petierunt. ne pythio quidem uulnerabilis erat, cum ingens magnitudo pro uastitate corporis solida ferrum et quicquid humanae torserant manus reiceret: molaribus demum fracta saxis est.

pithio V pithyo P pilo uel pilis Erasmus.

The readings of VP point to pithio. This cannot be right, for (1) the word does not occur elsewhere as a name of an engine of war, (2) ' quicquid humanae torserant manus' would be more appropriate of weapons thrown by hand than of missiles hurled from a ballista, (3) there is an ascending order—(i.) ‘sagittis fundisque,' (ii.) 'pythio,' (iii.) 'molaribus saxis.' If pythio-ballista, as Buecheler suggested, this ascending scale is destroyed and the climax molaribus saxis anticipated, for the ballista hurled stones, sometimes a hundred pounds in weight. Now when the arrows and sling-stones of the light infantry had no effect, what would be the next weapon tried?—The pilum of the legionaries. Then we get (i.) arrows and sling-stones, (ii.) iron-pointed javelins thrown by hand, cf. 'ferrum et quicquid humanae torserant manus, (iii.) big stones hurled by a ballista. Therefore Erasmus's pilis should be adopted. The plural is required to match the other plurals. The corruption pythio probably arose from some marginal note about the Python slain by Apollo at Delphi. [I find that Lipsius defended pilo by the argument' crescit oratio.']

86 § 10. sed, di boni, quam iuuat illa balinea intrare obscura et gregali tectorio inducta, quae scires Catonem tibi aedilem aut Fabium Maximum aut ex Corneliis aliquem manu sua temperasse!

.

quae scires .' is imbalinea intrare, quae scires

The combination of tenses in 'quam iuuat possible, for it is nonsense to say 'quam iuuat illa Catonem tibi . . . temperasse.' Read quam iuuaret would have been to enter a bath which you knew . . .').

('What a pleasure it

87 § 9. hic, qui inter illos apparatus delicatos cum maxime dubitat utrum se ad gladium locet an ad cultrum.

Gladium is of course the gladiator's sword, cf. Cic. pro Sex. Roscio § 118 'num aut ille lanista omnino iam a gladio recessisse uidetur? ... Of cultrum two explanations have been offered: (1) The Thesaurus Linguae Latinae, s.v. culter, has 'culter . . . i. q. pugio uel gladius breuis. . . SEN. Contr. exc. 6, 5 quidam Thracum cultris armati. SEN. Epist. 87, 9 dubitat, utrum se ad gladium locet an ad cultrum (i. ad officium gladiatoris, cf. SCRIB. LARG. 13 occidi hinnulum cultro, quo gladiator iugulatus sit). . . .' It is not clear how the writer proposes to distinguish between ad gladium and ad cultrum. If he thinks ad gladium means 'to serve as a soldier' and ad cultrum 'to fight as a gladiator,' he is clearly wrong, for, apart from other considerations, there is no evidence that culter was a technical name for any gladiatorial weapon. If he thinks ad cultrum means 'to fight as a Thraex,' he is equally mistaken. In Sen. Rhet. l.c. the MSS. have Thracum cultu '; cultris is only a suggestion of A. Kiessling. Thracum there means natives of

184

NOTES ON SENECA EPISTVLAE MORALES

Thrace. Moreover, the offensive weapon of the Thraex was the sica, see Daremberg and Saglio, s.v. gladiator, II. 1587, Friedländer, SG. II3. 539, cf. Mart. 3. 16. 2. The writer of the article in the Thesaurus must have been unusually careless, for he gives Varro R.R. 2. 5. II 'paulo uerbosius haec, qui Manilii actiones secuntur lanii, qui ad cultrum bouem emunt: qui ad altaria, hostiae sanitatem non solent stipulari' (he does not quote the passage in full) as an instance of culter in the sense of a sacrificial knife, though the antithesis of ad cultrum and ad altaria shows that it must mean a butcher's knife (see Daremberg and Saglio, s.v. culter, I. 1585). (2) Muretus, Lipsius, Daremberg and Saglio, s.v. culter, I. 1586, Friedländer, SG. II3. 393-4, 544-5, and Professor Summers take cultrum to be the hunting-knife used in the uenationes, cf. Mart. 4. 35. 4 ‘stupuitque superbus | uenator cultro nil superesse suo.' Culter is used of so many different kinds of knives that its precise meaning can be determined only by the context. Here se ad gladium locet at once turns our thoughts to the harena and hence serves to limit cultrum to the weapon of the uenator. Also it is known that men did hire themselves out as uenatores, see Friedländer's quotation from Ulpian: et qui operas suas ut cum bestiis depugnaret locauerit.' Hence there can be little doubt that by cultrum is meant the culter uenatorius.

88 89 doces me quomodo inter se acutae ac graues <uoces> consonent fac potius quomodo animus secum meus consonet.

6

I do not see how fac quomodo is to be translated. 'Cause my soul to be harmonious' would be fac potius ut. . . . Moreover, the antitheses which follow in §§ 9-11 show that the antithesis should be between what is taught and what should be taught. 'Fac potius <discam> quomodo . . .' or 'fac potius doceas> quomodo . . .' seems to be necessary.

UNIVERSITY OF TASMANIA,

AUSTRALIA.

R. L. DUNBABIN.

THE AFFATIM GLOSSARY AND OTHERS.

THE bilingual Philoxenus Glossary drew some of its materials from Festus de Signif. Verb. and occasionally mentions his name. Its Festus glosses have been collected in a Jena dissertation by Dammann. The Abolita Glossary (if we may so term the collection enclosed within square brackets in Corp. Gloss. Lat. IV. pp. 4-198) seems to have begun (in its original arrangement) with Festus excerpts. Before we can glean from these two glossaries every available scrap of evidence about Festus, we must try to complete and correct them. For of the Philoxenus Glossary (Philox.) we have practically only one MS., and that of the ninth century. It must have omitted many items and perverted the words of many others. The Abolita Glossary (Abol.) was associated (but not mixed up) with another, the Abstrusa Glossary (Abstr.), in Spain, and this composite collection (Abstr.-Abol.) passed into Italy. The MSS. are, we may say, only two, and of each a certain number of leaves are missing. The older and better MS. (Vat.), an uncial codex of the eighth century, was written apparently in Central Italy; the other (Cass.) was written two centuries later at Monte Cassino. They show us Abol. not in its original form, but rearranged in a not too strict alphabetical order (between AB- and ABC-). The St. Gall Glossary was compiled from (1) Philox., (2) the composite Abstr.-Abol., and supplies some of those Festus glosses of Philox. and Abol. which are omitted in our MSS. It seems to have been a Bobbio compilation (but see below). Our best MS. was written in rude uncials at St. Gall in the eighth century.

This St. Gall Glossary (Sangall.) is of a type which prevailed from the Carolingian Revival of Learning, the extract-glossary; for it offers us not Philox. and Abstr.-Abol. fully transcribed, but merely selections from the pair. The compiler selects now an item from one, now an item from the other, and often blends his two authorities into a composite item or re-casts the words of the one to suit the words of the other. It is a practice which is easily explained, especially in the earlier Carolingian period. A glossary was a book indispensable at a monastery. It could not be lent to another cloister, at least not for any long time. In most cases a visitor would be the transcriber, and the amount he transcribed would depend on the length of his visit. The result of this practice has been to obscure the history of gloss-tradition. We find many glossaries which resemble each other more or less, but we cannot say which collection is derived from which, or how precisely they are related to a

common ancestor. The nucleus seems the same or at least similar, but the rest, the wrappings, different.

The nucleus common to most of the glossaries in Vol. IV. of the Corp. Gloss. Lat. (and to many in Vol. V.) is clearly Abstr., and in none more clearly than the glossary with which this article begins, the Affatim Glossary (Aff.). In pp. 471-581 Goetz provides an apograph of a Leyden MS. (67 F) written somewhere in the north of the Frankish Empire in Charlemagne's time, and an apparatus of the variant readings in other MSS. from the same archetype, the oldest of which is a Ghent codex of Charlemagne's time, now at Leyden (Voss. Fol. 26).

At first sight Aff. seems to be nothing more than a compilation from Abstr., the Abstr. items being re-arranged to suit the arrangement in Aff., which is more advanced in alphabetical precision than the order followed in Goetz' MSS. of Abstr. (see Loewe's Prodromus, p. 107, for details). I need not illustrate the connexion between the two. It is patent on every page, except on pp. 524 sqq. (of Corp. Gloss. Lat. IV.) where there is a marked break after the hi-glosses (see the HI-section of Abstr. on pp. 86-87 of the same volume). Clear traces of compilation from Abstr. do not reappear until certain in-glosses (corresponding to some in the IN-section of Abstr., on p. 98). The break was due to the loss of four leaves of an exemplar, a loss duly recorded by the compiler of Aff. Hic habet minus (i.e. 'deest,' or rather desunt') inter H et I folia quattuor quae excisa fuerunt de exemplari (524, 45-46). Deprived of Abstr. for this portion, the compiler has given us a clearer view than we get elsewhere of his other source (or sources). We find batches of Virgil glosses which often retain their original order, the order of the words' occurrence in the text of Virgil: e.g. (p. 525) No. 41 Infandum (Aen. 1, 251); No. 42 Impar (Aen. 1, 475) ; No. 43 Iamdudum (? Aen. 1, 580); No. 44 In arce (Aen. 3, 531); No. 45 Infabricata (Aen. 4, 400); No. 46 Iam uertitur (Aen. 5, 626); No. 47 Iam ualidum minus (? Aen. 5, 716); No. 48 Imago (? Aen. 6, 695); and so on (cf. 527, 15 sqq.; 528, 39 sqq.; 530, 60 sqq.; 531, 21 sqq.). And we find many Abolita glosses.

Now a clue to the source of these non-Abstrusa items is furnished by an error of the archetype in the ca-words, the fusion of the two glosses Catax and Consentaneum (491, 35 Catax: claudus a coxa, Consentaneum : <conueniens, aptum). Both are Abolita glosses. On foll. 119-128 of Leyden 67 F is a fragmentary transcript (A-F) of a glossary which must be a representative of the source we are seeking; for in it the gloss Catax is immediately followed by the gloss Consentaneum. From the details furnished by Loewe (Prodr. p.171) we see that it consisted of Virgil glosses taken directly from the marginalia of a Virgil text (for Loewe mentions their use of hic in this passage') and of Abolita glosses. Where did these last come from? Must we infer that the (North Frankish ?) compiler of this composite Virg.-Abol.collection (or at least of the Abolita portion of it) drew from a pure Abol. MS., in which Abol. was not associated with Abstr.? The inference is not unnatural, but cannot be called

necessary.

Pure Abstr. MSS. were common. In the composite Abstr.-Abol. MSS. of the early period the two constituents would be marked off. It would be easy for a compiler, whose monastery-library already contained a pure Abstr. MS., to limit his selection to the Abol. items.

Whether incorporated with Abstr. or not, so early a record of Abol. items must be a useful check on the imperfections of Vat. Here is a list of some items of Aff. which might make a claim to be Festus glosses (or on Early Latin words). Those found in Vat. or Sangall. and, of course, those known to be Abstrusa items (for there is no evidence that Abstr. drew from Festus) are omitted:

I. (471, 27) Atta: qui primis plantis ambulat. This looks like Philox. (378, 46 of Corp. Gloss. Lat. II.) Atta: ó ăκpois toîs daktúλois èmißaívwv (on this and the following Philox. glosses see Dammann). 2. (472, 10) Anatum (leg. Anctum ?): miserum (leg. Anatem: miseriam ?). 3. (472, 35) Applare: cocla (leg. cochlear? cotula ?). There is a Philox. gloss (18, 32) Appia (leg. Applar?): eidos σKEÚοUS, ÉS Пaкоúßios. 4 (472, 37) Abstlatata: navis piratica. σκεύους, Πακούβιος. Cf. Philox. 188, 50 Stlata: TeιρатIкоû σкáþоus eidos. But there is rather a suggestion of Caper (Gram. Lat. VII. 107, 1) Stlataris sine C littera dicendum ab stlata naue piratica. And this intrudes a suspicion of the gloss Applare, whether it is not a mere wrong-headed transference from some grammarian's list of nouns in -are: applare, cochleare, etc. 5. (480, 38) Acinari (leg. Ag-?): tricari in paruo morari (a kindred glossary omits 'in paruo '). 6. (480, 50) Anticipassit: ante coeperit (cep-) uel prius prehenderit (found also, with the addition of ante fecerit', in the AA glossary, a South Italian compilation which made use of a good MS. of Abstr.-Abol. and has preserved many of the Festus glosses of Abol. omitted by Vat.) 7. (481, 22) Antiae: capilli a(nte) missi (? with a glossary's characteristic use of capricious suspension, on which see my Notae Latinae, p. 416. Goetz prints 'amissi'). 8. (481, 39) Armilustrium : quod armis locus lustretur. 9. (482, 14) Ancti: excruciati. 10. (485, 49) Actus: spatiumn agri CXX. pedes. II. (485, 52) Augur: qui per auspicia diuinabat. 12. (486, 26) Amussis: regula fabri ae<qualis> qua tabulae diriguntur (the correction is supplied by a kindred glossary). 13. (486, 27) Auentibus aud [i]entibus. There is an Abolita gloss (23, 29) Auentes: cupientes uel gaudentes. It suits Paul. 13, 17 Auere nihil aliud est quam cupere . . . cum significet et gaudere. The Affatim gloss suggests the (doubtful) emendation in Abol. 'audentes' and in Paul. ' audere.' 14. (487, 3637) Bardus hebes, stultus, ineptus, Bardus: carminum conditor. 15. (488, 11) Bacario: urceoli genus (Bec- in Aff., Bag- in a kindred glossary, etc.) There is another gloss (487, 41) Bacarium: uas aquarium. There seems to have been a Sangall. gloss combining the two (IV. 585, 24). 16. (488, 12) Blenni: taetri. A kindred glossary offers as the interpretation 'putidi, taetri.' There is an Abstrusa gloss (25, 3) Blennones: putidi, hircosi. 17. (489, 8) Bidental: locus [bis] de caelo tactus (a kindred glossary omits bis '). 18. (489, 28) Boues lucas: elephantes (-tos); quorum stridor barritus dicitur. Combining this with