teaching of the Poeticsandits terminology were repro-1Heylbut,Eustratii etMichaelis etAnonyma in Ethica Nicomachea Commentaria, Berlin, 1892,p. 320~1.36.Cf.Bywater,op. cit., p.xxi. 2Rabe,AnonymietStephani in Artemrhetoricam commentaria, Berlin, 1896,p.259.Cf. p. ix.3Bywater,op, cit., p.mi. ALostBookofthePoetics ducedinthelaterliteratureofcompilation,ltheamountisatbest small and comparatively late, so that the work must have been either ignoredorlittlestudied. Neither Dionysius ofHalicarnassusnor Quintilianknew the Poetics ata1112and Bywater confesses that in Byzantine times the second book was completely f ~r got t en. ~Strabo, who flourished atabout the same timeasDionysius,hasnothingto sayaboutAristotlewhichcanbe verified,notexceptinghisstoryd theAristotelianlibrary;andhe does not mentionthepoetic^.^ Thereis,indeed,apassageinThemistiusthatVahlenquotesin his footnote to Poetics 3, 1448 A 33,6 which seems to parallel Aristotle, atleast as far as the coupling ofthe names ofEpicharmus and Phormis is concerned.In this passage ofthe Poetics, however, as in some other ancient writers:theSicilian origin ofcomedy is asserted.In 5 ,1449 B 6,thenamesofEpicharmusand Phormisarejoinedinthis same connection.Thenames,however,werebracketedinthetextby Susemihl andsucceeding scholars,andthereconstructionofthe pas- sage, clearly ungrammatical, was effected with the aid ofThemistius? Whilethis process maybeinteresting intheannotationofAristotle, i tdoes notgofarinprovingthat Themistiusderivedhisknowledge from the Poetics, or that i twas known to him. Vahlenalso quotesa passagefromone ofthescholia onDionysius Thrax to parallel Poetics I,1447 B18. ~The point in question is indeed mentioned by Aristotle, but it is also mentioned by Plato and by other later authors who do not show any knowledge ofthe Poetics.Indeed, by reason ofthe languageused, it is more reasonable to suppose that ifthescholiastderivedhisideadirectly fromAristotleitcame from apassagetreating the same theme inhis dialogue Onpoet^.^I n any eventneitherinThemistiusnorthescholiastisthereanytraceofa second bookofthe Poetics. Roseacknowledges thatheisentirelyunabletoexplaintheloss, and falls backonthe explanationoffered by Alexander Neckhamfor Op.cit.,pp. xxiii, xxiv.Vahlen, op. cit., p. 8, n.2Ritter, op. cit., p. viii.Bywater,op. cit., p.123, Bywater, op. cit., p. xx.7Op.cit., pp.143, 144. Shute, op. cit., p. 68.Vahlen, op. cit., p. 6,n.Rose, Fragm., p.76.(No. 70; Diog. Laert.,8,57,) A.PhilipMcMahon thedisappearanceofanotherwork:'"Aristotelisviamuniversae carnisingressurussubtilissimascriptasuaiussitsecuminsepulcro recondi, ne utilitatiposteritatissuae deservirent." Thewholestoryofthecaveat Skepsis isdubious,andthesame considerationsthatmakeusdoubtitalsorenderunlikelythedis- appearance of so notableaworkas asecond book ofthe Poetics, con- taining a theory ofcomedy.It is, however, not so much the story itself asthe valueattachedto it that is unwarranted.While the ordinary causes for the loss ofclassical works are sufticient, when definite traces ofthematsome previoustimecanbediscovered,it has beenfound necessary, forsooth, to find a specific disaster for a unique manuscript to explain the loss ofa work ofwhich there is no definite witness else- where.Thattherearenosurvivingtracesofatheoryofcomedy derivedfroma second bookofthe Poetics will appear inthe course of thisinvestigation,andwehavereasonto believe that,whateverthe meritsofthenarrativeofSkepsis, noneofthescientific treatisesof Aristotle were lostto the Peripatetic school. Critics rely on Strabo and Athenaeus,withsome aid from Plutarch, tocomposetheexplanationthatrestsuponthetaleofthecave.2 Strabo veersslightlyinthecourse ofhisnarrative, forhebeginsby speaking ofthe libraries ofAristotleand Theophrastus,and then later talksofthesaleoftheworksofAristotleandTheophrastus bythe heirs ofNeleus to Apellicon ofTeos.These books might, indeed, have beenonlythecollections ofthesephilosophers,buttheremainderof histaletreats them as the original manuscriptsoftheseauthors.I n the face ofevidence elsewhere that Aristotelian works were in existence and that the Aristotelianschool enjoyeda continuouscareer,Strabo's remarksthatthePeripatetics lackedthegenuineworksofAristotle is manifestlyincredible. In one place Athenaeusspeaks ofa certain Roman Laurentius who collectedtheworksofGreekauthorsincluding"thoseofAristotle and ofNeleus,who preservedAristotle'sbooks, fromwhomourking PtolemyPhiladelphus,havingboughtthemall,putthemtogether with those which he had bought from Athens and Rhodes and brought themtofairAlexandria."TheinterestofthiscityinAristotleis I Op. cit., p. 134.aOp. cit., p. 30. Shute, op. cif., pp. 29 ff. ALostBookofthePoetics I7 certainenoughllandthepassageprobablyreferstotheworksof Aristotle.Athenaeusinanother place,however,does notagree with hisownstatement,forhesaysthatintheAthenianinsurrection Apellicon tooka leading part, a man who was originally a Peripatetic philosopherand had boughtthe library ofAr i st ~t l e. ~ The readyexplanationofthis state ofaffairs suggestedbyShute is that " no really publishedworks ofAristotlewere lost to the school meanwhile,"and after saying ofAristotlethat "Cicero mentions him overandoveragainasanauthorwellknownto all,andrepeatedly attacked by the Stoics and Epicureans,"he rightly exclaims,( (ifthis is oblivion, what is knowledge ?" AndronicusandTyrannion,theeditorstowhomtheworksof Aristotle were entrusted by Sulla, do not appear to have thought they hadtheautographsofthephilosopher.NordoesCicero?aclose friend ofTyrannion,mentionwhat would have been a great discovery ifthemissingmanuscriptsoftheworksonwhichthePeripatetics dependedwere suddenly recoveredand placedinthe handsofimme- diate friendsforediting.6Indeed, Rome wasthecentre ofAristote- lianismfromthetimeofCiceroforward:andthecharacterofthat philosophyseemsalwaystohavebeenmorecongenial to the Latin thanto theGreek mind.Not onlywere the editors Andronicusand Tyrannion residentsofRomeorRomans,butGalenandBoethius, among the most importantnames inthe study ofPeripateticism, also dweltinthatcity.Thus,withthegreatprobabilitythat,ifthere is any truth atall in the story ofSkepsis, copies ofall Aristotle'sworks were still in the hands of his students, the loss ofa second book ofthe Poetics cannotbeattributed to the damp and neglectofacellar. 4 The evidence forasecond bookafforded bythe cross-references in the works ofAristotle,notonlyforthe numberofbooks, but for the contentsofthesecond,ifsuchtherewas,isgreatlyimpairedbya criticalexaminationofsuchreferences ingeneral.Asafirststep in this directionit maybe observedthat,ifitis admittedthatallthe Op.cit.,p. 30.Op.cit.,pp. 35 ff.;Cicero, Fin.,4,28, 79. Op.cU., p31.Shute, op. cit.,p. 50. 09. cit.,pp. 33 ff.Op.cit.,p. 52. 18A.PhilipMcMaholz numerouscross-references aregenuine,it mustbesupposedthatthe philosopherhadaprearrangedschemeofhiswholesystemandits executionfrom theverybeginning,thathecarriedthisplaninhis headbothforcompletedworksand forthosenotwritten,andthat thetitlespreconceivedbyAristotlewereneitherchangedthen,nor havevariedsince.lIf,however,someofthemmaybegenuineand somenot,thentheirgenuinenessorspuriousnesswillhavetobe decided ongrounds quite distinct from theirmere presence in the text, and it ishardtoseehowtheycanwellbeusedasimmediateproof ofthe facts to which they refer.2This argument holds apart from my observation ina later connection,that neitherthe titles ofthe works nor the indications ofthe numbers ofbooks have any constant relation to one another in our special question.Thus a reference to matters as beinginthe Poetics is far fromproving, ifwefailto findthemthere, that they were originally in the second book. I tmaybeurgedagainstthe cross-referencesingeneral that they are over-elaborate;that whereas there are comparatively few to works wedonotpossess,theseareofavagueordoubtfulcharacter;that someofthembetweendifferentworkscouldnothavebeeninserted atthe time ofwhichever was the earlier;that inthem the same work is referredto by different names;that in some cases references inthe sameworktootherportionsofthatworkcontradictoneanother becausetheycite it as bothprecedingand following agivenportion; that in some cases the references which act as connecting links between two adjoining books occur atthe end ofone ofthemand atthe begin- ning ofthe other;that the references imply an arrangement ofworks inan artificialorderwhichcould not have existed until long after the time of Aristotle;andthattheycontainserious errors asto thereal meaningofthe doctrine whichtheycite orits relationtothematter underdiscussion. The attempttodefendthereferencesonthescorethattheywere insertedinasecondeditionbyAristotle,orthattheywererepeated andenlargedwithsuchadditions,ismoreorlessfutile:forthe sup- positionthat Aristotleeverpreparedaformal edition ofhisscientific works is unfounded, and it is d%cultto imagine a lecturer remember- ingsohugeacourseandat thesametimemakingminutecross- 1Op.cit.,p. 96. 2Op.cit.,p. 10.0 p .cit., p. 2 6 .ALostBookofthePoetics I9 references to questions far removed from his discussion both in thought and order.' Forafullexposition ofthesecharges againstthecross-references I candonobetterthanrefertoShute, whosefifthchapterentitled " OfTitlesandReferences " treatsthematterthoroughly,substan- tiatingbynumerousinstancesallthedifficultiesraisedabove.He concludes: " To sum upthen,wefindthetitlesoftheAristotelianbooks did not arrive ata fixed conditiontill some hundred years after the death ofthemaster;thatontheotherhandthereferences assumeallthe titlesas already fixedduring hislifetime;andthatevensotheyare notexplicable,unlesswegrantfurtherthathedeliberatelycalled several bookseachbytwoorthreenames;that hehadplannedout all his books before he beganany, and carriedall the detailsofbooks bothwrittenandunwritteninhis head.Eventheseliberalassump- tionswillnotgetridofallthedii%culties, andtheonlysatisfactory wayofexplainingthematteras awholeisto believethat allorthe greatmajorityofthereferences are post-Aristotelian,andthatthey proceed from editors neither ofthe same date nor altogether inagree- mentas tothenomenclatureandorderofprecedence ofthe books." From what has justpreceded wehave seen thatthe personal author- ityofAristotleisnottobeinvokedforanyofthereferences,and themotiveoftheeditorswhoinsertedmostofthemwasatbestto assureanacceptanceofwhattheybelieved,ratherthantostatea universallyacknowledged fact.The mode ofcomposition and publi- cation ofAristotle'sworks was, however, most favorable for the inter- polationofsuch references.Indeed for their proper understanding his works requiredsome such aid, so that in spite ofnatural reverence for thephilosopher'stext,scholarsearlyinsertedwhattheysawwas necessary for the intelligent reading ofthe works, and the practice once begun was continued withoutsystem and with diminishing success. The statement ofthemethodofcomposition by Case is one ofthe mosta de q~a t e . ~Aftermentioningseveral hypotheseswhich hefinds Op.cit.,p. 98.O p cit.,p. 96-116. Encycl opedi a Bri t anni ca,11thed., Cambridge,1911,ii,pp. 506ff. 2 0 A.PhilipMcMahon himselfobligedto reject,hecontinues:"Turning to Aristotle'sown works,weimmediatelylightuponasurprise:Aristotlebeganhis extantscientificworksduringPlato'slifetime.. . .Howeverearly Aristotlebeganabook,solongas he keptthe manuscript,hecould alwayschangeit.Finallyhediedwithoutcompletingsomeofhis works, such as the Politics,and notablythat work ofhis whole philo- sophicalcareerand foundationofhiswhole philosophy, -the Meta-physics -which, projectedin his early criticism ofPlato'sphilosophy ofuniversalforms, gradually developed into his positive philosophyof individualsubstances,butremainedunfinishedafterall.If then Aristotlewassomethirty-fiveyearsgraduallyandsimultaneously composingmanuscriptdiscoursesintotreatisesandtreatisesinto systems, he was pursuinga processwhichsolves beforehandthe very difficulties which have since been found in his writings." Onthe question of the publication ofAristotle'sworks, Case shows that printing has given us a wrong conceptionofpublication, and that Greek authorsthought ofworks rather than ofbooks issued in succes- sion on definite dates.Philosophersespecially had for their public the immediate circle oftheir students, so that "itdoes notfollowthat his ownworkswentbeyondhisownlibraryand hisschool.. . .There is. . . nocontemporaryproofthatAristotlepublishedany partof his mature philosophical system in his lifetime." Any chronological orderingofthe works isthereforeuncertain,and thehintsforsuchanarrangementgivenbythe cross-referencesare useless.Thedouble versionsareresultsoftheschool,theheads of whichwereprobablylessdaringintheiremendationsoftheactual text than the immediate successors ofAristotle, but found i tbetter for the prestige ofthe school that the scientifkworks should not be pub- 1ished.lThus the distinctionbetweentheesoteric and exotericworks grew up, thelatter,whichwereprobablythe dialoguesin aliterary form preparedforpublicationbyAristotle, beingcastinto disrepute comparedwiththeesotericworksinthepossessionoftheschool. That mostofthesescientificworksdidreachalimitedpublication later through the eager purchases ofPtolemy and the seizures ofSulla didnot preventthe growth ofthistradition,especially inviewofthe superior elaborationand originality ofthese treatises2 1Shute, op.cit., pp.8 ff.09. cit., p.37.21ALostBookofthePoetics Whilethe professors ofthe Peripatetic school could notcompare in vigorousoriginalitywiththeirmaster,naturalstagnationandsys- tematizationmadethemrelymoreandmoreonlecturesfromthe Aristoteliantext aidedby increasingfamiliaritywiththe whole ofhis works.In this way there would be evolved a set ofreferences to facil- itate their interpretation. Again,although~r i st ot l eabandonedtheuseofthedialogueasa meansofcomposition,it is probable,evencertain,that questionand answerconstituted a large part ofthe means ofinstruction,and prob- ably the occasion ofa more minute development ofmany points about whichquestions were raisedin the course oftime.'This explains the frequent anticipatory use oftechnical terms.If, then, he did not insert the questions to whichagoodmany passagesare plainlyanswers,it becomes still less probable that he would have inserted cross-references. AsShuteremarks:( (Therewouldbemoreover. . . anatural tendency among editors, who were themselvesusually Peripatetics, to exalttheesotericandunpublishedworksabovetheexotericand publishedones.Wefind,asweshouldexpect,thatreferencesto esotericworksaremuchmorecommonintreatiseswhichwerein vogue allthroughthe periodofdarkness than inthose which may be supposed to have remainedunpublishedduring that time.. . .Thus they(thePeripatetics)talkofthedialoguesunderthegeneraland somewhat contemptuous name ofthe external doctrine, without taking thetroubletospecifywhatspecialdialoguethedoctrineistobe found in." Butoneofthechiefreasonswhyany pluralreference,apart from thatofthe indices,isunreliableisthatnearlyallthe titles ofworks in the Aristotelian canon, both as drawn up in the indices and in other sources, are containedin morethan one book.With suchapractice in vogueit isnotda c ul tto see howany vague orgeneralreference to the Poetics couldeasilyemploythe pluralarticle,especiallywhen tothis invitationtoerrtherewasaddedtheexistenceofadialogue dealing withrelatedmatters,pluralityinthe numberofbooks in the I tis interestingto notethat the dialogue form was employed by Minturnoin hisItalianexposition ofthe Poetics.I n thefour books ofhis workherepresents himself as conversing insuccessionwithVespasiano Gonzaga, AngeloConstanzo, BernardinoRota, and Ferrante Carafa. ' Shute, op. cit., p.103. 22A.PhilipMcMahon dialogue,and aconfusionofnamesbetweenthetratiseand thedia- logue.Wearereducedthusforanyinformationaboutthesecond book and its contents to a search for any possible traces ofit elsewhere. Theinvestigation,however,canbestbemadeinconnectionwitha detailedexamination ofeach ofthe references to the supposedsecond bookor its contents. 6 Althoughthecross-references haveingeneralbeendiscredited,it is well, ifwe can, to account for them, and to examine each separately, asfarasitconcernstheobjectofourinvestigation.Amongthe possiblesources,thedistortionofgenuinereferencesisanunreliable hypothesisforanydefiniteresults.Norare weonsolid groundina hypothetical work nepiXkEews,'portionsofwhichappearto have been absorbed into the Rhetoric, and from which some ofthe passages in the Poetics, suchas that on the languageofpoetry,may have beentaken, witha generalconfusion ofreferences after this absorption,and after thisworkhaditselfdisappeared.Again,someofthegenuinerefer- ences may have originally alluded to the dialogue, and other references tothemmayhavebeeninsertedafterwards;theselastaremore practical possibilities. The citation fromBoethiusDeInterpretatione290,iscalledafalse translation byGerckein its use oflibris,and in the case ofBoethius it isclearthat heusedtheconceptionoftragedy which,I hopeelse- where to show, was contained in the dialogue On Poets.The reference ofBoethius,however,correspondstoourtreatiseofthePoetics,and yet,althoughhewasacquaintedwithit, hisdefinitionoftragedy,I believe,is notderived thence." I nlibris" may bea general expres- sionmeaning" inoneofthebookswhichhewroteaboutpoetry," and so may refer to the work On Poets.This possibilityis interesting in viewofthefactthatwhileTheophrastusisnotmentionedin the ConsolatioPhilosophiae,thedefinitionofthedramatic species which weafterwards findascribedtothatphilosopheragrees withtheidea ofBoethius.I tmight follow that the Theophrastian definition which Boethiuswouldhavefound" in libris" ofAristotlewasderivedby Theophrastusfrom Aristotledirectly. 1 H. Diels, Ueber das d7itte Buch der AristotelischenRhetorik,Berlin,1886. 2Gercke, Pauly-Wissowa, ii,I. col.1053. 23 ALostBookofthePoetics OneoftheproblemstosolveisthereferenceofthePoliticsto katharsis.Though Aristotle had promisedto discuss this topic more carefully in the Poetus,the treatmentaccordedit inthelatterwork isstillinadequate.Ofcourse,onaccountofthemannerofcompo- sition,apledgeonAristotle'spart,evenifthereferenceisgenuine, guarantees not fultilment but only intention;in fact Finsler find only onecase, and that doubtful, in whichageneral promisewas fulfilled. Next, a highly technical and obscure theory such as that ofkatharsis wouldbejusttheonetoattracttheattentionofananxiouseditor, whocouldeasily overstate the case, whenall he reallyhadto relyon wasarepetitionofthewordinPoetics6.Eitherhismemorywas poororhisintentionunscrupulous,foritwouldappearmuchmore learned to say " for a fuller treatment ofthis subject, see the Poetics," than," thereisanotherplacewherethisideaisinvolvedbutleft obscure."Aneditorissusceptibletojustsuchtemptations,andto allowanacknowledgedobscuritytoremainunilluminatedbyhis learning may befelt a reflection on his editorial capacity. That such a theory ofkatharsis was to be found in the second book andsurviveduntilthefifthcenturyisthehypothesisofBernays,' whoattributedtoProclusanimmediateknowledgeofAristotle's treatmenL2The passageofProclus he interpreted in the light ofthe Politics,3 butas Bywatershows: *'' It will beobservedthatProclus refers not only to Aristotle but also to other apologists for the Drama; it is quite possible, therefore, that it was from one ofthe latter rather than Aristotle himself that he derived his knowledge ofthe Aristotelian idea ofkatharsis.And in the context in place ofthe Aristotelian term hesubstitutesas synonyms dr+aiwaisand hnbpaars -neitherofthese wordsbeingfoundeither inthisorinanyothersense intheextant writings ofAristotle." Vahlen goes so far as to affirm that the discussion ofkatharsis came after the discussion ofcomedy in the second book:but Finsler shows thatthesource ofProclus'sidea wasprobablyPlato,and thisprob- ability is increased by the absence ofa full treatment which the refer- Bywater, op.cit.,pp. 94, 95.Cf. p.xxi. Bernays, Zwe iAbhandllmgen,p. 47. aFinsler, op.cit.,p.I .Bywater,op.cit.,p. xxi.Op.cit.,p. xxiii. ence inthe Politics promises.The scholarship ofProclus,moreover, is generally uncritical.' Farther,theexplanationofkatharsiswhichBernaysthoughthe discoveredrepeatedinProclusisdoubtful,althoughitsetment o thinkingontheproblemagain.In objectiontoBernays,Susemihl and Hicksargue:('If wesaythatthe' painfulemotion ' offear and pity is removed, we are reminded that the definition in the Rhetoric (2,C 5 , C8)makeseachofthese -fearitselfandpityitself -a sort ofpain(XQnqTLS) , although the emotions are generally defined as 02sgne~a~7j6ovfi."3 TheproperexplanationiscertainlyXQnq Ka' qualitative ratherthan quantitative, and the discussion by Susemihl and Hicks reviews the main positionson the question.4 There isthe view,whichisnotverytrustworthy,thatthe Poetics wasintendedonlyfortheuseoftheschoolinlectures,sothatthe explanationofkatharsiswasoraL6I tislikely, indeed,thatifAris- totleexplained katharsis,hedidit orally,but weneednottherefore rashlyjumpat theconclusionthatthePoeticswasonlyalecture- outline." Margoliouth,indeed,thinksthatthe wholequestionis adequately expounded in the Problems,7and even ifthis work is notauthentic, i trepresentsanancientview,andoneprobablyneareritssupposed sourcethanthatofProclusandIamblichus.AsFinslerpointsout, suchpromisesfrequentlyrefertolaterpassagesinthesamework.& ThePoliticstoamuchgreaterdegreethanthePoeticsdemandsa more complete development ofits topics, and ifwe admit the hypothe- sis oflostpartsofworks,orunfinishedworks,the Politicswouldbe one ofthem.Ifthe latter part ofthe Politics werelost orleft unfin- ished, alatereditormighthavechangedthereferenceofan earlier Finsler,op. cit., p.3; Sandys, op. cit.,i,p.373. F.SusemihlandR. D.Hicks,The Politics ofAristotle,BooksI - I Y, London, 1894, p.652, n.2. aRhet.,2, I ,8,1378 A2 1 ; Nic. Eth.,2, 5, 2, 1105 B23. Susemihl andHicks, Politics, pp.641ff.; pp.650 ff.; "Katharsisas an aesthetic term." Gercke, Pauly-Wissowa,ii, I ,col.1037.35. Op.cit.,ii, I . col.1053. 17. D. S. Margoliouth,The Poetics ofAristotle,London,1911, p.60. Finsler, op. cit.,p.8 andn.3. ALostBookof thePoetics one or of the author himselfto another bookwherethe wordatleast 0ccurred.lFinsler,indeed,believesthattheexpression8vroisn ~ p l TOL~TLK~~S, asit stands, refersto another partofthe Politicsin which the vexedquestionofthe relationofaestheticinterests, especiallyof poetry, to social interests and the state, was di sc~ssed. ~In any case, it is clear that a deliberate attack on Plato, such as some critics expect, wouldnothavebeeninharmonywiththegeneralcharacterofthe Poeticsinitsclosefollowingofthatphilosopher,andwouldhave better suited the P02itics.~ The significance ofthetheoryofkatharsiswassmall inAristotle's view,andthewholemodernassumptionofacompletetheoryofartin Aristotle is misleading.As Bywater points out:the idea ofa theory ofart in general is recent, and goes back to a date no more remote than Winckelmannand Goethe.Aristotle'sideas on aesthetics were mostly those current in his own time.The modern preconceptions,the recent extremeinterestinpsychologyofthephysiologicalkindat which katharsishints,andaboveallthesplendidopportunityforendless scholarlydisputesofferedbysoprominentanobscurityagainst thisbackground,explaintheexaggeratedpresentimpressionofits importance. The majority ofthe references to the Poetics in Aristotle's works are to be found in the Rlzetoric, and halfofthese are capable ofverscation. I n addition to this, it is to be notedthat none ofthe references in the Aristotelian works ever specify a secolzd book.But ofthose which are capable ofverification, all correspond to passages in chapters 21and 22.Chapter 20, immediately preceding, which begins the discussion ofdic- tionwitha passageon the parts ofspeech and other grammaticalde- tails, Butcher considers" probably interpolated,"and he accordingly bracketsit.%Chapters21and2 2 continuethesubjectwithspecial reference to poeticdiction, and in chapter21there is a passage on the genderofnounsthatButcheralsorejects.Although,aswehave seenfromthetheoryonAristotle'smethodofcomposition,thein- consistenciesthatarehereevidentdonotdisprovethegenuineness Op. cit., p.8 and n. 2.0 p . cit., p.6. 0 p . cit., p.8. Op. cit., p.vii. S. H. Butcher,Aristotle's Theoryof PoetryandFine Art,gd ed., I, ondon, 1902, P.3.O p cit., pp.71ff. 26A.PhilipiMcMahon ofthechaptersin whichtheyoccur, itmustbeadmittedthatsuch looseness ofconstructionmadeveryeasytheintroductionofsome-what irrelevant materials, and the task ofan early editor like Androni- cusmustoftenhave beencomplicatedbyshort, incomplete passages which were notclosely connectednorcapable ofclose connectionwith any ofthe set treatises.Such a conjecture is aided bythe possibility ofthe inclusion in these chapters ofpassages from an earlier rhetorical workaapi X8tews.Ritteraccordinglydoubtsthevalidityofthe references inthe Rhetorictosuchmatters inthe Poetics, andattacks the authenticity ofchapters 21and 22. l Gercke doubts the authentic- ity ofthe end ofchapters12,20,and the end of21.He thinks them insertedundertheinfluenceofthelaterStoicgrammarinthethird century,andholdsthepositionofchapter15do~bt f ul . ~Ritter, indeed, devotesconsiderableattentiontodevelopingatheory ofan interpolator and abridger inthepoetic^.^ Cicero, a tany rate, in his comment onthis passageinthe Rhetoric whereone ofthe references to the Poeticsoccurs, says nothing about asecondbookofthePoetics, nothingat allinfactaboutaparallel treatmentofthe matterelsewhere.4 Dielselaboratedthetheoryoftheworkaepi XCtewsspokenof above.5HeshowsthatAristotlemusthavebeentheauthor,and thatitwaslatercombinedwiththetwobooksonRhetoricintothe workaswehaveit.Thistheoryprovidessomeinterestingpossible consequences;namely, either that the same person who combined the aepi X ~ ~ E W S withtheRhetoricinsertedsomeofthematerialsinthe Poetics,andthatthisenlargededitionofthePoeiicswasnot,how- ever, the one which gainedwidestacceptance;or that early criticism rejected part ofthe insertions.If the insertions had been made before the workreached Andronicus,perhaps their partial elimination is due to him.These conjectures, however, only add to the reasons for hold- ing the evidence ofthe references fromtheRhetorictothePoetics in slightesteem. Ritter,op.cit.,pp. 230-243. 2Gercke, Pauly-Wissowa,ii, I . col.1oj3. 48. Ritter,op. cit., pp. xx ff. Cicero, DeOr.,2, 58. Diels,Ueber das dritteBuch, p. 34. 27 ALostBookofthePoetics The references to a treatment ofthe ridiculous, moreover, may have beenbasedonthe Ethics,as was largely the case withthe discussion in the Tractatus Cois2iuia.n~~.Indeed, this treatise, manifestly ofPeri- pateticorigin, and showing several different strata in its development to its present state, may have been thought to be a part ofthe original Poetics by some editors, and these references may actually correspond to it. It is,however,thepledge,foundinPoetics6,1449B 2 ; , todeal withcomedy laterthat, nextto the authority ofthe indices seems to impressmostcriticswiththenecessity forasecondbook.Butthis reference comes in atthe head ofa new division, in which the philoso- pherbeginsthediscussionoftragedyalone,sothatitwasprobably usedasaconnecting link,perhapsoriginally insertedinthemargin, but afterwards creeping into the text. Partofthe promisecontainedinthispassage ofthePoetics isful- filled to the satisfaction ofcritics, -that regarding " the poetry which imitates in hexameter verse,"but the remarks on comedy in chapter 5 do notappear to these persons s a c i e ntfor the purpose ofthe initial announcement.Aristotle says merely:"I propose to treat ofPoetry initselfandofitsvariouskinds.. . .Epicpoetry,andTragedy, Comedy also and Dithyrambic poetry, and the music ofthe flute and ofthe lyreinmostoftheirforms, are allintheirgeneralconception modes ofimitation." There is a further possibility that, although the part ofthe reference whichtouchesepicwascorrect,theinsertionofthewords" andof comedy"wasafurtheraddition, probably by a stilllater scribe after thereferencetoepichadfixedaplaceforitselfinthetext.This sentencecomesimmediately 'aftersomeremarksontherelationsof tragedy and epic, and following i tis the series ofchapters on tragedy. It isnotstrictlynecessary,andwithoutitthetreatiseasitstands would beself-suilicient and consistent. Diintzer pointsout that the beginning ofthe fifth chapter indicates that the philosopher does not intend to discuss comedy f ~r t h e r . ~This reference isnotameretransition,butcontainsanessentialstepof progress inthethoughtofAristotle,anddoesnotbearthemarkof 1 Poetics, I. 1447A 1,2. Dihtzer, Zeit.f.d. Alt.(1844,pp. 278 ff.Cf.p. 283. A.PhilipMcMahon the scribe.Herethewriterimpliesthat heclosesthesubjectofthe nature ofcomedy by connecting his observations with some preceding remarksontheobjectofimitationincomedy.Inthetreatmentof tragedy, which begins in the next section ofthis same chapter, he does not refer backto the deterqinationoftragedy'sobject,as he does for comedy, forthe veryreasonthathehasnotclosedthesubject, but developsitmorefullyinthefamousdefinitionat thebeginningof chapter6.I tisthereforeonlyreasonabletosupposethatAristotle would not have connected his points atthe beginning ofchapter 5with his philosophyofimitation, and then have stated his conceptionofthe ridiculousinthis placeifhe hadever intended to treat it later.This is a consideration which does not depend on the use ofmere references oftransition, but indicates a greater degree ofcoherence than is usual inmostoftheAristotelianworks,ameritwhichitcouldthemore easilyhavepossessed hadit beenoriginallyshort and self-contained, withoutthe additionofa second book. A further evidence ofAristotle's intention to finish up the treatment ofcomedyatthis point is his balancing ofvarious aspects ofthe sub- ject,andthus a briefstatement ofwhythereis no historyofcomedy is set against an outline ofthe evolution oftragedy.Since, however, to hismindenough hadbeensaidaboutthe aestheticsofadramatic speciesinwhichhewasnotmuchinterested,heparallelsthelater elaborationofhisdefinitionoftragedywiththecursoryremarkson comedya tthe beginningofchapter5.Stillfurther,attheendofchapter22beforetheconsiderationof epic, which is promisedin the firstwords ofchapter 6, we read,"con- cerningtragedyand imitationbymeans ofactionthismaysuflice." Now, this sentence, by whomever inserted, whether by Aristotle or by thecustomaryeditorwithhismethodofjoiningtogetherseparate portionsofthetreatiseatthispoint,establishes severalfacts.One is that the writer judgedthat the discussion ofall drama -" imitation by means of action "-hadbeen iinished when the treatise reachedthis point.Next, that more about comedy did not follow this observation. Then, except for the words" and ofcomedy,"this referenceinchap- ter6,whateveritsorigin,wascorrect,forthediscussionofepicis delayedbyituntilaftertragedy;whentragedyisdone,thereis anothertransition, summing upthedramaas a whole, andtheepic ALostBookof thePoetics istakenupaccordingtotheannouncement.Yetjustas thelatter reference a tthe end ofchapter22states that here ends the discussion ofdrama, -aepi pkvotv~payq6ias~ a 1 CvTGa ph~~e i vSO T ~ S pi p+ue~s,t he promiseabout epic, aepiplvotv~ i j sCv2[apC~poispi pq~i ~i j sis ful- filled,and a t t he veryendoft hetreatise wefindi t saidin con- clusion:'"ThusmuchmaysuiKceconcerningTragicandEpic poetry in general."2 AsepicrepresentedforAristotleallthe6iq-yqpa~i~+,SOdidtrag- edyrepresentalltheformsofpoetrywhichimitateCvTGaph~~ei v.There isalsotheobviouslyawkwardconnectionofcomedywiththe designation ofepic as the poetry which "imitates in hexameter verse " -oneaconceptionaccordingtomediumemployed,andtheother according to the aesthetic principle,if" comedy " meansanything in thiscontext.I tispossible,indeed,reasoningonthe basisof23,I, that,insteadofthewordK U ~ ~ ~ ~ C I S , originally theword8iqyqpa~i~ijs stoodinchapter6,3whileanotherscribe,misunderstandingordis- likingtheword,insertedinthemarginthe wordKwpq6ias, and that thiswas substituted forthe originaL4 Asalready noted,thereare notracesofthe supposed treatmentof comedyfromthesecondbooktobefoundinVarrooranyofthe grammarians.Thereis,indeed,theTractatusCoislinianus,which Bernaysheldcontainedsomereflections ofthatlostbook.He him- self, however,pointedoutthe secondaryandderivative characterof this reflection, and from that beginning,scholarly opinion has gone so far as to say, with Bywater:" There is no evidence to show that the Op. cit., p.283.Poetics, 26.1462 B 8. aForin23.1459 AI 7 iv[l] phpycorrespondstobifaphpocsbut6bt)-ft)fiarrtiis while moresuitable than rpayy6ias, is not its equivalent, although found in the cor- responding position.There has been a good deal ofdoubt, indeed, as to the proper reading in23.1459 A17.Butcher, on the basis of1449B 11 and1459 B 32, writes ivlfi hpy,although the codices have 4vphpy,and he notes the reading bv CEafihpy ofHeinsius.This last Vahlenalsonotes,butadoptsthereading ofthecodices. Bywater agrees with Vahlen in his text. As Aristotle seems to include all non-dramatic poetry under the term, 6r~yt)parcmj alone is not the equivalent of that division, while, if6~vt)parc~i j s were there instead of ~w&ias,it wouldthenpresentalogical indication ofthewholefieldofnon-dramatic poetry and ofthat part ofit which is actually treated in chapters 23 ff.DHntzer, op, cit., p.282. "water,op. cit., pp. xxi, xxii. A.PhilipMcMahon later grammarians knew ofBook 11, or ofthe theory ofComedy which must have formed part ofit.But there is evidence showing it to have been unknown to them.The so-called Tractatus Coislinianus preserves a definition ofcomedy, which has no doubt a certain Aristotelianlook; any one can see, however, by simple inspection that i tis nothing more thananadaptation,orrather,asBernayscalls it,atravesty,ofthe well-knowndefinition ofTragedy inthe existing Poetics." Theenumerationofthepartsofcomedyisthesameasthatin Tzetzes, whoappearstohavegotit fromEuclides.Butthen,if,as some scholars think, Euclides was a grammarian ofthe classical period,' it only provesthat neither the compiler ofthe Tractatus, nor the early grammarianhadseenanysecond book.Nobodyexpects Tzetzesto furnish any reliable evidence on such points. Nevertheless,whileAristotelianscholars havetestedtheTractatus andfoundi twanting,thespecialistsinAristophaneshaveinrecent times become aware ofit, and accorded it a more hospitablereception than it ever before received.Starkie relies implicitlyonthe findings ofBernays, and asserts:"The value ofthis fragment was not fully realizedtillBernaysdemonstratedthatitrepresentedasummary, mutilatedandmisunderstoodinparts,ofAristotle'sanalysisofthe laughter in comedy.. . .Rutherfordalone has shown a due appreci- ation ofits value."Later he reproduces with approval3 the definition ofcomedy found intheTractatus, and indulges ina laboriousclassifi- cation of" the various methods ofexciting laughteremployed bythe writersofoldcomedy,especially Aristophanes,"" accordingtothe division ofAristotle,"i.e.,theTra~t at us. ~ What Starkie terms a " due appreciation" ofthe value ofthe Trac- tatus by Rutherford is ratherimmoderate zeal.He says:" I tis not thatthelaughter ofcomedy hadnotbeenproperlyanalyzed.Even thescrimpand grudgingabstract,nowsole relicofthesection inthe Poetics concernedwithComedy,willconvince anybodywhokeepsit inhisheadas helistenstoGreekcomic~pbuwaa, thataGreekhad 0 9 . cit., p. xxii. 2W.J .M.Starkie,The AcharniansofArislophanes,London,1909, p. xxxviii. 0 p . cit., p. xl. Op.cit., p. xxxviii. 6W. G.Rutherford, AChapter i n the HistoryofAnnotation, beingScholiaAris- tophanicaVol. 111, London,1905, p. 435, l .19.31 ALostBooko jthePoetics indeed read for Greeks the most secret heart ofthe ' mother ofcomedy,' and probe in hand,had made clear wherefore i tbeatand whati twas made of.. . . But Aristotlethought too much and was too greatan observer to be loved by commentator and rhetor.Living atease within their pale ofwords, it was not likely they would ventureoutside to be exposed to the perils and painsofthinking." RutherforddilatesontheTractatwsanditsapplicationtothe methods and interpretationofAristophanes.' Except for Kayser'srecenttreatment, it is Bernays who has offered the chiefdetailed discussion ofthe Tractatus;a discussion which is the fountain-head ofthe whole modem beliefinthe second bookas a fact establishedtothesatisfactionofscholars.He" demonstrated,"as Starkie puts it, " that i trepresentedasummary " onlybyemploying inan extremely boldandoften unwarrantedfashion mostofthe evi- dence wehave seen reasonto reject. Bernaysfirstpremisesthat,inordertoestablishthefactthat Aristotledidtreatcomedyfully inthe Poetics,theannouncementatthe beginning ofthe Poetics to treat all poetry should be takenin con- junctionwiththereferenceat thebeginningofchapter6tospeak aboutcomedylater,togetherwiththeunsatisfiedreferencesinthe Rhet~ric.~Since these premises do not necessarily lead to such acon- clusion, hiscase is therebyalmost lost. BernaysnextgivesthetextoftheTractatusandmakesthefol- lowing observations uponit.First, there isno unity apparentinthe treatment except that ofsubjecL4Next, the beginning ofthe Tractatus isunpromising because its division ofpoetryinto imitativeandnon- imitative flies inthe face ofthe plaindeclarationofthe Poetics where Empedocles is even refused the title ofpoet because his verse does not imitate.In addition to this he discovers several other notable errors, from theAristotelianpointofview,intheTractatws.Firstthereis the employment of6 iOZKTOU~ a l 46pos. 6CovsforAristotle'sZXEOS~ a lSecond, thereis abalancing between fearand sympathy, whereas theRhetoricexpoundsatheoryaccordingtowhichsympathy oughtnott obe&KK~ OL J UT L K~ VTO^)&XS.ov.Third,i t isstatedt hattragedyFxri p q ~ Cp a T ~ V This assertion isbasedon the Rhet-Xirxqv. OP.tit., PP. 435-455. a09.tit., pp.137-139. Bernays, ZweiAbhandlungen, p. 135.09.cit., p. 140. 32A.PhilipMcMahon oric, but in the Poetics Aristotleplainlyspeaks ofaproper 48ovtjin tragedy.' When he comes to the definition ofcomedy contained inthe Tracta- ks,2 Bernaysexclaims:"Diese seinsollende Definition derKomodie ist nichts als eine jammerlich ungeschickte Travestie der aristotelischen von der Tragodie."The sentimentality ofthe statement that comedy has 76vy6Xo~aas pq~kpais glaringly repugnantt o all t hatweknow ofAr i ~t ot l e. ~ Thereare,indeed,asBernayspointsout,certainopinionson comedyexpressedby Aristotle,especiallyinthe Poetics,theEthics, andthe Politics,wherethe laughterofcomedyis touchedBut, as he also shows, later grammarians mistook Aristotle's views, confined to Middle Comedy, for the whole ofthe ~ubj ect . ~Thus the distinction betweenXo~Gopiaand~wpy8i amighthave beenderivedin t he first place from the Poetics, as the wording is not un-Aristotelian! except the expressionZp+aa~s,for whichAristotle'sequivalentis b~bvo~a.Section 5ofthe Tractatus, Bernays proves, derives from theethic^,^ andthe balancein section 6betweenykXwsand T ~ P+ L Sis modelled afterthe +bBosandZXrosoftragedyin thepoetic^.^ To strengthen his case, Bernaysatthis point digresses on the prob- able influence ofthe Peripatetics on New Comedy, and ofthe relation ofTheophrastustoMenander,sayingthatthePoeticsagreeinthe main withthe practice ofthis poet.n ReturningtotheTractatus,Bernaysnotesthatthesixelements ascribedtocomedyaremodelled onthoseattributedtotragedyby Aristotle.l0Theelaborationofthisnumberofelementsfor comedy, once they were obtained,shows a surprisingly ingenious thoughsuper- ficial manipulationofother Aristoteliantexts."Thepedantry ofthe achievementpuzzledCramerandothereditorsofthework,andthe resultsmoveBernaystotermit" eineVerkehrtheit " . . . ('je Op,cit., pp.141ff.Kaibel, Comicorurn Graecwum Fragmenta, i,I, Berlin,1899,p. 50, 3. Bernays,Zwei Ablzandlungen, p.145. 0 p .cit., p. 147. Op. cit., p. 148. Op.cit.. p. 150.Op. cit.,p.152. Op. cit., p.151.loOp. cit., p.153. 8Op. cit., p.151." Op.cit., p.154. 33 ALostBookofthePoetics deutlicherdermechanischeWegzuTageliegt."The expansion of pBhosis believed by him, however, to pointto a more complete work.2 Atthispoint,however,Bernaysfeelsjustdiedinassertingthat "Wasubersie(dieKomodie)daherderExcerptorinderjetzigen PoetiknichtNachweisbaresbeibringt,darffiiglichausdemvoll- standigerenExemplarhergeleitetwerden, wofern innere Griinde nicht dawider sind."But since, as our case now stands, genuine and inde- pendenttracesofatheoryofcomedy mustappearintheTractatzls, thecontraryconclusiontowhatBernaysstates naturallyflowsfrom his previous examination. Thedivisionanddiscussionofthecomiccharactersis,indeed, genuinely Aristotelian,butnotindependent, for i tcouldall befound elsewherethaninthePoetics,andeasilyaccessibletotheingenuity ofthescholarwhocomposedtheTractatu~.~ThusBernaysiscom- pelledtoadmitthat: b" Vielleichthatteeingliicklichspurender Scharfsinn, ohne weitere Hilfe, aber dann auch wohl ohne allgemeinere Zustimmung, blos aus diesen Stellen der Ethik und Rhetorik die nach Aristotelesan sich komischenCharacktereaufdie. dreizuriickfiihren konnen, welchederExcerptornennt.Diesem wirdnachdem Unge- schick das er schon zweimal bei BenutzungderRhetorik gezeigt, Nie- mandgeradehiereinesoglanzendeCombinationskraftbeimessen wollen;undwenninEthikundRhetorikjeneDreizahlangedeutet scheint, so ist das nur ein Beweis mehr,dass Aristoteles sie auch in der Poetik aufgestellt undder Excerptorsie von dort abgeschrieben hat." BernaysthenoutlinesAristotle'sprobabletreatmentofthewhole questionofcomedy, butacknowledgesthat"derExcerptorhat nur die Rubrikdesselben ausgezogen." Now when a critic makes capital ofhis deficits in this fashion he can prove almost anything.Dzerences in degree ofpedantic acumen and in accuracy are easily accounted for by the assumptionthat the Trac- Satusrepresentsnotthe workofone scholar, butanaccumulationof the work ofa number ofrhetorical investigators.Ofcourse the excerp- tormadegrosserrorsincombining hisscatteredhints,butit isnot necessary to suppose that he was the same who formulatedthe defini- Op. cit., p. I56.' Op.cil., p.159. Op.cit., p.I57.Op.cit., pp.163ff.Op.cil., p. 158. 35 ALostBookojthePoetics After thus disposing ofthe mainpointson which critics have depen- dedforatheoryofthesecondbook,thereremainonlyafew ofthe less important references.Among these is the reference ofSimplicius t o Aristotle for a treatment ofsynonyms.Rose, however, shows that this is derived from Porphyry, and in turn depends uponthe reference in the Rhetoric.' The allusionofthe Antiatticistis veryslight evidenceinany case, and it is not atall certain that this anonymous controversialist referred to the Poetics in citing Aristotle.Even allowing this unknown writer the meritofhonesty,weare not obliged to conclude that his memory was sound or his source genuine.The matter to which he alludes may possiblyhavebeencontainedintheworkm p l Xkt;~ws, afterwards absorbedintotheRhetoric,oreveninchaptersofi tinsertedinthe Poetics t hatlater editors rejected. Bywaternotesvariousanomaliesofthoughtorlanguageinthe poetic^.^Amongthese he mentions:the anticipatory use oftechnical terms afterwards defined;variations ofterminology;inconsistencies in the use ofterms; inconsistency ofthought; and lapses ofmemory.Yet he also defends the philosopher on the ground ofhis natural limitations, showing how the Greek play limited Aristotle's views by its conventions with regard to stage presentation, form and structure, motives and sub- j e c t ~ . ~ His ideal playwas a compromise between thedramaof the great periodand that ofhis own generation, seventy years after the death of Euripides.Among the evidences ofthis assertion that Bywater brings forward are:Aristotle'stheory oftragic diction, andthe silence about the chorus;his concessions inplot to the more sensitive feelings ofhis audience instead ofthe harsh situations ofthe older tragedy;and the fact that his theory ofcomedy would have been more applicable to the NewComedy than to Aristophanes.For the state ofthe text he has the usualarguments. Bywateralso recordshisopinion,however,that" the bookas it is withoccasionalsidelightsfromotherworksisintelligibleenough." Now it is much more intelligible ifwe do not look for something inthe Poeticswhichthere is noreasonto supposeeverwasthere,orto feel disappointedwhen we fail to find it. Rose, Arist. Libr. Ord., p.133.0 p .cit., pp. viii, ix. Bywater, op.cit., pp. xiv ff. ' Bywater, op.cit., p. viii. 36A.PhilipMcMahort In the course ofthisinvestigationI havetriedto deal impartially withalltheevidenceurgedfortheexistenceofasecondbook,and withthescholarswhoheldsuchanopinionbasedonthatevidence. Theonlydirectstatementsthatthereweretwobooks,thoseofthe indices, have beenconsidered;and those references havebeengiven which indicate thatthere was morethan one book, -fromthe Aris- toteliantextitself,fromAmmonius,Boethius,Eustratius,andan anonymouscommentatorontheRhetoric.I nadditiontherewere the matterssupposedto have beenin the Poetics, but not now found there,indicatedbythetime-referencestothePoetics:thepromise aboutcomedy;thecross-referencesoftheRhetorictothePoetics about the ridiculous;the incomplete discussion ofkatharsis;and the theoryon comedy,reflections ofwhichBernaysthoughthefound in theTractatus Coislinianus.Furthermatterswhichmighthavebeen foundthere,accordingto scholars,were:adiscussion ofsynonyms; a treatment ofthe drama in defence against Plato;and something on comic diction.This outline was followed by a briefstatement ofhow the tradition ofa lostsecond book began and was developed. Againstthiscaseitwasarguedthatallthethreeindicesareun-reliable;that there are also references elsewhere inwhichthe Poetics isspokenofascontainedinonebook;thattheinconsistenttales, especiallythatofthecaveat Skepsis,andotherexpedientsdemon- strate the d%cultyin accounting for the loss ofthe second book, which is easily avoided by a rejectionofthe suppositionthat there everwas one.Thencameadetailedexaminationofthecross-references in general,where it was shown that theycannotbeusedas direct proof ofanyoftheirimplications;andAristotle'smannerofcomposition and publication was discussed.The investigationwas concluded with an analysis ofthe various references individually, with negative results as to theirvalidity,includingarefutationofthe claim ofBernaysto have foundtraces ofthe complete theory ofcomedy.The final result isaconclusion thattherewasnosecond book.While i tislogically impossible to proveauniversalnegative,thereisnoreason forusto believe, in this case, that there ever was a second book;and the facts ofthe case are all harmonized and accommodatedto one another with- out such an hypothesis. ALostBookof thePoetics 37The dialogues were,as Shute observes, ( ( the compositions ofAris- totlewithwhichantiquitywas bestacquainted,andforwhich,next to the IIoAi~~iai,Since thereisabun-we havethe bestauthority." dantevidencefrommanysourcesthatAristotleuseddialoguesin thefirstperiodofhiswork?Roseattributesthisdisproportionate influenceofthe Politics andthedialogues to the factthat, while the Politics,byreasonoftherangeofsubjectstreated,wereespecially interesting to grammarians and historians, the dialogues, because they treatedthequestionscommonto philosophyafterPlatoinaneasy, lucidandpopularstyle,attractednotonlythephilosophersofthe RomanEmpire,butalsothelaterrhetoricians.AmongtheseRose mentions Panaetius, Posidonius, Andronicus, Didymus, Varro, Cicero, DioChrysostom,Julian,Themistius,Basil,andPlutarch.4Indeed, the researches ofBernays,provingthat bythe exoteric works refer- ence is had to the dialogues, show that, in view ofthe unbroken testi- monyofantiquity,thedialogues wereAristotle'sin asense that can beappliedtononeoftheotheracceptedworksintheAristotelian The references wehavearemostlyhistoricalnotes,and arouseno suspicionagainsttheirgenuineness.ItistruethatFragment76 contains astatementaboutHomer,apparentlybasedonatradition prevalentin10s.Thequestionablenessofthisstatementcannot provethedialoguespurious,fori tisclearthatanauthordoes not necessarilybelieveallthestatementsputintothemouthofthe speakers inadialogue! The example followed bythe masterin dealing with poetryindia- loguesseems nottohavebeenan isolatedperformance,andwefind 1J. Bernays, DieDialoge des Aristoteles imVerhaltniss zu sei en &rigenWerken, Berlin, 1863; E.Heitz, Die verlwenen Schriften des Aristoteles, Leipzig, 1865;Schlott-mann, Arsdialogorurn quas vicissitudines apud Graecos etRomanos subierit, Rostock, 1889,pp.19-25;R.Hirzel, DerDialog, 2vols., Leipzig, 1895.2 Shute, op. ci t . ,p. 7.aZeller, op. cit., i, p.55,n.z. Rose, Arist.Pseud., p.23. Shute, op. cit., p.105. Zeller, op, cJt., i, p. 58,n.I. A.Philip McMahonHeraclides, among the Peripatetics,writing a dialogue ~ a p i T O L ~ T L K ~ ~ S~ a i TBV TOL~TBVIndeed, Heraclides appears to have written a a'.' number of works on the subjeck2 He also probably showed theinfluenceof Platohere; and he is coupled together with Aristotle byDio Chrysostom, as being a writer in that form in which Plato wasthefirsL3One of the characteristics of Aristotle's extant dialogues in whichtheydifferedwidelyfromhisotherworkswastheir style. ThatAris-totle himself estimated very exactly the literary quality of Plato'sdialogueswe see from what appearstobe a fragmentof his dialogueOn poet^,^ where he says that they are midway between verse andprose. Adistinctanddeliberatelyplannedliteraryexcellencewasoneof the qualities that antiquity specially noted also in Aristotle's dia-logues. We may refer, for instance, to the passage in Ammoniuswhere thebeauty andappropriateness of thelanguageis mentioned:andtoanotherpassage in E l i a ~ . ~ Thestyle,indeed,was,asThemistiusobserves,theprincipalattrac-tion of the dialogues for the readers of Aristotle. Cicero, except forthe Rhetoric, does not, on the evidence found in his works, seem tohave read much else of Aristotle but the dialogue^.^ His remarksconcerningtheirstylesuggestthesameconclusion.8 Cicero's allusionstoTheophrastusestablishsimilarfactsforthestyleof Aristotle'smostfameddisciple,and,asZellersays," Inhiscase,asinAristotle's, thismerit belongs chiefly to his popular writings, and especially to thedialogues, which, likeAristotle's, aredescribed asexoteric." Itwasevensaid,thoughoninsd5cientgrounds,bysomeof theancients,thatTheophrastus receivedhisnameon account of hisgracefulstyle.11Rose, Arist. Pseud., p. 77; DiogenesLaertius, De vitis,dogmatibusclarorumphilosophorum librix,2 vols.,Amsterdam,1692,5,86; 5, 88.20. VOSS, De HeraclidisPontici Vitaet Scriptis. Diss.,Rostock,1896,p.31. 3Rose,Fragm.,p. 24; DioChrys.,Or.,p.634,Emp. *Rose,Fragm.,p. 78; Diog. Laert., 3, 37. 6Rose,Fragm.,p. 23;Arnmonius,Proleg.i nArist.caleg.(p.36B28Br.). 4Rose,Fragm.,p.23; EliasinArist.cat.,p.26B 35. 7Shute,op.cit.,p.64. 8Rose,Fragm.,p. 23; Themist.,Or.,26 p.385,l. 28,Dind; Hirzel, op.cit.,i,p. 280; CiceroAcad.#r.119.9Zeller,op.cit.,ii,p. 352,and n. I. Op,cit., ii,p.348,n. 3.39 ALostBookof thePoetics AnotherknownfeatureoftheAristoteliandialogueswastheir method.Cicerotestifiesthatthespeeches ofother personswereso introducedthatAristotlehimselfwasalwaysthe principalspeaker.' Thischaracteristic isotherwisedescribed byBasilasapracticeof direct statement in distinction from the contrast ofopinions employed byplat^.^From Cicero wealso learnthat in dialogues ofmore than onebook,each parthadits ownprefa~e. ~OnthisquestionProclus gives similar informati~n.~From such evidence has also been derived the accepted beliefthat the dialogues were essentially distinct in form and methodfrom the extanttreatise^.^ In supportofBernays'sopinion that exoteric andpublishedworks areidenticalandreferalmostalltothedialogues, maybeadduced thepracticeofreferringtotheotherworksinwayswhichindicate thatonlythedialogues wereconsideredfinished literaryproductions and so worthy ofpublication.'Thus the reference in the Poetics to one ofthe publishedworks ismostnaturallyassigned tothedialogue On Poets, as Zeller remarks:rather than to the Rhetoric, as Rosesuggest^,^ sincethereisnocorresponding passagethere.Vahlenagreeswith Zeller onthis point,1 and while exoteric may include some ofthe more popularofthereallyesotericclass,suchastheRhetoricandthe Politics,wecanreasonablyinfer,because oftheabsence intheseof passages corresponding to references which require an exoteric or pub- lishedwork,that thedialogue 0%Poetsismeant."Thus, whilethe works still entire in the accepted Aristotelian canon afford little dehi t e proofofthe genuineness ofthe dialogues, the unanimous agreement of antiquityshowsthattheeditorswhoinsertedsuchreferencesto exoteric or published works had authentic Aristotelian dialogues which theycouldhavecited,andfromthecharacterofthefragmentspre- Rose,Fragm., p.23;Cicero, Ep . adAtt.,13, 19.* Rose, Fragm., p.23;Basil, Epist., 135.Hirzel, op. cit., i,p.275.Shute, History, p. 64.Cf.Cicero, AdAtt.,4,16, 2;13,19,4.Rose, Fragm.,p.23;Proclus, In Parmen., t. iv,p.54 Cous. Zeller, op.cit., i, p.55. Shute, op.cit., p.7. ZeUer, 09,cit., i, p.58,n.I. Rose, DeArist. Libr. Ord., p. 79. lo Vahlen, op.cit., p. 36, n.on Poetics,I;. 1454 B18. l1 Shute, op. cit., p.21. served, we believe thesefragments to be Aristotelian and to show that thedialoguescontainedpassagescorrespondingtosuchreferences. Furthermore,itisanequallyreasonablehypothesisthatreferences to the Poetics whichare nototherwisedefinitelyaccountedfor,may, onaccountof the confusion of OnPoetsand Poeticsastitlesofboth the dialogueandthe treatise, correspondto passagesinthe dialogue. Since antiquity great confusion has been caused by the similarity in names betweenthe Poetics and On Poets, one ofthem a formal treatise andtheotheradialogue.Allusionstotheseworksareinextricably contradictory, for withtwo works,eachcalled by either oftwotitles, andwithreferencestobothofthetitlesandbothworksas in from onetothreebooks,therearetwelvepossible combinations. Besidetheeasyconfusionoccasionedbytheclosesimilarityof names, the Poetics and On Poets, and a resemblance in subject-matter, we must also reckonwith the possibilitythat the dialogue, as the only formally published work, received a title earlier, and that general agree- mentonthe titleofthe treatisewas not reachedinantiquity. That the workOnPoets wasadialoguecannot, aswehave shown, be fully proved, but ifthe treatise is by agreement entitled the Poetics, thenwemustcallthedialogueOnPoets.The realquestionisthen: how did the dialogue differ in method and treatment from the treatise, and what other facts are knownabout it ? The orthodox view, with its inherent d%cultiesmore or less glossed over,isset forth byRose,'andmaybesummarizedinthe following manner.ThatthethreebooksOnPoets,whicharementionednear the beginningofthe index ofDiogenes Laertius, were really writtenin the mannerofa dialogueis expressly stated intheVita Marciana of Aristotle.3In spite oftheconfusion oftitles,thedialogueisdistin- guished by having one bookmorethanthe t r eat i ~e. ~Inthe dialogue were also discussed the artofpoetry;its nature as a kind ofimitation, its functioninthe state, and so forth, -allthisincapable ofdemon- stration by Rose'smethod oftreatingthe evidence.If, however,one Rose, Arist. Pseud., pp. 77-86. Op.cil., p. 25.8Rose, Fragm., p. 76; Vita Aristolelis Marciana (cod.257)f . 276A; Rose, Arist. Pseud., p. 77. *O p cit., pp.77, 78; Zeller, op. cit., i , p. 58,n.I. Rose, Arist. Pseud., p. 78. ALostBookofthePoetics 4I oftheworkswasinatleast one bookand theotherinthree,and if there is no reasonto suppose that the treatise, called the Poetics, was in morethan one book,thenthedialogue OnPoets was in three. Thecharacterofthedialogueisascertainablewithtolerablecer- tainty.Hirzel shows that the dialogue discussed the relationofphilo- sophyto poetry,and contendsthat,whilecriticsusuallyconsideri ta purely historical work, excluding the theoretical discussion of poetry, they do not makeit clear howthe historyofpoetrycould suitably be madethematterofadialogue,evenofanAristotelianone.'The fragmentsthemselvesleadtoanoppositeconclusion.The examples wereprobablycitedtosustainpointsoftheargumentasinthe Socratic dialogues,and the fragment whichmaintainsthat whenthe philosopherrhymes or the poet philosophizes,either the poetryor the philo'sophyisinferior,indicatesaphilosophicaldiscussionofaquite Platonic~har act er . ~Indeed,itisapurelyphilosophicaldistinction which opposes historical to poetic truth as the diflerence betweenpar- ticularandgeneraltruth,sothata tthe end ofthe Aristoteliandia- logue Socrates, who first investigatedgeneralconceptions with worthy results, and Homer?who above all others deservedthe nameofpoet, couldbebroughttogether.Thisconjunction,aswelearnfromthe fragments,probablyoccurredinthethirdandlastbook;andthe criticaltheory involvedagrees withthat inthe Poetics and the Meta- physics.6 I thas been supposed that the dialogue also treated the art ofpoetry, asupposition whichwouldexplain howthe dialogue could sometimes beentitledtheP o e t ~ s . ~ Bywaterthinksthat,atanyrate,thedia- logue did not devote special attention to the question ofstage-effects? The relationofthe dialogue OnPoets to the treatise the Poetics has beenseveraltimesdiscussed:andithasbeenthoughtpossiblethat manyofthereferences tomatters poeticwhicharenotfoundinthe Hirzel,op. cit.,i, p.288 and n. Op.cit.,i, p.288. Bywater, op. cit.,p.19.Note on PoeticsI, 1447 B18.' Gercke,Pauly-Wissowa,ii,I,col.1052. 63. Hirzel, 09. cit.,i, p. 289. Zeller,op. cit.,i, p.58,n.I.' Bywater, op.cit., p.233. 8Diintzer,op. cit.,p.278. A.PhilipMcMahonPoeticsmighthavebeenintheworkOnPoets.Thishypothesisis surelyasvalidandreasonableonitsfaceasthetheorythatsuch mattersweretobefoundinasecondbookofthePoetics.Two ofthe parallelsinAthenaeustothe Poetics notedbyVahlenmight conceivablyhavebeendrawnratherfromthedialogue,aswellas another parallel, also noted by Vahlen,inthe Anonymus de comoedi~.~ Moreover,the passageinThemistius,possibly,but by no meanscer- tainly,anexpansionofinformationderivedfromthePoetics,might, inviewofhisacquaintancewiththedialogue,havemoreprobably beenderivedfromthelatter.VonChristjudges,finally,thatthe dialoguewasapreliminarydiscussion, followedlaterbyamorepro- foundand technicalwork? The matter ofthe dialogue isbroadlyindicatedinthe extantfrag- ments.Much ofthe surviving material is not assigned to any particu- larbook,butthereareanumberofcases inwhichthe bookisindi- cated.Thus, inthe firstbookthere wasadiscussion ofthedialogue form and areferencetoPlate: a fact whichisfurther establishedby the evidence ofAthenaeus6 The second book,accordingto Macrobius,introducedthe evidence ofEuripidesonaquestion ofAetoliancustom^.^ The thirdbook,accordingtoDiogenes Laertius,gaveananecdote aboutSocrates?Inthisbook,also,therewasthediscreditedstory aboutHomer'sorigin?and ,from theseindicationsthematterofthe other fragments canto some extent be grouped in the different books, sothatwemaynowagreewithHirzelthatthe placeofthepassage mentioningbothSocratesandHomerwasat theconclusionofthe whole work.9 Athenaeus,8,367B; ii,p.302,ed.Kaibel; cf.Vahlen,op. cit., p.53,n: andAthenaeus,g,433C; ii,p.442,ed.Kaibel; cf.Vahlen,op.cit.,p.51,n. * Vahlen,op.cit., p.13,n. onPoetics,4.1449B 7. Vo n Christ,Geschichte,i, p.674. ' Rose,Fragm., p.77;Diog. Laert., 3,48. Rose,Fragm., p.78;Athen.,xi,p.505C. 6Rose,Fragm., p.78;Macrob., Saturn., 5,18, 19. Rose,Fragm., p.79;Diog.Laert. 2,46. Rose,Fragm., p.79;fragment 76. Hirzel,op.cit., p.289. 43 A Lost Book ofthePoetics IV.TEIEOPHRASTUS VarioussourcesrevealTheophrastusasthechiefdiscipleof Aristotle.Hisinterestwasmainlyscientific,buteveninscience hestroveto completeandsubstantiatetheprinciplesofhismaster, and introduced no radicaldiflerence~.~AsBoethius bears witness, he advancedfurtherthanAristotleinfields whichthemasterhadbut slightlytouched;otherwiseheaccepted histeachings."In the same placeBoethiustellsusthatTheophrastussometimesusedthevery wordsofAristotlewithout~ha nge . ~Cicero,indeed,pointsout that Theophrastuswasmoreaccurateinhisobservations,andespecially developedresearchinnaturalsciences.6CiceroalsothoughtTheo- phrastus a closer follower ofthe masterthan other Peripatetics,while Galen seems almost neverto h d any difference betweenthem.-' Zellerpointsout that Theophrastus investigatedthepsychological effect ofmusicandheldthat certaindiseases could behealedbyit? The few fragments we possess ofthis discussion le