5EM]E@W]EEFil

JOURNALOF THE INTEIINATIONAT

ASSOCIATIONFORSEMIOTICSTUDIES

NEVUCDE UASSoCIATIONINTEENATIONALE
DE SEMIOT]OUE

Editor-in-Ch
ieflRddacteur en Chef
TI.{OMASA. SEBEOK

Offprint/Tir6i part

Mouton de Gruyter
[ferfin-NewYork
 Review articfe
Philod,emus"Designis': An important ancient
semiotic debate*

GIOVANNIMANETTI

This book is based on five lectures originally given by Marcello Gigante

at the Colldge de Iirance during the winter of 1985 and subsequently
published under the title La bibliothique de Philoddme et I'dpicureisme
romain (1987), rvith au introduction by Pierre Grimal. The French edition
was revisedand published in ltaly under the title Filodemo in ltalia (1990).
The book presentsan overview of the philosopher and writer Philodemus
of Gadara (circa I lG40 n.c.n.),a follower of Epicurus and proponent of
Epicureanism in Italy in the first century s.c.e,.The work is an outgrowth
of the intense study which has been devoted in recent decades to the
philosophical and p'hilological significanceof Philodemus and his work,
in conjunction with the archaeological discoveries and interpretations
offered by art historians regarding the so-called'Villa of the Papyri'. This
is a villa discovered in Herculaneum, where it lay buried under the lava
and ashesduring the eruption of Vesuvius in 79 c.e.. A large number of
papyri were discovered in the Villa containing works belonging to
Epicurean circles (either works of Epicurus himself, or of his followers,
Philodemus among them). The Villa, with its library of papyri, is
considered to be the center of what was the Epicurean school in ltaly,
headed by Philodemus himself.
The first chaptel of Gigante's book provides a cornparison of the
various interpretations which have been offered of the sculpted decora-
tions which were discovered in the Villa, whose dense array of symbols
seemscertainly to r:efer to tlte world of Epicurean philosophy. Gigante
is especially indebted to the treatment of the problem by Panderrnalis
(1911),who considersthe sculpteddecorationsto be arrangedin opposing
pairs so as to exprerisa symbolic contrast between the Epicureart universe
and everything thirt differs from it, especially the contrast betweeu
the Epicurean universe and the Stoic alternative, and thus between the

tMarcello Gigante, Philcdemusin Italy: The Booksfrom Herculaneum,Lrans.by Dirk Obbink.

Ann Arbor: University of Michigan Press, 1995.

Senriotica| 38- I /4 (2002\.2'19-29'l 003?- I gqR /O?/n | 1R -r'l)?o

280 G. Manetti
Epicurean ideal of the private contemplative life on the one hand and the
ideal of the active public life which was typical of the Stoics. Philodemus,
according to Gigaute (again following Pandermalis), played a prominent
role in determining the artistic tastes of Lucius Calpurnius Piso, the
aristocratic Roman owner of the Villa and a prominent representativeof
Philhellenismin Italy.
Gigante returns to the subject of Philodemus's relations with Piso in
chapters four and five, emphasizing the fact that Philodemus's book
entitled On tlrc Good King According to Homer is dedicated to Piso, fol-
lowing in the tradition of Epicurean teachings regarding the proper
relationship betweenthe monarch and the sage.A famous epigram attrib-
uted to Plrilodemus(PalatineAnthology,l1.44) is also dedicatedto Piso.
Moreover, Gigante establishes a connection between the contents of
Philodemr.rs'spolitical tract and a seriesof sculpted portraits which were
found in the Villa, which Gigante takes to be indicative of the relationship
between Philodemus and his patron Piso, the Villa's owner. Other ele-
ments of Philodemus's biography are analyzed in chapter three, using his
epigrams as a means to trace in outline various aspectsof life.
The heart of Gigante's book, however, is in chapter two, which presents
a masterful treatment of the structure of library which was found in the
Villa. Gigante directly confirms the hypothesis that it was Philodemus
who played a determining role in the library's establishment.According to
Gigante's argument, the library's older holdings were put together outside
Italy, probably in Athens; Philodemus presumably assembledthe library
out of books that had been handed down to him from his master(s)and
which he brought with him to Italy. This original core of the library
included the various books of Epicurus's On Nature, as well as the books
of Demetrius of Laconia, an Epicurean who was a younger contempo-
rary of Philodernus's own teacher, Zeno of Sidon. A subsequent, and
much rnore substantial, body of material was added to the library later
on, including the works which Philodemus himself wrote along with
other books of the Epicurean school, including books written after
Philodemus's death.
Relying on the paleographic studies of G. Cavallo (1983), Gigante
outlines the development of Philodemus's scholarly work. During the
period 75-50 s.c.r., Philodemus was occupied with philosophical histo-
riography. This activity on Philodemus's part was characteristic of the
Epicureirn school at Herculaneum, and continued even after his death.
One of Philodemus'sendeavorsin this field is his Synraxis ton philosophon,
a 'great philosophical textbook', in which the section dedicated to the
school of Epicurus was written not only to provide tlre followers of
Epicurean doctrine with a senseof history but also to present the core
Philodemus' 'De signis' 281
ideas of Epiculeanism to Roman and Italian intellectual society, which
had up until that pcint been shaped by the Academic tradition, of which
Cicero was about to become the key representative.
To this same period can also be assigned the beginnings of
Philodemus's philosophical reflections on the subject of ethics and the
emotions, as well as topics having to do with teachingsthat are formative
for the human spirit and for the sagein particular, as representedby such
volumes as On Music, On Rhetoric, and On Poems.These volumes show
that Philodemus off:red an interpretation of Epicureanism that was both
personal and critical, given that Epicureanism is well known for its lack of
interest in humanistic cultural pursuits, even regarding them as a possible
obstacle in the pursuit of happiness. Around the year 50 s.c.E.we find
Philodemus's major: work on the schematizttion of moral concepts,
entitled On the Yicet'and tlrc Virtues Juxtaposed, and the years subsequent
to 50 n.c.r. feature writings on Epicurean theology, which were also rtsed
by Cicero. To this same period, but at the end of Philodemus's philoso-
plrical itinerary, Gigante assigns the three volumes Ethica Con'rparetti,
On Death, and On Signs.
On Signs
The treatiseOn Signsplays a fundamental role in the reconstructionof
ancient semiotic thought becauseit presentsa wide-ranging, sophisticated
set of philosophical speculationsconcerning sign-inference.Gigante con-
sidersthe treatise to be a book on logic, at least in the ancient senseof the
word, which is much bloader than the modern sense,although On Sigtts
alsocontainsa numlter of topicsthat continue to form part of logic today.
Gigante emphasizer;that this was probably not the only book that
Philodemus dedicat,:d to the study of logic (as M. Capasso has shown,
1980),and there is evidencethat other papyrus scrolls might have been on
logicaltopics.In an1'case,as Giganterightly observes,On Signs'isa work
of considerable rnaturity zrndboth theoretical and historical consistency'
(42). Based on pale,rgraphic evidencediscussedby Gigante, the book is
supposed to date to the year 40 n.c.E.,rather than to 54 s.c s. (as earlier
suggestedby Philippson).
On Signsis a unique text in the sensethat among the topics of logic
that it considers, it provicles a specific account of the way in which the
Epicureansunderstood the plocess of sign-inference,based on the princi-
ple of analogy. In addition to logical themes,it is not surprisingto find
here many of the topics that were hotly debated within Epicurean
philosophy, such as prolepsis, memory, and the gnoseology of the gods.
282 G. Manetti
In this regard, the treatise On Signs has an exceptional importance, as
Gigante emphasizes(42), becauseit provides an explicit acknowledgment
of the theory of the deviation of the atoms, parenklisis (XXXVI.I3),
which was not, according to Philodemus, a sufficient basis on which to
explain Chance or Volition. This ancient evidence plays a key role in
refuting the hypothesisformulated by Jean Paul Dumont, which held that
tlre theory of parenklisis could not be attributed directly to Epicurus, but
was supposedto be instead an invention oflate Epicurean sources,such as
Plutarch. Relying on this evidencein Philodemus, Gigante still maintains
that Usener was not correct in detecting a referenceto parenklisis in the
Epistle to Herodotus ($ 53), while holding out the possibility that 'we may
expect to find it one day in the remains of Epicurus's On Nature' (43).
Plrilodemus's On Signseven attracted the attention of Charles Sanders
Peirce,the American philosopher who, beginning with his first Lecture of
February-March of 1865 (seePeirce 1982), evolved a general scienceof
sigrrsunderstood as a scienceof formal logic. It was a readingof On Signs
that suggestedto him the idea of an autonomous scienceof signs, semi-
otics, as well as a name for inference specifically by signs, semiosis.This
took place in l879-80, when Peircewas supervisingthe doctoral thesis of
his student Alan Marquand on 'The Logic of the Epicureans', including a
translation of Philodemus'streatise.
The first critical edition of the papyrus contaiuing this work of
Philodemus (PHerc. 1065) was published by T. Gomperz, who signifi-
cantly gave it the title Philodem Uber Induktions-schlisse (1365). The
current standard edition is the text published by Phillip and Estelle Allen
De Lacy, On Methods of Inference(1978). This is a revised version of an
earlier edition (1941), now improved by the contributions of Marcello
Gigante who, together with Francesca Longo Auricchio and Adele
Tepedino Guerra, made an inspection of the papyrus with microscopic
binoculars, allowing them to recover many previously unrecognized
readings of the text.
The length of the treatise is unknown, and the part of the text that has
been preservedis limited to the concluding portion of the papyrus scroll.
The original title of the treatise is a n-ratterof debate, but based on a
discoverymade by Daniel Delattre (reportedby Asmis 1996:158and n. 2)
it seenrsmore likely that the title should be On Signsand Sign Inferences
(Peri semeionkai semei6seon). Daniel Delattre has also discoveredthat this
is likely to be the third book of the treatise (basedon a Greek gamma that
has been detected under the title written on the papyrus), and that the
treatisecontainedat leastone additional book, sincethe conclusionof the
book that we possesspromisesa discussionof medical doctors and their
ideas.
Philodemus' 'De signis' 283
Given, howr:ver, the asystematic nature of the materials that are
preserved,Jonathan Barnes (1988: 92) has argued that On Srgnsis not so
much a methodological treatise as it is a sort of book of annotatiorrs,
perhaps for Philodemus's private use, providing an account of the
Epicurean posit.ionwith regard to inference by signs. This was a topic of
great concern among the post-Aristotelianphilosophicalschools,in that
semiotic inference constituted one of the cardinal elements of scientific
procedure.
The treatise is subdivided into four parts, which vary ir.rlength but deal
with the same topic" in that they all contain responsesto criticisms made
against the Epicurean theory ofinference by signs.The first part contains
Philodemus'sown account of the ideas of his teacher Zeno of Sidon
(155-75 n.c.E.).The second part continues to treat the ideas of Zeno
as expounded by one of his students, Bromius. The third part reports
the ideas o[ Dernetriusof Laconia, a contemporary, perhaps somewhat
younger, of Zeno, while the fourth part continues with a series of
responseson thc part of an Epicurean author whose name is not attested
in the text that has t,eenpreserved,although many have thought that they
too belong to Demctrius of Laconia.
These various representatives of Epicurean philosophy respond to the
objections that have been advanced against them. The commentators
usually assumedthat their opponentswere Stoic philosophers,relying on
referencesto a certain Dionysius in the text, who is regularly identified as
Dionysius of Cyrene, a student of Diogenesof Babylonia and Antipater
of Tarsus. Above all the text suggeststhat this group of opponeuts formed
a homogenous group, unified by their use of the method of anaskeui.
Although there is no direct evidence that the method of anaslceui was
associated with the Stoics, this traditional explanation is defended by
David Sedley (1982; 241), who maintains that these critics were Stoic
contemporaries of Fhilodemtrs who were themselvesproposing ideas that
differed from those of earlier Stoic philosophers.
More recentl l, Jc'nat hanBar nes ( 1988: 94, n. 18) has at t em pt ed t o
minimize any such assumptionsabout the identity of thesecritics. The
one assertion that he does make in this regard is that if it rvere in fact
Dionysius the Stoic who is cited in the text, then it would make sensethat
all of the critics ar,: Stoics, although their arguments might also have
benefitted from the ideas of other groups, such as the Acadenic or
Peripateticphilosophers.
Still more recently Elizabeth Asmis (1996: 159-160) has questioned
even the identity of Dionysiuri,and has argued insteadthat he is actually
an Academic philosopher. According to Asmis, it makes more sense
generallythat an attack againstthe Epicureanswould havecome from the
284 G. Manetti
Academic philosophers (a hypothesis she had already suggested in
1984: 198),given that they aligned themselvesagainst non-evidentsigns
in generaland especiallyagainst Epicurean inferenceby signs.As evidence
in support of this hypothesis, Asmis mentions the criticisms against
Epicurean inference launched by the Academic Cotta in Cicero's
De natura deorttm. which match the criticisms refuted in Philodemus's
On Signs.According to Asmis, the Academic philosophers were in any
case resporlsible for mobilizing a unified opposition against the Epi-
cureans, drawing allies from various philosophical schools: 'I conclude,
therefore, that the arguments against Epicurean induction in On Signs
were shaped in the Academic school. lt is plausible that Carneadesfirst
gave systematic form to these arguments. The Academic objections are
indebted to the Stoics and others, just as the Epicurean answers are
indebted to Stoics and others. It would not be at all surprising if Stoics
such as Chrysippus also attacked Epicurean induction. But i[ they did,
their arguments were reshapedby the Academics. The Academics may be
seen as the rnaster tacticians who led an army of opponents against the
Epicureans.The troops under their command include Stoics, Empiricists,
and others' (1996: 179).
Semiotic inference
Tlre discussion of inference by signs at the center of debate in On Signs
involves a host of important elements from semiotics, logic, and epis-
temology, which are variously intertwined, as are the positions of the
contenders in the debate. In one of the most interesting aspectsof his
contribution on this subject,Jonathan Barnes (1988) has shown that it
is possible to take the various phases of the prolonged debate on signs
between the Epicureans and their opponents and to consider each one
separately. There are basically six distinct phases, as follows (and as
alluded to in VII.5-7): (i) an account of the theory of Epicurean inference
by signs; (ii) the objections to this theory advanced by their opponents;
(iii) the resporlsesof the Epicureans to these objections; (iv) the technical
objections o[ Dionysius to the Epicurean responses;(v) the counter-
objections to Dionysius made by Zeno of Sidon, Philodemus's teacher;
and (vi) Philodemus's own version of Zeno's approach. Even though
this exact sequence,with its precisecast ofcharacters, applies only to the
first of lhe four parts of On Signs,its inventory of positions and counter-
positions is valid for the entire treatise. Following Barnes, we can
then proceed to analyze the contents of this series of arguments and
counter-argulnents.
Philodenus' 'De sisnis' 285
The Epicurean position
The Epicurean theory of inferencebeginswith a definition of infelenceitself.
A 'sign-inference',or semeiosis,is a process which allows one to pass
(metabafircin)from a sign to its meaning. This corresponds to a passage
from that which is known to that which is unknown or, lnore specifically,
from that which is evident to our perception to that which is hidden.
The logical fonn in which the semiotic inference is formulated is
the paraconditional (parasynemmdnon),'Sir-rce p, q' or, in other words,
'since (epel) p, then q', which implies, unlike the simple conditional
(synemmdnon),the truth of 'p' (Burnyeat 1982: 218-224). According to
Sedley (1982: 243). this formulation, typical of On Signs,is arr inrprove-
ment on the classicStoic fiormula, in which the sign is the antecedentin a
conditional statemeutof the type 'If p, then q', in which case it is not
possible to include any indication regarding the truth of the antecedent,
as it must be insofar as it is effectivelyfunctioning as a sign. In general this
is an abbreviated form of the modusponens,whose complete formulation
is 'If the first, then the second; but the first; therefore, the second'.
As ElizabethAsmis notes (1996: 157),it is interestingthat in Epicurus
there is no sort of argulnent which supports the conditional form of the
inference.Two generationsafter Epicurus, Zeno of Sidon and his colleagues
seemto formulate the inferential statementin the manner that we have just
seen.According to Asmis this leadsone to think that they had appropriated
their opponents' definition of the sign.
The sign (semefon) is that from which the inference begins, and
Philodemus generally refels to it with one of the following terms:
to phain6menon 'l6at which appears', td phaner6n 'that which is made
clear', to enarghis 'that which is evident', but above all with the term
to par'hemfir,'that which is present to ns'. Barnes (1988:96) maintains
that this final formulation is the most genuinely correct, in that it includes
not only that which is revealed to our senses,but also the personal
experience,the experience of others, and that which is apprehended by
means of testing c'r argument, while the other expressionsare lirnited to
direct and individrral perception.
The terms most frequently used to indicate that to which the inference
leads are td cidelon and tri aphands. The obscure object need not be
completely obscure,but can instead be something that is simply outside of
our own experierrce.Inferences often lead from a pheuomenot.r th:rt
can be apprehencled in otrr experience to 'all' analogotts phenomena
'everywhere'. The follorving are some typical examples: 'Since men in
our experience are mortal, then all men are mortal' (1I.26-28); 'Since
animals in our experience are mortal, then all animals which exist in
286 G. Manetti
Britain are mortal' (V.34-36). The general form of this semiotic inference
is described by Barnes as follows: 'Since all the Ks in our experience
are F, then Ks elsewhere/anywhereare F' (1988: 97). This formula
accounts for all the cases in which inference begins with a group of
individual observationsof a speciesto yield conclusionsabout the species
in general, but it does not seem to cover the casesin which the inference
moves from particular individuals to other particular individuals or from
one type of body to another of the same type (XVIII.23*XlX.4), as in the
following examples:'If Plato is a man, Socratestoo is a man'(XII.l9*21);
'lt is impossibleto conceivethat Epicurusis a man and that Metrodorus is
not a man' (XIV.25-27).
Leaving aside for the moment this aspectof the problem, let us proceed
to the crucial question:under what conditions is the semiotic inference
as conceived by the Epicureans valid? The answer, which is addressed
continually throughout the treatise,is that it is valid under the condi-
tion of 'similarity' which exists between the elements involved in the
inferential process,which is to say that the inferenceproceedsby means of
similarity, katd tdn hontoi|teta tr61tos,or that it makes a transition based
on similarity, he kath'lrcmoi6teta metdbasis(Asmis 1996l.162).
According to this model, then, once it is establishedthat two things or
two classesof things are similar to one another, one can suppose that a
feature which is observedin one must be present in the other as well. For
example, if it is observed that Plato and Socrates are similar, the
characteristicof being a man which is presetrtin the first cannot be absent
in the second.The sameappliesto the characteristicof beingmortal which
is observedin a group of men in our experience,which must also be
presentin the group of men who live in another geographicalregion which
is not directly observedby us.
The notion of similarity is invoked by the Epicureansin a double sense:
the observed objects are similar to one another, and the Ks observedare
similar to the Ks not observed(Barnes 1988: 134).Moreover, as David
Sedley l'rascorrectly noted (1982: 256-257), the 'way' or method of simi-
larity coversboth simpleresemblance(asin the exampleabout the mortality
of men in our experienceand the mortality of men outside our experience),
and also analogous resemblance(as in casesinvolving inferencesfrom the
perceivedproperties of bodies to the properties of the atoms).
The attack of the critics
The critics' opposition consistedof an open refutation of the method of
similarity, arguing that the presenceof similarity could not ever be a
sufficient basis on which to make a valid semiotic inference.
Philodemus' 'De signis' 287
Central to this attack is the idea that the correct method of inference is
not one of similarity, but rather that of anaskeud,a term which had been
translatedinto English as 'contraposition' (De Lacy 1978),but which is
now better translated as 'elimination' (Sedley l9B2: 244 and passim;
A smi s 1996:155 and passim ) or as'r ebut t al'( Bar nes 1988: 98 and
passim). The meaning and function of this notion is a matter of some
debate. In the most clear description provided by Philodemus (XL32-
XII. 14) we rearl that given a conditional of the type 'If there is motion,
then there is empty space', we by hypothesis eliminate the empty space,
then in virtue of this elimination the motion would also be eliminated.The
truth of a conditional like 'If p, then q' would then only be properly
establishedaccording to the method of elimination.
Barnes(1988:100)asksthe questionhow the anaskeudcould supply, or
seemto supply, a method of semiotic inference.One of the possibilitiesis
that it roughly corrr:spondsto what modern philosopherscall 'inferenceto
the best explanation'.We can supposethat in relation to a conditional
of the type 'If p, then q', there is an inferenceof the type 'Since p, then q':
the inlerence ra'ould be valid insofar as the corresponding conditional
holds due to the method of elimination. In other words, the lack of the
second term of the conditional explains the lack of the first term. Going
one step farther, Baruesexplains,one could imagine that the Epicureans'
critics thought that the secoud term explained the first.
It is also possibleto establish (seeSedley 1982:243-244) a link betweeu
the method of elimination and the desire of the Epicureans'critics (the
Stoics, according to Sedley) to exclude the validity of the common sign
(koin6n), acceptingonly the sign proper (idion), ulrderstood as a necessary
sign (anankastik6n)which cannot exist if the thing signified does not exist.
The common sign,on the other hand, is suchthat it can exist regardlessof
whether the unknown thing to be revealed exists or not, as in the case in
which one infers thirt a certain man is good becausehe is wealthy: in some
cases,wealth can br: taken as a sign of goodness,but in other casesthis is
not so (I.l-12), meuningthat in certain individualsthe first term can exist
(wealth), but not the second(goodness).The adjective 'comntou' was ttsed
to characterizeIhis type of sign becausethe sign is'commou'with regard
both to truth and to falsehood.The method of elirninationis a test which
excludes this type of sign, rejecting the case in which the sign can exist
without the existenceof the thing signified.
Sedley,howe\'er, goes further (1982:245), and suggeststhat the method
of elimination could be a test of 'cohesion,' syndrlesis,a criterion for the
validity of a conditional attributed to Chrysippus.Syndrtesrsrequiresthat
the contradictory (antike{ntt:non)of the consequent enter into conflict
(mtichetai) with the antecedent (Diogenes Laertius, VII.73; Sextus
288 G. Manetti
Empiricus, PH, II-III), as can be establishedby means of a method which
hypothetically eliminates the consequent in order to verify whether the
antecedent is co-eliminated together with it. In this way only the sign
proper is accepted as valid and owes its infallibility to the fact that it
is subject to the test of elimination in a manner that implicates what it
signifieswith all the necessityof a Chrysippean conditional. The common
sign, on the other hand, probably belongs to the class of convincing
arguments which are neverthelessfallible; therefore, from a strictly phi-
losophical point of view, the common sign is not really a sign at all
(Sedley 1982:256).
The Epicurean response
Tl-rereare three different versions of the Epicureans' defensein the text
(Barnes 1988: 102-103): one consists in the categorical refutation of
the method of anaskeudas discussed in the fourth part of the treatise
(XXX.32-XXXI.l); the second,prominently maintained by Zeno, estab-
lishes that the method of anaskeud itself depends on the method of
similarity, such that if the method of similarity is not valid, the method of
elimination is not valid either (VII.8-12; IX.3-8); the third maintains
tlrat the method of anaskeui only applies in certain cases,while in other
casesthe method o[similarity is what must be used (XIV. ll-22).
The secondtwo forms of defensivepositions are in some sensecoherent
with one another, and one can seehow Zeno might have thought that the
method of anaskeui dependedon the method of similarity. The argument
would go something like this for the usual semiotic inference that.'Since
there is motion, there is empty space'. The connection between motion
and the empty space must depend on the validity of the conditional 'If
there is motion, there is empty space'. But to establish this, as Zeno says
(VIII.26-IX.8), we must have inspectedall the things that can accompany
objects which move in our experience, and in the absence of which
nothing moves. In more formal terms, we must make an inferenceof this
type: 'Since the objectswhich move in our experiencerequire empty space,
everywhere moving objects require empty space'. This is equiv-
alent to having made an inferential statement based on similarity, which
is in turn the basis for an application of the method of anaskeui (Barnes
1988:104) .
An inferencefrom smoke to fire functions in the same way, guaranteed
by the fact that smoke is, by nature, a product of the fire, in such a way
that by eliminating the fire the smoke is likewise eliminated. Yet this
understanding of the nature of the smoke is reachedby means of empirical
Philodemus' 'De signis' 289
generalization. It is only the empirical discovery of the nature of smoke
that then allows one to use the method of elimination applied to a con-
ditional such as 'Since there is smoke, there is fire'. The Epicureans,
however, while they conceded to their critics that a part of the two-stage
process was based on anaslceud,reserved the name of sign-inference
(semeiosis)only fol the stage in which the method of similarity was
employed (IX.3; XXXII.8-10) and refused to consider the elimination
stageof the processas a sign-inference(XXX.33-XXXI.36; XXXII.S-10).
In effect, according to Sedley (1982:261-263) the two-stage process
functions in the following way: in the first stage the nature of a certain
object is established.for example'motion' - and the observationthat it is
inseparable from enrpty spaceconstitutes on its own a semiotic inference
by means of similarity: 'Since many and various objects within our
experienceshare the charircteristicof being unable to move without empty
space,then motion is impossiblewithout empty space'(VIII.32-IX.3). In
the second stage the inference is completed by means o[ the method of
elimination, ancl colrcludes that the motion implies empty space,without
meaning, however, I hat this further stageof the processcan be considered
in and of itself a further sign-inferencein that it does r.rotin and of itself
have the power to leveal anything.
The criticst coulrter-response
The counter-response of the Epicureans' critics consisted in what are
basically two lines of attack against the method of similarity. First, they
maintained that such a method was intrinsically lacking in necessity,as in
an inference of the following type: 'Since all men in our experience
die when they are lrcheaded,then all men die when they are beheaded'.
The critics objected that the premise does not necessarilylead to a true
conclusion; the premise can be true, but the conclusion can be false
(Barnes1988:106).The secondobjection raisedby the critics was against
the very notion of similarity itself or, more precisely,against the range and
homogeneityof the two classesof objectslinked by the inference(V.8-15).
They demanded that the Epicureans specify whether the inference was
supposed to proceed from type to type (for example from 'animal' to
'animal'), or frorn speciesto species('man' to'man'), or from body to body
('corporeal being' t,r 'corporeal being'). In this firtal example they sought
to drive the Epicureans into an internal contradiction with their own
philosophy of physics, insofar as it would lead to unacceptableinferences
such as the following (V. l*4):'Since all bodies in our experiencepossess
color, then the atorns, which are also bodies, possesscolor.'
290 G. Manetti
Moreover, the critics maintained that the term 'similarity'was ambig-
uous, oscillating between the meaning of 'that which is indistinguishable'
to'that which has a certaindegreeof similarity'.At this point one can ask,
as Barnes does (1988: 109), why both the Epicureans and their critics
insisted so much on the necessarycharacter of the inference,considering
that Aristotle had demonstrated the existenceof valid inferencesthat were
neverthelessuot necessary. Certainly the Epicureans could have had
recourseto this possibility iu order to refute their critics, arguing that an
inference by similarity was not true by necessity, but was no less
interestingon this account from a philosophicalpoint of view. Instead,
they seemto have acceptedthe presumption of their critics concerning the
necessarycharacter of the inference.The reason for this is epistemological
in nature and can be explained if one considers that all the Dogmatists,
of whom the Epicureans were representatives according to Sextus
Empiricus, held that sign-inferenceswere a way to increaseour knowl-
edge of the world. In this case, if the link between the premisesand the
conclusions were not necessary,we would not be able to trust in the
knowledgeobtained in this way.
Is Epicurean logic inductive?
A question that arisesat this point has to do with the fact that many
scholars have considered the logic used by the Epicureans in On Slgns to
be inductive. This was asserted in the nineteenth-century edition by
Gonrperz (1865),and also by Asmis (1984:197 and 179-180;1996:163
and passim) and Sedley(1982: 256 ff.), and also in Manetti (1987). Sedley
definesEpicurean logic as 'primarily inductive' (1982:256) as opposed to
Stoic logic, which is instead deductive. Discussing the Epicurean
insistenceon the necessityof the link in a sign-inferenceas they respond
to their critics (who are, for Sedley,the Stoics),and noting once again that
for the Epicureans this was a 'painstaking empirical matter', Sedley
reachesthe following conclusion: 'This looks like a head-on confrontation
between empiricisrn and rationalism. The Epicureans must have felt that
in claiming that sciencecould work purely by deduction from necessary
truths the Stoics were failing to attach sufficient weigh to the inductive
element in the human learning process to which their epistemology paid
lip service'(1982: 259).
Barnes, however, disagrees strongly, and maintains that Epicurean
sign-inferencecan not be compared to the notion of induction in either of
the two modern meanings of the term. One of the modern meanings of
induction is the inference'from the particular to the general or from the
Philodentus' 'De signis' 291
particular to the particular'. In this sense, it proceeds from singular
propositions and its conclusion is either a universal generalizationor
another singular proposition. Epicurean inference, however, proceeds
from a premise which is a universal proposition (for example, 'All the rnen
in our experience die when their hearts are split'). Yet here it can be
objected that in sonre casesEpicurean sign-inferencedoes proceed from
the particular to the particular, as irr the example 'If Plato is a man,
Socratesis also a man' (XII.l9-21).
The secondmodern meaning of induction establishesa probable link
between the premiscs and the conclusions. But even in this sense,again
according to Barnes, Epicurean serneiosisis not inductive, in that the
Epicureansdid not concede that there was a probabilistic type of link
between the premise and the conclusion; rather, as we have seen, they
insistedthat this was a necessarylink.
Barnes's argumerrthas siuce beerrrefuted by Long (1988: 140) who
proposes that the tr:rm 'inductive' continue to be applied to Epicurean
logic. In fact, Long rnaintainsthat the useof universalpropositionsas the
premises in sign-inferencesis not sufficient to exclude the possibility of
calling such inferen<:es'inductive',and he reports an example offered by
Salmon to illustrate that inductive inferencein a context in which it can be
contrasted with deductive inference: '[Since] Every horse that has ever
been observed has had a heart, €very horse has a heart.' This example
appears to be precisely parallel to those found in the Epicureans and is
also compatiblewith other accountsof induction found in contemporary
philosophy.
Further responseby the Epicureans
The Epicureans thur; reject the criticism launched by their oppor.rentsand
insist that their mt:thod of sign-inference is based on necessity. The
arguments that thelr introduce are linked above all to the reftrtation of
the attack on their notion of similarity as somethirlg vague. This is
most apparent in the third portion of the treatise,in which the thinking
of Demetrius of Laconia is expounded, as he lists five points in which
the Epicureans' doctrine has been misunderstood by their critics. As
enumeratedby Ranres,thesepoints are as follows (1988: Il2 tr).
Degreesof shnilurit.v
The first matter of rnisunderstandinghas to do with the selection of
sirnilarities.The Epicureans'critics object that the Epicureanshad not
292 G. Manetti
bothered to specify which of the types of similarities could lead to an
inference.The Epicureansdenied that inferencecould be basedon chance
similarities, maintaining that inferencewas establishedby means of those
sinilarities which are relative to the specific species taken into con-
sideration, bypassing similarities of a higher order. In other words, the
Epicureansapparently presumed that there was a structure of generaand
species, such that the only pertinent similarities were those between
objects of the same class or a subordinate class,whose individual objects
observed were members. So, for example, the properties of birds could
be inferred from the properties of the class of birds, and the properties of
animals could be inferred from animals, and the properties of robins can
be inferred from members of the class of robins, but inferencescould not
be rnade on the basisof similarities betweenmembersof the classof robins
and mernbers of the class of animals.
Sitnilarities and dffirences
The second issue involves the existenceof strange or bizarre cases,which
according to the Epicureans' critics would invalidate the possibility of
inference by similarity. For example, while proceeding by inference from
the classof known men to the classof unknown men, one could encounter
the dwarf of Alexandria whose head was as hard as an anvil. which would
invalidate the inference based on similarity. The same problem would
result from the existenceof the magnet, the one case of a stone that was
supposedto attract iron.
The Epicurean response is that unique or strange cases actually
reinforce the method of similarity. The argument is basically as follows
(XXXV.4-I4): if the inference begins from the observation of casesthat
are absolutely identical, it would not be an inference; on the contrary,
inferenceshould begin from observations of casesthat are widely different
from one another, even including strange cases,so long as they belong to
the same class.
Absenceof contrary proof
The third point has to do with the precautionarymeasuretaken by the
Epicureans that a sign-inferenceshould involve no possible evidence to
the contrary. For example, to infer that atoms have color based on the
observation that all bodies in our experiencehave color would be an error,
in that atolns are very different from the normal bodies that we perceive:
atoms are indestructible. invisible. and so on. Thesecharacteristicswould
Philodemus' 'De signis' 293
qualify as elementsindicating a meaning contrary to the possibility that
atoms have color. Yet as the Epicureans'criticsobserved,this becomesa
moot point in that it is difficult to establishwhat would in fact constitutea
contrary indication, and the Epicureans did not actually l-raveany valid
argument by which to do so.
Qua truths
The fourth issue treated by Demetrius involves a responseto one of the
most important contentions of the Epicureans' adversaries,who argued
that sign-inferenceshad to depend on qua truths (III.26-IV.2) - which
can be translated as 'truth inasmuch as' or 'truth insofar as' - and that
Ihese qua truths could only be affirmed by rneans of the method of
elimination (VI.3I-VII.5). For example, the Epicureans' critics would
reformulate the inference'Since men in our experienceare mortal, then if
there are men anywhere they are mortal' in the following way: 'Since men
in our experienr;eale mortal insofiar as (qua) they are men and inasmuch
as they are men, then men everywhereare mortal'. According to Barnes,
the Epicureans agreed that valid sign-inferenceshad to be based on
qua truths, but at the same time they maintained that these would be
established by mearrs of the method of similarity. The vocabulary used
to express the qua propositions is rather complex and involves cornfilorl
expressionsfrom the philosophical lexicon, such as hdi and kath6, as well
as other words, such as par6 and syn, which are peculiar to the text of
Philodemus.
The author of the fourth part of the treatise identifies four sensesin
which these 4uo truths can be understood. In some places the text is
illegible, but Barnes (1988: l2l) supplies a coherent reconstruction of
thesefour interpretations. The first holds that the expressionstnean that
given a type l( thert: necessarilyfollows a property F. For example: 'Men
are made of flesh arld are subject to sicknessand old age insolar as they
are men'. The secorrdholds that Ks are F insofar as F is the definition,
l6gos, or the pr\lep'sis of I(. For example: 'Bodies have mass and resis-
tance insofar as they are bodies', 'Man is a rational anirnal insofar as he is
a man'. The third type is stated in a corrupt portion of the text. It is
described in a rather vague way (synbebekinai t6de t6ide) and it is not
possible to reconstruct an1, examples with certainty. Barnes argues
(1988: l2l) that it rnight deal with that which is direct, but not a matter
of definition. There are then examplesof the fourth type of qao truth but
without a definition; it involves that which is non-direct, which is to say
a property which belongs to something which in tnrn belongs to K.
294 G. Manetti
For example: 'I(nives cut insofar as they are sharpened','A body falls
insofar as it has weight', 'Atoms are indestructible insofar as thev are
s olid' .
All four of these types involve the principle of necessity
(XXXIV.24-27). To say that Ks have the property F insofar as they
are Ks mealts that I(s are necessarilyF, a rule which allows inferences
accordingto which all Ks (eventhough which are not known) are F. This
rule has the advantageof blocking inferencesthat were unappealing to the
Epicureans, such as the attribution of color to the atoms. Their critics'
objection had beeu basedon the fact that the observation ofobjects in otrr
experience shows that all objects are endowed with color: therefore, if
the method of similarity were to hold true, then the atoms should also
have color. To this the Epicureansreply that the atoms, insofar as they are
bodies, can be said to have resistanceto the touch (and this in fact forms
part of the definition of a 'body'), but they do not have color: in fact,
bodiesin the dark also effectivelylack color (XVIII.3-10). The problem of
necessity,however, still remains, since the Epicureans do not make clear
in wlrat sensethe qua truths are establishedon the basis of similarity. In
other words, one must still ask how it is possiblethat the observationof
similarities between things that are K is able to establish a connection
based on necessitybetween K and F which the qua truths demand.
At this point, we must considerthe objection of Long (1988: 142-143,
seealso Asmis 1996:164),who maintains that the Epicureansassigneda
different, empirical senseto the qua truths, which is not the same as that
used by their critics. According to Long, the Epicureans do not use
formulations as 'insofar as such individuals are men'. but rather some-
thing like'insofar as the things familiar to us are of such a type'. This fact
emergesquite clearly in the following passagefrom the treatise:'When we
say that sincethings familiar to us are of such a kind, things outsideour
experienceare of that kind, we are judging that insofar as things familiar
to us are of such a kind, something outside our experienceis conjoined
to thern, as in the case of "since men familiar to us, insofar as they
are men, are mortal, if there are men anywhereelse they are mortal"'
(XXX.24^32). In other words, the Epicureans relate the qua formula
('insofar as') not to the thing in general(e.g.,'man'), but to the thing in
our experience.The Epicureans statement of the qua truth thus takes on
the following form: 'Insofar as Ks familiar to us are F'. In this way they
avoid the trap of essentialisminto which their critics want to push them.
The Epicureans begin from the fact that mortality accompaniesthe men
observedinsofar as they are men, and conclude that all men insofar as
they are nlen are nrortal. In this way the conclusion is not presupposedby
the premises,but is produced by the method of similarity.
Philodemus' 'De signis' 295
Inconceivability
With the fifth point Demetrius makes a serious attempt to establish a
necessarylink betweenthe antecedentand the resulting consequent.To do
so, Demetrius introduces the notion o[ 'inconceivability' (adianoes[a),
which raises the question of the modal connection between the class of
known objects (K) and their properties (F) on the one hand, and tl.reclass
of unknown objects and their properties on the other hand. The principle
of inconceivability is discussed at various points throughout the text
(XXVIII.I5-24, XXXIII.I-7, XIV.l7-21) and can basically be under-
stood as follows: the method of similarity establishes that it is
inconceivable that something evident exists and has certain character-
istics,while somethingwhich is not evidentwould not exist and would not
have such characteristics.
It is interesting at this point to recall a suggestivehypothesis of Sedley
(1982:257, n. 46) rvhich compares Epicurean adianoes[a and anaskeud,
following his assumption that the Epicureans' critics here are Stoics two
generations after Chrysippus. In this context, the Epicurean adianoesia
would be a test of similarity, basedon the fundamental elementof sign-
infierence,just as anaskeudwould be a test of Stoic .syncirtesis, that is, the
connection assertedby Chrysippus for a valid conditional.
Epicrrreans
Nature of the link syndrtesis hontoi6tes
Validity criteria anaskeui adianoesia
In any case, the notion of inconceivability seems to be the actual
principle which the Epicureanscan use to reclaim the elementof nece.ssity
advanced to defencl their type of inference.The principle of inconceiv-
ability can in fact constitute the link between similarity as attested by
experienceand logical necessity:similarity leads to inconceivability and
inconceivability leads to necessity(Barnes 1988: 124-125).Yet Bames
also points out at the conclusion of his essaythat there are two difficulties
with this notion of inconceivability. The first has to do with the fact that it
depends on two attitudes which modern thought rejects: psychologism
and conceptualism Psychologism, which finds its classic expression in
Hume, presumesthat somethingis consideredpossiblesimply becattsewe
are able to conceiveof it; likewise,it could simply be a matter of our lack
of imagination which rendetssomething impossible.Conceptualism,on
the other hand, holds that the non-existenceofsomething dependson the
296 G. Manetti
fact that our conceptual apparatus cannot anticipate its existence.Again
in this case it might simply be a problem of our feeble powers of con-
ception that are responsiblefor our conviction about the non-existenceof
something. The other major difficulty cited by Barnes is basedon the fact
that the Epicureans link necessityto psychological or conceptual incon-
ceivability and this in turn to empirical observation. In effect the
Epicureansmaintained, for example, that the guarantee of the belief that
all men are mortal insofar as (qua) they are men relies on an empirical
processof observation of many, varied cases,accompaniedby the lack of
contrary proof. But how is this possible?asks Barnes (1988: 127). One
possible way to explain, if not justify, the Epicurean defenseis linked to
the particular notion of prolepsis which the Epicureans had developed.
This prolepsis consistedin a sort of conception of something, albeit rough
and preliminary, that had to be supplied before that thing could begin to
be investigated.The processof investigation and of cumulative observa-
tion had two purposes, and yielded two types of results: (i) on the one
hand it enriched our conception of the object under study, for example by
linking two elementswhich had first been considered separate(such as a
class K and the property F which is attributed to all the members of I(),
with the result that these elementscould be understood as associatedby
necessity;(ii) on the other hand, it extends our empirical knowledge, an
idea which as Barnessuggests(1988: 128)would have pleasedeven John
Locke (Essay, lll.vi.4647).
Conclusions
Although Marcello Gigante'sbook is not explicitly dedicatedto semiotic
problems, it neverthelessmakes a significant contribution to the on-going
debate, both philological and philosophical, surrounding the figure of
Philodernus, a major figure in the study of the history of semiotics.The
historical study of semioticsis a relatively recent development, which can
in sone sensebe dated to a debate dedicated to this topic at the second
Congressof the I.A.S.S. held in Vienna in 1979.Major stepshave been
taken in the past twenty years, especially with regard to the history of
ancient and medieval theories of the sign, relying greatly on the con-
tributions of philological and philosophical research of the sort I have
discussedin this article. The importance of Philodemus is due to the fact
tlrat his treatise On Signsis the most complete and best preservedwork
which has reached us from antiquity on the subject of sign-inference.
Moreover, the philosophical debate which it offers is marked by a high
level ofprofound thought and speculation. It is safe to say that it has not
Philodenws' 'De sisnis' 297
yet revealed all its secretsto us, and iemains an extremely rich source for
further study arrd reffection.
Translated from Italian bv Laura Gibbs
References
Asmis, E. (1984). Epicunt' Scientifc Method. Ithaca, }rIY: Cornell University Press.
-(1996). Epicurean semiotics. In Knowledge Through Signs:Ancient SenrioticTlrcoriesand
Practices, G. Manetti (ed.), 155-185. Brepols: Turnhout.
Barnes, J. (1988). Epiculeans signs. ln Oxford Studies in Ancient Philosoplry,supp. vol.,
J. Annas (ed.), 9l-134. Oxford: Oxford University Press.
Burnyeat, M. F. (1982).'fhe origins ofnon-deductive inference.In Scienceand Speculation,
J. Barnes et al. (eds.), 193-238. Cambridge: Cambridge University Press.
C apasso,M. (1980).P H erc .67l : U n al tro l i bro' D e S i gni s ' ?C E rc 10, 125-128.
Cavallo, G. (1983). Libri scritture scribi a Ercolano. CErc 13, second supp., l-18.
De Lacy, P. and De Lar:y, E. (eds.) (1978 ll94ll). Philodentus. On Metltods of Inference.
Naples: Bibliopolis.
Gomperz, T. (1865). Philoelern Uber Induktions-schlisse. Leipzig.
Long, A. A. (1988). Reply to Jonathan Barnes, 'Epicurean Signs'. ln Oxford Srudiesin
Ancient Philosopfty,supp. vol., J. Annas (ed.), 135-144. Oxford: Oxford University Press.
Manetti, G. (1993). Theories of Sign in Classical Antiquity. Bloomington: Indiana University
Press.
Marquand, A. (1883).The logic of tlre Epicureans.ln Stttdiesitt Logk: by tlrc Mentbers of the
Johns Hopkilts Universitl', ChrrrlesS. Peirce (ed.), l-l L Reprint, Amsterdam: Benjarnins,
l 983.
Pandermalis, D. (1971). Zum Programm der Statuenausstattung in der Villa dei Papiri.
Mitteilungen des Deuschen ,4rchdologischen lrlstituts Alhenislte Abteihtg 86, 173 209.
[It.trans.byL.A .S c atoz z aH ori c ht. l nLav i l l adei papi ri , s ec onds upp. C E rc 13, 19-50.
N apl es, 1983.1
Peirce, C. S. (1982). Writings of Charles Sanders Peirce. A Chronologit:ol Ediriott, vol. l.
Bloomington: Indiana University Press.
Salmon, W. C. (196-l). Iogic. Englervood Cliffs, NJ: Prentice-Hall.
Sedley, D. (1982). On signs. In Scitnce and Speculation,J. Barnes et al. (eds.), 239-272.
Cambridge: Cambridge University Press.
Giovanni Manetti (b. 1949) is Professor of Semiotics at the University of Siena, Italy
(manettig@unisi.it). Hir; principal researchinterestsinclude the history of semioticsand the
theory of enunciation. F{is major publications include Sport e giochi nell'anticltitd classica
(1988), Theories of the Sign in Classical AntiEily (1993), Knovletlge rhrough Sigtt.t,Ancient
Semiotic Theoriesand Ptactices (ed., 1996) and Sr'gnsand Signifcatiorr (ed. with H. S. Gill,
1999 and 2000).