Professional Documents
Culture Documents
3 Geometry and scepticism IAN MUELLER ly Book mt of avers Mathematios (Against the Geomecers) Sextus Empircos directs his sxpiclarlery against the stom point of the Greck philosophic amy, Euclidean geometry." The [ssl iwlkely co imprese the modern reader who knows Kane for Hibere oe Poincate. Sextus’ arguments are repetitive, and ‘pend on afew, relatively simple, ideas, and some ofthe ideas set outrageous, Morcove, if one sce Sextus atack a part ofa ‘vat called philosophy, sshich began withthe Greeks an xsl ‘continuing, itis probably air to say that Sextus ison the losing ‘Sle, But if one is more interested in battles than in che whole war [think it can be sid tat Sextus won his with the geometes, or at least achieved alle ever aimed at, a stnd-off in sshich every ‘atioalparscipant acedes to the slogan mall (80 more’ a8 in'pisno more uve dan ts denials). Fra philosophical bates Fought in spec mllectl conditions. Sextus’ opponents could not invoke the work of Dedekind or Hilbert, andy a shall rg, ‘what they could invake was inadequate to avoid conceding ot lon, T do no dink i takes much mote than a general knowledge of| acli’s Blments and of standard defences (notably Arstotles) of the geometry in it to apprecate the fore of Sextus assault, But in the ns Sete rather leat both presupponing and attacking ident the provenance of which is wot immediately obvi. In this paper I wane to make some suggestions about the relation of what Sextus says Stoic and Epicurean philosophising. On the wholesuggestions will e eather general. shall point out where views expressed of attacked pale views found explicitly or impliciy elewhere, but [sll not make many specific aims about the relation of one text to another. In patil I shall have Tite ro say about the sources Tor Sextue' arguments, ether immediate or orginal, Te is gonctally agteed that Stoic and. Epicureanison are empiricist philosophies in 2 way i which Plaonism and Are torshanism are not. Seats, t0,telies on empiric dea, but 383" septic, he can ne the ideas of others without commiting hist forthem, Ava rule mathematics has not sorted well with emipinic= ssn. For mathematics pear 20 involve a knowledge of fet Which teanscends the limits of empincal knowledge in both certainty and accuracy. Since Hume empiricist have customaely tempted to explain this apparent ranscendence away by denying ‘hae mathematies has a factual content 3 well-known exception s 4J-S. Mil, ho dense tha the cereainy and accuracy of mathema~ tical knowlege are diferent sn kind from she certainty and accuracy of empirical knowledge. Both of these accounts are fbsiowsly attempts to ft recalcitrant facts into + philosophical theory. and neither i ulumatelysuecestl The very characteristics which make mathematic 2 problem for lhe empiist make ia paradigm of knowledge forthe rationals. He finds no dificuty inthe iden that there can be certain knowledge af a realty which is noc directly related to sensory experience. Mathematics may well be an important for Plato's postulation of ideal objec, Aristotle, of course, ejects Plato's ontology and insists shat mathematics tre ofthe sense Chings of tis world, not, he adds, que sensible but rather gut bodies or plane or lengths or indivibles, Although Aristotle's theory has a cetain empnicst cast ti it sno really compatle swith strict forms of empiri. For, according £0 him ur awateness of mathematical objects depends on an ability t0 “abstrac from the sensible properties of things, and mathematics 'Suruc ofthe sensible world only in very odd way our senses do not confirm that, fr example, a sphere ans sright edge toch by sharing single point? . Teg omer Se Ano Ansa} Bumesert feta, - Geometry and xepiiom 71 Ratinalse approaches t mathematics arc kly to concentrate ‘on the question of erth ad to deal ina ater schematic way with the problem of knowledge. Both Plato's and Aristotle's philo- Sophies of mathematics ae primaily attempts to specify the way imwbich mathematis i er. Empiticists, onthe other hand, ave Tikely to begin trom the problem of knowledge and to teat the problem of truth 3s 2 mere appendage. For example, on Mills [ccount mathematical wuth is no diferent from ordinary (emp faa) truth; and om standard theories which deny a factual con- tent co mathematics, the character of mathematical knowledge fzrantce ite truth: for example, mathematics may be sid to be tiue Because f the way terms are defined or conventions Id ov ‘Sextus focuses on the problem of mathemascal knowledge However, he does not focus so much om the question oF the reliability of our mathematical belief as on that of che cohercnce (oF basie mathematical concep. In this eespect his svestgaions tte continaous with dhose of Plato and Aristotle, who se the problem of mathematical truth as primanly the problem of providing objects for mathematics to be tue of! However, Sextus purpose isnot ro vindicate mathematics but to destroy (or eather to render asset toi philosophically imposible) by showing that its alleged objects are inconceivable. Once he hat Alone thst his ows satisfaction, he has no reason to rise Farther ‘gestions about mathemati ruth, Sextus procedure seems to me methodologically enobjection- able, but," When combined with the genenlly nominalst and ‘materialist tendencies of Hellenic plslosaphy, if contains the sods of a radical transformation of the PlutoiceAvistoeian conception of mathematics. For the chagge of inconceivailiy s fatually Faked with the view that mathematics in a human fnwention, 1 mental constuction. And the epy that mathematical objec are conceivable does nothing to undermine this view. Indeed, by itselt che reply reinforces the View by suggesting that the oniy issue is conesvabiity. There i, of course, no strict logical connexion hee, but I'do Believe that dhe notion that ssathemtis js. mental cretion eater than 2 dincovery of rthabout the world has its beginnings in Hellenistic empiricism and seepeiism.* ror purposes of discussion 1 shall divide Sextus’ arguments against the conccivabilty of certain geometric notions into wa Kinds which T shal ll dicot and rebtiona. The later mainly concern the relation of #=t-dimensional abject, point instants Si postions), lines, surfaces, co nedimensional ones. lines (times Sind notions), surfces, solids; in other words, they deal ith, problem about the contact between and composition af eons Direct arguments are most simply characterised negatively a5 che ‘non-relitionsl ones, Roughly they attack definitions ether onthe {gous that there i some incoherence the dbo tle, ¢g. [Stculary, or om the ground chat whats defined is inconceivable ‘The distntion betecen the to kinds of argument i by 20 means absolite, since Sextus sometimes invokes elation consid= rations ¢o argue thi 4 defined objet is inconceivable. 1 shall ‘onsider some of Sextus! purely dee anguments Fist and post pone considerition of mined cases untl I treat his relational eguments Te i aso convenient to distinguish berween wo kinds of| cometric notions attacked by. Sextus: the fundamental ones, Point, Ine, surface. solid; and the derived ones. straight line Ingle, cic, Sexe attacks dhe Iter i sections 94-107. In the ‘se of the cise he consdets slight but cra varan of Euc’s ‘etntion and points out that the defiins contains terms, 6 ‘point and lin which he has already “destroyed. He makes a shila point about a vasiant of Bucs definition of straight line (CA steph ine ain lying evenly (om) sith spares) and tbo stacks che obscure phrase ex ou, which he takes eeasomably ‘ough, to man in 4 strait ine” and ence o be unsuitable for ‘Sting straight line, Mader readers ae ikely to concede the ‘onrecess of Seta’ eiticisms of ex Haw, but thik of them 36 Seote ope themes - Geometry and ception insignificant. For fom 3 moder point af view Euci'satempeed ‘eintons of stsight line and ther simile concepts ate a waste of| time, since these terms are appropriate primitives for Eucid's Alevelopment of geometry. And even Fuck never uses the ‘etinions in bis proofs, However, the notion of a pramitive term seme foreign to the Elemmtr and 10 the Greek conesption oF tnathematics in general. The lange number of alematve defii- ‘pons of basic mathematical concept which survive from antiquity suggests 4 general awareness of the problematic character of EEucid’s definitions. Evidence also suggests that controversy bout such definitions predates Euclid, 50 that Sextus is snmp Continuing a tradition of what might be called foundational Stic albeit a eraction of litle ssentfe value fret oe point ot view The ater definitions of straight ine esidered by Sextus, that which eovolves uniformly with ts Tint’ or ‘that Which touches the pln with alls pact hen rote around ie mis (ep. Hero, Deft 18.2), ad the defnitons of ange he consider, the leas under the inclination of to srnghthines which are not pale onthe it ditance under the inclination? (op. the account scribed to Apollonius by Plutarch of Athens apd Proc, in Ee 135. 14-18), donot seom posible to trace! Het there sno een to think Sextus ie inaccurate of his eiicloms unjust, however ‘py the whole controversy about deintons nov sccm 0 Sextus has found » genuine soit spot in ancient geometry. and he probes i effectively. "The same kind of point can be made about Sextue diect critics of the definions of fandamenal geomeric concepts, Tut ere he rises shat ae perhaps pilonophically more smpor= tant sues The definitions considered by Sextus are the follow ing poine (tim): an unextende (adatom) selon Tine brendthles length or limit of suriceo Bax of point (oni mein)surface: limit of a boy ‘Body what as length, breadth, and depth Except for the definition of point and the lst definition of line, these definition are approximately Eucll's. Euclid wes the term ‘Znwion for point and deine ie somewhat disconcerting as that OF which pares nothing or of which noting isa pact ow meas ‘hen. Hehe idea ofthis deiniion i that points ate pads, then the definition considered by Sextus is perhaps am improvement, although the use of one woe for poin ih the definition af anaher is diticul to fathom. Sextus hire spears to use the to terme interchangeably, and his words in r00 (0 ba tus someon kat stg) imply tat the "genmeters ae both terms t refer to the Same thing,” Euli also icles in his definitions the remark hat the limits fa fine ae points, ad he defines a sarface a what his length and Breadth only. tn addition he uses the tera “sl (econ) shore Sextos speaks of body * ‘Sextus (28) sugaests that Eratosdhenes conceive the line a the flux of point. There are some interesting parallels Between the ‘other definitions considered by Sextus and statements seribed by Diogenes Laer to the Stoic Apollodorus A body whats extend in tse ways ia gh in rade, and in Aleph abo elled sod body. A aries 2 Ut of stay, of Sins fal rea only not dep Aline ht of = sorrento ht tol Aw it Here one notes the same use ofthe word body” instead of sod? and the characteration of sgn kindof som. Sets 00 Somiimes uss the phrase solid boy” (¥2, 84)? lis ute of the ‘word “unestended instead of las in the definition ofa pont may represent Epicurean influence, since in Epicurean theory 3 Test not» point, but an extended indivisible magnitade. On the ‘other hand, there would not seem tobe any serous Sto objection 1 ee ee Geamety an septiine 75 to the ase of the word ‘unextended’ in place of Apollodoras "ea Sextus’ cits of the notion of body ako suggest conne- sion with Stic and Epicurean ideas. Lsummarse the ctticisms tel 1 (83-8). Body i ther (something other than length, bead fn depth or (Gi) theit combination (ations), but) i Impossible becase body cannot be conceived separate rom the ‘hee dimensions and (a) imposible because the dimensions, being incorporel, cannot by combining produce body 2 (85-00) Fee coming together (modes) ofthe thee dimensions produces body, then either (i) the three dimensions are each ready corporeal or (i) corporalty results fom their comting fogether, buf () were the cise, the coming together of the Alimensions would not produce body. since the-dimensions would aeeady be bods! but i (3) holds, then either (i) the Intute ofthe dimensions remsins incoepotel afer che coming together or (ib) the dimensions become conporeak But (i) is incompatible wth the production of body: and (ub) isimpossi~ ble because (iba) if the dimensions change they mest be conporeil to begin with, since only bodies chang and (bb) 3¢ incorporeilsaelike bodies, which change without ceasing tobe adic, the dimensions wil not ceae to be incorporeal 5 (ot) Length, breadth, and depth have been shown tbe thing therefore, body, which x conceived through paeiipa= tion (sts courant dings will ot be “The first two ofthese arguments make use ofa radical distinc tion between the bodily ot corporeal and the ncorporeal. Re semis futur to astocate this distincion with Stoic materialism, but tually natural so question i relevance in the present context For ‘Sextus use ofthe distinction seems ro impor confusion beeen the geometric solid and a physial Body: the idea that» geomettic Sphere is constted by 4 certsin amount of extension in thice {Tensions seems substantially less objectionable than the ides76 tan murtiee that an apple # constituted this way Sextus’ ase of the phrase ‘soli and resistant body” (64 suggests chat he has in nnd, not his apparently mathematical definition whichis abo. Apollodor bot something like the definition he says (M #24) Epcaras used £6 distinguish body Eom words the dheee-imensional combined resitance (@ ich! dan met ata jowever, Tam not convinced that me ought to reject Sex criticisms as irlevant to even the Stoic conception of mathema-. tial Body. The Epicurean defstion ost quoted i ascribed tothe Stes in the reatse de Qualitas inopor sometimes assigned to Galen (xix 483 K). And i is posible that eh Stics made the presence or absence of resistance the bass for diingihing body Simpler trom smathemaiesl body, a8 Hero docs (Def. 3-17-19 74 ti-t6; ep. Hermanus oped Simpliis, Cat 134.338) ‘This possibilty gets ely weak contrmation from 4 emack by Plosinus in 2 discussion of che Stoic notion af body hey sy ha wha common ta body the heeimensona, they Speak mathemsocaly barf hey sty i th dco walk fisunce. they Me not speaking of singe thing. since sn ‘pay ordered fom as) ea. ehh. 0 (On the other hand, ome might read this ematk indicating that the Stoics had filed to disingush clearly between mathematic and physical body, perhaps considering the two definitions of| body a alternative characteisatons of dhe same thing, Sambutsky has turned this apparent tre ro make 1 distinction int 4 vrsoe by aging that he Stoics produced 'sphysicalisation of geomet by endowing geometrical gues with the elastic properties of material bodies? However, the evidence i inset to supe PNR Sea ari ato eso Pim eae sheds eon shal uty mye se Geometry and spticiom 79 port Sambursky’spictte of che Stoics as proto-dferntal geo- Intos cen very likely that the Seon nterwove mathemati~ sland physical considerations, just a Aristotle does, but sce no rounds to suppose that ether Adstotelians or Stoics had a conception of geometry any diferent from what We find in Euclid ‘Tam incline, then, to think that the Stoies did no dissing clearly between mathematical and physial body, bu doubt chat Soe obtide the Phonic eration di so. And inthe Timaces Pato himself eather bts the distinction, To this degree Sexes tgunsents aginst body may hit their mark. Bu thete i another Festare of the arguments which peshaps makes them inapprope- st. All hice ake for granted that Body nin some way compased Of the three dimensions. Euc's and Apollodoras™definions provide a formal justification for thi assumption, 4 justicaton thc peraps strengthened bythe arrangement and metnodol- ‘ogy ofthe Element. For Eacid develope plane geometey before “ereometry, ad he normally proves tercometsc propositions by reduction to plane ones. However, Lam inclined 10 think that Apolldonos ts more interested in ontology and epistemology than in forinal dfntion. The onde in which he characters the fundamental geometric objects suggest chat he is taking body as base and treating the others as mits of thee prdecertore. One ‘night then se the account of sy ay an informal characterisation tnd hat of the others at nication of the way in. shich fanental geometec concepts ae formed. In this ease dhe Stoic view of mathematieal concept formation would seem tobe quite like Aristo’, bur forthe Stoie the objects which ae charace= rise a mit exis omy in though (a pion). nn sin ni poem, 0 2-) dtya 2 i tn Sete inp tm a pa = 5h ay ek en opt et tesa eben oe ar ohh on ty“The most interesting of Sextus’ dec atacks on hese mits is his eaten ofthe ine (37-57) He begins wih a general account {tthe way in sthich concept (spunen) are conceived (pain And in general every concept x conse int priary ways ier Gyearnet expenence pnp or by tannerence (ma) om fle things ad this thee way by taney oma eeu toton af Socses may te formed fom kenya) by ‘omposion pine, «go ton of cau om coming ‘men and hort) by peoporionaey (oak © gue and Prams by analogy watordiarysed Greeks) 43) The sume account i ascribed to Bpicures by Diogenes Lact Fos ll our concepts (pin) are derived from perceptions by dist “xfevince of propotonaity or sina or cmpestion, wth enon ‘onerucing somthing Gamba wpm) (DLX 33) And the account which Diogenes ascribes to the Sti, ssng the same example as Sextus in the relevant css, i basicly an ‘extension ofthis account Some concepts (mune) are consived i) by dct expen, zany ami some by pointy ome by eens {peta eg etares wth efor onthe chet), one by compu onan some by contrat wane our ton of what ike sph dain terrence ong Some we coed {ocived marly and some af soncved by pivtion (ie) ft ‘eample, man wabout kane (DL vaso) leis toe noticed that whereas Sextus makes tab gens of concept formation with thece rudimentary empiric speoes, the Stoes treat i a a9 important epistemological link between the world of experience and things which are known by the mind slthowgh noe the object oF any possible experience. Th difference presumably reflects 4 nom-empirict element in Stoic epstemolo= iby, but Iam inclined to dose that ears om the guestion of mathematical knowledge. For the examples of things knovn by cabs, ek and place, suggest that metas involves quasi Seieniic inference to an explanatory concep, and sich inference docs not seem relevant tothe mathematical concept of line ™ Goomemy and xptiom 79 Sextus i convinced that we have no dine experience of Irendtsses, tha “whutver sb length we peeve, We wil aways peti with + cern amount of Brea (8) Others disagreed. According to Prot, We can get perception of [ie ie by looking at wha divides ‘nina cides om owen shad on ibe mom snd te fae For what hetwecn the resins uneatended rea, but Sine twee song the ge od shade tn eng rox, Be tcoueio) Modern scholars! have attributed this claim about the fine r0 Apollonigs becnuse he is mentioned By Procus afew ines carb. ‘The same caimn is made by Hero in his Definitions (16s). Tt would seem then thit some mathemabicans were aware of Scepicl attacks om their subject and actempred to respond (© them, itis perhaps worsh underlining tht tet esponse Seems x9 he aimed se establishing only the conceWabiliy of a kind of ssometric object and not the tat of geometry. For presumably either Apollonius nor Hero thinks of geometry a the theory of thondarice berwcen bright and dark surtaces, OF course, they could sre their account of he perceived character of these thundares ojos the claim that the ming could sso apprehend other lines “in mature’. As we shall sce shortly, there is some tvidenc that Apollonis and pethaps also Hero di proceed in hit wy. ‘Sextus ejects siiaity and proportionaiey as mem for con- ceiving lines essentially on the round that there is no basis for comparing the breadehess with what ha Breath (45-6, 48-30). He gives no real argument against composiion (4), Buti reasonably clear that nothing breathless en be composed out of| things with breadth, Ie would seem then that Sextus has provided a bass on which Epicurds or follower could deny che intellect respectability of geometry. The evidence suggets that Epicurs Aid deny is respectability." and it scems to me eeasomable to ‘uppore that he might have used arguments ofthe kind we Find in Sextus, Ihave, however, no ies whether Sextus Boreawed these arguments ffom the Epicureans or whether he and they derived then ffom 2 common source fs Be LE HatSextor does not mention the Stoic ideas of rearrangement ontarery, oF natural conception in his direct arguments against the conceivably of the Tine, but he dacs mention str Reaeangement and contrarity seem unlikely candidates for Plhining mathematical concept formation in any cis, And 2lthough on some philosopher of mathematic natural conception night bea likely one, dhete would sem to be no evidence that the Stoies invoked it for this parpose; moreover, if they had one, would have expected Sextus to bring his seepial armoury tober nthe poi. Starr bears an obviowsrsemblance to Antotlisn abstaction and docs seem a plausible way 10 explain how we ome to conceive breadthles eng, Sexton treatment (88-6) ot Js rather unsatsfctory. He says dhat it no mote possible to conceive length without breadth by srs than i is possible to conceive flesh without corruption (athe xara) o body without reste, In AF 8 407 he intodaces Further gonctl consierae ton, probably derived feom carly Academic discussions of the Forms,” according +9 which stile such as nonchorse and noneman do not exist in ety (hapa). Sextus goes om to consider an Aristotelian claim that we ca consider the length of 2 wallallby iself, but he eepics that, although we can conceive the length ofa particular wall without its particular Bradt, we ‘cannot conceit the length of any wall without any breadth ‘whatsoever (57-9). Interestingly enough, Apollonius appears to Taveadopted something ike Aristotle's position. For according Procius, the circle of Apollonius say that ‘Eos of sade wae be mented fr ed ot hen nde te lend but we cle the etemson in ne dimension, (Prk, Bue tooe-ioep Hero. Do 169-10) Sextus refers to another mode of concept Formation, epitais introduced by the goomeet' to account for out knowlege of the Tine “They sty that when we apprehend some lng wit certain amu of Ire we rece ths eth bye mare and moe iteming (at ln opin) naeonres, and ts we tat wh octet way by sat 3 Bread length, (2) Geometry and spticiom 81 Sextus constuesepiasis a6 a finite process: and although he sem to concede the possiblity of conceiving arbitral stall Breadth, he denies that tee cam ever make the wanston ftom length with bred to breadeles length, On Sexe reading, then, the ws of | siti iwolveea confusion beeen arbiter smal reads and Treadtlesness However, the words ei ka millon suggest that the geomncters may have had in mind a Limiting proces which breahlesner i conceived as the limit roward which smaller and smaller readthe converge. lt x unlikely that they could have ‘expressed such an idea precisely. and even if they had, Sextus ould invoke sic empiricism aginst them, "The sis of Sext critiiems ofthe concept of ines relatively lear He takes over certiin empiricist notions of eonceeability fn uses them ro argue that we cannot conceive anything breadth les. Sextus makes no similar lim in his attack on surfaces Gra}. presumably” because we do. seem to. have (Wor tlmensional images. He doss make sich a claim in rating the point (2-8), but there itis ssorated with the conception ofthe Points limit of line shall dherefore wea ie in connexion with fis relational arguments to which | now torn In general, what Iam elling relational considerations aise in suivant fms in connexion wth points nd ies, ith stants fin! time periods, with postions and motions, with lines and Surfaces, and with surface and solids although consideration of| Aierent cases sometimes helps to clarty ae obiseate diferent issues, I shall uae points and stsgh nes to give a prelinary ‘nsfication ofthese ise, bu treatment ofthe Greck text wil este consideration of othce pats. Trust that no dificates wl be in tansfering what is sid about points and lines to these other pars. The sues may be represented in terms of three protien 1 The limit problem, A straight line (sent) Limited by eo points. The eandard Greek notion seems tobe that these points be somehow included in the line, fe. et every stage ine ‘corresponds to what we cll a closed terval. The question Atises whether shes limiting points ae pare ofthe line, whore 1 pAb pent pcg Simp: No te pro ‘Sate ena peo se oho.boing» part would scent imply having size comparable with (Gearing a ratio to) that ofthe line 2 The composition problem. The problem of point as pats is intensified when supposed that line can beeut anywhere at Apoint, Le, that anyshere onthe line there isa point. I then sem reasonable o suppose that dhe lines composed of pits, However the standard Grek notion of composition appears 0 he juntapostion of concitenstion, and elementary geomeene , tmuths ae incompatible with the tea tha lines are composed of points lying next to one another, For example, Excl’ first postulate allows one to draw a straight line between any 60 Points and his tenth proposition allows one to bse the ine 0 Produce a point between the original ewe, 3 The contact problem. This problem arises when one considers 2 seright ine AB cut into two segonents AC, CBr imagines vo Scgmenss AC, DB joined to form a single staighe line AB, Deca ofthe conception ofthe line ay including ts ining points, the cutting of AB would scem to involve the division of 2 ingle point ino tw, and the joining of two seytens the Ineldng of two points into one ln my judgement it san open question whether Greek mathe- smatians ever managed to deal th these rhional problems in ny but an evasive way.” We nove know that the ati of ie, being poi, bens no rato to then, that ee contact problem is solvable by allowing fines correponding to open inervals, and that the composition problem can be solved by denying that ‘omposton #necesany justapostion. And we know tha hese Solutions ae compatible with the geometry in Euch’s Elen But che only sense in which there are any solutions in che Hlomens ‘hemelvess the sens indicated inthe description of the composi= ton problem: propositions are asserted which rule out a cern ser Clary such solutions are of ile value if the composition problem is taken to case doubs onthe propositions themselves. Inthe case of the limit and contac probleme there i nothing tht ‘ould call n answer inthe Element Encl’ techniques seem £0 1 Fars don of is qt we Ha. Wa Geometry and sepicion 8 ‘reli even rising these questions or similar ones about adding 2 poine 0 or subracting fom sine of about generating a point ‘by subtracting one line from another or about hove a ine pases through a point. In this eannexion one might think of modern Formalied theories in which only certain lations ate expressible A plilosopher might very well hve questions about the objects oF ‘the sandard interpretation ofthe theory. questions which cannot fev be raised within he theory and therefore are no birance to its devclopment. Some of these questions may Well be of mo mathemadcal inerest whatsoever, eg. whether sets are real ‘objects or artifical constrictions, but other need not be, 6 ‘whether dhe notion ofallsets af mtegers make sense Tam inlined to think that the relational problems {have introduced flint the Second cass. They are resonable questions task abot points En lines, even though the mathematician doesnot need vo aneer ‘then anda fara case inthe cise of Greck mathematicians, cannot answer them). In this sense Sextus relational seeps thout geometry can be said to be entirely relevant But exsily cradled Tefore considering this scepticism I want to indicate briefly what [ake to be Aristotle’ solution these relational problems tn to sggest hat the Stoisesentially took these solutions over ‘Avstore avoids the composition problem by making the fine prior co the point, a viw implicit in Apoldorus” definitions Points exist at Kimits of ines, but these mts are noe parts i the relevant sense af the word, Points can be Brought into existence by making cuts in the ine, but geometey shows chat one cannot rake two consecutive cuts in the lie to generate succeasive prints Geometry also shows thatthe straight line is ininstely Evin, so that no sense can be made ofthe ide of aividnga fine Simo indivisible segments or ino lis points. Arstouke's solution tothe contact problem appears to sole maintaining that whats le point when the segments are together becomes to points Wen they are separated, although the pois not divisible This Solution is perhaps the weakest lik in Arstock’serextment oF reitonsl problems, Bur i sho slko be clear ea his treatment84 aN vee presupposes he rth oF geometry: and the ides that pons only eae when they ate marked out in sone way although ingenious, ‘Shandy likely to seat rex he mind ofa person troubled by the ‘composition problem ‘Our major source of information forthe Stoic approach to relational problems i Pltarch’s wtih ‘efation (Comm. a to78e-1081¢) of Chrysippus and the Stoic” denial of stor least for parse or indivisible pars ‘ivsblty of magnitudes which alsa plays 2 fundamental role in Aristo's chery. In denying the existence of such parts, Chry= Sipps apparenly drew the tii consequence that Bodies could not be composed of either finitely oF inintely many of them. Pare Esters on this consequence by pointing tothe absence of 4 third altemative between finite and infinite multiplicity. This Objection i clearly 4 red herring.2 since Chrysippus is only denying the existence of Ist pats, not ascebing 4 curious tultipiity 10 them, Ptarch also assigns to the Stoics a version of an argument against contact between parless things, an argoment hich is already found in Aristotle's Pps (21, 231b2-8 ep. LE gptane bg) and is also ased By Sextus (35-6; ep. AP nx 5%64): such things mast be in contac either 3 part with pare or 38 whole with whole; but the former i impossible for parts things. and che laer represents lending (las), not contact. The Stoic view oF contac according Pltarch, tha things touch at nt (ht eras), nota very informative phase, but presumably one which -— i the mame of the infinite Geometry and rpticom 8 Snes the denial ha limit ofs hing is part ofc Pluarch's critica disussion ofthis view of contacts not entirely cea, 50 | ‘ure i in Fl a init no body.) Therefore,» boy il eauch 3 Hody by a8 incorporel [asus and age owl or eauch since there a incorporelnberwen. (8 Bats weet touch s body we act and Sct on by av ncorporel For bodes marly touch and 3 09 aad treated omby dante, (6) Anda body were to have contac y 30 iMcorporc, sould. have comesion. snd blending and growing Together Then se neconry tha thee connesons sd conaets he limes of the odes exer fel) pert (a) do hot pes but we tdsroyed teach of then incompuoie wit hr cneepion For {Gone} they themes do oe ame tha thee geberaton Sd Listeactn of mcospre But nec) there anno be Being and ‘inion f Bade whith old ono thesis. Fo ait Stns and frseve she mau ots body Barf bs ne not ate of acing Fars bee ports but the fise together wha re ended ap wholes ith whats they sy, ivnecsey to admis eseration of ten ‘ended norton of ham ben hinge are scart. But ho onc oie tse dhgs cay (Comm, nr ttoe-totrs latch goss on ‘ovdcvdop sbjccton ban tclmw of, hae sphres ling mone pines) Pluarc’s discussion here presupposes the Stoic distinction be- ‘ween the justaponition (putas) and blending (ris) of physi ‘al substances, so thie iis impossible eo exeae seth certainty ny definite Stoic description of geometric touching. However, the phrase kate peas suggests that touching bodies share a it this in Plaats words, thet enim and blending and roving together. Patarch argues that this notion commits the Stoes to the Aistoelan ies of destraction andl testion ofits, alebough ‘they are inno postion to admit such possiblity. Plutarch may be Fightin atempeing to cont the Stoiest0 the unsstsictory Arstotlian position on the contact problem, but his (ery ‘opaque discussion of Cheysippus reply tothe Democrtean cone problem suggest a leas the possiblity of core Stoic WaY ott The Democrtan problem concerts the cuting of cone By 4 plane pralel ss base. The ut appears to proce se0 suis 2s Aga egy a po sn a sin iinDemocricus argued that f the two surices ae unequal, the cone will have indeneations, and if they ate eg, che orginal gore wvll be cylinder not a cone, According to Pltarch, Chrysippus responded by saying that the surfices are nether eal nor ‘nego nthe light of Plutarch’ earer treatment of Chrysippusie iD oem i aa 7 Fone since des machines 101 description des qutee termes du mouvement, puis Texpost des ‘les de proportonnalte acc les differs cust les exceptions ‘avinvoque Arsot ‘Le moteur, Je mG, Vespace, le temps: Cest entre ces quatre fermes que tout se joe. La, proposiion inkile du chapitee nonce sos Forme sbi Le mom too gle hse tds gl hve tinge ajc chose Aristote précise ensuite le sens des prépositions "ans" et us- {qw": Fetroinéme terme est le temps "dans eguel” Tie Te Inouvement, ele quatieme est Is longueur determing (poo 1 inch) que parcourt le mouvement. Les deux premiers termes ormentun couple, qui eget successvement d'ateesappliations an cours du chapitre. On + dabord moteur. (Aim. nowmenon: pus kin est emplac par dais free) deux termes cant employes dans un sens tes Voisin, pulsgQARsto%e paste de "to Ato Kou" 3 "he it dais oph ow A. Enis Te Couple moteur-md exe designe pa es mots iets (his prt hrs, 2500). Tes précisions.apportées 3 Ténoneé initial s'appliquene au rmosivement loc: “usgu' sgafie une longueut, par exemple [A mecut B d'une longveur C, ete. Cependant ln proposition _ginrale ot abstesiteénonede a debue deci aussi bien le schéma ‘Pun mouvement d'aleation ow d'accronvemen, comme le p= ‘Sse Aristotelian capt, Les ute termes sont lor ce qs sccroit ce gui est accra, en quel temps et de quelle quansté (exon, “ianomonon,en posi shoni, pon, cu ce qui alte, ce gu est alr, de quelle quate selon le plus ct le min, em gol temps (alos, alioumenon, por katt 10 mall ksi ton, em post ‘Irons Le mat dane semble appliqué ici uniquement au 8d mouvement local, cependne Ariote prise ailleurs (Phys vat to, 266126) gion parle aust de force (amt) pout un écautle- ‘mnt ou un adouctsement, comme pour un ancement ‘Ges analyses de Dhys Vrs peuvent eae eésumées dans le Tableau ‘Ce chaptre du live vi west pas le seal endroit of Aristote ssn de dgager les prinopaus “paramétes™ dy mouvement. 1 ‘Se méme Gonnant de trouver dex passages ater nombreux sat {ette question, et de constater sn acon assez net enue ces textes.“Tableau x Les quate termes dx mouvement selon Py vn § confement Vanlyse présente en vt 5. Le Tableau donnera tine ee synopeique de ces dlfrents textes, en indiquane quls ‘ocable correspondent respectiverent su quate termes énoncés en Phys os. “Tableau: Comparaison de Pips vn 5 avec d'autres textes wu - tage rn ec as a" Roe =o Le Tableau u est certainement loin d'éue exhaust ai norm iment issé de ce laste de termes ndiguée en Phys 298413. parce que Ia signification da passage ne nese pos palitement late et semble saccorder aster mal avec les ares textes, On ‘emarguetacertsines variants In prépostion “dae” peut dig ner aus bien Te eomps que Tinervalle de variation le suet de Is phrase peut ue le corps mi erp ee Fespae peuvent figures soit séparment, sot combings dans la rpidté dumowvement. I Fore et sien der madhines 105 Yy des places vides dane plusieurs rangées, en de positions Uilrencs, ten alt toutes les combinaisons penis ic ne comportent que trois termes. L'aalyse de Phys. st § es done la plus deailée ec la plus compte: cependane on surat pu la Feconsttuer en confrntant routes les aus, L'analyse da mouve- iene en Py its pparat ainsi comme le point culminan des flor repetes ec diversies UAvistote pour fepérer les varbles ‘sseniles dy mouvement Dans notre chapte, les quatre termes seront nsuite insénés dans des relations de proportionnalité Il en est pas toujours de tdme dans les textes parle. En Phys, vat toe de Calo 3 a distinction des termes permet un rasonnement de comparaison fete pls grand ot pl pest. Ces deus textes soa done assez proches de Phys. vu. Mais les tess autes textes cits dans Te “Tableau sone bien dffrets, En Phys is agit de savoir de combi de manigres un mouvement ex divisible. En Phy v4 et ‘ula question ex elle de Tunité ds mouvement oul titel mouvement eat veiuablement un ct contin, on doit s¢ ddemanderd'abord selon queles sores de “dimensions” le mouse ‘ment peut tre dun ou mulple. Aritote indique dane ce cis trois "dimensions" pour sivorr i le mouvement ext us, ot vériers quil eat un quant au mobile, quant 3 Tistervalle p (Mm! 3.t3) Les deux demites scion (stone) —> (Mmns/ata) (hme) — (Bt/aimi338) ‘ow selon la deusitme lecture qui ven fit qu'une rg unig: (Mont) & (Mmalasl3) > (Ml...) Air ela, on est tnté de eanscie les telations sous forme de proportionnlite seule. Les propries de a relation (,«, } sont atte pr tne proportiomnaite Art On retrouve ce gus a Ge affresh Mise aloes Momzsaer et Mamfzestl2 ot de tnéme Abmsfzts ee Miaim/zs Cependant se tlle Tormels= tho ex top prise et top forte. D’abord ell nent pas compe des restctons qu'apporte Anstote. En second let, fs auteurs Smterieurs aw Xe necle n'gerivent jamais de semblales pro- portionnaltes, parce qu'lles mettent en apport des termes non homowens. Fone since es mathines 107 Aussi Aritote Siti! appari, non pas une proportonnaleé lobule qu le les quate tees, mais par exemple celle qui combine, dans Te denier cs, des vigucnrs ce des poids la Geusitme vigueur est Ts premitre comme le deusieme poss 38 premier (ie Mi2:M-m2m) en grec homotes dé cus ba analog Ie ics presto bara (2508). Le texte paral dv de Cao 2 procéde dela méme maniee. Avstte énonce @sbord une rele pele (Cece plo pst ph iger et mi davanage parla mm puisinee Gob m i prc onsite Leap’ da pls pe ser elle da pls grand comme ke ps grand ‘cups eta pos poe Gerben) (On patains une elation gui eos termes, omen ixe wn ei pissnce qui reste la mine), ctl dot y avoir proportionnalie Fovsgue Ton fat varie les de termes non fx: item: vitesse alle: ale La relation entre le moteur, le mobile, Pespace et le tmp nest donc pas exactement ne. proportionnalté, mas elle permet ‘Pobeeircertaines proportionnalits, Ta suite du texte apporte une resection importante ce gsi vient te énonee: dane certains ca a proportonnalté ne joe plus, Siune puissance donnée peut mouvoi un corps, ise peut {uune puissance plus petit (ob phis petite pae rapport au corps) te produie ps Tee gu corespondrat- Dans Te jargon ge Jntroduit ps hau, cete thse se tradult pa (fons) > (Atlas (hong) (tam La premire justification ext ordre experimental. Ariscote pron exemple des haleurs qui treat an batea Un seul des hilours sepuseritnutlement, si espéeait fire patcourt a batea la feaeion de parcours qui correspond 4 Ls puisance un esl homme. La proportionnalit n'est donc pas satis ‘Cex exceptions ont aus une porte théorique esentille, Ane. tote répond Sun argument de Zénon, qu'on pourrait consiérer femme le prototype des sorts auront tant de fave ans es08 FRANCOIS DE caNDT sites suvants.11 Un boise de mil it brit en rombant, tai i se pet gum grain hole qu tombe ne Fase stcan bra parce quil ne eeusia pas 3 mettre en mouvement, méme ts EEntcment, Pair gue le Boiss ener avait rant,” La soltion résde dar ls notion tre en puissance: une force peut nexistr q’en puissance tant quelle ne sst pas fondue dans tine orc ples grande gui englobe et active. Le mot Susan, dans ce texte, digne done 2 la fos le moteur ("pussance™ rotrce) et wn certain mode de rea (tre "en puesance). La reponse d'Arisiote 3 Zenon nous oblige 1 sisi cere double shianis comme une notion unigue Il fat envisage la vruaite fomme ue force: Tete en pussance est tnda vees Tate, et pradit de ce fi un mouvement ft, dans Taste sens, voit on [i force une seta et admettre que [a force eae ne pas paser ‘ute etite dans son eft ql yun seul force, ou pot tun sui au apport farcelcorps 3 mouvoin seal en deh duguel a force reste porement vir Le lire vit de Ia Physique sate rusquement apres énoncé de ce igs de proportionate Nous nesavons dane pis qulles 1 Yeoman ay CNG 0 9 pen Ze, Protger opera: Prtguen eis aan Sle selegatcha Sans fase Sell el ds ees cell ns ese Circ es es gt a a a sa "+ Gileenens Tao anne Br mee: He SACude: spor Saber owe ea at igen a ge pa ae dn cr pr Ee ‘Ss ue cnc et ono de fos detauvement Av cng €oppats ‘Siac pe mune he pans Spouse le See es pn ey Fone et scence des machines 109 fins elles ont été formelées. Cependant il existe plusieurs autes psages of Arstote présente et utlise de serblablsrepes, dans tke argumentations relatives a8 mouvement. Par exemple, en de Gil 9 2b) Arte pose I pie sean pur un Un moccur da, apisnt on un temps Spal, modi queue chose ‘perm mt pt ne lege it gus Feige a popordonnie sed il ves maser los grand pur eapport au pls pot Ce principe ex rts proche des eéples que nous avons examines, iat et plus fble ow plus vag. Il permet ict 3 Aritote de jeter Thypothise un comps ini ae const Pf ne ses mi par aca moteur Heo ae Tempe inne eee nest dane aac rapport (re) De me, dans le passage da de Cielo qu a invogué pasiurs fons, Targumentation et dinigée contre ide d'un corps "sans poids, Exane donne que “ce gu est pls ptt et pls lege est mi Svante pa la meme poisance en rslterae que ce quest “Ss poids parcourrait une dintance gus depsse route distance” (de Galo ut 2.375210), Lede Mo Animalium conten in risonne= ‘ene mois explicit, mais analogue, qu s'ppligue 3a sevation ‘dels tere en repos au cntee du monde ‘Si guelg’an acest urate le repos de I tere ps pane da trounce eg reunies pa ce mouvement Teer da Ghe,Or lew ear quel vig, dnt see res puissance est fs nd ate elt mest pb fie done ho ps son po {a4 sabia) Par conséquen il est possible, au moins en principe, quest un ‘movement capable de déplacr ls terre. Comme dans ls textes prccésents, la proporionnalit dw moteur et dam est ublisge de Inanire indéerminge, vague &t non quantitative, C'est simple= tent une affirmation de principe qut assure quil aura une ‘ertune proporionnaité, du moins rant qu'on reste en présence de termes fii| | Le méme rasonnement est employé pour sft la possibile «tun moueemene dans le vide (Phys. 8). On notera sine variate import it de dv dopperen ete os end et pe tun acre corps mas le miley Ii-mtme, Pespacetraversé qui ft ‘obstacle au mouvement, Le Tablea mm met ce talsonnement en paralleavoc les quatre termes de Phys. i Tablean mz Comparaison de Phys vt 5 t Phys. v6 § Fact cope dgiemen mir sah op logit Le moteur et cet fis Ie corps en mowwement, Ie mide le tie, c Is mémes proportional doen continue jouer: fle corps en mouvement cts Tongucur pucourue rest le. tities, ktm sea proportional Ts esance done ut {en clogion tow ompeisonts soma, 24963) Ce. pine nonct wie de Fappiationssivante das le vie te pro Dortomnalés ne powvee subs, "carl vide ns acum oppor a leucl lexctde Te comp, toxt comme le tien vier du hombre (215613). Une nouvelle fo, Avstte nels prox proionalits ds mouvement dane le veal but caer es ese Fits. de bonnes ents demewurcs, sae apport Danse meme pasge, Arte exrime le prince geal qu le guide dan des ca de ce sorte Poor ean on ot dena cane de egy pase ot Fe be mon ps dnp ot ey Gaeagh PPO LL proportional ext done, avant tout, une proportion de principe, On ne fait aucune asion 3 des mesores fective, 8 des tigations de grandeur, 4 des rapport précsément dternings ‘ui pourtaen, Son es situations parties, tempi Ia place A termes On a contente dalioncr qu'il doit y avo pom portional et donc rapport cls excl par conséquent es et 08 tine proportion serie impossible, pare que un des termes te pourrait enter en rapport vee Pasi. Gace exgence que les teres en presence sont ins x tout & ‘it fondamentale dans la penée d-Arstor, ct cle et lige 3 la notion pus gence d'un ordreet dune nature. Ces Is posse one since bs machines 138 imme d'une nature gui exten caus, ainsi qu la posible de la Counassance. Par exemple, a est absurde de parler d'un repos ‘Fume. durée infine, auguel soectderait sane justification un mouvement, ct cette absrdité ext x consequence d'une thise séncrle relative 3 Torde ea fnitade dans a nature Th ny a ren de dévondonné dans lx choses ui sont por mane ct over 1 Isrts ct sgt cx on toutes cae ordre Ot Fini aac rapport vested fe tow ordre et rppoet (Pigs vant asta) Parce qui ya une natute, et done wn certain onde, le démesur€ ou le sansrappore sont exch Les fondements dela méconigne atiqne (1): Arcee Ho cst grand temps aborder le deusiéme versnt de notre ques tion: ete analyse dirmouvement-t-le inspire développement tela science dex machines? “Yat djs cite au debut Tétonnane eémoignage de Simplicius 3 propos de Phys vn § ist on vt de cote proporonmalnt ene e maeu, le mobile tle ‘Sei porous oy Alchomee ucomponeFinstrament dein 3 pest Spl sharin Phys (CIAG x) 130) Ge document ne nous est malheurewscment pas dun grand secoury bord case des date ts tardive (ve 5002p. J.C) En second lie, om dseutecoujours sur la nature de Tinstrament vogue Te demice auteur! qu ait abordé le probleme conse le mot chariston comme Ia transcription dun vocable armeno-persn Grotn, qui désignerat quelque chose comme le chadouf du Moyen-Oniet . et pemonne ne croita quArchiméde at invent le bom veux proctde ds chadouf D'autre parton peut se Alemander st Simplicis évoqve wa derit théorique e'Archimede, ‘ou simplement donne sa propre explication 8 propos um insrue tment cosérit par Achim, Enfn n'est pas imposible que Ke i dQ Tae Qe ph ‘Fase Chuan eae ese PekSimplicius invogue ici un ouvrage pend dArchimdde (par exem= Pele pert Zugon mentionné par Pappu). ‘Dan les ouvrages 'Architde qe nous possédons il n'y a en ‘ous eas rien qui ressmble un sulonnetneatinspisédesreles Tarte. Les propeats des levies ct des centres de ravite sot derives d'un ensemble de postales od le mouvement nitersent nullement. Le aisonnement pose sue a combinaison du poids et de a dstance au point &sppus [Now demandons que des poids Ggiux squllient 4 de distances lsc que de pos gn ee dence gales ne ub ps {cows demand se) si es poids son on quire 2 cranes ‘Seances et que on jute 4 Tan ds dene poet pnd me Sea ‘ene pas maw iy nt inclion dete dx pide sequel on 2 tou (edi. Eguie Pns Poses tee En commengant de cette manitre, Archimde prend simplement comnme point de depart la_propriété fondamentale di lever. Dulin deerit la démarche d'Archimede en ce domaine comme tune "méthode de démonsrxton”, mas non une "méthode {Tinvention™ (Origins 112) les pencipes du raisoanemeat sont “ogee Ta surface de phenoménes, et non pas drains ds fond nme des choses” (i). Par opposition 3 este archiméden. une ematche inspinée @ Aviso devrat eemonter em dee des pro- prictcs du lever, et chercher ly cause de ces phnoméncs din es Felains ent a force moteicee les autres paramres du mouve- sent. Cee méthode despliation aunt Pallets Tsvantage de assembler plusiurs sortes de phénomnes sous des prinipes commune Ie ois des comps flotant, peut-etee meme celles des projectiles, poureaent dependze des mémes hypothises fou thmentales gui repiesent les levier, le pouies ct les plans intings'© AU contre la théorie ’Archimde nest aullement tine enguéte sr les causes des phénoménes mécaniqucs, C'est smame d peine dela physique, platde une géomric des pois, tssce prothe de ce qui deviende In géometre barycentriqhe: on sssociea chaque point de apace un nombre post que Tam pest appeler “marsc"s ou de rout autre nom.!” Les rages Aristote “ometes en Phys i § non rn 1 ies Cte sorte de tori Sits eae ier came Sean neo oreo since des machines 003, Louvrage de Hérond’Alexandsie, Les Mécanique, qui ne nous cst parvens qu’ version arabe, est asezdifétent pat som style des Egulibes Pons eArchiméde. Héson analyse diverses site ‘ons mécanigues concrete, et cherche a cause des phenome cn termes dymamigques. Il ouve cete cause dans les propriets des fercles concentric, rédit es autres machines simples! acs des cree: "Le levier qu met es corps graves ope parle meme ase qui agit dans ler dene cercles" (Les Mc. 9,p. 19) Talnceseramne aussi au cele." (i. p. 1H}, le trea est pare coe ue dee conceit. 10 ‘Mais sion examine de plas prs les lignes oi il expose le Principe inital de toute son argumentation, om wapercot quien Telit le ceteles concentriques ne sont pa le veriable point de ste sip ne yo un» i 7 esse sie ee ep la 74 See Ply. Optio mau aor as ME, a Observational enor in later Grech scene x68 used, to produce stating or apparently paradoxical effects. These Ae, indeed, mentioned by Heto in the opening chapter of his Cayce a6 one of te sims of his inquiry, and sn ch 18, for inatance, he shows how to place + mitror so that anyone Approaching isces neither his own image nor that of another bat ‘nly whatever image one selects But the een pont i ht, hostever startling chey may sem co he Bystander or vic, such cfccts depend, for thei production, on ule onthe part of the individuals who set them up.”” They had to have em ‘enough grasp ofthe principles of optics and know what they wete ‘doing. Admtedly the point a issue in Sextus t PH vag that dolmals with ferent shaped eyes may se things diferent, and in PH tig that postion may make a uiferencet how an object ppeats. AS in the weight example, the ater i certainly point thitmay be agrcd, except hat the sient wilt goon to dea ny radical scepica conclasions abou the suspension of jodge= ment. On the contrary, the firmer his understanding ofthe las of Felli and sefraction, the es inclined the student of optics il be to draw sceprical moras, the more confident he will be oncemning the mvestigabity of the phenomena, Thos in the ‘second book of the Opts Prolemy goes through a long list of ilsory phenomens and is able to give stsfictory accounts of many of them sofa front concluding from wc cass tha sight het whole i deweptive, he insists onthe contrast between the ‘xceptional and the normal es “The types of problems concerning peceprion that trouble dhe scientirelt eset sight or hearings such and in general than to particular obstacles to observation that ase an_particlar ‘orcumstances. General points, about the need ofthe observer 0 be shld, and careful, are made, bu they are co be understood within the framework oF the investigation i guestion. The tstronome, for instance, a8 we hae seen, may be concerned with the datrsing effect of eran atmospheric conditions: he should be aeare of possible lt in the construction and positioning of| inatsuments of errors that may arisen the tming of observations, 7 re iy. iw of orga, fr te, Pad, Bc os Shame lOin estimating wie angular distances.” and s0 on. Certainly the move to simplify and ideals che problems invesigated is com ‘mon in’ Gresk, 36 in all, cine, and this move often, even hormully, involves discarding some of the pereptble phenomena a irrelevant to the inquiry. There are already certain such “Mealiations in Anstore, a for example in his beng prepared ‘reat of point sources of igh in hs discussion of haloes 2nd che rainbow The study ofthe lever involves discounting the eles of friction (a6 te call #1 and the principle of Ube absence of ‘laecty of sencapuated in Porat 1 of Archimedes’ On Flowing Body. The astmption tat the earth is as point in relation tothe fixed stars Is common in astronomy’ for the purposes of his discussion of the longitudinal movernents af the Planets Polen explicly leaves of account thir deviations in latitude: ane many other examples could be given. But ie ane thing to discount some ofthe perepeble daa as irrelevant eis ‘quite anther 20 discount the all as unzclisble To be sure, the Timer move is made in certsin contexts within one stand of 3 Phitonising tradition. For Plato himseif it was thought, astro= nom should be reduced to a branch of geometry acoustics to smumbertheory. But Arstoxenus is clear dhe such 4 reduction is ie Triage ite ee Sencar FO (Obervtonl erin later Gree sence 165 foreign to acoustic: ‘some of those (predecessors) introduced extraneous ressoning, ad ejecting perception as inexact fabri ‘ated ration princpis, 2 theory utely extraneous © the Subject and. quite at vartnce with the phenomena! Again Slthough Prosi, many cntutes er, began his Ouiner wih 2 phous reference to Pat's recommendation that the tre philo- oper should bi the senses good-bye and concern himsel with the study of slowoners itself and "sped self "n tae umber the ‘whole of the rest of is work deals with the investigation of che Pile moving objects inthe sky. ‘ealacons and simplifications there may be, indeed must be. Burt the key question, whether or not it was made exphc, was, rather, which perceptible phenomena to dscard. Those that femained, that provided the explnanda of the Science in ques- than, had co be as comprehensive and rable as possible, and ‘comsierable efforts rere expended, at est eran quarters, £9 achieving that end "The tnustworthines of perception in general was not the main fsaue in the exact othe natural sciences: the problems elated. father. to pater context and tothe particu circumstances oF fv inguiy. The writer we have been studying offen show quite 2 eon sense of sthere, sm their particular subject, the practical probleme of securing reliable dat-base ly, They exercse, on ‘ccison, some ingen n meeting or geting round thedlficu- Ses, and they are, at list somtimes, carefel to emphasise the pproximate nature of approwimate rests. The linitations of sch methodological discussions as they engage in are, however, fpparent, While Rirst-orde recognition of particular problems and ‘ficulties x common enough, second-order analysis ofthe sues ‘connected with eror and approximation procedures sare, thas fnotably of sich questions as she ieterpoation and extrapolation techniques to be usd, and ofthe methods to be applied to extract rests from sts of discrepant data ~ wile there litle or no use til ofcourse, oF caster of concept ssoiated with probabil ‘Sonh G2 hy) sewer es fe gh oFlity in che tarsi sense. let alone a theory ofezor ofthe oem familia since Gauss. On thee and similar questions, not mich awareness, let alone sophistication, shown by practising Grek Scientists In this respect, they could have done with being beter epistemologints and philosophers of cence. Yee snot ae they ould have leart much on many of these topics from the wsual {pstemologial discussions in contemporary pilosophical i= Ings: fori the pointe about the problems connected with errand probability were not made inseience, they cannot besaid to have” bon pat ofthe philosophies! debate either ay ise aa 6 Astrology: arguments pro and contra ALALONG ‘One of the special sciences which Sextus Empires attacked was astrology. From a modern viewpoint it may Seem surprising that Ihe regarded astrology 3+ worthy of serious reitation alongside such subject a geammar, mathematics, and music. Buti ater Snguity, though there wa consideriBl opposition 0 astrology by philosophers, its defendants could ince scents of the fst rank, most notably the astronomer Ptolemy. Arguments for and Seainst the min principles of the subject have a continuous history from the second century .c. up to the time of St ‘Augustine. Orso Neugebauer, the grate ving authority on cont astronomy, ist gaod guide to the ight historical perspee- sacl development [i horoscope astrology] mut ke conser 6 2 importa compongnt of Hellas soence To Grok pit phe ind wstronomere the univene was a wel defined sructre of {ity tlacd Posies, The concept of profile sara Berwcen these bess in principe ot a all" diferent om any. moder Inschane tony 2 Compute ik the background of religion, trap anys, the finamental doctrines of tology ae pare Forsunately the vat of astrology isnot my deme. I sal be concetned rather with three stages of 4 complex debate, which began with Stics and Academics and cominsed ito Neopaton- jam and. Christianity, Following some histonca preliminaries, Pare Il will be "the controversy according to Cicero’, which ‘stables the chief foundations ofter argument. In Paris and1V Lill consider Promy's defences of astrology, and his sucess in meting Seepricleriticn as represented by Cicero, Favorns, and Sextus Empirics. glancing for comparison at Manus. Finally, in Part V, some later developments wil be sketched: the positive aguments of Firmicus Mateus, she qualified ertisms DF Plotnus and the atacks af Augustine” 1, Historical preliminaries ‘The experts seem to age that horoscopic astology or genethlal- ‘oy, based upon the positon of planes inthe zodiacal signs, was ‘rtally unknown ot noglected sn Greece before the tirdcearary bec! Lite can be infeed from a diapproving comment about ‘Chaldaean predictions, aterbuted to Eadoxus by Cicero (Div. 7) and a Satement on thee “most amazing science (esi) ~ predicting indiviials hives (ag sell 3s weather) ~ ascribed Theophrastus by Prosts (i Tim. 2851). The calc known horoscope i Babylonian, of 410 bc but dhe ease Greek horoscope isa rele im Northern Sys signifying an event oF July fe nec the apotheosis (or coronation by Pompey) of Aniocus T fof Commagene.® A firmer datum for Greek famibarty with Babylonian asttlogy isthe history of Babylon waite i Grecia ‘Soper "a TSeog de meae rma coe psp ep ata moon ert ef hn hp a oy tip segs me ey nd Cand Aone Ears cieee be eceinee fe hedetge ioeg tir pey Seem e jn Fue ad. Der Wr ting, Pl -h R3, Asoegy: armen pro and cnn 167 about the year 280 by the Babylonian eros. Berosus sete on the islnd of Cos, and his work ceraialy gave an account of Babylonian astrology which stimulated farther reseaech In che Naural Questions 2.1 (cf. Stoie Heraius, Hom. ally 53) Seneca reports conexions posite by Bross between une Neral conflgeations and dlages andthe movement of the planes. “Earthly things wil urn when al th planets which nose ove in dferent orbits come together the sign of Canc, and ae So distributed inthe same path that a staigh line cn pase ‘rough all their spheres” In the Stoic theory of eka the “world conflagration’ occurs when the planets rears to the sme position they occupied atthe begining of cosmogony (Nemesis, SVE 1 625). The theories of Heros are sulicenly similar € fggest that he or doctrines lke his may have Been taken up tatusastclly in the early Sto eros’ dates are attractively close to Zeno later te in Athens. According to Cicero, Zeno sid that he seers god that 2 eational piniple” (ts) which pervades all are i fuenshed ‘wth divine powers and tat this power belongs to the stars, dt the yeas, months, and seasons (ND «50. Alli cogether with the general principles of Stocim, might incline ws to regard Zeno as strongly sympathetic to Babylonian astrology. But caution "The Stoic concept of universal “sympathy” was to become the firs axiom of philosophical astrology and is constantly stated by Manilus (et 247-s4. 1 60-81). There is every reason 20 think that Zeno himself esabished chs principle of Stoke cosmology, $0 Fundamental to de causal sheory and ethical doctrine. No dob, 100, he and his successors eagerly drew upon all valle support for natural signs which reveal the ordered naare of the common and provide material for Hving accordingly. In Stoieismn the cewstence ofthe gods requires the validity of divination,” Bot no texte actually associate Zeno or Cleanthes with abolegil Aviation’ and the only evidence which seems 10 give Chrys 6 Yee ae pe are Racin ‘ett amps mong swe gypus’ bei in horoscopes isthe famous argument in Cicero's de Fao concerning the analysis ofthe condicona:"fsomeone i bora tthe rising of the Dog Star he will not die 25.” The context in ‘which Cicero sttes this conditional, as 4 paradigm cise of "stologialprineple, ie concerned withthe modalities of stte= ments refering to the fre, ad Chrysippus daagecement ih Diodorus Cronus. Cicero wants to develop a contradiction be- tseeen Chrysipps’ chim that shat wl ever happen may stil be postibleand his support fr divination, since divination appears £9 {Exclude the possibly of Fabius dying ase fe was born a ehe Tsing of the Dog. Str. For Cicero’ purpose any” prediction, expressed ava conditional, woud serve ‘Certainly, fom the de Fato om ts ow, one cull suppose that CChrysippus” defence of divination rested in patt upon serology anu the weight of moders scholarship has impressed upon us the Aecsive oe ofthe Stoic as defenders of astrology Buin the de Divinatine self ee poedctons with which Cheysippus i ined aye exchsively oracles and dreams. 1 was on these. and not on tstrology, that he wrote his two books dealing with divination ( 6. The cise for divination, see out by Quintus in Book 1, ‘enionsstrology onlin passing athe very end, andthe mare "sociated with there is Posidonis (130), ‘Cicero's counterarguments in Book th, owing much to Cat- neades, focus equally on dreams, acs, and portents interpreted by augury. But he docs nce one passage (87-99) om Chases apparitions (Chalitonam mons) sting our and then seeking £0 demolish the principles of astology.” Nothing in. Book + coresponds to ths. He prefaces it by saying dat Panaetins ithe Base ee men ened as gah pe Sy oe st rt te i ctnee emetle eae tred oe cy Sitesi ates Erich mele nea ob ait Sak Sop ear aes i ae Asrolgy:arnets pro and cons 169 ‘only Stoic who has rejected astologicalpredicions (8); and he ‘onclies his primary critcams with the question: ‘Do you fbserve that a not stating Carneaden’ arguments hut those of the Stoic presdent, Panactus” (Dio. n 97). Within this section mention is made of Diogenes of Babylon (o0), who suceseded CChrysippus a head ofthe Stoo. He, says Cicero, gave ualiied support fo astrology, holding tha predictions of dispositional Ghurscterities could validly be given but aot parsculae details of Sn individuals frre Ie Since Cicero explicitly drew on Panaetas, not Carmendes, for his primary retitation of astrology t seems probuble that astolo~ {gy tes a8 mort a minor topic sm Cameades many arguments Sesinst Chrysippus on divination, Certainly that is hot the only Possible explanation for Cicero's use of Pamati here: for he is Sbviously enjoying the opportunity of sting one Stoic against ‘others, Bue the genetal conclusion seems to. me inescapable astrology was at most a subordinate feature of the cates Stoic Snteret in divination © The most likely explanation of this = Historica, If Bouché-Lecereq and other experts ate right, the Grecks did not simply tke over Babylonian astrology. Bouché Leclerc (p61) thought that the Fl orm of the Greck zodiac swas not Teched before the time of Hipparchus, mid second century sc. Later research seems to show that the arafical Aivision into ewelve equal segments of hit degrees was already in ove in Babylonia tn the fourth century! But Neugebavet agtes thatthe evidence for direct borrowings from Babylonian concepts remains exceedingly thin. The main structure of the Astrological theory is undoubtedly Hellenistic TE thsi correc, i helps to exphin why astrology became of substantial interest to che Stoa from the mid cond century 9.6, 1 Bat tet ent Bea {ap lo aor Wi op gr te Bp iene70 A.A. LONG the period of Hipparchas, but not apparently carl." Greck astrology differs from esely Babylonian in is concern with ‘ordinary individuals” horoscopes, The calculation of bith times, Panetary angles and specs, andthe rlevan degree ofthe zone {yasa highly complextask, Development in ellensticastronomy ‘cen 1g have bech primi determinant of dhe theoretical founda- tions ohoroscopic astrology which also drew heavily om standard “concepts of later Grosk physics In short, he system as we know «| iein Manis and lace writers was probably a cresion of the second century tac. fot later This ems the most economical fxplanation ofthe virtual sence of Chrysippus and Carneaes, the qualified approval of Diogenes of Babylon, the opposition of Panaetue andthe wholesale suppor of Posidonis Tsay ‘wholesale support of Posdonivs. But even this puative fact may be based more on th convenient assumptions of modern Scholars than on hed evidence, Certainly Augustine, following ‘Cicero presuanaly (btn im our Cicero), calls him “uch given to astrology’ and fe atacs him for holding that simultaneous iinet of two brothers pointed to their Being born and conceved Inder the same configuration of planes. Posdonias was willing force conentons Between celestial and human events as evidence for universal sympathy" But no technial deta of astrology are linked with hisname, I tht omy the acident of tansmission? Posidonis’ general approach to divination seems to have been an fpenended interest which fll shore of eliming a causal cone= Sion berveen celestial of other ataral evens and particular human foreunes. He Believed that valid predictions could be made fiom ‘signs in ature” (Di. #129). Bue what sve may cll hand Strologyeluinied more than this The aspects” of the planets ‘within a nodiacal sign were not jus indiaions ofan individuals Atralagy arguments pro a conta 171 Party of Avgoxinc's agument are alrexly found in Aula Gell Summary of Favorinsn (01-19, 25, and many other cbjesons nce in the City of God 17) Belong tothe sn sraon Sepia critic Sach antiguas, doen part presumably to the sources he sed, i characte of Alanine’ method 6 ‘Veontoversi. Thee no evidence that he took account of eee BPlp inr'see Sorc ‘hon rfl tne Dv we Aaa pal Sean re ‘Geese mir anving ear op 00 gute host Bde sgn ogy bot Dre Tatas foray ee Asroligy: arguments pro aad coats 191 ‘Prolemy’s refinements or even knew chem. Heredity and ensiron~ iment, causes whove importance Polemy admited (see above P 113}, are the postive weapons Augustine uses o rete atologie- tl explanation ofthe sary and diferences of twins (v 2). ‘Augustine has two surprises, though, to remind us that is primary concer is theology, He admits tha astrologers oem give fre predictions, but the explanation of this les in the power of fil spires and noe the are of horoscopy’ (V7). Secondly. he Temper his pase of Cicero's refitaton of astology By acting him af denying fore-knowledge to god: Cieero mistakenly feared thtivin foreknowledge would undermine freedom ofthe wil Bat even astrology i preferable to 2 denih of god! fore knowledge (v9) Contain Plto’s astral theology and Stoic determinism and pantheis are doctrines which helped to give astrology a theortiel Foundation ‘within the Greck philosophical tration. Some Stois and some Phatomists ook astrology seriously, but war nota subject which ‘mattered greatly even #9 them upto the ume of Potints. On the tvhole the philosophers were hostile, especially Epicurean and Sccptics, Astrology is importa wo later Greck philosophy largely becnuse of the soptica argument provoked. Thee began in the sccond century ne, ith Panactis, probably under the stimuls fof Cameades, But astrology coms to have been ony 2 mina tem tn Carmeades cise of Chrysippss om divination: and. even Cicero treats it briefly and rather perfunctory “The arguments he seconde were probably developed before astrology ha obtained recoreplacein the Graceo-Roman word Bat they continued ta be repeated by critics up t0 the time of ‘Augestine. Dialectical eather than technical, chey are effective Seainst the primitive versions of astrology which offers stl 252 ‘omplete explanation of terrestnal evens. Astologial wtes au i,t rs fet, it ‘io sean eee Ce7 such as Manilus and Firmices Matermas made ony hal-hearted ‘flor wo meet chem. But Ptolemy went mach farther in defend- ing aterology aguinet the standard critic, His disclaimers and Imoshicaions Gogether with the guied criticism of Potnes) {re porhaps the high points of» protracted debate mse par Which ook up entrenched postions 7 The origins of non-deductive inference M. F, BURNYEAT. God, a8 we know, did nor lave i 40 Aristotle to: make men rational, But he did lave it9 Artal to make philosophers take $ccoune ofthe fac hat there more to men's rationality than the silty to consouctsyllogss. The ability to reason ffom signs, fr amore pencrally, the use of evidence, came to bea central topic in the Hellenic philosophers’ discussions of rationality. 2nd it ‘vas Aristotle who first proposed an analy of the notion of sig (amen) His teestment set both a precedent and a standard of 'ccomplishatent for his scessors. So its woth Aristotle eat we Should begin “The notion of sign itself is of course vineally a6 OL ay the rock’ habit of giving grounds oF evidence for their assertions. “The term semen” may be found in tragedy, nthe orators, inthe historian, inthe medic writes, in the philosophers. Reporting the illegal burial of Polyneies, sentry sips, ‘There were no signe fof any beat or dag having come and mauled the body” (Soph. ‘hw 357-8). Nese Heracles i place “where they now show the Signs of Heracles descent to Hades’ (Xen, Anah v2.3). An orator pleads, ‘Don’ seck any other test of my good will but the signs Funished by my preset conduct” (Andoc. 2.28). The accused args that the fice hata man se nt sepped is noe sign that he ‘eas ot murdered foe his thing (Ansiphon 2s). Any wumber ‘of persons marshal grounds for claim by saying “Here ate the Signs for if, or words to that etlect fe, Astoph, Nuh 360; Dog. Apoll. fr, 4 Hipp. VAT 18.1-2 Ise, Pang. 86 PL. There. 183A), where the signs which follow ate as likely eo be abstract and argumentative as concrete and observational, There + ho fixed preterence for uring scmeiow’ of observable things oFog Mo BURNYEAT “observa tates ofaffais, amy more than theres the Engh teige of igh’ for ‘a token or indistion (ule or othowise) of Some fictor quality” Tn any one of the examples cited evden’ Would be as good a canslation a5 ‘Given this bickground, we naturally assume, when fst Aristo= teand then liter the Stics propose an analysis of sgn, that ie wil be a technical analysis of 2 notion in common use, not the stipulation ofa techn concept. We expect no restriction on the fatge of things that can serve a sign or evidence of something, forevisting usage displays none, Iris not even correct say that 3 sgn f what we Would call empiieal evidence for something. (Often this is xo, but in the Elatie tration, when Parmendes' “signposts” (ma 8,2) became Melsus signs ome 8.1). they were intended to give demonstrative proof ofan ineseapable ‘inclusion, Likewise in Seaton Empires eis regularly reported that demonstrative proofs one species of sign (PH 1196, (23,134 134: M vit 140,180,278, 280,293). Iso, ign” covers any kind ‘of ground, evidence, of reason for heheving something, including ‘demonstrative cexdence, we might expect that a rough, general firseskerch of the notion ay ie functions in everyday dncourse ‘ould take the following simple form: For X to be a sign ot tvidence of Y ruin) chat should be Pde oe manifesto us in some appropriate way, (i) tht ie should be evidence of something ee in that ¥ ean he inferred from it. The task ofthe Technical analysis would then be 0 explain the eelationship between X and Y which sustains and jusies che inferring ofthe second from the fest Lotus sce how far Aristotle Falls these expectations 0 The offical acount is Prion Analytis B 27. Aristotle's stating points not so much dhe ordinary’ man's notion of sion as the Inferences to which it applied. People sy? for example, She is pregnant becuse she has mall’, "Wine men ate good, for Pacos food, “She i pregnant beciuse she ie sllow The forms of ef ta rr io mine SHA Sawn ar Sse ep tS Siren sneha Sore a pees ‘The oigins of wondebucive nfrenee 195 inference expe in there unstied Ioctions ar leap ‘Arto thik of winning rath * Bot none of ther fray {ald as they stand and, move portant ith many of them Pnsbltilingout the ressoning wil be formally ai ter. Tecus use the phrase the econtastion ofan ference tone ‘rerything thas involved in supplying onexpresed ssmptions Soa seanging premises and conehson in proper logic form > “Then Avs sees his che akin Pr Ba a5 ha of sorting tis exampls into thoue tat do and show that do oe admit of forolly vad reconstruction” “Aru ony means of exibiing formal vaity was sylo- gsc So with some strain (eg sdting singular terms) ylgie ees serve? Ths fr the milk example we soppy Al who ave ml ae pregnant’ and cone "Ths woman basil therefore ths woman popes to got the form "AI are the Ca Byiherctor hs Cea = vad ist gure syllogism. By amlogous procedares che Pitacus example Beomes "acs ‘egos Pitacts uve therfore all he wise ae good, which ‘Ausote ascribes she foe an vd thin figure slog me oe Sy no is es ERGe lipemia il, aegis te pcan of iy elo mga iP et Ag era hace e ot Bia en hi shes chs ct Sea eyecare ee redone Sarre ace nnc ee196M. BURNYEAT Finally, "All who ate pregnant ate sallow; this woman is sallow trefre, this woman is pregnant has the form of an invalid Second figure syllogii. Taking these examples to be representa tive enough to cover al reasoning from signs (7oat 1-13) Avistale onelades thatthe only examples which can be made formally ald ae those with a Bs fgune reconstruction (7032837). Since the scond ad third Figure reconstructions are invalid, in he other faser the conclusion reached may be tu, but it doesnot Follow tncesurily (70435, 57) fom the premises Sct out in the recone Tanto agg that shold be perfeey plain thatthe purpose ofthis technical exercise nat torject he inferences which do not ‘Minit ofa Formally valid reconstruction. The invalidity isto be noted and appropriated forthe classification and understanding of ertain comnton and wel inferences of ordinary lif. When the technical work done, Aristotle eserves the word know for thei conmexion with examples which do adm f formally valid Ge, fst figure) seconstracton, Javing icopen whether ‘emion ttould collec the rest (those with second std third Figure recon~ Steuctions) or should continue to stand av the gente of which Iskonrion tone species (701-6). This is ary to dismiss scion sno more than an ineaiiey. Ie is an act of linguistic regimenta- tion which finds its justification in the sim (7ob3-3) that Tor ‘ordinary parlance that which makes us know something isthe Inemiin a yor inthe claim (Rit. 1357-10) tha ermal cally chmavon" connotes conclisivents. These two conentio whether correct or incorrect about the usage of “thm belong together andthe implied contrast surly coerce, thatthe tvidence indicated by sfmron need not be conclssive enough for Knowledge, Someone who infers tht woman i pregnant fom ‘he fice hat she i allow dso Know tha she regnane; but eee rear ete gy eres Sait hata ik sme se mere ie The ais of nonsdecsive iniene 197 hisbelis reasonably based on a sefil piece of evidence (7022 takes i serve that sllowness i concomitant of pregnaney: in ‘ase of doubt, substitute morning scknes), There are diferent trades of evidential support and Arstodeexplicidy states (7064-6) hat inference trough the fst Figure, ie. fom a terion, the ‘most reputable (ndexoaan!and the most tue (matte alti, fre. the most productive af teucconclasons. This imple thatthe inferences reconstructed in other figutes do have some, though + lesser claim to eputability and rath ° Rec race” descent Hades and the signs listed eal: they were evidence Bat not ll of them seee conclusive evidence, Aristotle's thesis in An. Pr B27 3¢ that they would be conchae evidence (teed), ullient for knowledge, ifand only if ere universal generalisation (70830) canbe supplied to give the infrence a wali st igure recomsrace "What, then 2 sign according ro Aristotle? The question ma be answered at wo levels. The non-technical dictionary di tion soto speak, suhich Asso gives a 7oay-9, records simply that one thing (tate of afar, event) X, is sgn of another. being the cate or having come about and only if, given that Xs the cise of that X came about, Y isthe cise or has come about bere ov affer X. This mount toying that X isa sign of and onl if given X, 1s being or happening (eater of later} may be interred, It gives no guidance om the question with sehat warrant or assurance itmay be infereed~ that investigation comes lhewhere nthe chapter ~ but does recognise thatthe ign not the bare fact as such that she has mil (chat itacusis good, hat she i sallow), but that face a the Basis for an actual ot possible) inference ta something fither. The point i highlighted when the ‘non-cchnial definition supports (ga in 7) logiian's technical 1 De ingens of rg by mle pce wh She ec a a ae ‘al tn Aree Nom lc) gs See ‘From es tne Ree cae on bee seins syoS Mors punNvEAT explication of what a sign is (9046-7): prowsis pode, a premise Fran inference oF» proposition used to show something [K proponition used o show something 3 proposition asserted 3s true, a0 Arete isnot moving 38 fr from ordinary language 3s hhe igh scem to be doing when he consttues the sign as a proposition rather thn a thing (state of afr, event) which is oF Fappens. Ordinary Greck and ordinary English license both Xia sigiof "and That pieasigh of. with the choice repeated forthe banks Bur given the point we have been emphasising, thot signs belong in fnferences, ie mc clearer #0 proceed in terms of propositions, Tht why Leled 7007-92 definition but Joao=y an explication, in other words 3 “consrl «which s Intended to replace familar but vague and ambiguous notion by nore precsely characterised and systematically fil and ‘llminaeng one “The fll explication of sign’, howeve, is hi: prota gpadek- tik ¢onankait#endexoe, a proposition, either necessary of rept= thle, ured to show something. The fuer characterisation either necessary or reputable sounds, puzlingly, like a comment on the nodal satus of the proposition ill. But what we need 10 Complete the technialexplction is 3 comment on the Warrant Which the sign-propostion, a+ we may calli confers on the Conclusion inferred fom iI cll those signs necessary (neni) Sh aica aa eee Shs i ekerecmn coe Soak apie eee Sn ee The ovgns of non-deactioe infrone 199 fiom sich a syllogism can be constrcte (hin gets ims). 4 valid syllogism ia sehich the eosclusion allows recesaily (rom the conjunction of the sigh-propositon sigh the tue generalisation supplied inthe reconstruction (Rhet.1357b5-6: ff. 140ss0-13) Thus Cauankss ¢onoser antspates te distinction ‘Asse will draw and defend berecen rebrand other signs Some evidence is conclusive. Given that she ha milky it 6 necesary that she is pregnant (ct lt. r3s7bL4-i6} cist the highese degece a respectable thing to believe (onesotton, 7063) ‘More often evidence fs no in this way conclusive o slicer for Knowledge I rely makes a conclision 3 respectable or rept able ching endoso) to beeve “That comiplees my accoune of Prior Analytics B 27. If che Icerpretation ie correct, it would be churish not to hail the ‘hapter ea pioneering star tothe study of on-dedctve logic.” Bat to back up this concloson [need to set tn 3 wer context and quell an akernative interpretation! acording to which Ar Fete mcins us to think tht, when an inference il the test oF syllogntie vay, i fails a a inference tnt cut Le has probative force whatsoever, nthatcaseall here isto be si about Brguments from signs other than toni that they are bad frguments, The moral will ber so much the worse for the ‘theta contexts in shich such arguments abound. Mt Bla ts ie th in pe seer oman he ft Ste neal a HAAR sa cas ag ta we aay eee sina tea a Ses achi fence Ea ele ope i SDN te ea st ins Ot ry SsuD on ante ramet nelson Sh Bn 8 nw dca, a He[Notice is tha if logistically inva inference from signs is tobe condemned withoue reprieve, a good deal else goes with ‘The chapter we have Been discussing spar of 2 projet begun a An. Ph By We mst now sate shat noe ony are diet and demonsttive Thc fg th oad sense deine na. Pre syt8-0) ‘ced by wy of teaforril gue [ne te edie logis ‘hemmow seme definad bythe gues thew sendy argc sp. te DB: Ass} buts ako ate torical ones soda genera ny pesuasive Stgunent> whatever ite proedare For all comsction cames oUt ‘ier by ay of ection ox om induction (pain). (9-14 “The projet ito shove that sllgitic sa universal test of logical alii, That i the meaning of “All hese arguments ae effected Trough the gues; i is not che absurd claim that ehetricl deductions are already in spllogistic farm. Likewise uggest, is by way of promoring syllogistic as 2 universal test of logical validigy tha in B 2} Avstode goes on to a syllogisticreconstrac- ton of induction (epagie), finding thatthe conelasion follows recesaily only ifn extra condition i satisfied, over and above Isha is given by the premises of the sllogisn isle. The etea ian ta a ets SILVEIRA Sethe hr atc Stnmeta See Seed i ce ha he sc le ge ef Sa an PONG rg ee fied aN Beh ce gh Pe ae te ‘The erin of nowdetcive infiene 208 ‘condition i thatthe enumeration of species Fling under the genus ‘whichis subject of the conclsion mast be complete) Now, if [Ariston condemns syllogstialy invalid sign-interences, because they are ayllogstcally invalid, he ought equally to condemn inductive argument, even in the fivourale creumstance of Compete enumeration, and rte in the more commen cae, ‘which includes many ofhis own indctve argent, where cis tats condition ir nt satisfied. No such condemnation is utered in B 23 oF esewhere. On the contrary, Aristotle regulary teats induction a a respectable, indeed essen source not merely of| conviction but of knowledge. ‘Much the same story cin be told of the analysis in B 24 of| argument fom example (paradigm, Le. extapoaton fom one puticular cage to another, Aristotle offers a two-stage sylogistic Feconsraction, the frst stage of which is exactly parallel to the {hind figure sigo-argument in B 27.2 That, fom 4 Formal poine fof view it commits the fallacy of lit Minor. But there not 2 ‘word to imply that rational persons wil eschew example, and it ‘Would bes fobsh politcan who di so “Examples are the most Suitable for dliberative speaking, for it by conjecering from bist evens that we jodge the fare” (Phe. £368129-31). “The truth of the mater, I believe, i that Anstole dees think that sllogitc i universal test of formal or deductive validity, ‘bur he does nos think that formal or deductive validity sche only test of orether an argument i intelecally respectable or has 2 {stable eaim on rtional minds. Various forms of inference ae inuse and enjoy good standing with those whose busines i 8 to frgue a ease Let them be cassified, reconstructed in sillogistic form, and texted for validity: then We shall sce how far their 2 ft ea ats many oh a, Te mci ey ite fae Ps snstrength i strictly logical (eeefuable if ru’, An. Pr. 70229-30, [Rho 1357617), how far and where leaves room Foran opponent to counersteick by objection or refutation. That the project, land Arstoil’s Ristori shows how usta tis for the pratce of Propounding arguments and replying to the arguments of one's Sidversay in jc and poltel debate. Most of che argument= forms atembled inthe Rheori are taken tobe open to objection for refitation (ef, exp. t4oabi2-03433)* Debate about human ‘Mbit of argumentative strength in which the outcome s Secied not on necesite alone but on Hkaiboods (1403b32-4 cf ‘ysraaa-3i, Ifthe forms wete pruned to meet she requirements cof sllogini validity, Aristotle's stemps eo build a inelectally respectable rhetoric would crumble. Tonchide that the wider contest of Prior Analyis B27 confieme the interpretation given ealet. Not only 18 the very ck rien tieestenciaorts Seem eps Docu dea The ois of nondedcive infiense 203 notion of Formal validity Asstotle's invention: he hs farther seen that there aferespecuble arguments whose probative force docs not derive from thet formal validity, because they ate in fact invalid. Teall this 2 pioneering advance because tthe decisive insight prior co which no one will dink to ask about when an ngument which is formally sald is nonetheless a good! argi= tment and when iia poor one, of about how the good specimen ‘an make a rightfal claim on faional minds. Avtotle docs 20t dees these questions But he let he subject poised for further ‘evelopment by any successor with the logical acumen to discern the challenge that they pose He also left some indications a5 to what might be gained by taking up the challenge. Prior Analytics B27 conelades with an appendix (eb7IF) on the logical requirements Tor seting up 2 science of physiognomics, + systematic method of infereing ‘mental characteristics in human inividene from bodily Features fake a5 tei ‘signs This may strike us today a8 an unpromis- Ing field of appieaton for a logic of evidence, bat the idea ofan applications there nonetheless and in fet the general orm ofthe problem is both interesting and comples. les 3 problem aboot [stablishing 2 one to one correlation beeween two lems $0 28% Infer a conmon cause and then using the creation to argue om fone tem to the other in significantly differen case The goals to provide a theory which wil yild generations for the reconstruction of such inferences at "He is 1 cowed because his yer are weak and blinking and his movements constrained” (cf. TA) Physiog. Sorbs-13). ‘We shal return to physiognomics later. Meanwhile a cone- sion between sigs and our overall vcw of human rationality smade in the fist chaper ofthe Rletovi, dhe treatise hich ims t0 Study the all ange of everyday reasoning, whether ibe rigorous 14. Deng ot io ac pect gh a i ‘Sekatorvmont hath dave espe nel c se a an Roe, Spr inns ya oe se wake Dr, sR es Sep ee Se sh eyo he Tage of meg whchclad? and inadng 28 one ingore division rexoning from signs. The toductoryapaloga for ta ty eomais the remark tha reasoning in defence of nes eave pot of ew tno charactont alow of nan than ysl sete (iossaittay Again, “All men share in some: way in both {thse and dlr forall wp seein pin endeavour 0 Exc or uphold an argue to defend themselves o ese (Ghsaay-0) God id noe atin vain Indo, wah 30 uch Selo eth iinet cn i mre rejected rey to overlook some important omsons. Two raters in particle oman untouched which wil be athe cee Sie plosophiel ducsson and which would hae ben well ‘itn the compas of Aros poncerng stat ‘One hing theming any cession nhs context ofthe covering gevcasaon whch maker 2 aig into a tintin teconsile in he form of is igure slog Song it ia tue une proposition (oao), the contusion flows teensy, bar fe might want Take questions sbowt the pati ov che moda sas ofthis waives. Artes ov ample “All who have mila pregnant shows one ofthe problems that can are. Av sated is Gt (ke 90 many Flilsopbers examples, north cay toy pesely ‘Bhar gencal uch that on rcs om in making the renee "Shes pregnant because she hs mike Would mater sone sea ore cran ofthe inference than of the exact speteation of > er macy te ele in rs, ene eto Scenario woe me or shorter nae “The ain of omebutiveinfence 208 the covering generation? There may bea wesk spot here for Iter seeps probing ~ or aermavely fr fl philosophic exploration "The tee noteworthy absence it compensated for by Ros. wren he report the conesion betwen sig and what sigs, 2s connetion between really sly poied character Sa’ ka ay pried ove simulans previous, oF sb: {hen oi relesing vo 7-10 (he montecnia deiniton) = thick, owever conse aboltly nothing corsspending ' {heconra Ihave fai maybe temping spp tat ‘Arle mast have somthing ofthe sort view he presume Sbiy commited toby spdebk), bt nat the hap makes xpi comment on the epstemi tats os pretence ony Sten not on the sigh premise bt on premised the reconstruction: in she Paces example people omit to sat "Daracs ir wie’ Beco i i common knowledge (019-30; ‘Bh 1)5yaso-ss) As with the covering generation, so ako on ‘his pomt, Arsote's account of igs makes no use of eitenc totons. Has he made s deco of pin o kxp the lagi of tridentate, ors separate sx posable, fom te eptmolo- fy oF se of ereencel Ths contrst with ter developments (re ow is so sing (nd Ron's misprespion of the text tua) a ne ould easy thank 2, 'Atany rt, this loks to be the only substantial point om hich Arto his led ou inal expcations (tog above). He ht analysed, ona lest given an expition af the ordinary mans motion of meon ant has done prety tay by i He has put no Tetrion on the ange of things that can verve 2 gn oF rience, Por example for al hesays to te ontay. propeany Could be gs of acon) But although fe ha dtingised ‘vince which suficentfr Knowledge frm evidence which tothe hes ne exactly matched ou inl, ough characterisation ‘of sign For he has Hot rguired, nor even mentioned that the ‘Son proposition, ttng the evidence fr something, mist lf [becvtent or known On theater and, igo a he study of ertsign-arguments has brought him within teach of the idea of 2 hon-deductive logic, he has gone momentously beyond those imple fst expectations in section 1 we shall compare and contrast he toi approach. Bat fist, because ofthe nite of on source mteral om Stoic sigs, theres work todo to gain the right to make the comparison find to ser the terms in which i should be made W “To caabish the reality of signs, the Stes assert eis not aed speech but intra speech by which man ie from {rnionsl nals or ros aad pares and jas utr seuss sounds Nor‘ it bythe nee fr of hang impressions uc for he C00 freee impression The ference ue man hae npession aang om inference and cninuton Thi amount this posesing the ee [keonsequene los ml eey thereby grasping Se oncep of Sen For sun ie ial ofthe so thw hem ue Therefore the ‘EStence pee) of flow fom the ate and conseraction (dete) of man exter Empurcae MW 75-9) “This isa remarkable passage. Just how remarkable emerges when we supply, fom the reprise later in Sextus discussion (Mf ‘285-6, the extra premise without which the argument 16 10 Srgument at all man is providentally constructed (rents intaskevatha.* We only have one ofthe peemisses ia the claim that ‘the ably to think (to discourse wth oneself, to fame foncepts and fo deaw inferences is part of man’s nature’? or FRE ore tuning up ated eto, SER it eg ooey apt pe ekas ri yee a = er eect ce lowed by he wal cme of Kayser (er NF 7 HT) fe ae neem The ois of nondedsive nfreme 207 beter, that ie i the distincvely human aspect of his nature, constitative of his tionality. And hie aim ells hardly news Ie'is she Stoic’ inheritance from Plato (chinking 36 internal discourse — Theat 1891-1908, Soph, 2636) and feom Aristotle, ‘whose poston sea already, as we have seen, chat the philosophic 2 account of human natre must inci reasoning of every 30% including reasoning from signs What is original to the Stics the appeal fo providence andthe argument built upon i ‘Manis providentlly constructed. Man ic resoning, infering creature, which means, fist and foremost, that he draws safer fences from signs, Therefor, thete mus in realy be the connie sions that reason taker self 0 be discovering. 1 nothing 5 sbjeccvely evidence for anything ese, mans poorly equipped for the sword he has to lve in; which cannot beso i his cognitive guipment (hth sehe endowment of providence. ite a old argument, Sestn sentir justied when he points cout (MF wt 285-4) that ts conclsion, the existence or realty of Signs, isa good deal les controversial thn the premiss which invokes providence to establish i, That's jst bullying, he 373 (phos hisin), But I hirard the opision that Arsotl could not ford to beso sceptical. His Physi is founded onthe no Tess bold Contention the che fundamental concepts needed for the under Sanding of nature ar tobe obtained by probing examination of the ideas ofthe orcinary man anda handfil of previous seers Mier sath. The ideas have to be clafied, made consistent and thw cat es, car ‘omen, hen Sco Rico Sor Te CGN Bok th Rest inde tbo tae208M unnvear generalised. Bur the large assumption (contrast Newton is that this the place to stare, Wha anurance do we have that man’s tind and ature ae so attuned to each other? Once the epistemo= logical question becomes a major isue a eypiel Hellenistic development ~ Stoic providence (which is not, of course, an gency outside mare seems the ansiver closest tothe spt of the Plstonic-Anstotelian tation.” “Thus the concept of sign comes ro take a central pace inthe Hellenistic version ofthe established conception of men's ational ity, Whac has happened t the concep of sign seo) co enable to bear this tole What exactly has been proved to be real iit ‘suablished that signs are real? “The pasige belore us might suggest the following answer. Signs ate conditional proponitions ofthe form IF his, then that so the argument eatblshes the real existence of 3 clas of Propositions. This s quite certainly wrong, doubly so, and its worth secing why. First, 1 proposition in Stoic doctrine sam immaterial eto, a sere sayable' an doce not exist, Along sith place, time and the oid, the leon defies 3 celebrated Platonic axiom in that i Somedhing bueno thing that (exists): ie merely subsists with underlies our thought." Admicedly, various passages (ours among them and others more polemical have besa thought t© make tific to acope thin oil categorisation a dhe whole Story, but Ican bypass the complexities of tht issue? if ean show tha i has no relevance to the passage we are examining. ‘offer two reasons, each of which setms to me suficent Om is {@) "The conctasion of our patsage is tha signs huprchen, and this docs not contradic the conchsion ofthe seep’ argument ta nl ie el sh ot aa aes ra Sk poi ti, haw 2 te niin Nema oh CPU NE asap gun ef has Sen et mara as ‘sh tevtorsoneyeeros nue M's. nd hora wera ‘iene sae ts ‘ ‘The origins of ondebucive inference 309 that signs mé einai (275, just before my quotation begins) ‘unless parce = cna. Bue is quite eleae what means in ‘the contex of this debate o say that sgn ena or that they do tot, The rival theses are that nothing isa sign of anything (eg ‘ym 279) and tha a least one ing ia sign of another (eg Vl 278), The ontologie status of propositions dems Not the peimary issue at all (by The scond reason why the ontologies status of propositions themuclves hus no bearing on the present passage fs that ‘would reflect badly on providence. Merely to ensute the ral existence of 3 whole lr of propositions would leave open the possibilty tht they ate, ne and al false. What is needed for {he human mind t be n tne withthe worlds that something should actually be evidence for something, not that there shold be 3 “syable’ to that effec, [Now forthe second mistake in che wrong inerprtaion which the wording of our passage might sugges. Sigua are not cone ‘tinal propositions. The offical Stocaecount state, “Asin = 4 raponitionsohich Forms the antecedent in 3 sound conditions, ‘ing revelatory ofthe consequent” (M van 245) Not the whole onitonal butt antseeden the ign. But geting that righe s ‘nly a prelude to the serious problems Granted tha the passage should not be read as equating signs swith conditional propositions, dacs it sy or ply that signs are expressed oly in conditional propositions, as antecedents thereoP > Novy hati Sb nh Moe ‘sie tts mag Sa pe he eo nap tt eg ome ey Tae Scie antpon tnlpe eo Shoe yrds oi se Ach Gh 5) nie no nh ht oie ik she fleur whee the pcos he 48 Rieravey, Het ad, auld Be judged ics one pine ceone means speak ofa sgn, does one have to start out with che sword If" That cannot be right either, or ele proof would nok bea specs of sig, “The standard explanation ofthe claim that proof ia species of sign points tothe fact that na demonstrative proof the premises serve 0 revel i give knowledge of, the conclusion (PH 113 134M vit £49, #80, 399). The conjunction ofthe premises 3 sign of the conclusion (iM vit 277) leis thus by 4 sort of retonymy that proof comes to be cssifed under ‘sign’. The oficial genus of proof is ogo, i.e argument, Le. premisss plas onclison (eg. AL vit St, in such cose proximity to the elim st S99 thatthe genus of proo x sgn that we must suppose the v0 ‘nsfctions are not thought to be at vatance with each other tn try to understand them accordingly. Stiey, not the whole proof butte premises make the sign. The metonymous csifia- bonis best expined at M vir 14 ie by participation ign, itis Because te premises area sign ofthe conclusion, that a proof serves to revel is conclison, Proof, i seems, derives i Fevelstory character from tat of sig Bue news a proof does not begit With che word I (save per sscdou). To every argument (hence #9 every peool dete cortes- ponds conditional proposition with the conjunction of the $gument’s premisses is antecedent and the conclusion ofthe argument 3 i consequent, and Stoic logic declares that the "rguntent vali and only ithe associated conditional s sound (HT 137), But that famous contribution to logical theory presupposes a clear distinc between an argument, which ss 4 equence or ‘sytem’ of propositions (PHT u 35-0), and a com Alina eich sa single complex proposition (DL vn 68-9). Ina proof in which the premises ae 2 sign ofthe conclusion they are fot encloned within a conditional It follows that, even i every Sign it 3 proposition which forms the antecedent of 2 sound Conditional, it should nor be Stoic docteine that the proposition ‘outs or serves a sigh ly when occurs as che antecedent ina ound condiional, Righily ead, our passage ought to claim an Jmporeant association benween signs and conditional an associa sion related tothe sociation between proof and conditional): The ovgies of nodeuctive inforone 211 should not imprison signs within ee conditional form ofexpres- ‘We mas Labour our way through ane more thicker and then (L trust) aight will dave Tis generally assumed that two kinds of sign ate recognised by the tous, the commemorative (ape fo) sigh and the indicative (endithon sign. Te is controversial ‘whether both ae meant, or only the second when ts sad hat 3 Sign is 2 proposition which forms the antecedent of + sound ‘conditional. being revelatory ofthe consequent. In due course | ‘sll angue that boty are meant. The immediate problem i that ‘when the distinction betwen the ewe kinds of ig turns out that signs of foth inde ate observable, could ever be, ‘There should really be no dispute about thi. The com= memorative sign is (by definition) something observed in com junction with whats signifies, as smoke i observed i conjance ‘ion with ie (PHT w 100. Mf wut 153). The indicative sign 8 not something observed in conjunction with what it sgn, but the reason for chi is that what sgafiss unobservable (M vin 154), ‘Thesig cf, by contrat, mune be immediately evident, 8 when bushing a sigh of shame (M van 173) or the movement of 4 person's body sigh of the soul within (M vu ts4es). The stinction is expressly tailored (PH i yp) M Nu 151, 156) 10 a8 “epistemological distinction between twa clases of non-event thjeet The commemorative sgn i for getting Knowledge of things which ae temporarily (os kai) non-event as the tty of Athens happens noe in present circumstances to be event {ous now. The indicative sign is for geting knowledge of chings ‘hich are naturally (phure) non-event, one important category ‘of which comprises tatemal sites of the human body. eg. the prsrence of “ineligible pores’ nour lsh through which sweatows (ef also M vu 219-20); but remember that for Stoic ‘materialism psychological sates Hike shame are bodily as well ‘Accordingly, just athe thing sige is a non-event object of ‘onditon, sn these examples and many odhers whats spoken of | 1 the sign (commemorative or indicative) is 2 hing (state of Mis, event) observed, not 2 proposition which i an abstract objer of thought 'Ac this pont it may be suggested thatthe problem is unre “There ie no hard evidence tha the distinction herween indiative ani commemorative nga goes back to Sto sources, indeed no ‘explicit formulation of the catinction before the tnt of Sextus Epics.” My reply shat the distinction ae Sextus dacs andthe terminology in which iis cast, may ell be ate and may ‘owe ae mach to medical as to piosophica ceces but that 3 oreesponig distinction with the same or similar Function Firmly embedded in Stocinm atleast a ar back as Chip. Fire, it indisputably a Stoic thesis, presopposing their pini- ple of concitionalation, that the premises of proof ate a sigh "which reveals the conlusion; Sextus interweaving ofthe topics of sign and proof, with the sume examples recrsng (se belo), nue eflet Stic sources, And we know tit, or the proof be - Sacer Sp es ec pe Daten ye st, wand pp om ert ae ety io ‘ carta “cso mena ee wr tere The origins of non-dedacoeinfrance 213 4 valid proof, dhe conditional in which the conjunction of prem festa as anteedent tothe concisin a consequent has to bea necessary truth Hf must Satay the strong extrion of soundness Called smarts Cconnesion’ or “cohesion, which the genera ‘onsenstis of modern scholarship apsociates with Chrys. Bat sccondly, no one who teas through the Hist book of Cees se Divination wll doubt thatthe Stocs have a mase of theory ‘concerning + ype of ign whose conditional expression could not possibly moc this strong criterion. In divination, an medicine fur knowledge of whats a sign or evidence of what grows out of | long record of observed and remembered conjunction, gradtl= ly corrected for error (Div. 116, 24-5, 10). 127) That there ae sch dvintory signs fot us to lea and use the Stois exablish By fn argument from divine providence which i simply a more fdaborate and specialised version of the argument we have been ‘isctssing (Die 482-4 1101-2; ef DL wt 49) and this argument is exprenly ature to Chrysippus, Diogenes of Babylon and Antipater (Di 184,10) "Now given that there iin any case independent evidence that CChrysipps was Keen fo distinguish Berean 2 strong (sar) ‘conditional, Which states a neeesary connexion, and a weaker {Philonian or material). conditional to be employed when the Former is inappropriate," and given that the de Divination’ signs (observed items hike entrils and cocks erouting) obviously go swith the weaker conditional, all we ned 1 complete the four-part Jigs purses o find that other observed tems, such 3 blushing {an example used by Cleanthes: SY 1 518, n- 80 below). ate Supposed to stand ins evesary connexion with that which they fea sigh of, This we shall Bnd shortly. Thus the dstincion between two types of conuitional lads naturally toa correspond ing distinction between two types of observed sign, Westall need rome terminology to mark the Lauer distinction. Since we knowof none that Chrysippus used, owe might as well cal the fistneion between indicative” and commemorative’ sign, idding sare-quotes when we mean to speak ofthe Stoic origin ‘tthe distinction which was later 40 called. So the problem remains of squaring the account in terms of propositions with the deinons of the go kinds of sign, Feflecing a8 they do everyday tlk about the condition of the ‘entails Being a sgn of an abundant hay crop (an example from Cie. Dis" 30) or Bushing sign of shame. But T submit that we havenow assembled enough material to sce that both this problem sand reali dificlies dissolve a8 soon 33 we recognise that there are two levels inthe Stok account of signs, as there were in Assos. The distinction between ‘commemorstve and n- icative signs, andthe explanation ofeach of them, covresponds to the non-technical definntion of an. Pr. yap-g. The statement that a sign sa proposition which forms the antecedent in a sound conditional, being revelatory of the consequent, correspond %0 the logician’ echnical explication at An Pr. 7oa6~7-" Thac is out Firs gain from sctng the Stoic account of signe side by side with Aristole's: inorder tobe rd ofan issue which has caused much perplexity nthe scholaeyltrstare ll we need tse than the move from the not-techsical level to the technical the ‘ordinary man's idiom Xs sign of Ys epaced by logician’ talk of propositions ~ Arstor's recommendation i accepted by the SriesBecase isa prerequisite for analysing the inferences in Which signs esentally Belong 3 But of course the Sto alysis ill use propositional logic rather than syllgisic ‘When epson’ ally brome bali cea Serna apaceot arokele seer i ‘Te eign of nodebuciveinfree 213 “Thus the milk example returns, but Aristo’ universal gener- alsation “All who have mull ae pregnant gives way to the Conditional lf she hus malin er breast, she bas conceived” (PH ft tos; van 253)% We surmise hat, a dhe Arstoelan recon Steuction ofthis example tok the forn of a first Figure slog (ara), othe Staeteconstracton ia proot in the form of thei first indemonstrabl (dur pone) Ise has maki her breasts, she as conceived: she has male in her breast; therefore, she as conceived And this very argument daly uens up in the discus Sion of proof (M vit 423) Similly, che ‘negible pots’ which ine stock lustation ofthe sor of thing See need the ndative Sign for (PIT i 98; M vin 146), secur in the Frequently ited proof | ‘sweat lows through the skin, there are atligible pores dhe flesh; bu the frst therefore the second (PH 140-2: vt 306, i300). These pores, morcover, are carly stated to ea necessary onsoqenceof thee sign, on Broadly conceptual grounds (PH 1 "4a; M vu 309) That one ctr desideratum scoured (p. 213 hove). Notice aso hat reconstrection i modus pone allo, indeed it requires, the sign proposition ‘She has milk’ tobe able to fppest as an independent asteron, outside the conditional of which it is the antecedent. Thereby 4 second desideraums i {eeurel and confirmed (p. 210-1 above). Can we go farther and Shes hat the Stoic ace wih Arttle that ehisis dhe mode in ‘which sign propositions ordinarily appear? Te stems from Sextus’ exposition tht we cn. Not one of the concrete examples in Sextus (ether commemorative or indicative) fas the ordinary man asserting of thinking a conditional, On seeing 4 seat, one says “There tara wound there before (PH 1 toa; M vin 153), not If ehis man has a Sear. he has had a wound “The proposition which forms the consequent of che conditional appette ts an independent assertion, presuposing the same status ‘SEI Rt pan Salt Fe ad aes oe SC rictercee sch tett eaforthe proposition framing the antecedent, Again, on observing a ‘man wouruled in the are ve free chat death wil come (M 8 153) not If this man s wounded inthe heat, he wil dc Thus far the examples are of ‘commemoraive signs so pechaps the onditonale are missing because the technisl expiation ad Inadus poner reconstruction ae not meant foe them, only or the “indistive” sgn (cp. 211 above)? Buc no, ehese very conditional stim up among the examples used sn Separate context later — when the discussion has cued tothe technical Stoic explication (Mf vin 252-6; ch PHT n 106) Ie the same with ordinary le ‘examples ofthe indicative sign. We reason, “What produces such Inovements as these a power within che person's body” (M a ss), not If-a body moves thus and so, it as a soul. The onsesuent of the conditional is asserted as the conclusion ofa Inference. And, ofcourse, as our example of proot in which che ‘injunction of the premisscs is sign ofthe conelasion we have TE Imoton exits, void exists motion exists therefore, void exis! Pa ere a— The origins of on-dedustve infrnce 217 not the associated consiinal I, if motion exists, vod exists, and ko motion exists, then void exists (M vi 277). Only afer the “ffs explication of sign has been introduced do we mest "Ise Tas nile in her Breasts, she bas conceived and the ike (M it 252 with 244; PHT 106 with 10), The contents where sign (ot~ Iemorstive™ of “indiative) and signifiate appear unasserted inside a conditional are ust those contexts where de discussion 8 atthe technical level ‘Thus from the organisation of Sextur! discussion, fom che _mannerin whic he presents hi examples, and above ll from the ‘euiemients of oder pone zeconstrucion t would seem to be thelogican whe brings out thatthe ordinary man's signinference relies exsentally on the soundness of the associated conditions. “The conctional I this, chen that gives expt expresion tothe ‘connexion of consequence (eelouis) which che infrence uses. Like Aristotle's universal generalission, i belongs more to the technical reconstruction of the inference than to the everyday Language of the doctor's consulting room: "This brings back t our passage Tes che voice of a logician turned metaphysican (very Soie, and bie of a bully) which brgues that since mans prone #0 make inferences, providence ‘Would have quipped him badly forthe world he as to live in tnless the assacited conditional were realy and teal rekable For that is what he is saying when he conclides ‘Hence, signs exit, changing a the subjeet but the level of his discourse. His tmesning isnot thee are antecedents i the aforesaid condtionals (eign as expat), But rather: cerein empirically observable items vi. thore whose existence or occurrence is asserted when a Sign proposition is taken ut of te conditional and used to show Something) relly are evidence for what is inferred fom them (= ‘Sign’ as nonechnialy defined), w Ie was once sid ‘La théore des signs aa pas analogue chez ‘Anntote™ hope tit by nove the aalogy bectcen Aristotle ardthe Sto i looking sulicieny close to suggest that a comparative assessment would be Both meaningful and rewaeding. We shall Find that dhe assessment leads to qualifications and a change of ‘emphasis in the anogy, 8 so fr deawa, which makes compare son possible. ‘One important matter which Arstode is lear ahowt is thatthe sign-inerence sam argument (enthymeme, Ax. Pr 70st, be. 4 thewrial spin, Rie. 133836) i which an inferential parle joins two independently asserted propositions. This is especialy tvident inthe example Wise men ate good, fr Ptacs is ood’. where surface grammar presents premiss and conchsion 3 di Unc statements Aniston's other examples use locitons of the form 'g eeausep’. which have the surface grammar of single complete statement, bt is whole approach determin tha hese Shall be construed, lke the Peacus example, a8 expressing 3 inference ofthe form, 504. Both the sigh-proposiion and what isinferred fom tate asserted as tue. Did he Stois appetite the pointe clearly as Aristotle? We have so far given them the benefit fF the doube, but i took a bit of work (pp. 245-17 above). Was fr effort 3 mistake? ‘We may approach this question by way of another, ‘The dsadvanege oflustating ordinary language sigh-inferences by Toations ofthe Form "y Because pi that ‘because’ can ineodce considerations (of causality. explanation, etc) other thin the putely inferential, Aristode i not misled, ba for caity we might prefer a location such as ‘Since p,q" which presents 9 as inferred From p with nothing farther ether stated or implied. Adit 0 happens that the Stoics also take the trouble to. divsingish hetveen “Since p, gang becuse p The following is reported from the Artof Dsl of one Cins (DL. vi 71-4)"Since pss trae if and ony if (4) Ip then dite, (2) ps tre; 4 beetse str sand only if condition (1) andl (2) hold and also () “lg then ps fle (the idea, presumably, that because” imps an ty fi te ree hei ‘ hein can a ony a a ee The ois of nonsdective inftene 239 explanatory dependence of g on p 3 relation which ought to be symmetrical = cE Ar An. Pr'B 16), Now. has this contrast Snything to do withthe study of signs? Would a Seoie be wiling ince p, 8 the coma representation of 2 sign eStocs clearly wou, atleast for purposes of iscusson, iF iti Stocs with owhom Phiodemus is debating in de Sign, fox ‘Since p, 43 the most common form of sgh-nferenceslustrated ters” Sextus sna help om the pont, sine im fet he never spells fut an example filly enough to heed ether because’ or ine” of ny other inferential particle (fp. 215-17 above). Let us be ‘tious therefore, For the moment let i be mere hypothesis fof our imagination that Stoic would want to accept “Since > a the canonical representation of 2 sigasinfrence. Then, Tinks we are hound to observe thatthe Crinis analysis teats ‘Since. 36 single complex proposition, ale subeonditional (paranménmeno)? To sserts subcondisonal is equivalent 10 sorting» conjuction ofthe form 'p and fp then 4 and 2 man tho docs that has not yet carried out an deen, No inference lakes place unless he proceeds to ase ‘So much the worse, you may si fOr the suggestion hat the subcondtional hae anything to do wih sign-infeenees. But let = te no ce sn ey a eldinnagination persevere a while longer. One solution which twene tieth-cenury experience might offer our imaginary Soi, could he have foresen it, is. 1 dntincion between the content Of at serion sie set and what presupposes Instead of holding, ‘vith Cen, chat Since p,q asses the conjoint uth of" and TE prthen the right anrwer i might be suggested, is that "Since p, asserts the cut of'Irp then othe prenppson tha the ath ‘Of pis already given, The apesker who uses the word sinc” father than is (understood to be) presuppoting~ so that he does noe nove need to arsrt = tha the cise where 7” fle ‘ued out Would some such account as this ie Sine p to be ehe Canonical representation of a sign-infeence? ithe question cems anachronistic, demanding more promis ‘han even Stoic wisdom can aspire to, may nonetheless open out tyes fo one of those divergences terminology and formltion ‘which Jacques Brunschorig has aught us to sce ae sgn thatthe birth of important logical notions ra slow proces of refinement, rota sudden emergence of fully armed wisdom. Brunschwig Acer sn Sextus three distinc, progressively refined accounts fof proot% I ean now reveal that they are preceded by three Aistncr, progressively refine attempts to formulate a definition ff sign to match, To see tis, we must attend cael (following Beumchwig's example) to differences between the PH and te Mt versions of the Stoic thoy. ‘A sig sa proposition forming the antecedent (athigounenn) in a sound conditional, being revelatory ofthe consequent.” So bogine the technical section at M vn ays. The next matter ist select one of the competing sets of truthconditions for the conditional: the Philonian conditions (material iampliaion) are selected, without argument. These Being stout, they yield three {ass in which the conditional i sound (19 T, FF > 7). Bt reference back othe inital definition cll us that sign ought to bea uth which establishes, reveals, another ath (249) cet gael She gna Te Snel armel yk hited me ‘The origins of omdeduaive infrence 226 ‘So when itis sid that a sign is proposition forming the antecedent (Aahégeumenon) in a sound conditional, we shal ove 0 terond (dite kouin) this inthe marower meaning. the antecedent (kithégounenon) proposition in a conditional which both bine with + truth and ends wich a tut (2) The fall detniton 5 finally reached by dividing the conditional so spec fed into the case where the antecedent (howe) is revelatory fof the comequent and the cae where if ix not (250-1) ~ the technical Stoic definition of sign unlike che Anstotelian, has an epistemic component, The antecedent mast be evident and must fe us knowledge ofa non-evident consequent, whether this be fhe mon-evidence of things that ate by mature unobservable (indicative sign) o- that of chings that are temporarily Beyond the reach of observation (commemorative sig). ‘We ate now ready to draw the conchston’ ‘A sig, therefore, rst nor only be the antecedent thigenmensn) in a sound con ‘ional, hats (2), one which both begins with 3 uth nd end ‘with a euth bat must also possess amature such as to reveal the ‘consequent (252) The comment we aall have to understand” Sn "that, together with the backoreferncesa¢ 248, 29, 250 [ive the game away. This sa definition by division designed to clucdateand make more precise a pre-existing definition, in very rch the syle chat Bch revealed inthe vit discasson 6f proot The tenminology confi ithe original definition tne ehégumonan for antecedent, while the commentary on ses the standard term ‘higounonon Janigek’s index shows that ‘athigounenon occurs in Sexts only sshen he Hs reporting oF refering directly to this very definition (AVN 245,248 350,286, 268, 271, 272, 265, 269)Ieremsins to comment on the choice ofthe Philonian conditions for the soundness ofthe sign-condiional. The choice ls presented asa deliberate choice from the many available (24s: ef M vt «28 fn the list of four sets of soundness conditions at PHY w 10-1), ‘The intention must sueely be to fix the minimal, mose general conditions fr signhood, 0 a 9 cover ‘commemorative’ 3x wel 435 “indieative™ signs (ef pp. 20%~14 above). For the examples in ‘his indubtably Stoic dncussion include two (scae-wvound, heat ppuneiuing-desth, ase-s) which ave paradigmatic for the cone ‘memorative sig (PH 102; M vn £53), a8 Well asthe indicative silk example (283). ‘When we tr tothe definition of sign at PH n 101 and 104 we Sights bird of greater arty stil: sign s2 proposition sehich = ‘prokuhégoumenon in sound conditional, Being tevelatary of the omaequent.™ The term ‘pokathegrmenon” evidently presppors “taagornenon, So thatthe PH definition, ike the M definition by Aisin, s posterior tothe original definition, ‘Prolarhgounenen means, we ate old (PH 11 106,115). "The antecedent (egonoon) in conditional which both begins with a trath and ends with 3 tut In other word, the PH definition encapsulates the results ‘of Mus clarifying work: the tath ofthe antecedent ui nto the definition of sign from the outset. Where the M vit division laboured to embed the sign-proposition na condisiona satisfying just those conditions that Crintsasocates with Since py the PH 1 account handles the same material wih noticeably greater sssurane and tes ell up in a newly appeopiated tre of a ‘But why choose the prefix ro for this erminological innova tion? “Kathigoumenon (ht. leading classe’ already sncates the oposition which comes fst in the conditional and "guides" you 4 The mk cal ks inne pe Rho Sey, Rae ye yu Ba a ann he ans Sea be see ‘BE ces uly pace ign kc age toe 7 arc Dis eg ads happens as So wai he na Somos wp. bt oon a tens SReTEe sort hy ek ess Woe a ne aT The origin af won dedusive inirene to the proposition which fllows.”" The only thing. for ‘pre’ to 2M is tha che th of the antecedent precedes the conditional ~ jst as knowledge of the uth of the antecedent supposed 10 precede any ure of the conditional to gain knowledge of the Eonseqient (PHT 1115-20) This suggestion i specalave. "Pn aheeoumcnon may be no more than a erm borrowed from a verb of ordinary language fora technial purpose. But ithe borrowing floes bring something of 3 ordinary Language meaning with the addition of ‘pr can only serve to emphasise that the fntecedent i ke aide leading fom ahead.” We are, i scems 20 fe, lose to the thought that 4sign-condtional preeppots the truth ofits antecedent. Ana this would bea additional mote for turing to Since p aan economical and appropriate expression forthe sgo-inferences on which so mich analytical effort hasbeen spent.® We may ad tha the diference berween iP and “since fad been unde discussion since Theophrasss, whose explanation was that people se “since” when the antecedent not only true butako evident and undisputed." In modern terms, we might Say that Theophrastus has since’ contributing to speaker's meaning, CCrinis to sentence meaning (euth-conditions), whe extending the PH presuppostional analysis to "Since p. 4° would connect it with pragmatics. Ie seems clear that ‘since i the focus of a problem. We se that ce problem is when we se where al Uiree ofthese analyses go wrong. Quite simply, none of them catches the argumentative force which makes ‘Since p, the expression ofan inference. All three inalyser leave q so fir umasserted, ‘They locate the dilTerence as 7 Seal mts wise te a a Sh cote ee ‘wc ic a abs apt mS en ea * pen di fe on Seon eagerbetween “Since p, 4° and ‘Ip, in the fice thatthe antecedent of the formers somehow asserted as tue, ad they fss about ro find 2 way (speaker's smpliation, conjunction, presupposition) in ‘hich the antecedent of a conditional can acquiteaserore force Buc this is both insicen, as leaving 9 unarsred, ant doomed to fllre, because there is no wai which she antecedent of 3 ‘ondtonsl ean acquire ascrtorc force, The mistake, however, Hes notin thinking hat "pn ‘Since p,q asserted, whichis correct The mistake st thik that Since pt condtional of any shape ge form, when sin set an expression ofthe inference ‘pesog® “This mistake, if the Stoics made it, is of far-reaching sgnti= ‘ance, Itmay scm outrageous to sugacs such an caro, given that the distinction betwen argument and condtonal lies atthe ver¥ for of Stoic logic (p. 210 above), Bota distinction sich is ite fear in one type of conext may get lost in another. and ‘unforeunately, the hypothesis thatthe Stok saw a signinfrence 25 somehow more ikea conditional proposition than an argent ‘would explain quite alot more than the uneasy dealings we have just surveyed v (0) The point about ‘Since p, andy because p' for which Criss proposes the same syle of analysis) Was that they have the surface 1s hotest tee te ay Sede Ti idan ed ‘wine es The vgias of nodose infione 235 grammar ofa single complete statement. Hence the tempation To construe the aa spel sore of conionsl. But the other sie ‘ofthat con is that, i You yield to the tempeaton, you then face the problem of wha say about Arto’ loetion For p' oF the equivalent ‘p, so 4) oF any other where surface grammar suggests we have two distinc statements joined by an inferential parte these do the same job as Since p,q and the laters nat an argument, will you say chat ‘p, so qi not an atgument citer Stoic orthodoxy ss that indeed itis ot. One premiss does aat rake an argument (Mv 443; PH 167; Alex. in Top. Stall it ‘An, Dr 7st; Apulia, de at 18g.10-23). Why not? Te follows immediately feom the Stoie definition of argument (logan) an what is consracted From premises (plural) and conch Sion (o sess ot ema ka pips, Met 301; cl PH 135, Cris aud DL. v7) that someane who sysp 30 has not yet constructed an argument. Accordingly, the orthodox view, x Sextus express it tht here are’no one-premiss arguments, while» hereal view championed by Anuipater maintains that ‘one-premise arguments can be constructed (hat smiths, Avan 443). Examples wo illostate Angpater’s poston include “You sce, 30 you are ave’ (Apuleis), "You ate brexthing, 0 you are alive "is day. 30 ab is Hight (Alesander). These are Aistinguished.” both by the definition of argument and by ‘Alexander (An. Pr23-2}~), fom the category of ‘unmethod tally conclasive™ arguments ike “ais equal to; b is equal to 6 therefore, is equal to. ¢, that, abel argomenss with plural premises but requiring supplementation t0 get them into the nomic form of a Stoic allogsm™ here the addition of the ondiional premise I i equal to b and bis equal to then ais ‘equal to ¢ i neded, nok to make the argument vali, sil lest ‘ke iam argument, but, a the terminology shows, to make i “mechodial’, se. formally 5 well as deduceely valid Se eed aa Spare Cie Dy 11a) Aer ‘porn ah um, ya copa pen ‘Soon dee ste Om ete bee a Sc ip 1 She pate me ede a236 Mors muRnvEAT [Nor are Antipate’s examples like the Aristotelian enthymeme, arguments which are invalid without supplementation (Alexan= der, in Top. 9.91, does bring inthe entaymeme, but aa distnet item for comparison, for common grousd to bach parties the dispute the they ate examples whete 4 fllows of necessity feom p (Alex in An. Pro 1726-1841 in Top. 9.$-8). ust this, indced, sets the problem Alexander argues againse Ancipater that on though es ight falls of nest from Tei day 8s not, fedundan to addi is day, tight, and one must 24d to rake an argoment which i nor deficient but complete (on Top. agro. pla conco Apsley) Bue ste ext premiss is nor needed to ensure thatthe conclusion holds of necessity, the ‘only further thing it can be needed for i to ensure that the onchison holds of aecessiey beans the premisses hol 3 com= plete argument rouiree not only premises such dat the conelt= Son holds of necessity but, in adton, premisses such thatthe eonclsion holds of nevessiy becuse the premises hol where the "beens" explains not why the face 650 but why one must cope that the fact 50 (Alex. in A. Pr 17. 23>4;Phlop. in An Dy 3548), Thos what the Stoic onthodoxy denis that, given snan who is brething, hes here shown te ave, Le jus ‘inne of he brute fac tha he is Breathing. Rather. hes shown alive by bis breathing taken in conjunction wath the face hae he is breathing, he ss ave, ie withthe fact chat there is 2 connexion between bis breathing and hs being ave” Wh ee ‘conditional adds i a explicit mention of the consexion;"conne- 19, Wien Asner wy sh ne com ina 8 4 ble & oe ma § eto the ong i nth seepage mre ee Speake lh Saree La ent hie in of eeany y eepemonr «lap ee ie, ‘The ovis of non deductive infiense 237 sion, indeed, is the literal meaning ofthe standard Stoic ter for “eondional, samcnmenon (Lain cones) ‘We have sen that the elim that ex day, sts ight is nota onmplete argument andthe clin that "Its day, it igh ies fay; thetefore i is light suffers from redundancy are the vo ‘opposite sides of a single dispote; which explains why, when Sextus mentions the dispute, it sto dragoon Antpaterinto bearing witness to the scepie aeceation that even the fist indemoor strable hat leading igh of Stoic logic, s guy of redundancy. Buc i we wane co know whst Aneipters examples ate on the orthodox view that dey ae deficint a8 arguments, the only assistance we get is Alexander’ statement (in An. Pr 1725-33) that the concept of following necessarily fom is wider than the concept of fallowing sllogisiclly fom, Thatch fe Binds in the Aristorelian work heis commenting pom An. Pr A 33). Not So the illustration he gives ofthe point, nanily, that i sound ‘onion such as “Its day, is light the consent fills recess att lagi fom he antecedent. This suey is to fuse argument and conditonal statement inthe most distros ‘manner. One could find no beter illustration ofthe dangers ofthe pereasive Stoic habit of using the notion of following or couse nee (sloutcn, hepestha) to cover both the elton of the consequent of a conditional tot antecedent ad the relation of the onclson of an argument to its premises. Ii indeed hard to See, given this fasion ofp, 0 q with Ip. cheng ovat Ie day, tis light can add tos day, seis ight the ation not redundant, as orthodoxy insist, ican only be becuse, just row suggested, it helps 0 exhibit in perpiovous form the connexion betwoon the material already given “This rings us back to signs It has emerged thatthe Stoic 2 Spite, np mato ea Chey ‘pend sme ftp seo ‘Say i nwo pt pi wt iy eckast Mr. puRNYEAY logician will think of the eechaical reconstruction of a sign inference, notin the Aritoelan way a8 iling oat am enaymema- ‘ie argument, but as making a9 argument whete stil speaking there was none before, only the materials for one. Regardless of the Tocution he stat from, bei ofthe form Fp, then fo Since ined ot'p, 404) the materials he has to Wark wih are invariably 2 Conditional and its anecedent the later having Been asserted in ‘conjunction with the conditional or somehow presupposed 35 true. So ineariably the aggumene comes out a: am argument in modus ponons, Mota ponensrequtes, a8 we sad carer (P. 215 hove); thatthe sign proposition be able to be umcondiconally Ssered, but pethaps we ean understand now why it was hard svork in sttion Il so extricate Stoic signs from the conditional Tora of expression, Where Aristotle supplies a universal genet- slisation sb ex the Stic logician finds is conditional already present in the ordinary language location from which he stars. In this sens, che gap Between the technical and the non-technical levee sls, hence less evident, than st was wth Aristo Tt nother sense, however, the gap is greater, for it ithe gap between assertion and argument (2) A carious confirmation ofthis diagnosis may be found in che Sroic approach to physiognomics. Anstote, as we saw, took Physioyomics co be a matter of empitical investigation and Inference, Cleanthes discerned instanly, tom che manner of his Shecre, that 2 rough-looking fellow was in fact 3 homosexual ‘leminae, thereby vindicating Zino’ lai chat man's charac fer ean be grasped (ktlin) from his appearance (DL vit 173) {sus ape that the story whether Facto Beton, i eminently phisible. People often eam tll ast by looking. The question is tocs this “ellng’ have an inferential basi, even fi tt one that ‘he person cos Formate? T want to suggest thc the Stoic nse to this question i "Yes anxd no" Recall hit one ofthe functions of the “indiative’ sign fo give ue information about other people's mental sates {blushing i sgn of shame and in general bodily movements of the soul within). We reason (lizometha), writs Sextus (M M0 1). Wht proces sich movement ay these power within the person's body’. Vet pethapein a way thisishardlyeemoning 3 All. oe Sextus ako wits thatthe indieative sign as a pecliat yy ‘The eign of momeluv nfrence 229 ature uch that practically speaks right outloud t signily the presence of soul (bid 154: ct PH 1 101). (Chaplin silent ove would be an exelent aston.) The logic content of this characterisation will concern us later. In the present context the story about Cleanthes suggests tht we might connect with 2 surprising doctine of Cheysippas that both feclings hike pain oF feat and virtues or vices of character can be pacived alongwith people's appearance (Pia. Stoic rp. togath ck Comm no esac). Teis noe supposed that everyone equally good 3¢ this kind of seeing!” ~ Cleanthes nearly filed he challenge, whe theres independent evidence dat with pestce ad abst the Sige cn tel differences, betwcen two eggs for example, thatthe lunttored eye cannot se (Cie Ae w 7), But the Very concept fof knowledgeable seeing scoms to override the distinction ietween drs bservation and inference which for many plilo- sophers isthe epistemological analogue tothe distinction between dttetion snd sogument. Teis tue and important forthe overall assessment of toi, that various modern philosophers have made 2 strong case for recommending that we shouldbe sceptical about the epistemolo- teil diinctiom between direct observation and inference, where ‘cher minds are concemed int asa short-cut pregnancy test, But the commendation dacs not eatry over into lie the dintion hetween assertion and argument i fundamental and unassailable pine th boing Po repeat per tents toch Etre {ida an ait iin ty ee CahUnfortunately, i all too characteristic ofthe Stoic east of mind to teat the evo cases a parallel, oF even {So deep and inradiable isthe tendency to fuse epistemology and logic) 3s to aspects ofa Single ssue. Coin funher, 8 pps not too bold to suggest that both che Epicresn a che Sov eistemologics area perfect tnateh or model for their respective universes. Eplutcan epte: mology i inferential though and through. starcing with ineren= fal and combinatorial operations on mental images ~ the “atom, ‘of the mind. Ini worlds away Flom the Stoic emphasis on seca Sid grasping conncted wholes. For the Sto. the wine andthe invisible ae connected in 3 thoroughgoing, organic unity gove ‘med by the cosmic reason. Man ty par ofthe system, hence eovidenially endowed with reson. We come into harinony with ature when reason fines with sonseperception 10 grasp the Conexion (amommenon) of the visible wich the iavisile as 3 Unitary whole. To begin with the connexions may bea matter of Inferential reasoning but we can arn to se ther ab natualy at ‘we se fear ina man’s fice, “God has broughe man nto the world to bea spectator af imei aed of bis works, and nt merely a spectator but also an interpreter (Epic. Diss 16.19). (3) The inses explored under (1) and (2) pertain almost exclusive- Ty to the ‘indicative’ sign, whose conditional expression states 3 rocesarytrth, We mst now Broaden the snguty 4 take inthe ‘commemorative’ sigh, picking up ftom the dlagnoss reached at the end ofthe discussion under (I) ip. 238 above), For we haves teo:meation the most important consequence of al. Moss ponent anguments are akways formally vali. So if, for the reasons give, in the tole reconstruction sigirinferencsinvati- {bly come out 5 arguments medhs pores, thee wil be no room for invalid forms in thee logic of evidence. Try Anstode’s oid crea ed a The orgs f mndductiveinfrnce 290 inference fram sllowness to pregnancy. “Since she sallow, sheis preymant, Stoiclly construed. invokes the Wrong conditional ‘Since shi pregnant, she i sllow” gets the inference the wrong sway round. The conditional which would apply the empiscal content of Adstoue's ‘AI who ate pregnant are sallow” to the ‘Stoic econsractionis'lfshe is pregnant, she issllow’, but fone sels thin 3 the bans for inferring he nner condition Fron the Colour of her face, one must be prepared to cleate an argument ‘which cannot be made formally val" To people who thinkin the manner Ihave been tying £0 describe, the idea of non Sedative logics bound to seem absurd. Om thei analysis, signs fare conclusive (the reconstruction isa formally vali argument in modus pon), tek, suficent for knowledge, or they are ‘signe sail, The Stoics are of course aware that in ordinary life we are prepared to call something 1 sgn which is aot in this way Conclusive. Bue they remain unimpressed: “What can be more Sbsurd than co say, "This sige (quo) or proo eumentum) o ‘thi, and I therefore fllow it, but could be hae what signex iseither ise o nothing tal?” (Cie. Ae. 36) A rationale fortis radical sounding claim can be constracted by putting together + passage fom Sextus (AI vit 201-2) with a passage fom Phic Todemus (Signs 11-20) s follows: (0) Take ny sign S which ‘common! (ino) toto things X and Y ~for example, 2 mian’s fll tom wealth to poverty might Ibe evidence of his having lived ie of dsipation, but might equally be evidence of is having met with disaster ase (S. X.Y fe repeatable event-ypes, not rks), Considered in abstraction {rom the circumeences of particular case, Sis no more evidential (FX thin is of Y. That unc should be uncontroversial Bat i) ‘uppore that Ys or involves the absence of X. Then S sno more evidential OFX than tis of ot-X, in which cae it fe no relly 2 ‘sgn (f 8) all We only eres tas one Bocas we ae implicitly {elying on farther information about the particular circumstances Which we donot (perhaps could not) formulae Taking this ine ‘edge tots to cover cameras well 39 meee mEaye munnvear ‘of thought finther, one could restore some respec to the tnrigorous everyday us of sign’ by construing it asa three-tem instead of two-term relation: can bea sgn of X fra person A sind sgn of fr person B3F A and B ave learned to respond 10 “ferent features of the cicumstances im whi occurs ~ only thre must in principle be an explanation for thei ability to infer. fonectly we ire to understand, different things from the same "gn Gusta here ie an explanation forthe diferent effets that fire has on diffrent materials, and the groond of the explanation * will ead us tothe sign srcly and peopeey so called.” Be that at may, s0 fara (and i) ae concemed iti tal to appreciate tha, ule the argument requires hat any genuine sig= relation inseaniste an exceptions, crcumstanee-independent fgeneraisition, it dors not require that the genenisaton be self a necesary trith, Philodemus reports his opponents 35 thinking thatthe inference "This man is good beease he cht is unsound Because the generalisation ‘All ich men ae good is fake. He ds not have them ay, i the paste under considerae ‘don, tits unsound because dhe gonetalstion s contingent? {=n tl puree popu comma ge pre ECD AEine gion cst sia sepmms meme 7 icc hy eae a te ga hr oreo Sob teele Sethe dateore tera iusto esr Soke Sk at cn ne aa ae i ee oe “The oign of av debutive infence 333 ‘Thus the rejection of "common signs, the insistence that genuine signs ate peculiar just one sigan, quite compat- ble withthe Stoics having maintained aversion ofthe distinction between indicative and commemorative signs, That divinetion will show up as dilference i che modal seats of the generis tions instantiated by ‘indicative and ‘commemorative sig Conditionals respectively Lets stat fom atypical prof examples. a) He is blushing soe is ashamed (6) "He hay sar, aoe has hada wound’, The Stoic reconstruction formulates eich of these se 4 modes omens segument fom a vingular conditional (3) Ihe slashing he {shamed bu the ist therfore the secon (b) Whe has se Berge tprimolaneenees ita ae Wien ‘Roma LidaPs some eae) emt 4M. rpURNYEAT Iheas had a wound but the ist therefore the second’, The logic Urthe two eximples isthe same, Episteologically, moreover, has one atezoriclpremis which taken to be established Br ebsercation kis the condional premises which differ: in) 3 Som (omen) conditional which states a meessary tr. &) ‘Say he weak Philonian conditional Hence the conditional in (a nlc orcourse entails the corresponding Philonian conditional auniater + generalisation “IF someone blushes, he i ashamed hich con be bnown to be true a prior, on broadly conceptual Jreaunds 2 The condisonal n (8) on the other hand, can only be Enown co hold om the strength of a contingent generalisation Ting scars to wounds which we have antecedent established by observation and memory “Tis, indced, san outcome we could have foreseen long when the Philonan uth conditions for sign-conditionals were ‘etntine ith the epatemi reqivement tha the antecedent be ‘velstory ofthe consequent. [rknowig the truth ofthe antece- {Eon int lead us to knovr ce euch of the consequent hen we shave seedy Know that the conditional 8 sound, andependenty of nowy the tuth-values of ts costents. The Stoxs expressly Mule thse If py then 4” oka sigo-condiional fs an be Enbw 9 be sound by the Philonian enterion simply from its eing evident tht p anc evident that q (A vu 2sont) This aes po peablem with the contol in (a), which is itself necessary, Bo doe ti eath of the antecedent can of tel by its ow matute) ‘eit th the consequent aso i ru (PH 1 10, AF 154) But with () there 0 option bu to say tha its on dhe strength Gran antecedent Know generation that the singular con Siioal is Known 0 be eve ‘Nevertheless, 30 long a the generalisation is indesd a tue exeeptonfee wenelisition, the proposition "He fis had + ‘Sound? il be established conclusively and of necessity ~ the recuuevnscgaae of the Aristotelian ekrion. Infect, the The eins ef nondebuive infence 235 Sic thesis that che only leita eon ia ekcion™ ~ hat the burden of the argument about "common’ and “pecubiar ‘em The upshot shit Stoic logic guarantees 0 Sto epistemol- Se hat the only warrant which one proposiion can confer om ae ache warrant of conclsive proof. Aristotle idea that ‘here ferent grades of evidential suppor: rejected. If one foes aor hive grounds sich make it absoluely necessary that see pas no geounds at all, and had better Keep quit with Radgcinent suspended. Ani this, 36 everybody knows, is exactly ‘ehat the Stoie Sage s meant to do, No daub tae exceptonlss contingent generalisations are hard to ye by, a harder in divination than in medicine, which i SS etson forthe prominence in these discussions of medical and SUnfamedical examples.” But the providenil ordering of the ‘SOL which everything connected ith everything, guaran- eth they ate there ta Be found and that man as the cognitive ‘Sfhetues to fnd them. The Knowledge which can dagnose these ‘its he medial metaphor is zeveahing) belongs to the Stoic Be fea cone se divine (SF tog) The rest of us, wo Sree Sages, ate prone to error! we take something ro be an 2 dhuonfes continent generalisation when its not, oe We MISS STedbvane dierence betwcen two Gcumstances of application Mee Diet ri8, eds But an principle it is possible for 2 (fetnatory generalisation 40 stand 36 certain, and is enough © inet exstene of divination ft happens ost once that we ate Pra onion to ny, Chance cannot have played eve the sightst furtr this prediction’ coming tue’ (Die. 124-8), One sound Pipa Suicient to prove the existence of the art. Mistakes rid hacks, however numerows, no more prove she conta than ey doin dhe fed of medicine or navigation (Div. 124. Afterall, ‘Mhcwe take to ea necessarily re generalisation may’ be wrong se The texdbook example iustraning the Stoic definition of the eave ithanon, something gave bi to something, es Merimother (DL vi 73) i 4 universally quanti’ conditional2x6 BURNYEAT precisely designed to bring home to people the flibility oftheir onceptual intuitions (hey have to be reminded that a bed is not the mother of eg) thes correct the ference in modal stars between trong and werk conditions does noe automatically Bring with i 2 ference in epistemic stats. Phlonian conditionale may be no les certain than snare conditions. Tis is abvious where che Prhilonian conditional is entiled by a sina conditional, oF where antecedent and consequent ate both evident im their own” right (Mit ast), But we have ust confiemed chat itholds equally SF Philonin generations (oniverslly quantified material ‘onditionas)” Wht the modal ference does mean i thatthe ‘etinty inthe two eases fas ifeent ground inthe one case we ‘onsult_our “preconceptions”? sn he other observation and Inemory, We need the second because we are hsman: only God ‘an sce things whole and grasp the entire interconnected sequence fof pan, present and furuce (Cie. Div. 1126-7). Irs only when the Secon! metho to has led ws hat we mae resort to generalise tions that are only “forthe most part tue (Div. 1120) Ths third type of generalisation brings ss fn hack to Arist. "The picture thus fir his boon chat the logic of our resoning is allways dedctve, What varies the souree of the materials fom which our arguments are constructed, Accordingly, the thesis be defended hs been tht fr the mot part good divince is right about what he takes to be a tue exceptonlss generalisation (C Div. 1 35.118, 124-5), That 8 quite differnt fom the more Aisorlan topic drswa, i should be noticed, from Posidonis (uss) of generalisations which themsslves hold only for the rmost pare or about which one can be certain of no more than hat Sate tr alacs ont gine Se oe The ovine of rondedcsive infeme 237 they hold forthe most part. Given a generalisation ofthis hid ‘pe (imagine that 'Sdllow women ae pregnant 8 an example), ‘what i the statis ofthe singular conditional which nseantaes ina particular ete? Does it presuppone that we can formulate, and have elite she kinds of exception to which its lable. p. aut abovel? With what asurance OF probity can we dase the ‘onclision that she pregnant from he allow featuges2"= What tole played in sch inferences by background knowledge ofthe {Geeumstances which the generalisation i applied? Such gues~ tions as these could have led Soi-tained logician to develop 3 non-deductve logic witha diferent, and pethaps more promi- ing structure than Ansoselian logic could provide (¢ p- 231 howe) But sty quite cleae, Tein that all inguiy I eis Aluccton was spumed by the exablishment figures of Stoic hiosophy “The Stois recognise that we voyage through hfe fllowing, for rach of the time and even in the matters most vital to" our Welfare, nothing beter than the pithnon: that which we happen to Find convincing or persuasive. We owe ito providence tha, for the most part. we getaway witht, But tothe extent that we rly fom the pidanon, we are fools, ceatares of uneeason, fling abysmally 10 use correctly the rational Geulies with which Providence has endowed us!" There i no loi 0 be discerned here, only persuasion. And indced, iso the art of persuasion, 3 represented by the tetorc testes of 3 Cicero oF 4 Quintin thar we must look if we want to study the later history of Anstole’s distinction between conclisive and non-conchsive Tue when we compare these works (the most sophisticated Aiscusson is in Quint It ¥ 9) seth Asatode’s Reo all238 sroauanveaT too obvious that thir authors are not logicians. Nor are the Epicurcans much interested in logic (inthe proper sense of formal logic lthough they have some good things to say about what lakes to establish a generalisation om inductive grounds, and they ‘git rightly charge the Stoic with paying insufficient atetion to ontimation dheory."™ All dough the Hellenistic period serious logics the preserve ofthe Stoic establishment the poltial image 4s pethaps not inappropriate), which means, as I have ered to exphin tat the whale masive weight of the Stoic system stood against any farther development of Aristor's pioneering sae. ‘With tht at resalt we have come fl ctl, Hone believes that an adequate philosophy of science must find a place for non- deductive a well for dedotive logic, one wil conclude that 35 logicians. Avstote was a beter fiend to the sciences than Zeno and Chrysippus. IF Arstot's wisdom i these mater. dis fppested int the theron tation, rather than being taken wp tnd developed by philosophers or scientists, a lage share ofthe blame mus es with the authority of Zeno's work Ow Signs (DL v4) and the Stoke tradition generally." "0 Neal ne da de Sane in tara need Be ar ad aml mn 8 OnSigns DAVID SEDLEY 2. Phiodoms, de Sigis With the appearance in 1978 of Philip and Estelle De Lacy's second edition of Philodemus, de Son, one which unlike its predecessors a sed on adequate papyrolgical information! the time is ripe for rencwed discussion of this work and ts place in Helens philosophy. les real tle is Philodemus, On [| ond Sign Irons. The missing word may be phanasa, pressions oF phenomena, "appearance. but this need not concer HOM "Gagner fe Can atin pig Sag ee, ‘NG gel Gud: ola cay adda el ya he ese nh a toe a Ee outer Para ‘eg Satbecause the surviving final part of the book seems concerned purely with signttrence (mein, that i the discovery of hhon-evident truths by means of evident signs.* Writing inthe mid first century 1c the author repors from his distinguished Epicurean master Zeno of Sidon (e155-<78 B.C) the argaments ‘of some contemporary adversirs, followed by Zeno's reply (11-283), There follows a farther brie summary of the Epica Fean position taken from the weidngs of an equally renowned slightly youngee, Epicucean. Demeteis Lacon (38.15-29.19) Finally he reconds an oral contribution on the same topic, prob ably fom 3 thid Epicurean ashore name isnot preserved in the {ext 29.19-38.22). ‘The conventional identification ofthese opponents a Soi has never received the fll defence tha it deserves’ Ie must be 24dmited that none ofthe terminology or philosophical examples fributed zo dhem iso unmistakably Stoieas wo sete dhe matter Nevertheless, I Believe the convention to be correc. It aor philosophical sect were contributing t0 this debate 3 medical School would be the kes candidate, bur chi sued out bythe losing sentence of the book. in which Philodems seems, with lie enshusitem, to defer discussion of medical andlings of the Secs bie Liar 2 Se On Siew 261 topic to subsequent book (38.22-y2), The only contemporary Piblosophical opponente alae to Zeno and Demetris would be the Stoies, the Peripatetic, andthe Academic septics. Uhave i tre erat |i (On Signs 259 concomitant ofa property, £6, [A man, insofar ashe sfotsh, ‘S ucerly unhappy’. The author observes that any ofthese four {types ca undrlcsign-nferences,* and thr not only (1) bu so (3) G) and lu) pick out some kind of necessary connexion. The fevson why the Stoics think that an “insofar as clause must be Incl inthe premis and thatthe ference wil ten Be by the ‘imination meted, is that they have given no thought to the precise senses of insofs a" and to how an insolr at” premis is Sablished#? (34.20-35.4). In fact, he goes on, establishing the recesary connexion ‘which insofae anal its senses marks can fonly be a painstaking empirical mater ~ even for apparengly ‘Setnional properties like man’s morality (5-4-2). “This tooks ke a head-on confomtaion between empiricism and rationals, The Epicureans most have fle hatin chiming thaseienc could work purely by deduction from necessary tuts the Stoss were fling go tach suliient weight tothe inductive ‘ement in the human leaning proces to which cei epistemolo~ iy pa ip serve (SUF 83; Cicero, Ac 031, 30). And one can BeeMihe point of their tfusal to bow to Sioic pressure by feformuliing the morality argument Inco a deductive. wad fone. ether way it rested primarily on the inductive similarity Inethod, and nothing was gained by adopting 3 formulation which Aiguised this fact. “This leads on tothe central problem: why, in spite ofa, do the Epicureuns concede that some sign-inferences, including that tt vod from motion, do sey onthe eimination method (121-14 Te rinig: 3539-7629)? fargen Mau, following 2 suggestion by Daiischs interestingly argues thatthe objece was to tease che Stoes by making the rpugnane inference from mation to vox! Took logically valid even on thei own estenon. And the De oe torn me miso pey SEE sPivtagaaaetgure nan * we oe ‘es iio n= Se er Sree Nal let eg se Hing Gaye eee ain “360 pavIp septey Lacs may seem to lend support 10 thin ies of 2 merely a ston ve he ey cee to hr scion the id IMgumene ened 8 purely ndtv oe (ing 80-98 a BEhiyecy © Bathe tee theory founders when we nae at ater itrenceicading the nonpartisan one rom smoke 0 fea ened on por wth the vond one (627) and i theeore suey beter ook or an nerpretaon which wil wt have Zeno bang tc wo ange on fe nny metodo fe moment and on the eliinaion metod the next without tony oF explain, Pre ofthe soon cn be sen a SeasS63, where we are cold sh the void and reinfrences rst, arcy on te mnsion met and yet exe he cones Eom som empiri obervation of te regular dependence of trian on empey sac ht of make ome nded is ‘ae scr the de Sgt tha the iminaon method ges fe conan om te stn method (bate. ef 308 Toh or a more pewregue langage thi what the former “Ghar heer ps unlocks ey G1) The postion truth beta force here ie moton tees vou goes rough evily bythe cimnation method thanks tthe nature mat impossible excep thonsh empey spaces, Bu hat Sfurtaare sone whih we hav sre nds So 00 ¥en Shore comoversaly I or'Since] tere smoke ete res {uraeed by the naar smoke as he produc of re but once Sams nae can only Ye sme through empire gener te The alleged Epacean wuvering onthe satus of te void Stgument ison frtove page which sem to make indore ae unmskably deserting how we dacover the nature SFmerton ss someting possible without empey apace (at= Si 3535-362), whe thous which ink eto te ciation Inethod ar einige nference'Sice hte is motion heres ‘ood which thar dcovey Hees "When and why do the Epicureane adopt sis method, and how apache ty coneting tie Sox bys doing? The ase the Bat question con Be found at 7-38 be ex propery Eero hc te she monary rms the wld pene SSM Sera bot mens On Sign 261 understood. The anonymous Epicurean criticises the Stocs for Eling to make two lstinctions beeween diferent Kinds of implication. The fist distinction separates cases Where the appa- rent thing i peculiarly connected to the non-appacent thing because of some sort of causal selationship, for example heceuse fhe non-apparent thing isthe materal origin ofthe apparent ching, fiom cases where the basis of inference i resemblance ether direct or analogical. Now che second kind of case exactly matches ‘hose inferences for which, ab we have seen, the Epicuteans insisted on retaining the similarity method of inference. The fst Kind of eat ss expreily cated withthe inferences for which the climinaton method is sanctioned, exemplified elsewhere by those from motion to void and from smoke to fire; and the reason is clearly tht in ll these the premiss sled tothe conclusion not by any resemblance beeween thei subject terms, But because the conclusion is explanatory ofthe remiss. I is evident that such inferences could not go through by the similarity method. But why did the Epicueane choose to assimilate ther to the eliminae tion method in particular, even though, asthe smoke example tives away, che price was a much more hospitable interpretation ‘fat method's scope than the Stoic permite themelver? The faswer, suppose, i hat the simply did not have svalble any method of scent inference apart from similarity and elimina Hon (@E 8.7710). A Stole would have pt the amoke inference, 38 ‘one based on pis, into the clas of pita, thus denying i any scentiSevslcty. But an Epicurean could not have fllowed him fn thar path withoue weakening many basic tenets of atamisn, inluding the proof of void. Besides, the stength of the Epicurean infeence lay inthe supposition thatthe conclusion was sniguely ‘Since fon gene tae 1) tel ab ade ene ea coa nec lescxplnatory ofthe promis, wheres implicston founded on pa Could cam nothing more than regu observed conjunction So the ncesuaryconnesion chins by the climinstion metho vould pally commend sel to thea ey Tet Tt we Spread to ees wre the neesty bard looked Toa hat was itray siping n view ar sre of he boemary Eee Concspeal md empl th "To ogai som of the gromd conceded 0 the Stoic, the Epicuents make ther rane valuation ofthe two stages in such ‘neers gly vor oso. Only tothe ce Sager in wth the similarity metod is employed, do ehey rane {htnume of sigmnferene, stung 9 ea boo), exes iy watbolling trom the stage gover by the eiminaon Method (ners ne) mtrpe hi flows The {Empire process of eablhing che are of motion, ncuding impart oon empry sae, in el gmanene By the smarty methods Stel the many and varied moving thje within ou experience shoe the character of ein SIE me wi emp, mon ie Sec empey space 43-83). The al rence whieh tars $s way und Boson y he clmiaton metho to concude that motion atthe lcorcpie level imps empty Pa {ead} al he more, sigrinfrence, Bt he ets cement © detinstiat asad rhe gon le es bey ae [35 atu opti wd pice ane nae ee coir Ste ea Steet BG ety Sa ge een cinwieesss On Signs 263 ‘of staightforward irene by the simination method shich i sled a this second stage is noe in ea Farther signintorence, Since on is oven i is povwerless to reveal anyehing 4 Erion ou norcomestion Eplcueus is agreed to have atiched particular methodological importance o the ewin principles ofepinrtarsts doa atin tress. choose the almost teal translations “tesiion” and ‘non-contestation’, which provide the convenient cognate verbs “anested' and ‘oncontestad (The verb contest must be thought fof here in the sense bring evidence against) Our only fall account ofthese methods i provided by Sextus Empiricis at M Vi 21I-16i the course of his historical survey of theories concerning the criterion of ruth, I fllows very fall nd acd summary ot [Epicuras’ doctrine that all sensations ate tre (203-10) chal begin with some speculation sbour the historical creden- tial ofthis account, bated om close look at 213-14, which eeads 2s follows [Non-conestation she following (slut) fom hat which i apps feat of the nomapparnt thing pouted and bcheved. For cxampe Epica, sang that dre vid which is wonrappaen, cones tthe through the sevident Fe of maton For oa dos ote ther uit noo Be mation ther sine the moving body woul Ik pce to pars into ar tou of creythings beg al sai ‘Therefore nom apazer thing bled cote by that which ‘Wapjrent, since there's mes Contteton, op the othr bands oath which conflicts with na-contetaton. Fo te elimi ‘Son (anata ofthat which appar by the posing othe non Sppuen hing. For example, she tess that ond des at ei FES pt ce te ge ce See “Sed maton sla sg