David Sedley



1. Introduction

The section of Sextus, Adversus mathematicos, vii, devoted to the

history of theories on the criterion1 divides up as follows:

46-7: views on the criterion: 1, there is no criterion; 2 (a), the criterion

is in logos·, (b) the criterion is in enargeia; (c) the criterion is in logos

and enargeia.

(1) Those who say there is no criterion 49-52 Xenophanes 53-4 Xeniades 55-9 Anacharsis 60-4 Protagoras 64 Euthydemus and Dionysodorus 65-87 Gorgias 87-8 Metrodorus, Anaxarchus, Monimus

(2) Those who say there is a criterion... logos “Physicists” 89-91 Anaxagoras 92-109 Pythagoreans


110 Xenophanes (again!) 111-4 Parmenides 115-25 Empedocles 126-34 Heraclitus 135-40 Democritus enargeia (with or without logos) “Post-physicists” 141-4 Plato 145-6 Speusippus 147-9 Xenocrates 150-8 Arcesilaus 159-89 Carneades 190-200 Cyrenaics 201-2 [Asclepiades] 203-16 Epicurus 217-26 Peripatetics 227-60 Stoics 261-2 retrospect

Thus, in the event, 2 (b) and 2 (c) are conflated, and we end up

with three principal divisions. It is instructive to note that the

philosophers conventionally known to us as “atomists” are distributed

between these three divisions: Metrodorus of Chios and Anaxarchus fall

into the no-criterion group, Democritus into the logos-only group, and

Epicurus into the enargeia group. This serves as a reminder, if one were

needed, that atomism is to some extent a doxographical fiction, in-

deed one to which modern doxographers are more wedded than their

ancient counterparts. Although the names of Democritus and Epicurus

are often linked in the ancient sources with regard to their basic physical

tenets — “atomism” in the strict sense — they are equally often con-

trasted with regard to their theories of knowledge and numerous other

doctrines. Sextus himself is no exception to this pattern2. In most


respects his treatment of the two principal atomist philosophers is rather

conventional. There is little evident consciousness of the special relation-

ship that you might expect a Pyrrhonist to feel with either the one school

or the other — either with Democritus as a forerunner of Pyrrho’s scepti-

cism, much revered by Pyrrho himself, or with Epicurus as an admirer

of Pyrrho’s ethical outlook, who even shared the Pyrrhonist official moral

goal of ataraxia, tranquillity.

With regard to Democriteanism, the furthest Sextus goes is at Out-

lines of Pyrrhonism,
i 213-4, where it appears among the philosophies he

deems close enough to Pyrrhonism for the difference to need under-

lining. And he maintains, sensibly enough, that although the ou mallon

dictum is superficially common ground between the two movements, the

Democriteans (a) use it in order to deny the reality of sensible properties,

and (b) positively assert the existence of atoms and void, while the

Pyrrhonists retain an open mind on the existence of both. As for

Epicureanism, he alludes to its kinship with Pyrrhonism only once, at

the opening of Adversus grammaticos. There Epicurus is described as a

disciple of Pyrrho’s pupil Nausiphanes. But Sextus proceeds to make the

sharpest possible contrast between the Epicureans and Pyrrhonists as

regards their grounds for opposition to the mathemata3.

A question which would reward close examination — on some other

occasion — is how far Sextus’ perception of Democritus has been shaped

by the Epicureans4. One likely symptom of such mediation is the total

absence of the name Leucippus from Sextus’ works. This may reflect

the studied silence about him in Epicurean texts generally, stemming

from Epicurus’ well-known denial of his existence5.

But by far the most intriguing passage in Sextus bearing on the


atomist tradition is the one I mentioned at the outset: the history of

theories of the criterion at M vii 46-2626. For both Democritus and

Epicurus it contains what are arguably the richest treasuries of surviving

data on their attitudes to empirical cognition. Yet when it comes to evalu-

ating those data we may seem to know little of what kind of source we

are dealing with, and what kinds of filters or distorting media they have

been passed through.

The section on Democritus is our primary source of his sceptical utter-

ances. It would be easy to jump to the conclusion, as some have done7,

that this massive emphasis on his cognitive pessimism is to be explained

by the fact that Sextus is himself a sceptic, anxious to maximise his indirect

forerunner’s endorsement of his own school’s philosophical outlook. This,

we will see, radically misrepresents the character of the passage.

In the long doxographical passage on the criterion, Sextus is himself

an entirely transparent figure. Whenever you try to focus on him, you

find yourself looking straight through him, and what you see instead is

the early first century B.C. You find yourself in the hands of first-century

B.C. interpreters like Posidonius and Antiochus. The reason for this is

not very far to seek. Sextus will be repeating here, as often, material

from the writings of his principal forerunner and authority, Aenesi-

demus, the founder of the neo-Pyrrhonist movement in the mid first

century B.C. And Aenesidemus himself, we may speculate, compiled his

own account of the dogmatist theories of the criterion largely by consult-


ing recent historical or critical textbooks. Hence we find ourselves, in

this passage, more in Aenesidemus’ philosophical world than in Sextus’.

This practice of relying on the latest available studies is not particu-

larly shocking or surprising. If we want to know the position of Kant

or Aquinas on some specific issue, we often cannot afford the luxury

of turning to their works for enlightenment: even if we happen to have

their collected works on our shelves, we may not have the least idea

which of them to turn to, or which ones are deemed the most reliable

guide to their views. How much easier to browse through a convenient

encyclopedia article, a pre-digested masterpiece of synthesis by someone

much better informed (we hope) than ourselves, or through the latest

critical study to appear in the bookshops. The absence of indexes in an-

cient books made such shortcuts even more irresistible. To arrive at the

principles of Hellenistic epistemology by checking through the 300 books

of Epicurus, the 700 of Chrysippus, and thousands more, would have

been a massive and perhaps a foolhardy undertaking. We can hardly con-

demn Aenesidemus if he preferred to rely on the very latest surveys,

some of them written by the most eminent philosophers of the day.

The trouble is that such surveys are not always an innocent guide

to history. The philosopher who plays historian of his subject will proba-

bly find it difficult not to impose his own prejudices on the material

he reports. He may even be writing the history for that very purpose.

2. Metrodorus and Anaxarchus

Aenesidemus might have appreciated this, if only because himself was

a practitioner of just such creative reporting. For there is every reason to

believe that the first section, on those alleged to deny that there is

any criterion (M vii 49-88), is the work of Aenesidemus himself. This

can be inferred especially from the treatments of Anacharsis and Gorgias,

both of whom are recast in the rigorously dilemmatic form of argument

characteristic of Aenesideman scepticism8, and from that of Protagoras,


which assimilates his relativism to the style and content of Aenesidemus’

own fourth Mode9. The presence, in the same list, of acknowledged

forerunners of Pyrrhonism, like Xenophanes, Metrodorus of Chios, and

Anaxarchus, points the same way10.

In theory, the same features would be compatible with the author’s

being not Aenesidemus but some follower of his, even Sextus him-

self11. But as I have already suggested, it is Aenesidemus’ direct in-

volvement that best explains the reliance on first-century sources in the

following sections, and, if so, it becomes more plausible as well as more

economical to imagine him as the compiler of the entire doxography.

The atomists Metrodorus and Anaxarchus earn just a brief mention

at the end, along with the Cynic Monimus (87-8): Metrodorus for saying

«We know nothing, and we do not even know this very thing, that we

know nothing», Anaxarchus and Monimus for comparing existing things


to a stage-painting and holding them to be the objects of delusion. I

do not want to dwell on this passage here12, beyond one remark. The

words attributed to Metrodorus clearly correspond to the opening of his

book reported rather differently by Cicero (ac. ii 73): «I say that we

do not known whether we know something or nothing, nor do we know

that very thing, knowing or not knowing, nor in general whether any-

thing exists or nothing». Should we regard this, or the Sextan version,

as more authentic? Given Aenesidemus’ practice in the preceding chap-

ters, especially those on Gorgias, I have little doubt that it is he who

is guilty of adjusting Metrodorus’ words13, to make them neater, more

schematic, and more Pyrrhonian14.

3. Democritus

I now turn to the long section (89-140) on the “natural philosophers”

(φυσικοί), who are said to place the criterion in logos. This has certain

recurrent features which distinguish it from the other two divisions. First,

while the other divisions discuss their respective lists of philosophers

in roughly chronological order15, this one makes virtually no effort to do


so, but adopts the chaotic-looking sequence: Anaxagoras, the Pythagoreans,

Xenophanes, Parmenides, Empedocles, Heraclitus, Democritus. Here

only Democritus’ position at the end captures his chronological relation

to the others. Nevertheless, I believe that there is an overall rationale

to the order. Anaxagoras is placed first as the archetypal natural philo-

sopher (φυσικώτατος, 90). Thereafter each of those listed is presented

as selecting a different type of logos as criterion16, and the order is so

constructed as to maximise the contrast between each philosopher and

his neighbours in the list. Anaxagoras opted for logos “in general”, the

Pythagoreans for a specific type. The Pythagoreans’ use of mathematical

or scientific logos in turn contrasts with Xenophanes’ reliance on merely

“doxastic” logos, which is itself then counterposed to Parmenides’ “epis-

temonic” logos. Empedocles invokes orthos logos, but this is primarily

“human” logos, whereas Heraclitus’ logos is “divine”, as well as univer-

sal. After this series of careful antitheses, it is almost an anticlimax to

come at the end to Democritus, whose criterion is simply said to be logos,

with no further refinement.

One might try to read the series of antitheses as reflecting a sceptical

perspective, the chosen order being aimed at highlighting the conflict

(diaphonia) between the warring parties. Certainly that aspect of it will

been more than welcome to Aenesidemus and to Sextus, since they do

indeed call the entire catalogue of views on the criterion a case of diapho-

(261). But it is hardly how the passage comes across when taken in

its own right. On the contrary, it repeatedly emphasises the continuity

between these thinkers (as I shall try to show shortly). A more accurate

reading would be that the series of antitheses is meant to bring out how

each of these Presocratics focused on a different aspect of logos. Then

Democritus, placed correctly at the end, may be viewed as standing above

the series of antitheses, and as combining in his notion of logos all the


aspects which his predecessors had isolated17. If this is the point of the

passage’s structure, it was (at least in its original context) designed to

give Democritus an especially prominent place in the story.

A second distinguishing feature of this whole division is its extraor-

dinarily high content of verbatim quotations, including (quite unusually

for Sextus) substantial extracts of prose as well as verse. By contrast,

the other divisions of the passage on the criterion, much in keeping with

Sextus’ usage elsewhere, rely mainly on paraphrase.

This itself goes hand in hand with a third feature. The attribution

of the logos criterion to these thinkers is an undisguisedly creative piece

of interpretation. Only for Heraclitus does the author claim to find any

such doctrine expressly stated, and he distinguishes this case by telling

us that Heraclitus says so «quite explicitly» (134, ῥητότατα). His practi-

ce of regularly quoting the passages on which the interpretation is based

serves him as a safeguard, making sure that his readers are not misled

about its speculative status.

A fourth feature is the author’s special interest in the theme that

like is known by like, to which he recurs in the sections on the

Pythagoreans (92-3), on Empedocles (116-21), and on Heraclitus (130),

invoking the further support of Democritus (116-8) and Plato (119).

This kind of invocation — the establishment of alliances between dis-

parate thinkers — itself constitutes a fifth distinguishing feature of the

division as a whole18. Anaxagoras’ strictures on the weakness of the senses

are amplified by parallel remarks quoted from Asclepiades (91). On the

Pythagorean like-by-like principle, Empedocles and Plato are cited for

comparison (92-3), while for Empedocles’ use of the same principle Demo-

critus and Plato are invoked. And for the alternative interpretation of

Democritus, taken from a certain Diotimus, which he appends at 140,

parallel remarks are quoted from Anaxagoras and from Plato’s Phaedrus19.


A sixth and final hallmark of this historian is his readiness to juxta-

pose alternative interpretations, without insisting on an exclusive choice

between them. His message is, again and again, that you can read these

philosophers as making logos the criterion, but that other readings are

available. Xenophanes can be read as rejecting all criteria (as already at

49-52, the “Aenesideman” section, so omitted here), but on another in-

terpretation (110, κατὰ τοὺς ὡς ἐτέρως αὐτὸν ἐξηγουμένους) he makes

doxastic logos the criterion. Empedocles, on what appears to be the more

straightforward interpretation (115, κατὰ μὲν τοὺς ἀπλούστερον δοκοῦν-

τας αὐτὸν ἐξηγεῖσθαι), has six criteria of truth, namely the four elements

plus Love and Strife, but according to others (122) he makes orthos logos

the criterion. Likewise Democritus can be seen as abolishing all cognition,

or as making logos alone the criterion (135-9), and there is also Diotimus’

interpretation of him, according to which there are three criteria (140).

In none of these cases is it insisted that the logos-only interpretation

is mandatory. The pluralistic style of history-writing makes an interesting

contrast with the monistic reporting in the other sections of the passage.

Then who is our historian? He cannot be dated before the late second

century B.C., since he quotes Asclepiades20 ; nor much later than the mid

first century B.C., if he was himself used as a source by Aenesidemus. His

readiness to create alliances between diverse philosophers is itself sugges-

tive of the syncretistic tendencies so characteristic of the early first century

B.C. And there is one outstanding candidate: the Stoic Posidonius. That

he may be the source of this entire passage (89-140) has been suggested be-

fore21, but never, as far as I know, worked out in detail.

It is, at least, a matter of virtual consensus that he is the source

of the long Pythagorean part of the passage (92-109)22. The grounds


include the following: Posidonius’ interpretation of Plato’s Timaeus, as

relying on a like-by-like principle, is cited explicitly (93)23; the section

includes a story (107-8), otherwise not recorded, about the Colossus of

Rhodes, the island where Posidonius lived and taught; and it uses Stoic

doctrine in expounding the Pythagoreans24.

If that attribution is correct, it need not necessarily follow that the

rest of the passage is Posidonian25: its author could be himself drawing

on Posidonius merely for the Pythagoreans. But any such possibility re-

cedes when we bear in mind the many distinguishing features which the

Pythagorean section shares with other parts of the passage, especially the

syncretistic tendency and the fascination with the like-by-like principle26.

Moreover, Stoic doctrine surfaces occasionally elsewhere in the pas-

sage27, especially in the emphatic reading of logos in Heraclitus as


divine universal logos — an interpretation of Heraclitus with no pre-Stoic

antecedents but for obvious doctrinal reasons beloved of the Stoics28.

And that is not all. There is an even more distinctive Posidonian

fingerprint on the passage. For as it happens there is one case in which

we know something about the kind of motive which led Posidonius to

rewrite philosophical history. This concerns his decision to adopt Plato’s

tripartite psychology, even borrowing from the Phaedrus the comparison

of the soul’s two irrational parts to a pair of horses driven' by a human

charioteer29. Now in appropriating this theory he was not declaring

himself a Platonist. Rather, his ultimate authority was Pythagoras, and

he set out to show that it was from none other than Pythagoras that

the Platonic theory was itself derived. According to Galen, he did this

by tracing the theory back to the master via Pythagoras’ own pupils30.

How striking then that our passage, in commenting on Parmenides’

proem, interprets the horses drawing his chariot as representing «the

irrational impulses and desires of the soul»31. Parmenides was himself

regarded as a follower of Pythagoras32, and it is hard to resist the


conclusion that we have here caught Posidonius in the act (if only en

) of bolstering the Platonic theory’s Pythagorean pedigree by find-

ing it, in the form of the Phaedrus charioteer simile, already present in

Parmenides’ poetic imagery.

Cumulatively, these clues seem ample to confirm the hypothesis that

the original author was Posidonius. If so, what was the context? For-

tunately the old view of Schmekel, that the Pythagorean section, at least,

came from a commentary by Posidonius on Plato’s Timaeus, is now widely

rejected, and I need not repeat here the arguments of Mansfeld, Kidd

and others33. There is, it seems to me, a much better candidate: Po-

sidonius’ work On the criterion (Περὶ κριτηρίου). Not only is this the

most apposite possible title, but the conjecture sheds instant light on a

much puzzled-over sentence from Diogenes Laertius’ Stoic doxography:

«Some of the older Stoics allow orthos logos as criterion, as Posidonius

says in his On the criterion» (vii 74). This is so out of step with the

remaining doxography on Stoic criteria that modern interpreters have

been compelled either to ignore it or to let it substantially affect their

interpretation of Stoic epistemology34. But if the Sextus passage derives

from the same work by Posidonius, we can see exactly what has hap-

pened. When Posidonius, as source of M vii 122-5, attributes to Empedo-

cles the criterion of orthos logos, there is no pretence that this is a straight

doxographical report: it is, quite openly, a speculative reinterpretation

of certain remarks made by Empedocles. It is surely in that same spirit,

and not by way of formal report, that Posidonius attributed orthos logos

as a criterion to some of the early Stoics35.


We should note that our Posidonian section in Sextus is explicitly

limited to the early “physicists” (89, 140-1) — a standard designation

of the Presocratics. It must represent only one part of Posidonius’ origi-

nal work On the criterion, which clearly included similar analyses of his

forerunners in the Stoa and, we may be sure, of Plato, among others.

My own guess would be as follows. Posidonius’ contemporary An-

tiochus had published a work, the Canonica, which inter alia classed the

Stoics among those who make enargeia the criterion and deny logos any

independent criterial status (I shall supply the grounds for this supposi-

tion in section 4). Posidonius, I suggest, published his On the criterion

as a reply to Antiochus, arguing that it was in fact common ground

between the founding Stoics and all their most illustrious forerunners,

right back to Pythagoras and Heraclitus36, to assign independent criterial

status to some kind of logos37.

Having identified Posidonius as our source, we can now turn to his

account of Democritus (135-40). It may seem surprising that Posidonius,

a Stoic, should give Democritus the special prominence which, as already

noted, attaches to the final position in the list. It can hardly signify

special doctrinal authority. On the other hand, Posidonius was only too

likely to approve of Democritus’ exceptional intellectual range, so similar


to his own — including physics, mathematics, ethics, geography, astronomy,

meteorology, and that general concern for aetiology which is so strongly

associated with both philosophers38. It may be this special affinity that

underlies Posidonius’ preferential treatment of Democritus.

I shall divide the passage into six parts, labelling the citations with

the letters a-i:

I. «Democritus at times (ὅτε μὲν Usener, ὅτι μὲν codd.) eliminates

sensory appearances, and says that none of these appears truly but only

in opinion (δόξα), and that the truth in the things that there are is that

atoms and void exist. For, he says, (a) “By convention sweet and by

convention bitter, by convention hot, by convention cold, by convention

colour. In reality atoms and void (νόμῳ γλυκὺ καὶ νόμῳ πικρόν, νόμῳ

θερμόν, νόμῳ ψυχρόν, νόμῳ χροίη· ἐτεῇ δὲ ἄτομα καὶ κενόν)”. That

is, perceptibles are objects of belief and opinion, and they do not exist

truly, but only atoms and void do» (135).

II. «In his Kratunteria, despite (b) having professed to ascribe command

over evidence to the senses ὑπεσχημένος ταῖς αἰσθήσεσι τὸ κράτος τῆς

πίστεως ἀναθεῖναι), he none the less is found condemning them. For he

says: (c) “We in reality have no reliable understanding, but one which

changes in accordance with the state of the body and of the things which

penetrate and collide with us (ἡμεῖς δὲ τῷ μὲν ἐόντι οὐδὲν ἀτρεκὲς συνί-

εμεν, μεταπῖπτον δὲ κατά τε σώματος διαθήκην καὶ τῶν ἐπεισιόντων

καὶ τῶν ἀντιστηριζόντων”». And again he says: (d) “That in reality we

do not understand what each thing is or is not like, has been shown

in many ways (ἐτεῇ μέν νυν ὅτι οἷον ἕκαστόν ἐστιν ἢ οὔκ ἐστιν οὐ

συνίεμεν, πολλαχῇ δεδήλωται)”» (136).

III. «And in his Peri ideon: (e) “Man must know by this yardstick: that

he is cut off from reality (γιγνώσκειν τε χρὴ ἄνθρωπον τῷδε τῷ κανόνι,

ὅτι ἐτεῆς ἀπήλλακται)”, and again (f) “This argument too shows that

in reality we know nothing about anything, but seeming for each of us

is an influx [or ‘reshaping’] (δηλοῖ μὲν δὴ καὶ οὗτος ὁ λόγος ὅτι ἐτεῇ

οὐδὲν ἴσμεν περὶ οὐδενός, ἀλλ’ ἐπιρυσμίη ἑκάστοισιν ἡ δόξις)”, and also

(g) “And yet it will be clear that to know in reality what each thing

is like is beyond us” (καίτοι δῆλον ἔσται ὅτι ἐτεῇ οἷον ἕκαστον γι-

γνώσκειν ἐν ἀπόρῳ ἐστι)”» (137).

IV. «Now in these [i.e. (c)-(g)?] he is virtually (σχεδόν) rejecting all

cognition, even though it is only the senses that he attacks specifically»

(137 fin.).


V. «But in his Canons he says that there are two kinds of knowledge

(γνώσεις), the one through the senses, the other through the mind. Of

these, he calls the one through the mind “genuine” (γνησίην), ascribing

to it reliability for judging the truth, while the one through the senses

he names “bastard” (σκοτίην), depriving it of infallibility for the discern-

ment of truth. His precise words are: (h) “Of knowing there are two

forms, the one genuine, the other bastard. And of the bastard kind this

is the complete list: sight, hearing, smell, taste, touch. The other is

genuine, but separated from this one (γνώμης δὲ δύο εἰσὶν ἰδέαι, ἡ

μὲν γνησίη, ἡ δὲ σκοτίη· καὶ σκοτίης μὲν τάδε σύμπαντα, ὄψις ἀκοὴ

ὀδμὴ γεῦσις ψαῦσις, ἡ δὲ γνησίη, ἀποκεκριμένη δὲ ταύτης)”. Then,

by way of judging the genuine one superior to the bastard one, he adds

these words (εἰτα προκρίνων τῆς σκοτίης τὴν γνησίην, ἐπιφέρει λέγων):

(i) “When the bastard one is no longer able either to see smaller, nor

to hear nor to smell nor to taste nor to sense by touch, but finer

(ὅταν ἡ σκοτίη μηκέτι δύνηται μήτε ὁρῆν ἐπ’ ἔλαττον μήτε ἀκούειν μήτε

ὀδμᾶσθαι μήτε γεύεσθαι μήτε ἐν τῇ ψαύσει αἰσθάνεσθαι, ἀλλ’ ἐπίλεπ-

τότερον)”. Therefore according to Democritus too reason is the criterion:

he calls it “genuine knowing” (γνησίη γνώμη)» (138-9).

VI. «Diotimus used to say that according to Democritus there are three

criteria. The criterion for the cognition of things non-evident is appear-

ances; for “Appearances are a sight of things non-evident”, in the words

of Anaxagoras, whom Democritus praises for this. That for inquiry is

the concept (τὴν ἔννοιαν); for “Concerning every topic, my boy, there

is but one starting-point, to know what the inquiry is about” [para-

phrasing Plato, Phaedr. 237 b]. And that for choice and avoidance is the

feelings; for what we have an affinity for is to be chosen, what we are

alienated from is to be avoided» (140).

Step I seeks to establish the significance of passage (a), Democritus’

best-known statement about the metaphysics of cognition. It is presented

as evidence, not for his acceptance of logos as a criterion, but for his

firm rejection of the senses. Sextus might be thought open to criticism

for failing to add b 12539, the ensuing reply of the senses, attested by

Galen: «Poor mind, you get your evidence (πίστεις) from us, then you

demolish us. Our fall is your demolition». But it seems much more likely


that b 125 is the very remark that he is summarising as citation (b)

at the start of n. This is always translated as telling us that in his

Kratunteria, despite having “promised” to ascribe τὸ κράτος τῆς πίστεως

to the senses, he none the less is found condemning them. I doubt if

this notion of an unfulfilled promise can be sustained by the Greek.

Ὑπισχνεῖσθαι plus future infinitive is “promise”, but that sense seems

virtually unattested with any other tense of the infinitive40. The rule

is observed elsewhere, incidentally, not only by Sextus, but also by

Posidonius41. Here, used with an aorist infinitive, it surely means

“profess”, “claim”42. If so, in his work the Kratunteria, Democritus,

claimed (not promised) to ascribe command43 of pistis to the senses, but

none the less condemned them. Surely this “claim” is the very one im-

plied in the reply of the senses (b 125): the mind depends on the senses

for its pisteis, so is in no position to condemn them. If that is right44,

we can infer that the famous reply of the senses occurred in the Kratun-

, and that Sextus, or rather Posidonius, has not after all suppressed

it, even if he has chosen not to quote it verbatim.

Moreover, although the expression τὸ κράτος τῆς πίστεως does not

occur in the reply of the senses itself, we can with reasonable confidence


attribute it to Democritus himself, not to Posidonius or Sextus. For

κράτος clearly picks up the theme of the Kratunteria, which seems to

have been a work devoted to adjudicating the struggle for “command”

between intellect and senses45. As for πίστις, although this is the term

used, along with its cognates, throughout M vii 89-140 for evidential

reliability46, it also features in the reply of the senses with the weaker

meaning “evidence” or “assurance”, without any obvious connotation

of ultimate reliability. In the power struggle, it seems, the senses arrogat-

ed evidential supremacy to themselves (b), but were then nevertheless

condemned as untrustworthy (c)-(g).

No contradiction need be involved: Democritus could quite consis-

tently hold both that the senses do indeed command the evidence availa-

ble to the mind, and that we know nothing for certain, because the senses

are themselves unreliable47. Posidonius’ point is not that Democritus

contradicted himself within a single work (let alone that he broke a

promise), but that his remark about the priority of the senses was not,

in its context, an attempt to make perception an acceptable criterion.

Posidonius’ overall strategy in the passage is clear enough. In i he

quotes Democritus’ best-known condemnation of the senses. In v he will

go on to present him, in accordance with this, as elevating reason above

perception as a criterion. He could if he had wished have restricted him-

self to those two moves. But in between he chooses to mention a possible

obstacle: Democritus’ restoration to the senses of control over evidence

in his Kratunteria. To accommodate the obstacle, he points out, first, that

this restoration was not enough to prevent Democritus, in the very same

work (n), as well as elsewhere (m), from numerous pessimistic utterances

about the possibility of any knowledge (texts (c)-(g)). He thus succeeds

in showing that, even if Democritus in a way privileged perception over

reason, it was not as a criterion.

But the price of the achievement is to have drawn attention to pas-

sages (c)-(g), in which Democritus appeared to deny any route to know-


ledge, i.e. any criterion at all. How then is Posidonius to rescue the logos

criterion? In iv he concedes that passages (c)-(g) amount virtually48 to a

complete denial of cognition — although, he adds significantly, Democri-

tus’ only specific attacks are on the senses. Clearly it is part of Posidonius’

damage-limitation exercise that (c)-(g) should be seen as primarily directed

at the senses, and not at the powers of reason as such. This in fact fits

(c) well, and is at least compatible with the general disavowals of

knowledge in (d), (e) and (g). The only serious doubt concerns (i), in which

the expression ἐπιρυσμίη ἑκάστοισιν ἡ δόξις can with some plausibility

be read as questioning the validity of all “belief”, on the ground that

it is nothing more than a mechanical realignment (ἐπιρυσμίη) of the soul

atoms49. But the alternative reading, which ties it solely to sense-

perception («seeming for each person is an influx») has had more than

enough supporters over the years (starting with Hesychius50) to make

Posidonius’ assertion a perfectly defensible one.

Thus Posidonius insists, in support of his interpretation, that

Democritus’ primary attack is always on the senses. He concedes that

in some works the dependence of the mind on the senses allows it to

be dragged down with them. His final move, in v, is to draw attention

to an important exception: that in one work, the Canons, reason was

permitted to be independent of the senses, and was granted the status

of a criterion. Once more we are witnessing Posidonius’ pluralistic ap-

proach. He is quite happy to leave intact the “no criterion” interpreta-

tion of Democritus, which we know was current in his day51. He even

supplies us with the evidence for it. As with Empedocles, so too with

Democritus, what matters most is that somewhere in his writings the iden-

tification of logos as criterion can be found.

The crucial text from the Canons comes in two parts. The first, (h),

establishes Democritus’ separation of the two forms of knowledge. The


second, (i), supplies the further evidence needed to show that he prefers

“genuine” knowledge to “bastard” knowledge. Notoriously, (i) presents

a crux by apparently breaking off in mid sentence ἀλλ’ ἐπὶ λεπτότερον.

I shall not recount the numerous ingenious emendations that have been

proposed, but shall try instead a new approach.

We have already identified Posidonius as the source of the entire

passage, 89-140, and noted his predilection for verbatim quotation. What

can we learn from the rest of the passage about his methods of quotation?

Two points in particular. First, like many ancient collectors of quotations,

he is prepared to break up a continuous passage in order to interject

glosses of his own. A clear case is his citation of Empedocles at 123-4.

The two excerpts are metrically continuous, the interruption having oc-

curred in the middle of a line, and editors have no hesitation in reattach-

ing them. Posidonius interrupts merely in order to point out the lesson

to be learnt in the next bit. He introduces the first excerpt by saying

(123): «Concerning the fact that the discrimination of truth does not lie

in the senses, he says the following». Then comes the first excerpt, after

which he introduces the second by saying (124): «And concerning the fact

that the truth is not totally unattainable, but attainable as far as human

logos can reach, he makes this clear by adding to the previous words (τοῖς

προκειμένοις ἐπιφέρων)» — whereupon the quotation continues.

Second, note that in the passage just quoted the regular verb for

“add” in citations, έπιψέρειν, is used to indicate a directly adjacent addi-

tion. When there is a gap, as between the two Heraclitean quotations

at 132-3, he scrupulously indicates this with the expression (133) ὀλίγα

ἐπιφέρει, «a little later he adds»52.

If we apply these two lesson to the Canons quotations in v, we can

see that there is a strong prima facie case for reading the two excerpts

as continuous, since they are joined by a simple έπιφέρει, with no qualifi-

cation to indicate a gap53. As with the Empedocles passage, Posidonius’


interjection signals not an interval in the text, but the need for a guiding

gloss. The first excerpt establishes the existence of the two forms of

knowledge, and pauses after ἀποκεκριμένη δὲ ταύτης54, since this is

the firmest indication that the two forms are separate and independent.

He then glosses the following excerpt: «Then, by way of judging the

genuine one superior to the bastard one, he adds these words». And the

second excerpt, which ensues, does indeed imply something about the

superiority of γνησίη γνώμη to σκοτίη γνώμη, since it says that the

former takes over where the latter can no longer cope.

I propose, therefore, that we should try our hardest to follow these

clues, and read the two excerpts as continuous55. Suppose that we

place nothing more than a comma between them. The sense at the join

(starting at ή δέ γνησίη in (h) can then run: «The one which is genuine,

but separate from this one, (is) when the bastard one is no longer able

either to see in the direction of greater smallness». The harsh ellipsis

of the verb does at least fit the overall style of the passage, which has

already suppressed it in the equally laconic μέν clause with the words

σκοτίης μὲν τάδε σύμπαντα56.

That Posidonius’ gloss should thus interrupt the grammar does not

seem to me problematic. What I, like many others, do find incredible

is that in the second excerpt he should have broken off the quotation

half way through a subordinate clause, thus obliterating the grammar and

coherence of the sentence as a whole. Scholarly discomfort on the point

has given birth to many ingenious editorial supplements to the text of

Sextus to complete the sentence. But an easier solution, I suggest, would


be to leave the text intact, except for a tiny repunctuation and re-

accentuation near the end57. That is, we should simply delete the com-

ma after αἰσθάνεσθαι, and accentuate the next word ἄλλ’, “other

things”. The whole saying could then read:

γνώμης δὲ δύο εἰσὶν ἰδέαι, ἡ μὲν γνησίη ἡ δὲ σκοτίη. καὶ σκοτίης μὲν

τάδε σύμπαντα, ὄψις ἀκοὴ ὀδμὴ γεῦσις ψαῦσις. ἡ δὲ γνησίη, ἀποκεκρι-

μένη δὲ ταύτης, ὅταν ἡ σκοτίη μηκέτι δύναται μήτε ὁρῆν ἐπ’ ἔλαττον,

μήτε ἀκούειν μήτε ὀδμᾶσθαι μήτε γεύεσθαι μήτε ἐν τῇ ψαύσει αἰσθά-

νεσθαι ἄλλ’ ἐπὶ λεπτότερον.

«Of knowing there are two forms, the one genuine, the other

bastard. And of the bastard kind this is the complete list: sight, hearing,

smell, taste, touch. The one which is genuine, but separate from this

one, is when the bastard one is no longer able either to see in the direc-

tion of greater smallness, nor to hear or smell or taste or sense by touch

other things in the direction of greater fineness».

On this reading, our incapacity to see ἐπ’ ἔλαττον in the first limb

is balanced in the second58 by our incapacity to hear, taste etc. “other

things” (i.e. other sense objects than those of sight) ἐπί λεπτότερον. This

distinction between smallness and fineness embodies a perfectly reason-

able point about perceptual thresholds. For vision, the threshold below

which perception fails is standardly one of size: hence ἐπ’ ἔλαττον. For

the other senses, the relevant threshold is not that of size. Rather, certain

flavours, odours, sounds and textures are too “fine” or “subtle” to

detect: hence ἐπὶ λεπτότερον.

I do not claim that this is an immaculate specimen of Greek prose

style. But it is, I suggest, convincing enough to render emendation both

unnecessary and risky59.


Encouraged, then, by confidence in our source Posidonius and in

his meticulous methods of textual citation, we can perhaps, along these

lines, see our way to vindicating the integrity of the text as he has

reported it.

It now remains only to consider vi, where Posidonius adds from

one Diotimus a quite different interpretation of Democritus. There is

an old debate as to whether this is the Democritean Diotimus of Tyre,

or the Stoic Diotimus. Of the Democritean, we know only that he had

his own distinctive formula for the ethical telos60. Of the Stoic, we

know that he was a detractor of Epicurus, on whom the Epicurean Zeno

of Sidon was said to have taken lethal revenge61 — an anecdote which

places him in the late second or early first century B.C.

Majority opinion has long favoured the Democritean Diotimus as

the source cited in our passage62. Who, after all, is likelier to have

propounded an interpretation of Democritus than a Democritean? But

the case for the Stoic is much strengthened by the identification of our

source as Posidonius. As a Stoic contemporary of this Diotimus, he is

almost bound to have known him personally. Moreover, in Diog. Laert.

x 3-4 Posidonius is listed directly after Diotimus as a fellow detractor

of Epicurus. One of the charges brought by the detractors listed there

is that Epicurus plagiarised his doctines from, among others, Democritus.


And the interpretation of Democritus quoted from Diotimus itself be-

trays signs of a directly anti-Epicurean motive of just this kind. It looks

like an attempt, on the flimsiest evidence, to show that Epicurus’ three

criteria of truth — αἰσθήσεις, πρόλήψεις and πάθη — were not original

to him but anticipated by Democritus63. He even listed them in the

same order as Epicurus64. What more likely, then, than that Posidonius

knew the novel interpretation of Democritus from his fellow anti-

Epicurean65 Diotimus66, and added it for good measure. This weakness

for citing contemporary sources is one which we have already noted in

Aenesidemus, and which Posidonius shows himself to share when he

quotes an obscure parallel from Asclepiades in exegesis of Anaxagoras (91).

4. Epicurus

In dealing with the report of the Epicurean theory on the criterion,

I shall try to be briefer67. It occurs in the final doxographical section,

vii 141-260 (see table supra, p. 22), which covers first the entire Academy

from Plato to Carneades, in chronological order, followed by the Cy-


renaics, Epicurus, the Peripatetics, and lastly the Stoics. Again, it has

certain recurrent features which distinguish it, as a whole, from the

preceding two divisions.

(1) Its author has a consuming interest in the notion of enargeia,

which recurs numerous times and in connexion with every school dis-

cussed68, despite the slightness of the interest shown in it by Sextus in

the subsequent critical section of the book.

(2) There are practically no verbatim quotations. In common with

the Aenesideman section, and in contrast to the Posidonian section, the

author’s main resource is paraphrase. But he differs from the Aeneside-

man section too, in his very heavy reliance on technical terms as a key

to interpretation69. Of course, to a large extent this difference reflects

the fact that his philosophers are later in date and use a more obviously

technical vocabulary. But it is instructive to see him applying the same

method, quite inappropriately, even to Plato. In his sole verbatim quota-

tion (142), he cites Timaeus, 27 d in order to extract from it the word

περιληπτόν and to convert it, with little plausibility, into the technical

term περιληπτικόν, “comprehensive”, i.e. inclusive of both enargeia and

truth, which he then reconverts into the Stoic technical term καταληπ-

τικόν (144). This contrasts with Posidonius’ way of exploiting the

Timaeus (93, 116, 119), as a source of ideas, not terminology.

(3) The author is far more addicted than either Aenesidemus or

Posidonius to illustrating a point by the use of examples70 and


Who is it this time? There are strong reasons for suspecting our

source to be none other than Antiochus of Ascalon72 — yet another


contemporary of Aenesidemus. One initial clue is that Antiochus is twice

cited by name in the passage. At 162 it is for his elucidation of Carneades

on one particular point. At 201 Sextus quotes a cryptic passage from

book ii of Antiochus’ Canonica, about an unnamed doctor who placed

the criterion entirely in the senses, adding that Antiochus seems here

to be hinting at Asclepiades. If Antiochus is the source of the entire

passage 141-260, why should he be named just at these two points73?

In the first passage, it would be because he had explicitly presented the

elucidation of Carneades as his own addition. In the second, it looks

as if Aenesidemus, who was clearly exceptionally interested in his con-

temporary Asclepiades74, was struck by a passing allusion75 in An-

tiochus’ book, and preserved it verbatim as evidence for his conjecture

that Asclepiades was meant. The two named citations may not prove the

hypothesis that Antiochus’ Canonica was the source throughout, but they

are fully consistent with it76.


Although these mentions of Antiochus’ name are a useful clue, and

the title of his work, Canonica, is entirely apposite to a history of theories

on the criterion, the strongest reasons for identifying him as the source

are doctrinal. Enargeia (Latin perspicuitas) is a central notion in Antiochus’

epistemology as we know it from Cicero’s Academica (ii 11-62)77.

Some facts, he holds, are self-evident or self-certifying, not in need of

rational proof — especially, though not exclusively78, those grasped by

the senses. Reason is derived from these self-evident cognitions. It can

in turn protect them from sceptical assault, but is assigned no indepen-

dent criterial status. Such is the position which Antiochus adopted as

the best available bulwark against the fallibilist epistemology from which

he was setting out to rescue the contemporary Academy.

It is only too easy to see how the passage at M vii 141-260 could

have been written to trace the historical evolution of just this concept

from Plato, with whom it opens, to the Stoics, with whom it closes —

a line of descent which Antiochus acknowledged as his own79. It shows

Plato offering a complex criterion, in which reason and perceptual enar-

are interdependent. The Academy then develops this criterion inade-

quately. Plato’s early followers, Speusippus and Xenocrates, separate the

two criteria, even though they both acknowledge some kind of interaction

between them too. And his later followers Arcesilaus and Carneades try

to abolish criteria altogether (including an attack on enargeia, 160-3), but

significantly they too are in the end «virtually compelled» by the prac-


tical demands of life to accept one for themselves80. Meanwhile outside

the Academy Plato’s philosophical relatives the Cyrenaics are defending

an over-condensed version of his position, in which enargeia is emphasised

at the expense of logos (201); and that tendency is further developed

by Epicurus. A third, and more respectable, line of development is then

added. The Peripatetics offer an enarges complex criterion, whereby both

perception and intellect directly grasp self-evident objects, emphasising

that logos, although dominant, is itself derivative from perception. This

third tendency is then seen through to its culmination by the Stoics,

with their doctrine of sensory phantasia kataleptike, which, once they

have finally resolved their internal quarrels about it, is recognised to be

enarges to the point of irresistibility (257)81.

This is a typically Antiochean view of philosophical history. That

Plato’s own position had already gestured towards Stoic katalepsis is im-

plied, as we have seen, by interpreting his περιληπτόν as embodying the

Stoic concept of the καταληπτικόν. And despite the wrong directions

taken by his earlier followers within the Academy, the story seems to

continue, the Peripatetics and Stoics did find the right path82: they

eventually established a sound Platonist epistemology — very much like

the one Antiochus’ follower Lucullus is found defending in Cicero’s



What will be thought surprising is his inclusion of the Cyrenaics

and Epicureans, who were never regarded as belonging to the Platonist

tradition. His wish to borrow them as honorary allies clearly arises not

from any general philosophical approval but from his conviction that

they had, if nothing else, recognised the supreme criterial power of enar-

In the case of the Cyrenaics we are explicitly offered the justifica-

tion that they are Plato’s close philosophical relatives, fellow disciples

of Socrates (190). In the case of the Epicureans, no such kinship could

be invoked. But the fact that Epicurean hedonism was seen by An-

tiochus, as by others, as a revised version of Cyrenaic hedonism84 may

have been sufficient to justify the linking of the two schools here. And

it is instructive to note that virulent anti-Epicureanism is not an apparent

feature of Antiochus’ outlook85, as it is for the Stoics and the New

Academics. In Cicero’s Academica his Antiochean spokesman not only

invokes the Cyrenaics as allies (ii 21), but also treats Epicurus with a

relative lack of hostility: Epicurus rightly required the wise man to

separate opinion from enargeia (n 45), even though he failed to see this

aim through to completion (ibid.), and went too far in insisting that all

perceptions are true (ii 19, cfr. 101).

There is therefore no ground for doubt that even the Epicurean sec-

tion, 203-16, originates from Antiochus’ Canonica. There are, I believe,

no positive reasons to prefer Natorp’s derivation of it from the

Epicurean Demetrius of Laconia86. And in fact there are very strong

grounds for holding that the source cannot possibly be an Epicurean at

all. Since I have argued this at length elsewhere87, I shall endeavour to

be brief.


In the closing paragraphs of the Epicurean section, we are offered

a summary of the twin Epicurean principles, epimarturesis and ouk an-

the object no doubt being to show how even in scientific

discovery enargeia was the sole Epicurean criterion88. In the course of

this, the latter principle, which I translate “non-contestation”, and its

converse antimarturesis or “contestation”, are illustrated with the stan-

dard Epicurean example of the inference from motion to void. The

description includes the following89:

«Non-contestation (ouk antimarturesis) is the following (akolouthia)

from that which is apparent of the non-apparent thing posited and be-

lieved. For example, Epicurus, in saying that there is void, which is non-

apparent, confirms this through the self-evident fact of motion. For if

void does not exist there ought not to be motion either, since the moving

body would lack a place to pass into, as a result of everything’s being

full and solid. Therefore the non-apparent thing believed is uncontested

by that which is apparent, since there is motion. Contestation (antimartu-

), on the other hand, is something which conflicts with non-

contestation. For it is the elimination (anaskeue) of that which is apparent

by the positing of the non-apparent thing. For example, the Stoic says

that void does not exist, judging something non-apparent, but once this

is posited about it, that which is apparent, namely motion, ought to be

coheliminated (sunanaskeuazesthai) with it. For if void does not exist,

necessarily motion does not occur either, according to the method already

demonstrated ».

It is well recognised that this anaskeue terminology does not go back

to Epicurus, but belongs to the debate between Epicureans and their

probably Stoic opponents reported in Philodemus, De signis90 — a

debate datable to the late second and early first centuries B.C. In that

debate, most of the Epicureans accept that the “following” (akolouthia)


of something non-apparent from an apparent sign can be either by

anaskeue or by “similarity”. “Similarity” sign-inferences rest on the sup-

posed resemblance between two items, whether direct, as in inferences

from the properties of human beings we know to human beings we do

not know, or analogical, as in inferences from the properties of

phenomenal bodies to those of atoms. But other purportedly cogent sign-

inferences, which do not rely on resemblances, are attributed to anaskeue.

The inference passes the anaskeue test if it is found that to deny

(“eliminate”, anaskeuazein) the non-apparent thing signified is ipso facto

to deny the existence of the sign. A standard example of anaskeue in

the De signis is the inference from the existence of motion to that of

void91: the inference is held to be sound, not because of any “similar-

ity” between motion and void, but because of some kind of conceptual

or physical dependence.

In the Sextus passage that very same motion-void inference is used

twice, to illustrate first ouk antimarturesis and then antimarturesis. And

at the second occurrence it is explicitly presented as an anaskeue infer-

ence. So far so good. But unfortunately our author has completely missed

the point. The De signis Epicureans say92:

«That if there is motion there is void we apprehend in no other

way than by the method of similarity, establishing that it cannot be that

motion is accomplished in the absence of void. Thus having surveyed

everything that accompanies moving objects in our experience, in the ab-

sence of which we see nothing moving, in this way we claim that every-

thing which moves in any way moves similarly, and by this method we

make a sign-inference that there cannot be motion without void. Hence

if this method has no probative force, the elimination (anaskeue) method,

which is wholly confirmed by and through it, has no cogency either».

Although the Epicureans are unable to call the motion-void inference one

by “similarity”, they insist here, as elsewhere, that the real work of “con-

firmation” is done in a separate stage, in which the similarity between


numerous observed cases of motion establishes that all motion requires

empty space. The anaskeue inference, «Since there is motion, there is

void», is a further purely formal step, applying the lessons learnt by the

similarity method, with no independent probative force.

Our author, the putative Antiochus, has made the mistake of placing

all the emphasis on the anaskeue inference itself, even saying that it is

this inference from motion to void which “confirms” the existence of

void. Philodemus’ Epicurean master Zeno of Sidon insists, on the con-

trary, that all the work of “confirmation” is done in the separate

“similarity” inference. And although other contemporary Epicureans

mentioned by Philodemus, including Demetrius of Laconia, differ among

themselves about how to regard anaskeue — some affirming, others deny-

ing, that it is any kind of sign-inference at all — they are all agreed

that it has no power to “confirm” anything93.

As for ouk antimarturesis, although the term itself is barely visible

in the De signis, the concept of “no counterevidence” features prom-

inently there, and it is consistently presented as a way of confirming

inferences by “similarity”, not those by anaskeue94.

The conclusion is irresistible that what we have in Sextus is the

product of a faulty reading of one of the Epicurean works whose contents

are reflected in Philodemus, De signis95. Our source knows that Epicu-

rus has a scientific principle of ouk antimarturesis, and in order to find


out about it he has followed a procedure which should by now be becom-

ing familiar: he has turned not to the writings of Epicurus, but to a

contemporary Epicurean textbook. Failing to find in it the actual term

ouk antimarturesis, he has mistakenly identified it with the anaskeue

method, exemplified by the inference from motion to void.

Since our source turns out to be a non-Epicurean observer of this

late second-century B.C. Epicurean theory of signs, the hypothesis that

he is Antiochus is much strengthened. His explanation of ouk antimartu-

is the only formal one that we have, but, sadly, it must be discounted

as almost completely wrong. Does this mean that we must also disregard

the earlier part of his report, 203-10, on the truth of sense-impressions?

Not necessarily. But we must at least treat it with all due caution, pre-

pared to find not only misunderstandings, such as we have witnessed,

but also distortions due to Antiochus’ own quasi-historical purposes. One

likely case of this is his failure to mention the Epicurean criterion prolep-

— just as he subsequently omits any mention of prolepsis as a Stoic

criterion. For anyone attuned to Stoic thought as Antiochus was, prolep-

will imply logos, which according to the Stoic definition is composed

of them96. Hence I would guess that Antiochus’ reluctance to draw at-

tention to their criterial status for either Epicurus or Chrysippus reflects

his determination to discount logos as an independent criterion.

One worrying passage is the opening of the Epicurean section, 203-

5. In outline, the argument attributed to the school is:

(1) All pathe are true; e.g. what causes pleasure in us eo ipso really

is pleasant.

(2) Phantasiai are themselves pathe.

(3) Therefore what causes a phantasia (the phantaston) must really

be such as it appears97.

(4) This result conforms to the definition of a true phantasia as one

ἀπὸ ὑπάρχοντος and κατ’ αὐτὸ τὸ ὑπάρχον.


The terminology of (4) is directly taken from the Stoic definition

of phantasia kataleptike, suggesting that Antiochus is setting out to

present Epicurean enargeia as pointing towards the Stoic criterion to

which we know that he himself adhered98. This tendentious reporting

leads on to a further worry. Do steps (l)-(3) really contain an authentic

Epicurean argument? Their inference is from (1) the objective truth of

pathe, to (3) the objective truth of phantasiai, by way of (2) the premise

that phantasiai are themselves a kind of pathe. As Gigante has pointed

out99, this premise is not attested Epicurean doctrine. Pathe are an in-

dependent criterion (another fundamental tenet altogether omitted by

Antiochus!), with sensory phantasiai ranked alongside them, not sub-

sumed under them100. But we should not seek to remedy the problem

by emendation101. Rather, we should note that the definition of phanta-

as a kind of pathos is an item of Stoic doctrine102, and suspect that

Antiochus is supplying the premise himself.


What would Antiochus’ motive be? It is not hard to guess103. The

Cyrenaics, as presented in the preceding text, had got as far as recognis-

ing that pathe are enarge in themselves, but had denied them any cognitive

hold on external fact. To make the Cyrenaics into worthwhile allies, An-

tiochus needs someone to continue that process, first giving pathe cogni-

tive access to the external world, then going on to attach the same objec-

tive enargeia to sensory phantasiai, thus pointing the way towards Stoic

phantasia kataleptike. The Epicureans, it seems, were chosen for the job.

There seem to be grounds for pessimism. The Epicurean section

starts with what may well be pure historical fabrication on Antiochus’

part, designed to suggest steady progress towards the truth as he con-

ceives it. And it ends with a well-intentioned but hopelessly bungled at-

tempt to explain Epicurean scientific principles of inference. Can we sal-

vage anything? Miraculously, we can. The central section of the passage

(206-10) is a brilliant, albeit incomplete104, defence of the Epicurean

dictum that all sense-impressions are true. While its accuracy cannot be

directly proved, it appears to contain no blunders, anachronisms or wilful

misrepresentations. On the contrary, the terminology, the concepts and

the examples are all authentically Epicurean105. Antiochus, it seems,

did not always invent history or misread his sources.

5. Closing remarks

There are numerous further passages in which both Democritus and

Epicurus are cited and criticised by Sextus. An adequate investigation

would need to examine individual passages, themes and contexts in detail


comparable to that in which I have treated the long survey of views on

the criterion. And that would require not a single paper, but a book.

I have little notion of what the results of such an investigation would

be. Much would depend on general conclusions about the Pyrrhonists’

handling of philosophical history when compiling diaphoniai and other

refutations. But I am prepared to wager that in any such study the con-

temporaries and near-contemporaries of Aenesidemus will continue to

loom as large as they have done already. It is, for example, well known

that, apart from Epicurus himself, the one named Epicurean spokesman

to whom Sextus pays attention is Demetrius of Laconia, yet again Ae-

nesidemus’ near-contemporary.

I am also conscious that I have said very little about Sextus himself.

Of course, I do not mean to reduce him to a mere copyist of Aeneside-

mus. There is plenty of post-Aenesideman Pyrrhonism in his works, and

much of the medical input could well be his own106. But so far as con-

cerns the reporting of doctrine, outside the medical and Pyrrhonist tradi-

tions the latest named thinkers and sources in Sextus’ works are again

and again writers of the first century B.C.107: Posidonius, Diotimus,

Clitomachus, Antiochus, Philo of Larissa, Asclepiades, Charmadas, Ae-

nesidemus himself, and Demetrius of Laconia108. When we read the

history of atomism in Sextus, these are the people through whose eyes

we must expect to be viewing it.

My suggestions about this passage, which will constitute the bulk of the

present paper, owe much to a seminar held in Cambridge in 1985, with Myles

Burnyeat and others. I am also grateful to Margaret Atkins, Jonathan Barnes, Myles

Burnyeat, Ian Kidd, John Procopé, and Harold Tarrant for their helpful written

comments on an earlier draft.
Linked: PH iii 32, Μ ix 363, x 45, 181, 318. Juxtaposed: PH ii 23-5, Μ vii

265-7, 321, viii 139, ix 24-5, 42-3. Contrasted: M vii 369, viii 6-9, 62-3, 184-5,

355. For discussions of Sextus’ treatment of atomism, see especially F. Decleva

Democrito e Sesto Empirico, in Democrito e l'atomismo antico, Atti del

Convegno internazionale, Catania, 18-21 apr. 1979, a c. di F. Romano, Catania

pp. 393-410 and M. Gigante, Scetticismo e epicureismo (“Elenchos” iv), Napoli

esp. pp. 109 ff. Among earlier treatments, P. Natorp, Forschungen zur Ge-

schichte des Erkenntnisproblems im Altertum
, Berlin 1884, pp. 256-85 is outstanding.
M i 1-6. The attack on Epicurus here stems from Timocrates, his renegade

pupil; see my Epicurus and his professional rivals, in Etudes sur l’épicurisme antique,

ed. by J. Bollack, A. Laks (“Cahiers de Philologie” i), Lille 1976, pp. 119-59.

At M i 272, 281-5 the contrast between Epicurus and Pyrrho is less pronounced.
For a possible example, F. Decleva Caizzi, art. cit., p. 402.
Diog. Laert. x 13. Cfr. M.X.G., argued to be a Pyrrhonist work by J. Man-

, ‘De Melisso Xenophane Gorgia’: Pyrrhonising Aristotelianism, «Rheinisches Mu-

seum», cxxxi (1988) pp. 239-76, which speaks of the «so-called logoi of Leucippus»

(980 a 8-9).
I shall not be considering the concept of a “criterion” as such, on which see

G. Striker, Κριτήριον τῆς ἀληθείας, Göttingen 1974; A. A. Long, Sextus Empiricus

on the criterion of truth
, «Bulletin of the Inst. of Class. Studies of Univ. of London»,

xxv (1978) pp. 35-49; J. Brunschwig, Sextus Empiricus on the “Kriterion”: the Skeptic

as conceptual legatee
, in The Question of Eclecticism, ed. by J. M. Dillon, A. A. Long,

Berkeley 1988, pp. 145-75.
Thus e.g. R. McKim, Democritus against skepticism: all sense-impressions are

Proceedings of the First International Congress on Democritus, ed. by L. G.

, Xanthi 1984, pp. 281-90. Cfr. C. Farrar, The Origins of Democratic Think-

, Cambridge 1988, pp. 205, 207. The latter book contains (pp. 197-215) a particu-

larly powerful and coherent defence of a non-sceptical interpretation of Democritus.

My own aim as regards Democritus will be limited to showing how Sextus’ evidence

should be read, without prejudice to the interpretation of any overall position he

may be thought to have adopted. But for the record, I am not fully convinced that

there was any such position.
I go along with, but cannot here argue for, the view (see e.g. G. Calogero,

Studi sull'Eleatismo, Roma 1932, Firenze 1972(2)) that the M.X.G. version of Gorgias’

On not being is closer to the original than Sextus’. But one aspect which remains

constant in both versions is the concessive structure of the overall argument: p, and

even if not p, q, and even if not q, r. This device is a direct legacy of Gorgias

to Aenesideman scepticism, which uses it widely (I know of no serious philosophical

use of it between Gorgias and Aenesidemus, other than in Plato’s Charmides). Aenesi-

demus’ authorship of this section would thus help explain the massively dispro-

portionate space allotted to Gorgias in it (65-87).
Cfr. Sext. Emp. PH i 100-13. I find this more believable than the suggestion

of J. Annas and J. Barnes, The Modes of Scepticism, Cambridge 1985, p. 85 that

the material is originally Protagorean and has been borrowed by the Pyrrhonists:

the ancient tradition on Protagoras’ theory of truth stems almost entirely from

Plato’s Theaetetus, and it would be remarkable if Sextus or his source had alone

had access to substantial textual material of independent origin. Cfr. the immediately

following account of Euthydemus and Dionysodorus (64), which recasts along similar

Pyrrhonist lines their portrayal in Plato’s Euthydemus.
For Xenophanes as forerunner of Pyrrhonism, see M viii 325-6 (itself very

like vii 52), Diog. Laert. ix 72, etc. For Metrodorus and Anaxarchus, see Pyrrh.

frr. 1 A, 23-7 Caizzi.
However, one reason for doubting that Sextus is the compiler is that he is

much less keen than Aenesidemus to acknowledge forerunners of Pyrrho as genuine

sceptics: PH i 210-25. Whether those who rejected all criteria could be Sceptics,

rather than negative dogmatists, was debatable. At PH ii 18 and M vii 443-4 Sextus

suggests not. But at Μ vii 26 and viii 1 he implies that they can: hence Mutschmann

may be unjustified in excising the Sceptics from the list of those who deny all criteria

at M vii 49
The stage-painting motif is the subject of a revolutionary forthcoming paper

by Myles Burnyeat, too complex to summarise here.
In saying this, I correct my own previous preference for the Sextan version,

The protagonists, in Doubt and Dogmatism, ed. by M. Schofield et al., Oxford 1980,

pp. 1-17, at p. 10, and The motivation of Greek skepticism, in The Skeptical Tradition,

ed. by M. Burnyeat, Berkeley 1983, pp. 9-29, at p. 14.) Cicero’s Latin, ne id ipsum

quidem, nescire aut scire, scire nos,
is clumsy, but would work well in Greek, thanks

to the articular infinitive: οὐδὲ αὐτὸ τοῦτο, τὸ εἰδέναι ἢ μή, εἰδέναι ἡμᾶς. Eusebius’

version (praep. evang. 14.19.9), οὐδεὶς ἡμῶν οὐδὲν οἶδεν, οὐδ’ αὐτὸ τοῦτο, πότερον

οἴδαμεν ἢ οὐκ οἴδαμεν, could easily have arisen as an alternative attempt to simplify

the same original.
Neater: simplification of the language. More schematic: phrased to define

Metrodorus’ relation to the supposed Socratic dictum «I know that I know nothing».

More Pyrrhonian: the doubts about the world’s existence, omitted in Sextus’ version,

are not part of the usual Pyrrhonist repertoire.
The enargeia division may appear non-chronological at first glance, but I

shall try to show below that it in fact takes three separate traditions, each in chrono-

logical order.
H. Tarrant, Scepticism or Platonism? The Philosophy of the Fourth Academy,

Cambridge 1985, p. 104, points out that two of these pairings — epistemonic/doxastic

logos and divine/human logos — also surface in Middle Platonism (Alcin. Didaskalikos,

4). I doubt if we know enough about the earlier history of these antitheses to help

us identify Sextus’ source. The former has obvious Platonic antecedents (cfr. M

vii 145-8), but it must be remembered that the doxa/episteme contrast is Stoic too.

Thus it cannot do much to strengthen Tarrant’s preference for Antiochus over Posi-

donius as source (see infra, note 72).
That Democritus should have acknowledged the “divine logos” which Her-

aclitus, according to 127-31, identified with air, may stretch credulity too far; but

it might have been argued on the basis of Democr. a 78, a 106 and b 30 D.-K.
Not quite the same practice, but consonant with it, is the use of the poets

to expound the thought of Heraclitus: 128.
This may be a special case, to the extent that Democritus himself had

reportedly already invoked Anaxagoras on the point. There is also the question

whether the parallels had already been added by Diotimus rather than deriving from

our principal author. Cfr. infra, note 63.
For Asclepiades’ dates, see E. Rawson, The life and death of Asclepiades of

, «Classical Quarterly», xxxιι (1982) pp. 358-70.
It is mentioned in passing as a possibility by I. Kidd, Posidonius, Cambridge

1988, ιι, p. 342, and as an unpalatable idea by H. Tarrant, Agreement and the self-

evident in Philo of Larissa
, «Dionysius», v (1981) pp. 66-97, at p. 80. Prior to them,

it seems to have occurred only in A. E. Taylor, A Commentary on Plato’s ‘Timaeus’,

Oxford 1928, pp. 35-6 note. But it was, in addition, the unanimous verdict of the

1985 Cambridge seminar mentioned in note 1 above.
See especially W. Burkert, Lore and Science in Ancient Pythagoreanism,

Engl. tr. Cambridge (Mass.) 1972, pp. 54 ff.; J. Mansfeld, The Pseudo-Hippocratic

Tract. Περί ἐβδομάδων
, Assen 1971, p. 156 note.
If the whole passage is from Posidonius, why name him specifically only

here? Presumably because the original said something like “On my interpretation

of the Timaeus...”.
Especially 102 on συναπτόμενα, ἡνωμένα and διεστῶτα, cfr. S.V.F. ii 366-8,

and Μ ix 78 ff., a very Posidonian passage, where this conceptual scheme is the

basis for cosmic sumpatheia.
I have only noticed one detail which could be suspected to reflect Aeneside-

mus’ own mediation. The antithesis ἀνέξοιστον/ἐξοιστόν attributed to Empedocles at

122 might be thought to reflect Aenesidemus’ preoccupation with Gorgias (65 ff.).

But even here a likelier explanation is that it represents some dogmatists interpreter’s

attempt to synthesise Empedocles with his pupil Gorgias.
The Democritean application of this principle to the stratification of peb-

bles on a beach (117) itself finds an echo in Posidon. fr. 229 E.-K.: see F. Decleva

art. cit., p. 398.
Cfr. 119, where the Timaeus (67 a-c) definition of sound as a blow caused

by air is rewritten in Stoic terms as ἀέρα πεπληγμένον (cfr. S.V.F. ii 139-41,

etc.). 129-30 also has a strong Stoic ring to it (cfr. S.V.F. i 141) — stronger, at any

rate, than the Aenesideman echoes which led Diels (Dox., pp. 209 ff.) to assign the

Heraclitus doxography to Aenesidemus himself. J. Mansfeld too, Doxography and

dialectic: the Sitz im Leben of the “Placita”
(“Aufstieg und Niedergang der röm.

Welt”, ii 36.4), Berlin 1990, pp. 3056-229, at pp. 3066-7, 3164, sees in it the Ae-

nesideman interpretation of Heraclitus recorded at Μ vii 349. But there is no need

to infer that Aenesidemus is the ultimate source of the Heraclitean section, rather

than its transmitter. As Mansfeld notes, Aenesidemus’ interpretation of Heraclitus

is already strongly mediated by a Stoic source, and the hypothesis that this whole

passage came to Aenesidemus from Posidonius may help unmask the latter as that

source — as was already argued by K. Reinhardt, Kosmos und Sympathie, München

1926, pp. 192 ff., and by U. Burkhardt, Das Angebliche Heraclit-Nachfolge des Skep-

tikers Aenesidem
, Bonn 1973, pp. 81 ff. One critic has pointed out, as an objection

to my thesis, that τὸ περιέχον (129-30), used of the atmosphere, looks Aristotelian

rather than Stoic. Maybe so, but it is certainly Posidonian: see frr. 49.71, 49.331,

169.92 E.-K.
Cfr. Marc. Aurel. iv 46 = 22 b 72 D.-K., ᾧ μάλιστα διηνεκῶς ὁμιλοῦσι [λόγῳ

τῷ τὰ ὅλα διοικουντι] τούτῳ διαφέρονται, where the bracketed words are now recognised

as Marcus’ gloss — see e.g. C. H. Kahn, The Art and Thought of Heraclitus, Cam-

bridge 1979, p. 30. For the thesis that the Heraclitean doctrine of logos as a govern-

ing cosmic principle is in its entirety a Stoic invention, see M. L. West, Early Greek

Philosophy and the Orient
, Oxford 1971, pp. 124-9, endorsed by J. Barnes, The

Presocratic Philosophers
, London 1979, i, p. 59. If that is right, as I believe it is,

our passage is the main source for transmission of the fiction to modern scholarship,

and Posidonius has much to answer for. For instance, without his testimony, who

would have thought of introducing logos into 22 b 50 D.-K. by way of emendation?
Frr. 31.16-31 E.-K.
Gal. PHP v 6.43. No doubt Plato’s Timaeus, with its supposedly Pythagorean

spokesman Timaeus, was one item of evidence he exploited.
M vii 112.
Although formally regarded as a pupil of Xenophanes (cfr. M vii 111), Par-

menides was said (Diog. Laert. ix 21) to have been more influenced by Ameinias

the Pythagorean. One tradition even made Xenophanes himself the pupil of a pupil

of Pythagoras (Diog. Laert. i 15).
A. Schmekel, Die Philosophie der mittleren Stoa in ihrem geschichtlichen

Zusammenhange dargestellt
, Berlin 1892, pp. 405 ff., followed by Taylor (supra, note

21) and others. Contra, see Mansfeld, Kidd (supra, notes 21-2).
For a judicious discussion, see I. Kidd, “Orthos logos” as criterion in the

, in The Criterion of Truth, ed. by P. Huby, G. Neal, Liverpool 1989, pp. 137-50.
Consequently, for Diog. Laert. or his source to append the ascription to a

list of formal doxographical reports was highly misleading. It can be safely discounted

as evidence. I thus arrive at a similar conclusion to Kidd’s (art. cit., previous note),

though by a different route.
Through the Timaeus, Empedocles, the Pythagoreans and Parmenides, at

least, he no doubt wanted to trace the logos theory back to his ultimate authority,

Pythagoras. Xenophanes and Democritus could have been used for the same purpose

(cfr. Diog. Laert. i 15, ix 38). Heraclitus could not be regarded as a follower of

Pythagoras, but was revered by the Stoics in his own right. Of those on Posidonius’

list, only Anaxagoras seems hard to explain along these lines; conceivably he was,

on the strength of Plat. Phaed. 96-9, judged an ancestor of the “Socratic” and Stoic

doctrine of cosmic intelligence. (At M vii 89 the Presocratics are called oἱ ἀπὸ Θάλεω

φυσικοί; this is a standard designation, and need not imply the inclusion of Thales

himself in the list.)
That only “some” early Stoics invoked orthos logos would be perfectly com-

patible with others, including even Zeno, having invoked other kinds of logos: cfr.

the diversity of logoi invoked in M vii 89-140. For the suggestion that Posidonius’

On the criterion was an appeal to ancient authority, see already, I. Kidd, art. cit.

However, I prefer to reserve judgement on his further hypothesis that Posidonius

had an anti-Chrysippean motive, and that the whole criterion doxography at Diog.

Laert. vii
54 comes from him.
Democr. b 118 D.-Κ., cfr. Posidon. t. 85, fr. 176 E.-K.
E.g. G. S. Kirk, J. E. Raven, The Presocratic Philosophers, Cambridge 1957,

p. 424, «It is odd that Sextus did not quote it», with the possible further implication

that this casts doubt on its authenticity.
See L.-S.-J. s.v.; K.-G. i 195-7.
“Claim”, with present infinitive: PH ii 148, M viii 283. “Promise” with

future infinitive: Posidon. fr. 60 E.-K.
Admittedly ὑπισχνεῖσθαι + aorist infinitive is quite hard to parallel at all,

the “profess” sense usually taking the present infinitive, since it describes a regular

or continuing action, as in Plat. Prot. 319 a, where Protagoras professes «to make

men good citizens ». The force of the aorist will be that Democritus’ claim to prove

(or to have proved) the point referred to a single completed act.
Tὸ κράτος τῆς πίστεως is most commonly translated «the power of persua-

sion», vel sim., meaning “the power to persuade”. A T.L.G. search on the most

directly comparable prose authors — Herodotus, Thucydides and the Corpus Hip-

— has turned up no parallel for this construction after κράτος, with which

an accompanying genitive is usually objective (“power over...”: 11 occurrences), and,

failing that, subjective (“power exercised by...”: 2 occurrences).
Since writing the above, I have found the same suggestion made by H. Lan-

gerbeck, ΔΟΞΙΣ ΕΠΙΡΥΣΜΙΗ, Berlin 1935, p. 117, although without discussion of

the meaning of ὑπεσχημένος. So too perhaps E. Asmis, Epicurus’ Scientific Method,

Ithaca/London 1984, p. 345.
Diog. Laert. ix 46-7.
M vii 89-90, 111, 124, 126, 131, 134, 138.
The observation that the two assertions can be consistent I owe to remarks

by Myles Burnyeat.
I here take σχεδόν to modify the whole clause. If it were taken to modify

πάσαν alone (“nearly all”), the sentence would have more relevance to (a) than to(c)-(g).
Thus e.g. J. Barnes, The Presocratic Philosophers, cit., ii, pp. 258-9.
Hesychius s.v. glosses ἐπιρυσμίη as ἐπιρρέον.
It was held by Antiochus (Cic. ac. I 44), to whom I have already suggested

Posidionius may be reacting, and by the contemporary New Academy (ibid. n 73).
What follows is textually disputed, but it seems to me that the correct punc-

tuation must be ἐπιφέρει διὸ δεῖ ἕπεσθαι τῷ κοινῷ (ξυνὸς γὰρ ὁ κοινός)· “τοὺ λόγου

δὲ ἐόντος ξυνοῦ: «he adds why one should follow the common logos (because the

common one is [what he calls] xunos): “Although the logos is xunos”».
Εἶτα, “then”, has no such connotation. Cfr. Athen. 188 b, where two

directly adjacent portions of Homer are linked with εἶτα.
In defence of this over the v. I. ἀποκεκρυμμένη, see Μ. M. Sassi, Le teorie

della percezione in Democrito
, Firenze 1978, p. 214 note.
Ibid., p. 213 does print them as continuous, although by placing a full stop

between them, and not translating, she leaves it unclear how the grammar could

work. I have not found other scholars committing themselves on the matter.
Moreover, at least two clear advantages are won. (a) Unless the text con-

tinues in this way, we are forced to read ἡ δὲ γνησίη at the end of (h) as «The other

is genuine» — a rather lame repetition of what we have already been told. With

the text continuing, on the other hand, we can read it as resumptive, «The one

which is genuine», (b) We avoid the problem of why the second excerpt should,

in addition to breaking off in mid sentence (see below), also start in mid sentence.
Neither punctuation nor accentuation in the mss. is likely to have ancient

authority, so that changing them is not, except in a trivial sense, emendation at all.
The construal involves an asymmetry between the first two occurrences of

μήτε, which are co-ordinate, and the subsequent ones, which are subordinate to the

second: “Neither a, nor b or c or d or e”. For Ionic prose use of this subordinate

μήτε, cfr. Hipp. epid. 5.1.14, v 214 L., οὔτε ἐφθέγγετο, οὔτε ᾐσθάνετο οὔτε ἔργου

οὔτε λόγου. Likewise at prorrh. 2.17, ix 42 L., and diaet. 9, ii 296 L.
Ἄλλ’ is the least satisfactory word. Perhaps at least read unelided ἄλλα.

The accusative after αἰσθάνεσθαι is acceptable (cfr. e.g. Critias 88 b 39 D.-K.). But

the fragment would nevertheless read better with ἄλλ’ deleted.
76 a 2 D.-K.
Diog. Laert. x 3; Athen. 611 b = Zeno of Sidon fr. 4 Angeli-Colaizzo

(«Cron. Erc.», ix (1979)). In the latter text, where the correction of “Theotimus”

to “Diotimus” has met with general acceptance, he was arrested at Zeno’s instiga-

tion, and condemned to death. It seems absurd to suppose that the actual charge

was defamation of Epicurus — what legal basis could that have had? The text permits

the less implausible guess that the defamation merely constituted the motive for

Zeno’s vendetta.
His identification as the Democritean started with R. Hirzel, Der

Demokriteer Diotimos
, «Hermes», xvii (1882) pp. 326-8. Since then it has regularly

been presented as established fact, e.g. H. Langerbeck, op. cit., pp. 119-20. Most

recently, it has been amplified by M. Gigante, op. cit., pp. 82-5, bringing in evidence

from the Herculaneum Papyri. In favour of the Stoic, however, see E. Zeller-R.

La filosofia dei Greci nel suo sviluppo storico, i, 5a parte, Firenze 1969,

p. 318 note 86; P. Natorp, op. cit., p. 190 note; H. Tarrant, op. cit., p. 106.
Cfr. P. Natorp, ibid.; F. Decleva Caizzi, art. cit., p. 405. The loose cita-

tion of Phaedr. 237 b as a gloss on the ἔvvoιαι criterion may be another characteristic

Posidonian invocation of a parallel, Phaedrus being one of Posidonius’ favourite two

Platonic dialogues (cfr. fr. 31.16-30 E.-K.). But if its author is Diotimus, it perhaps

represents an additional swipe: Plato too had hit on this criterion before Epicurus,

as Epicurus himself conceded (Cic. de fin. II 3-4).
For the order in Epicurus’s Canon, see Diog. Laert. x 31. This in itself

casts doubt on Diotimus’ identification with the Democritean: we know of no

“Democriteans” late enough to have responded to Epicurus.
For Posidionus’ virulent anti-Epicureanism, see I. Kidd, Posidonius, cit., ii,

pp. 977-8, with frr. 22, 46-7, 149, 160, 187.
Ἔλεγεν (140) slightly favours word of mouth over a written source — Dioti-

mus «used to say». But that reading is not compulsory: ἔλεγε(ν) can be used in literary

citations too, possibly including, in the present passage, 92 on Philolaus.
For a highly positive evaluation of this passage, see M. Gigante, op. cit.,

pp. 118-48, who judges it «un modello di acribia e di profondità» and «un raro esem-

pio di precisione storiografica». I am much indebted to Gigante’s observations, espe-

cially on the terminology of the passage, even though my own evaluation will be

more negative.
Academy: 141, 143-4; cfr. 160-1. Cyrenaics: 200. Epicureans: 203, 211-2,

215-6. Peripatetics: 218-9. Stoics: 257.
158 on Arcesilaus and 169 on Carneades are particularly striking instances,

but the feature is ubiquitous.
162, 170, 176-8, 180, 186-8, 192-3, 208-9, 212-4, 244-5, 249-50, 254 ff.
146, 163, 179, 182, 184, 220-1, 226, 239, 251-2, 259-60.
I am not sure that the case has been argued before for precisely our division

(141-260). R. Hirzel, Untersuchungen zu Ciceros philosophischen Schriften, Leipzig

1883, iii, pp. 493-524, argued at length for Antiochus as source of most of it, though

not including the Cyrenaic or Epicurean parts. I have myself argued for Antiochus

as source of the Epicurean part (On signs, in Science and Speculation, ed. by J. Barnes,

J. Brunschwig, Cambridge 1982, pp. 239-72, at pp. 263-72, summarised in A. A.

Long and D. N. Sedley, The Hellenistic Philosophers, Cambridge 1987, section 18).

H. Tarrant, art. cit., developed in his Scepticism or Platonism? cit., defends An-

tiochus as source of the “entire” doxography 89-260. (My reasons for dissenting

from his view as regards 89-140 consist in the distinguishing features of the two

passages, which I have listed above, plus incredulity that the readings of Plato at

119 and 141-2 could have come from the same source.) J. Barnes, Antiochus of Asca-

, in Philosophia Togata, Essays on Philosophy and Roman Society, ed. by M.

Griffin, J. Barnes, Oxford 1989, pp. 51-96, at pp. 64-5, is sceptical about attribut-

ing any significant part of the passage to Antiochus.
This is an objection raised by J. Barnes, ibid., I have already tried to antic-

ipate a similar objection about Posidonius in note 23 above.
Witness his decision to preserve from Posidonius the rather marginal pas-

sage of Asclepiades at 91, and more especially the frequent listing of Asclepiades’

doctrines elsewhere in Sextus — a rare honour to accord to a thinker of so late

a date, and barely matched in Sextus by the treatment of any other first-century

B.C. or later thinker (except Pyrrhonists and medical schools).
The text makes it clear that the doctor, whether or not he was Asclepiades,

was mentioned briefly in passing, and hence cannot have had a formal place in the

book’s doxography.
Cfr. H. Tarrant, op. cit., p. 95: «unless it is to be supposed that this un-

memorable passage of Antiochus [about Asclepiades] had for some reason stuck in

Sextus’ memory, there is little alternative to supposing that the Canonica is the work

he had been using, to a greater or lesser degree, in other parts of the doxography».
Even if (as maintained by H. Tarrant, art. cit., p. 81 note 50, and Scepticism

or Platonism? cit.
, esp. pp. 89 ff.) the Canonica belonged to the phase in which

Antiochus still professed allegiance to Philo of Larissa’s fallibilism (see Cic. ac. ii 69),

enargeia will already have been a central epistemological concept for him, as it was for

Philo (see Tarrant). But the association of Plato with κατάληψις convinces me that

Antiochus was already disloyal to Philo at the time, even if this predates their formal

schism in 87: Antiochus was already regarded as a virtual Stoic in the 90s (see my The

end of the Academy,
«Phronesis», xxvi (1981) pp. 67-75, at p. 70). I cannot agree with

Tarrant, op. cit., pp. 53 ff., that Philo ever accepted κατάληψις.
E.g. Cic. ac. ii 24.
Very probably Antiochus’ book covered the Presocratics too, since Au-

gustine, Cic. ac. ii 15, tells us that he veterum physicorum [...] implorabat fidem. He

was no doubt rebutting the sceptical interpretation of them current in the Academy,

as Lucullus does at Cic. ac. ii 14, when he asserts that their sceptical pronouncements

are no more than occasional outbursts of frustration by committed dogmatists.
166, cfr. 158. A characteristically Antiochean remark: cfr. Cic. ac. ii 34, on

Academics who, convicio veritatis coacti, allow that some things are ἐναργῆ. For the

suggestion that the Carneadean arguments here served Antiochus in a more positive

role, see H. Tarrant, op. cit., pp. 89-94.
For Antiochus’ use of this same doctrine from the “Younger Stoics”, see

Cic. ac. ii 38.
Cfr. Varro’s rather similar Antiochean history of epistemology in Academi-

I: the early Academics put the criterion in the mind, not the senses (30-2); Zeno

“corrected” the system (35, 40-3), making κατάληψις the sole basis of know-

At ac. ii 142-3 Cicero echoes a (presumably) Philonian attack on Antiochus:

regarding the criterion, Antiochus can hardly agree with Protagoras and the Cyrenaics

and Epicurus and Plato and Xenocrates and Aristotle; he in fact follows Chrysippus

alone. Apart from Protagoras, this list coincides entirely with M vii 141-260, and

gains added point if these were all authorities invoked by Antiochus as more or less

on the right side.
E.g. Cic. de fin. ii 35; Aristocles ap. Eus. praep. evang. xiv 18, 31 ( = Aristip.

IV a
173 S.S.R.).
Luck’s collection of Antiochus’ fragments, G. Luck, Der Akademiker An-

, Bern/Stuttgart 1953, and the supplemented collection by H. Mette, «Lus-

trum», xxviii-xxix (1986/7), offer no evidence of anti-Epicureanism half as strong

as Cicero’s or Plutarch’s.
I have argued against Natorp’s grounds for the attribution in On signs, cit., p.

264 with note 60. Cfr. also infra note 91. But this does not preclude the possibility that

Antiochus himself drew information from Demetrius’ writings: see further, note 95.
Cfr. supra note 72.
Hence the closing remark, πάντων δὲ κρηπὶς καὶ θεμέλιος ἡ ἐνάργεια (216).
Μ vii 213-4.
See the invaluable edition by P. and E. De Lacy, Philodemus, On Methods

of Inference
, Napoli 1978(2); cfr. my discussion, in On signs, cit.; E. Asmis, Epicurus’

Scientific Method
, cit.; and most recently the outstanding study by J. Barnes,

Epicurean Signs, «Oxford Studies in Ancient Philosophy», suppl. vi (1988) pp. 91-

134, and comments by A. A. Long, ibid., pp. 135-44.
See especially de sign. xii 1-14. Since all the Epicureans reported in De signis

use this inference as an example, Natorp (op. cit.) was on weak ground in using

its occurrence in M vii 213-4 as evidence for Demetrius as source.
De sign. viii 26-ix 8, Barnes’ translation, adapted.
See J. Barnes, Epicurean Signs, cit., pp. 103, 133, for the evidence that some

Epicureans accept anaskeue as a sound form of semeiosis. But note that these texts

allow only that some things are “captured” (ἀλίσκεσθαι) by anaskeue: that term is

in the Epicureans’ usage contrasted with “locking up” (κατακλείειν, xxxιιι 8, cfr. xv

37), i.e. confirming, a task consistently assigned to the similarly method of inference.
E.g. viιι 1-13, xιιι 1-8, xxi 12-6, xxxii 24-7, xxxv 20-2, xxxvi 7-17.
My guess is that the source used was the work on sign-inferences by De-

metrius of Laconia. Sextus’ citations of Demetrius show that the latter was an ac-

knowledged source for Epicurean doctrine, and since his account of the Epicurean-

Stoic debate on signs was very condensed (Philodem. de sign. xxviii 13-14) it could

all the more easily have misled Antiochus. Philodemus’ De signis (even supposing

it to have been a published work) was itself written well after Antiochus’ death in

68-7 B.C. probably in the early 30s B.C. (A.A. Long-D.N. Sedley, op. cit., ii,

p. 263). As for Zeno of Sidon, his discussion was apparently an oral one, recon-

structed by his pupils Philodemus and Bromius (Philodem. de sign. xix 4 ff.).
E.g. Aetius iv 11, 4 ( = S.V.F. ii 83), and, for Antiochus himself, cfr. Cic.

ac. ii 30.
For a plausible emendation of this difficult Sentence (203 fin.), see M.

Gigante, op. cit., p. 122. However, I feel the transmitted reading (retaining ὑπάρχον

for ὑπάρχειν, with most mss.) can be tolerated: «what is productive of each of them

[the phantiasiai] is completely a phantaston, which, being a phantaston, cannot be

productive of a phantasia without being in reality such it appears».
See vii 248 for the Stoic definition. Since (4) is attributed to “the

Epicureans”, it might be suspected that these Epicureans are merely adding to Epicu-

rus’ argument the observation that his account obeys Stoic requirements too. But

the Stoicising tendency is being attributed to Epicurus himself too, since the same

Stoic terminology is already built into (1) as well: 203, ἀπὸ ποιητικῶν τινῶν καὶ

κατ’ αὐτὰ τὰ ποητικά.
Op. cit., p. 126.
Cfr. Diog. Laert. χ 31, Epicur. ep. Her. 63, R.S. xxiv, for αἰσθήσεις and

πάθη as coordinate criteria. There is, of course, also a more intimate link than that

between them: all πάθη are generically pleasure or pain (attributed to Epicurus not

only in our present passage but also by Demetrius of Laconia quoted at Μ x 225);

pleasure and pain are identical to good and bad respectively; and all good and bad

are found in αἰσθησις (ep. Men. 124). It follows that all πάθος is found “in” αἴσθησις.

But this in no way entails either that the two are identical or that one is a species

of the other, just that all πάθος accompanies perception. It would be hard to maintain

that the representational properties of φαντασίαι are varieties of pleasure and pain.
203, καὶ ἐπὶ τῶν φαντασιῶν, παθῶν περὶ ἡμᾶς οὐσῶν, «so too in the case

of phantasiai, since these are pathe belonging to us». M. Gigante, op. cit., pp. 122-

126, proposes to read καὶ ἐπὶ τῶν φαντασιῶν <τῶν> παθῶν περὶ ἡμᾶς οὐσῶν, «per

quanto attiene alle rappresentazioni delle interne affezioni», which is syntactically

unlikely and syllogistically weaker without even being any more orthodox doctrinally.

Cfr. D. Fowler’s review, Sceptics and Epicureans, «Oxford Studies in Ancient

Philosophy», ii (1984) pp. 237-67, at 247-8.
Aetius iv 12,1-6 = S.V.F. ii 54. Another Stoic intrusion into the passage is

the term φανταστόν: cfr. M. Gigante, op. cit., p. 127.
Here I owe to a remark by M. Gigante, op. cit., pp. 127-8, the insight

that this part of the Epicurean doxography is drawing on Cyrenaic themes. Note

especially how, in (2), περὶ ἡμᾶς πάθη echoes 194 on the Cyrenaics.
No mention of the incommensurability of the different senses (Lucr. iv

486-96, Diog. Laert. x 31-2, PHerc. 19/698; see further, A. A. Long-D.N. Sedley,

op. cit., section 16).
Especially the use of Epicurus’ term στερέμνιον for what has in the previ-

ous part of the passage been called by the Stoic name τὸ ὑπάρχον: see M. Gigante,

op. cit., pp. 130-7, for a careful survey of this and other authentic Epicurean details.
In the passages I have been discussing, that includes possibly the medical

analogy at M vii 179, and certainly the reference to his own work Ἰατρικὰ ὑπο-

μνήματα at 202.
The one possible exception I am aware of is the Stoic Basilides at M viii

258, who may be identical with the teacher of Marcus Aurelius.
If Philodemus influenced Sextus in M vi, as maintained by M. Gigante,

op. cit., pp. 215-21, or in M ii, as argued by F. Longo Auricchio, Epicureismo e

scetticismo sulla retorica
, in Atti del XVII Congresso Internazionale di Papirologia

(Napoli, 19-26 maggio 1983), Napoli 1984, pp. 453-72, that too would surely be

through his contemporary Aenesidemus.

David Sedley . :

This page is copyrighted


  • There are currently no refbacks.