1. WHY A LACUNA?

At the beginning of §133 of the Epistle to Menoeceus the text is unproblematic. Epicurus says:

The next sentence, on the other hand, introduces a complex problem. I begin by outlining the situation of the relevant manuscript tradition, drawing on the work of Dorandi.Footnote ¹ The three main manuscripts are P (Par. gr. 1759, eleventh/twelfth century), B (Neapolitanus III B 29, twelfth century) and F (Laur. plut. 69.13, thirteenth century). These are the oldest continuous manuscripts through which the Vitae of Diogenes Laertius have come down to us, and they are independent of each other (apart from the contamination of F with P). They all derive, either directly (P and B) or indirectly (F, via the lost γ), from a manuscript Ω which, according to Dorandi,Footnote ² was already written in minuscule script, but probably in scriptio continua, without accents or breathings.

I now quote Dorandi’sFootnote ³ text (apart from a few minor punctuation changes), which largely corresponds to the paradosis of these manuscripts, and attempt to show why it does not work and how it could be emended. I shall also provide a selective apparatus criticus, based in part on Dorandi’s.Footnote ⁴

1 πάντων ἀγγέλλοντος BF² (πάντων om. F¹) : πάντων ἀγγέλωντος P¹ (-ῶντος P^mg : -οντος P⁴) : πάντων ἀνελόντος Kühn : πάντων διαγελῶντος <εἱμαρμένην καὶ μᾶλλον ἃ μὲν κατ’ ἀνάγκην γίνεσθαι λέγοντος> suppl. Usener : πάντων ἐγγελῶντος <εἱμαρμένην; * * * * * ὧν ἃ μὲν κατ’ ἀνάγκην γίνεται> Bailey : πάντων ἂν γελῶντος <εἱμαρμένην ἀλλ’ ἃ μὲν κατ’ ἀνάγκην ὄντα συνορῶντος> Sedley : πάντων <εἱμαρμένην οὐκ εἶναι νομίζοντος, ἀλλὰ γίγνεσθαι κατ’ ἀνάγκην ἃ μὲν πάντων> ἀγγέλλοντος Hessler

Immediately after βραχεῖς—the reading of MSS B and PFootnote ⁵ —the witnesses have τὴν δὲ ὑπό τινων δεσπότιν εἰσαγομένην πάντων ἀγγέλλοντοςFootnote ⁶ ἃ δὲ ἀπὸ τύχης, etc. As it stands, the manuscript text is untenable, first of all because it is impossible to make sense of its grammar. For anyone who has just read the preceding lines, it comes naturally to take ἀγγέλλοντος as part of the second term of comparison, along with the other genitive participles (δοξάζοντος, ἔχοντος, ἐπιλελογισμένου, διαλαμβάνοντος), but this move alone is not sufficient to solve our problem, both because ἀγγέλλοντος cannot govern all the accusatives surrounding it, and because the δέ in ἃ δὲ ἀπὸ τύχης seems to require a preceding μέν, which is not present in our text.Footnote ⁷

Nowadays, scholars unanimously agree that something is missing, either after πάντων (Hessler) or after ἀγγέλλοντος (all others);Footnote ⁸ the discussion arises when we try to assess what exactly is missing. In the following exposition, I first consider the elements that most contemporary scholars believe need to be supplemented.

The first element needed in our supplement is ἃ μὲν κατ’ ἀνάγκην, since the infinitive introduced by διὰ τό mentions ἀνάγκη, τύχη and παρ’ ἡμᾶς.Footnote ⁹ The second element, which is highly plausible, though not fundamental, is a clarification of the mysterious τὴν … δεσπότιν, an expression which is deliberately allusive, but perhaps too allusive in our context. Since in §134 Epicurus refers to τῇ τῶν φυσικῶν εἱμαρμένῃ as if it were already present in the minds of his readers, it is reasonable to suppose that he mentioned εἱμαρμένη in the part of §133 that we are missing.Footnote ¹⁰ Thus, so far we have <… τὴνFootnote ¹¹ εἱμαρμένην … ἃ μὲν κατ’ ἀνάγκην …>.

These are the two elements that are more or less common to all the proposals listed in the apparatus criticus above. Another thing these supplements have in common is the syntactic interpretation of the passage on which they are based. Practically every scholar thinks, on the basis of the presence of ἀγγέλλοντος, that the sentence after βραχεῖς extends the second term of comparison (begun with δοξάζοντος) by further genitive participles—namely, ἀγγέλλοντος (or its emendations) and another participle that is missing from the text and needs to be supplemented. Such an interpretation of the syntax of the passage is necessary for scholars, such as UsenerFootnote ¹² and Sedley,Footnote ¹³ who think both that the lacuna begins immediately after ἀγγέλλοντος and that it depends on a saut du même au même caused by homoeoteleuton, and who therefore have to choose a genitive participle ending in -οντος as the last word of their supplements.

This syntactic interpretation is highly problematic. The next section is devoted to questioning it and to proposing a new interpretation. This will also help us determine where the lacuna begins, whether after πάντων or after ἀγγέλλοντος.

2. A SYNTACTIC DIVERGENCE

There are three main reasons why the syntactic interpretation given by previous editors cannot be sustained, none of which is conclusive on its own, but which together make a compelling case. Μy arguments are also intended (i) to suggest that the lacuna begins after πάντων, (ii) to show that our supplement should include a present indicative third-person singular verb, and (iii) to cast doubt on the correctness of the reading ἀγγέλλοντος.

First, the participles forming the second term of comparison are correlated by καί, whereas ἀγγέλλοντος would be the first to be introduced by δέ. The correlation καὶ … δέ is strange: if ἀγγέλλοντος were on the same level as the other participles, we would expect καὶ … τε or καὶ … καί instead. This strongly suggests that a new sentence begins after βραχεῖς and that ἀγγέλλοντος is not part of the second term of comparison.Footnote ¹⁴

One might respond as follows: after βραχεῖς there is indeed a strong syntactic pause, whether a question mark or a semicolon,Footnote ¹⁵ but the next sentence ‘retains the memory’ of the syntax of the previous sentence, that is, it is a kind of afterthought that extends the previous sentence. To this we may reply that such a solution results in a less than straightforward syntactic construction, and that this is particularly undesirable since the style of the epistle up to this point would not lead us to expect anything so complex. Moreover, this hypothesis is likely to leave us without the question we would expect after τίνα:Footnote ¹⁶ if we suppose that τὴν δὲ … πέφυκεν is attached to ἐπεὶ τίνα … βραχεῖς, then it is more natural to place the question mark only once after πέφυκεν.Footnote ¹⁷ But this is problematic, since πέφυκεν is followed by what is supposed to be an explanation (ἐπεὶ κρεῖττον … ἔχει τὴν ἀνάγκην) of what Epicurus affirms in the very phrase τὴν δὲ … πέφυκεν, so that it would be abrupt enough to pass directly from the end of a long question to the justification of one of its parts, the last.Footnote ¹⁸ Rather, it would be less artificial to place the question mark after βραχεῖς and to regard τὴν δὲ … πέφυκεν as a separate sentence; after all, Epicurus reserves a specific explanation for it. Apart from πέφυκεν, the only other candidate under this hypothesis after which we could place the question mark are the words διὰ ταύτην in §135, but that is far too remote.

To sum up, if we do not place the question mark after βραχεῖς we must imagine that Epicurus ‘forgot’ the initial syntax by extending the second term of comparison, resulting in strong anacoluthon. Consequently, it is not advisable to adopt this solution if a more fluent and straightforward text can be attained by some other way.

Second, until βραχεῖς Epicurus recapitulated the previous content of the epistle (gods §§123–4, death §§124–7, supreme good and end of life §§127–32) and laid down a version of the famous tetra-pharmakos (RS 1–4, Phld. P.Herc. 1005 col. V). It is therefore natural to suppose that Epicurus concluded the sentence with βραχεῖς and then moved on to topics left untouched in the preceding sections.Footnote ¹⁹

Third, there seems to be an intentional parallelism between τὴν δὲ ὑπό τινων δεσπότιν and τὴν δὲ τύχην in §134.Footnote ²⁰ Both sentences begin with an emphatic accusative that sets up the discussion which follows in the paragraph. In §133, Epicurus begins by mentioning the ‘mistress’ of all things (fate, if my and other supplements are correct); he alludes to how such a mistress is believed in by some people (or, literally, has been introduced by them); and then he says what the Epicurean sage thinks of the mistress—presumably that he rejects it—providing as justification the threefold distinction within events and the characterization of one class of events as ‘without a master’ (on this, see §3 below). In §134, Epicurus begins by mentioning ‘chance’; he then says that the Epicurean sage, unlike other people, does not regard chance as a god or as an unreliable cause; he justifies why he does not; and then he offers the correct Epicurean view of chance, which consists in preferring reason to luck. So the accusatives at the beginning of the respective paragraphs divide Epicurus’ agenda in these lines of the epistle into two significative sections, ‘fate’ and ‘chance’, each of which consists of the same components (what other people think, what the Epicurean sage thinks, why he thinks so) in roughly the same order. However, if the structural parallelism is rightly detected, then it is also likely that τὴν δὲ … πέφυκεν is an independent sentence which, on a syntactic level, does not merely extend the second term of comparison, just as the sentence beginning with τὴν δὲ τύχην is.

Hence the sentence after βραχεῖς cannot be an extension of the second term of comparison by further genitive participles. Now, why should this have any bearing on (i) the question of where the lacuna begins? Essentially because (as most scholars have thought) the lacuna can be justified mechanically, that is, by postulating a saut du même au même.Footnote ²¹ This means that, if ἀγγέλλοντος were the last word before the lacuna, the last word to be supplemented would probably be a genitive participle in -οντος, which is incompatible with our findings about the syntax of the passage.Footnote ²²

The lacuna, therefore, begins immediately after πάντων, and a second πάντων is the best candidate to be the last word of the supplement, since it explains why the omission occurred.Footnote ²³ The second πάντων echoes the first and has a rhetorical value: if, as will be argued below, the expected meaning of the passage is indeed that the Epicurean sage rejects fate, the mistress of all things—which might stand for the claim that all things happen by necessityFootnote ²⁴ —then it becomes meaningful to say that only some of all things happen by necessity.Footnote ²⁵ A second πάντων is also an appropriate choice because the close repetition of the adjective πᾶς is a clear stylistic feature of the Epistle to Menoeceus.Footnote ²⁶ So far, then, our supplement is as follows: πάντων <τὴν εἱμαρμένην … ἃ μὲν κατ’ ἀνάγκην πάντων>.

Now, it is time to explain (ii) my interpretation of the syntax. In short, I believe that after βραχεῖς a syntactically independent sentence begins, governed by a new verb, more precisely a present indicative third-person singular. This is where my proposal differs most from all the others.Footnote ²⁷ But which verb should we choose?

As anticipated above, it should emerge from our supplement that the Epicurean sage rejects fate, that is, the view that everything happens by necessity. Many verbs can express such an idea, and we can divide them into two classes: (1) verbs that can govern an accusative + infinitive construction, and (2) verbs that can govern an accusative. The first class includes verbs of utterance and verbs used to express thoughts and opinions (ἀποφάναι, [οὐ] φάναι, [οὐ] λέγειν, ἀρνεῖσθαι, [οὐ] νομίζειν, [οὐ] δοξάζειν, [οὐ] διαλαμβάνειν, [οὐχ] ὑπολαμβάνειν, [οὐκ] οἴεσθαι, [οὐ] συνορᾶν, [οὐ] δοκεῖν, [οὐ] θεωρεῖν), while the second class includes verbs that mean ‘to reject’, ‘to do away with’, ‘to eliminate’, etc. (ἀπωθεῖν, ἐξωθεῖν, ἀφαιρεῖν, ἐξαιρεῖν, ἀποβάλλειν, ἐκβάλλειν, [οὐκ] [ἀπο]δέχεσθαι). It is not possible to pinpoint the correct verb, but several of those listed are either not attested or poorly attested in Epicurus, or are attested but with different meanings. We should seriously consider ἀναιρεῖν, which is used by Epicurus in Ep. Hdt. 52.2 and Ep. Men. 123.10 with the meaning we need here—namely, ‘to do away with’.Footnote ²⁸ Moreover, the combination ἀναιρεῖν τὴν εἱμαρμένην occurs in some relevant passages which may be related to ours, viz. Joseph. BJ 2.164 and AJ 13.173,Footnote ²⁹ Alex. Aphr. In Top. 570.4–7, Nemes. Nat. hom. 39.113. In the choice of ἀναιρεῖν I have been anticipated by Kühn,Footnote ³⁰ but he wrote ἀνελόντος instead of the manuscripts’ ἀγγέλλοντος.Footnote ³¹

To complete the supplement, we may add a pronoun, such as οὗτος, to make the subject explicit (though this is not strictly necessary), and an infinitive, γίνεσθαι,Footnote ³² governed by ἀγγέλλοντος. Before reaching a conclusion, though, a natural objection to this interpretation of the syntax must be addressed: how can the presence of ἀγγέλλοντος be accounted for, other than by taking it to be an extension of the second term of comparison? Accordingly, the next section will be devoted to showing why (iii) ἀγγέλλοντος may not be the correct reading.

3. THE READINGS IN MS P AND ITS DESCENDANTS

According to my reconstruction, the only way to retain ἀγγέλλοντος is to suppose that a genitive absolute has taken the place of a conjunctive nominative participle, agreeing with the subject. There are similar cases, for example Thuc. 3.70 ἐς λόγους καταστάντων [sc. Κερκυραίων] ἐψηφίσαντο Κερκυραῖοι,Footnote ³³ so that ἀγγέλλοντος is not untenable.

Several considerations support correcting ἀγγέλλοντος to ἀγγέλλων: first, we have ὑπολαμβάνων in §134, which seems to us to be syntactically parallel to §133 (see §2 above); second, it would be easy to justify the error: a scribe, finding a text already affected by the lacuna and lacking a verb in the indicative, might have tried to emend it by changing ἀγγέλλων to ἀγγέλλοντος and taking the latter as part of the second term of comparison. Therefore, the problem of correcting ἀγγέλλοντος should not be treated separately from the problem of supplementing the lacuna.

Lastly, the situation we find in MS P might lead us to consider ἀγγέλλων as the original form, since there we read ἀγγέλο(ex ω)ν τος in the main text and ἀγγελῶντος in the margin. A reproduction of MS P illustrates the situation:

MS P, fol. 246v, line 5.

Both ἀγγέλωντος and ἀγγελῶντος are uoces nihili. What seems hardly questionable is that the scribe of MS P at first did not understand that -τος was connected with ἀγγέλων, but rather regarded it as the beginning of a new word, probably τόσα or τοσαῦτα (or again τοσάδε, as already suggested by UsenerFootnote ³⁴ in his apparatus criticus). This is clear because he did not abbreviate τος, as he usually does when he takes τος as an ending, and also because the breathing and the accent on ἃ are certainly laterFootnote ³⁵ than the rest of the word—the colour of the ink and the squarer shape of the breathing are telling in this respect.

While writing, however, the scribe of MS P realized his mistake; this is where the margin becomes important. He understood that τος has to be combined with ἀγγέλων, and so, in accordance with the basic rules of ancient Greek accentuation, he wrote ἀγγελῶντος, with a circumflex accent on the omega.Footnote ³⁶ The fact that he takes the trouble to offer a different, more thoughtful interpretation of what he found in his antigraph Ω—which lacked accents and breathings and was probably written in scriptio continua—suggests that he is trying to reproduce it faithfully. This testimony is therefore of great importance to us as regards the original presence of an omega in the antigraph, which probably had something like αγγελων^τοσα or αγγελωντοσα.

But what about the absence of a second lambda? And how do we explain the fact that both MS B and MS F clearly have ἀγγέλλοντος? The scribe of MS P may have copied the omega exactly but failed to notice the presence of a second lambda in the antigraph. It would be strange, though, to suppose that he was both accurate and inaccurate in relation to the same word; so let us instead argue on the assumption that his antigraph Ω already lacked a second lambda. Could this indicate the presence of a form related to γελάω? Drawing on the emendations made by previous scholars, namely Usener’s διαγελῶντος and Sedley’s ἂν γελῶντος,Footnote ³⁷ could we write either διαγελῶν (‘deriding’) or ἂν γελῶν (‘would deride’)?

Neither is plausible, however, either palaeographically or linguistically. Regarding διαγελῶν, corruption of δια- into αγ- is unlikely in minuscule and not straightforward in capital letters either. That is, even if we assume that the corruption took place at an early stage in the tradition of Epicurus or Diogenes, when, in the papyri, ΑΓ- before Γ could have been written as ΑΝ-, and therefore it was likelier than in minuscule script to mistake ΔΙΑΓEΛΩΝ for ΑΝΓEΛΩΝ, such a mistake is not at all common. It is not surprising, then, that Arrighetti’s apparatus criticus does not record any case of ΔΙΑ- corrupted into ΑΝ- or ΑΓ-.Footnote ³⁸

As for ἂν γελῶν, it is problematic in terms of meaning. Sedley himself explains his ἂν γελῶντος thus: ‘The point of ἄν is that this idealized sage need not actually have encountered determinism, but would deride it if he did’Footnote ³⁹ (my italics). However, it is not easy to say why the idealized Epicurean sage should actively embrace all the other major Epicurean ethical views, while only being unaware of the rejection of determinism. Moreover, such a rejection and the discovery of the three classes of events are closely related; so if we print ἂν γελῶν, we should probably think that this discovery too is only potential, which would be an unpleasant consequence, as well as being inconsistent with the rest of our passage. Finally, the use of ἄν with a participle is all the more striking when we remember that in general the participle with ἄν is not common in Epicurus.Footnote ⁴⁰

It is more prudent, then, to think first that the antigraph of MS P, or one of its ancestors, had already made the mistake of writing ἀγγέλων instead of ἀγγέλλων: this may have happened either because geminates were not pronounced in the Byzantine period,Footnote ⁴¹ or because a word like ἀγγέλων has obvious resonances for a Christian. Second, we should also think that P’s antigraph Ω presented τος either above the line or already in the text (deriving it from its antigraph), which was added to turn the nominative into a genitive, thus attempting to make sense of the syntax (which was already affected by the lacuna) by aligning this participle with the previous ones. The scribe of the antigraph, however, or the scribe of one of its ancestors, did not correct the omega of ἀγγέλων to an omicron, either because of inattention,Footnote ⁴² an ambiguity caused by the scriptio continua, or because of some other reason. Lastly, we should also believe that both of the other apographs of Ω besides P, namely B and γ (F’s lost antigraph), seeing ἀγγέλων^τος or ἀγγέλωντος, were able to conjecture ἀγγέλλοντος, probably independently of each other—which is certainly possible, since the conjecture is an easy one—or to get it in some other way.

At this point the following objection might be raised: it could be that it was the first hand of MS P that corrected ἀγγέλωντος to ἀγγέλοντος, and this arguably because the antigraph had the omicron, not the omega. In other words, by checking Ω, the scribe of MS P may have realized his mistake and corrected the text accordingly. I have two answers to this objection.

First, if this were the case, how could we explain the presence of ἀγγελῶντος in the margin? Why should the scribe of MS P have corrected ἀγγέλωντος but left ἀγγελῶντος if the antigraph had only one form with omicron? There is no easy answer. Second, and most importantly, there is reason to believe that the correction of ἀγγέλωντος to ἀγγέλοντος is not due to P’s first hand but rather to the well-known fourth hand, responsible for an intense phase of revision.Footnote ⁴³

All those manuscripts which Basta DonzelliFootnote ⁴⁴ claims were copied from MS P before the revision of P⁴ have forms with an omega, whereas those copied after this revision have forms with an omicron. In particular, in the first group of manuscripts, Q (Par. gr. 1758, very early fourteenth century) has ἀγγέλῶντος (fol. 200r),Footnote ⁴⁵ W (Vat. gr. 140, early fourteenth century) has ἀγγέλωντος (fol. 172r), while Co (Seragl. 48, early fourteenth century) has ἀγγελῶντος (fol. 153r). In the second group, on the other hand, H (Laur. plut. 69.35, c.1419/20), E (Pal. gr. 182, fifteenth century) and I (Marc. gr. 394, late fifteenth century) have ἀγγέλοντος (fol. 242v, fol. 188v and fol. 195r respectively), Y (Ang. gr. 97, sixteenth century) has ἀγγέλονοντος (fol. 200v), while Jb (Barb. gr. 21, sixteenth century) probably has ἀγγελόντος before correction, but in any case a form with omicron (fol. 280v).Footnote ⁴⁶ The distribution of the forms with omega and those with omicron between the two groups of manuscripts is too striking to be a coincidence. We may infer that ἀγγέλοντος was not written by the first hand of MS P but by the fourth—which explains why it only appears in the manuscripts derived from MS P after the revision made by P⁴—and is therefore not the result of an attempt to reproduce the reading of the antigraph Ω.

Not every step of this reconstruction may be straightforward, but at least two points seem hard to deny—namely, that MS P provides us with evidence that the antigraph had a verbal form with an omega, and that we cannot hypothesize that the original verb was γελάω or one of its compounds. So, although my proposal is not linguistically incompatible with ἀγγέλλοντος, we should print ἀγγέλλων, a correction already proposed by Diano,Footnote ⁴⁷ since the form ἀγγέλλοντος should not be considered independently of the presence of a lacuna, disregarding the fact that the nominative, of which MS P seems to preserve some trace, could easily have been corrected to a genitive after the omission occurred.

Here is the text I would print:

The form of the sentence thus supplemented resembles that of Ep. Men. 125.5–6 τὸ φρικωδέστατον οὖν τῶν κακῶν (the description of something) ὁ θάνατος (the thing described) οὐθὲν πρὸς ἡμᾶς (the Epicurean judgement). In our text, Epicurus claims, or literally proclaims,Footnote ⁴⁹ that some events are due to necessity, others to chance, and still others to ‘what is because of us’, thus doing away with fate. The infinitive clause governed by διὰ τό, although it does not present an argument, is intended to justify why such a distinction within events is relevant to the overthrow of fate: some events, those that happen because of us, are ‘without a master’, ἀδέσποτον, and this is clearly incompatible with the position that fate is the δεσπότις of everything.Footnote ⁵⁰ The rejection of determinism is thus achieved.