STOP Early Journal Content on JSTOR, Free to Anyone in the World This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in the world byJSTOR. Known as the Early Journal Content, this set of works include research articles, news, letters, and other writings published in more than 200 of the oldest leading academic journals. The works date from the mid-seventeenth to the early twentieth centuries. We encourage people to read and share the Early Journal Content openly and to tell others that this resource exists. People may post this content online or redistribute in any way for non-commercial purposes. Read more about Early Journal Content at http://about.istor.org/participate-istor/individuals/early- journal-content . JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people discover, use, and build upon a wide range of content through a powerful research and teaching platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please contact support@jstor.org. Classical Philology Volume XVII October ig22 Number 4 THE TEXT TRADITION OF DONATUS' COMMENTARY ON TERENCE Bt Chables H. Beeson The Donatus Commentary on Terence has been subject to many vicissitudes both in its manuscript and its printed form. The early editions were imsatisfactory because, with one or two exceptions, the editors used only inferior MSS; the last of these old editions — there were about fifty in all — was published by Klotz in 1838-40 and was based entirely on earlier editions. The first attempt at a critical edition was made by Schopen, whose dissertation on Donatus (1821) was followed in 1826 by a Specimen emendationis of a small part of the text. Schopen realized that hia manuscript foundation was insufficient and in 1851 he began a search for new material and enlisted the services of Vahlen for this work. Upon his death in 1867 the material he had collected was given to Reifferscheid, who used it in publishing parts of the text of Donatus and Evanthius (1868, 1874, 1875). The necessity of further study of the MSS became evident and Wissowa imdertook to examine the Italian MSS; the death of Reifferscheid in 1887 put an end to the projected edition. Before this Ritschl had discussed the MSS and early editions of Donatus in connection with his edition of the Vita of Terence by Suetonius, prefixed to Reifferscheid's edition of Suetonius. The next important contributions were made by Dziatzko in Rheinisches Museum (1874, 1879). The greatest ICussiCAL Philoloot XVII, October, 1922] 283 284 Charles H. Beeson advance was made by Sabbadini, whose researches in the Italian libraries made it possible for the first time to form a fairly accurate picture of the text tradition of Donatus, the value of the MSS, and their interrelationships.' By way of illustration he published por- tions of the Commentary represented by various groups of MSS — several of the important MSS are incomplete. The ground had therefore been pretty well cleared when Wessner undertook the difficult task of preparing a critical edition. On the basis of colla- tions made by himself, Wissowa and others, with the help of Sab- badini's contributions, he published an admirable preliminary report in Rheinisches Museum,^ but it was not until 1902 and 1905 that the edition appeared. The result justified the long delay; we now have a trustworthy text (it deserves to be called an editio princeps) with a critical apparatus that satisfies the demands even of the paleographer — something that cannot be said of most of our recent editions of Latin texts. The earliest mention of the Commentary in the Middle Ages is found in a letter of Lupus of Ferrieres (Ep. 103) to Pope Benedict (855-58). In this letter Lupus asks for a MS of Jerome's com- mentary on Jeremiah "post sextum librum," a work which, left unfinished by Jerome, comprised only six books; but Lupus evidently thought his own MS was defective. Lupus asks further for MSS of Cicero's De oratore and the twelve books of Quintilian; of these authors he has only parts and he desires complete texts. His final request is for a MS of Donatus: "pari intentione Donati com- mentum in Terentio flagitamus." The reason why Lupus is asking for a MS of Donatus is probably the same as in the other cases — ^the copy that he has is defective and he wants a complete one. Lupus' MS probably came from Fleury, the famous Benedictine monastery situated only a short distance from Ferrieres, with whose treasures he was undoubtedly well acquainted. Our oldest MS of the Com- mentary (Paris 7920, eleventh century) came from Fleury and there is no reason to doubt that it is a copy of an older Fleury MS. It is possible that Lupus' acquaintance with Donatus dates back to his ' Museo iialiano di anlichiti claasica, III (1890), 319 ff. and 381 ff. ; Studi italiani di filologia clasaica, II (1894), 1 ff., and III (1895), 249 ff. »LII (1897), 69 ff. DoNATus' "Commentary on Terence" 285 stay in Fulda, where he was a student under Rhabanus Maurus for a number of years; Mayence, the home of one of our lost MSS, was within the sphere of influence of Fulda and used the same script, the Insular. Our two oldest MSS attest an interest in the Commentary in the eleventh and thirteenth centuries in France, and here too toward the end of the fourteenth century Nicolai de Clemangiis cites Donatus.^ The Commentary, however, did not become widely known until the following century. The Council of BMe (1431-49), like the Council of Constance before it, gave a great impulse to the study of classical authors and the search for their MSS; among the MSS discovered at this time was a codex of Donatus. In 1433 Aurispa writes from Bdle to Tebalducci at Florence that he has discovered a MS of Donatus at Mayence. About two years later Pier Candido writes from Florence to Pizzolpasso, archbishop of Milan, who was in attendance at the council between the years 1432 and 1439, that he is sending him the text of the Phormio, which he has excerpted from his (i.e. Pizzolpasso's) MS, at the same time complaining of the difficulty of his task.^ Pizzolpasso replies that he is making him a present of a copy of the Phormio which his secretary has made (from Candido's copy). The situation is made clear by two letters of Pizzolpasso to Candido written in 1437.' Nicholas of Cues, it appears, had gained possession of the Mayence codex and had given it to Pizzolpasso; the latter had brought it, or sent it, to Milan and commissioned Candido to make a copy of it. Candido copied the Phormio and sent it to Pizzolpasso but it is not clear whether he con- tinued his work, though he probably did. Complete copies of this codex were made later in Italy as will appear below. In addition to these MSS we must assume that there was in existence another copy of the Mayence codex, made by Aurispa, or for him, probably immediately after its discovery. Aurispa returned to Florence at the end of 1434; he followed the papal court on its travels to Bologna and Ferrara and on its return to Florence and probably carried his ' Sabbadini, Storia e Critica di Testi Latini, p. 206. ' A copy of this letter is prefixed to the text of Donatus in the Oxford codex (C) ; it was published by Dziatzko, Jahrh. f. class. Philol. Suppl., X, 692, and Sabbadini, Mus. Hal., Ill, 407. 'Sabbadini, Mus. ital.. Ill, 411 and 415. 286 Charles H. Beeson Donatus with him; Sabbadini assumes that this was the beginning of the diffusion of Donatus MSS in Italy.' Aurispa's correspondence* reveals the characteristic activity of the Humanists in seeking new MSS, copying and correcting them — and often failing to return them to their owners; in this activity Donatus plays an important role. In the year 1451 we find another important MS of the Com- mentary in Italy, again in the hands of Aurispa. As early as 1447 Laurentius Valla had written' from Naples to Giovanni Tortelli requesting him to ask the Cardinal Prospero Colonna, or anyone else that had a copy of Donatus, whether there was a complete copy of the Commentum and whether Donatus wrote a commentary on all the plays of Terence (none of our MSS has a commentary on the Heautontimoroumenos). A friend of his, he added, saw a Donatus "apud Carnotum sine tertia comoedia 'Eavroj'Ttjuopoujuecw et non Integra quinta 'Y.Kvpq. itemque cum defectu in sexta, quae dicitur ^opixiuv. Aurispa writes from Rome in January, 1451, that he has known for a long time that there is a Donatus "Carnuti in Gallia, in bibliotheca ecclesiae maioris." He has had it copied and has received the copy; he is now having it transcribed, and when the work is finished he will send the original to Panhormita, not as a gift but that others too may make use of it. Three weeks later he writes again, repeating the statement that he is having the MS transcribed "ut ipsius copiam amicis facer e possim ne forte mihi eveniret quod Guarinus, Karolus et tu mihi fecistis' ' (Aurispa refers to the difficulty he had had in recovering MSS he had lent to these friends). We have therefore information about three MSS of Donatus brought to Italy: Aurispa's copy of the Mayence codex (probably to Florence in 1434), the Mayence codex itself (to Milan in 1436) and Aurispa's copy of the Chartres MS (to Rome in 1450). We find a reference to one other Donatus MS in Aurispa's correspondence. He writes to Panhormita,* probably in 1446, that the monk who brought Donatus' Commentary on Virgil to Italy has found a com- mentary of Donatus on three plays of Plautus and at his request has written to France for it. But in spite of Aurispa's assurance that the 1 Sabbadini, Siudi Hat, II, 17. ' Sabbadini, Biografla documentata di Giovanni Aurispa, Nolo, 1891. » Sabbadini, Mus. ital.. Ill, 387. ■• Sabbadini, Aurispa, p. 107. DoNATUs' "Commentary on Terence" 287 monk is "doctus et solers antiquitatis indagator, quamvis Gallus," he has probably mistaken, as Sabbadini suggests, Terence for Plautus. A Paris codex (7921, saec. xv) contains a commentary on three plays of Terence and this MS, or a similar one, may be what he had in mind. There is no evidence that Aurispa ever received the MS. The purpose of this article is to determine the character of the various lost archetypes through which the MSS have come down to us and to point out the disastrous effects of their script on the text. The outline given above, based largely on Sabbadini's articles, furnishes the necessary background for discussing the relationship of the MSS of the Italian group. No other profane Latin text offers such abundant and convincing evidence of the stages through which it has come down to us. This is all the more surprising because most of the evidence is derived from MSS of the fifteenth century, MSS in which we expect to find the traces of their descent obhterated or confused through the activity of the Humanists. No other text exhibits such bewildering paleographical confusion or betrays such helplessness on the part of the scribes in dealing with a puzzling script and still more puzzling abbreviations. In discussing the various stages of transmission I shall follow Wessner's stemma,' though I believe it to be incorrect in important particulars. The evidence given can be easily appHed to the re- arrangement I shall suggest at the close of this article. The text of the Commentum in its present form goes back to the sixth and seventh centuries. In regard to the script of the original MS Wessner^ contents himself with calling it majuscule; he points out that in several places in our oldest MS Latin words or letters adjacent to Greek words are written in majuscule, i.e. the scribe who copied the majuscule archetype in a minuscule script failed to distinguish between the Greek and Latin text and transcribed some of the Latin text as well as the Greek words in majuscule (Greek minuscule did not make its appearance in Latin MSS until the ninth century). ' Sabbadini's classification is as follows: Class I is composed of ACRTVB; II, aO; HI, {)=«?. pr.; IV, PcdLMNrstxz. Especial precaution is necessary in regard to the signatures of the MSS; Dziatzko's P =Wessner'8 A; Sabbadini's P = Wessner's F; Wessner's P= Paris 7921. * Praef. p. viii. 288 Charles H. Beeson Now the majuscule letters in the Latin words in the passages cited by Wessner are uncial, except that half-uncial r occurs once along with the uncial form. Sabbadini is more explicit than Wessner; he asserts that the archetype was uncial. He cites^ as evidence the wrong division of words (283. 16 sibi noscat at for si vino scatat) and the confusion of p and r (340. 12 de se rapuit for dicit se parvi). By comparing the lacunae in the Phormio and the Hecyra he recon- Archetypus s. VI A» XI «. xv< LedJfPrbst/ structs an uncial codex with thirty-seven letters in a line and twenty or forty lines to the page.^ None of these arguments is conclusive. The failure to separate words is more likely to occur, perhaps, in an uncial codex but it is common enough in minuscule MSS, especially if they are carelessly written; in fact, as will be shown later, the archetype of AB did not have the words clearly separated. Sabbadini fails to draw the proper conclusion from his clever emendation of dicit se parvi for the manuscript reading de se rapuit; it is true that the confusion of p and r is quite possible in uncial, but it is extremely common in the Insular script also. In • Studi ital., Ill, 335. * Op. cil., 11, 75. DoNATus' "Commentary on Terence" 289 this case, however, I am inclined to believe that we have an instance of metathesis, like rede for certe (90. 6).^ An uncial codex of forty lines to the page would certainly be unusual; Sabbadini's calcula- tions would apply equally well to a minuscule MS and the lacunae may not have been in the uncial codex at all. The original MS was undoubtedly uncial, probably with a mixture of half-uncial letters. This is the script one would expect to be used in the sixth and seventh centuries and the use of uncial for technical works is especially common; over forty of the four hundred uncial MSS that have survived belong in this category. The confusion of i and t, of which there are several examples, may be due to this script. The lemmata were written in rustic capitals to distinguish them from the commentary, as the following errors prove: 88. 20 MEUM] at eu A; 148. 5 MORA] moni A; 168. 14 ACCURATE] accu te A; 82. 16 meo (incorrectly included in the lemma)] AAeo A, adeo TCV. That the MSS go back to a minuscule archetype was recognized by Sabbadini and Wessner, but they did not attempt to determine its character. Lejay in his review of Wessner's edition^ makes a cautious suggestion that the archetype was perhaps in the Insular script but the two examples he cites as evidence are not to the point.' The confusion of letters (u and a, u and n, t and c, ni and iu, and especially r and point to a MS in Caroline minuscule intervening between the uncial archetype and the immediate archetype (which I shall call X) of a and j3. Other errors make it clear that X is descended, directly or indirectly, from a MS written in the Insular script (which I shall call Y). The most striking proof that Y was an Insular MS is furnished by the confusion of the words post and propter in X. There are several forms of abbreviating these words in the Irish and the Anglo- 1 A parallel case occurs in ii. 249. 13; here the codex Cuiacianus, which belongs to the /3 family, has parlum, the right reading, while the other MSS, all of the fifteenth century, have raptum. Since this error originated within the fi family it could not have been caused by a confusion of uncial letters; it must be due to a confusion of Insular letters or to metathesis. The archetype of the /3 family was an Insular MS. The reverse process has taken place in ii. 337. 13; here partum is found in all the MSS for raptum. Other examples are: 388. 2 mirari for rimari; 402. 20 nam sit for mansit; ii. 45. 14 captio for pactio; ii. 81. 23 sentiat for nesciat (V). 2 Revue critique, LVI (1903), 168-72. ' "De singulaires lapsus [ante pom- enim, 172. 18, duo pour hie, 136. 2] pourraient peut-&tre permettre de supposer une mauvaise lecture d'ecriture insulaires." 290 Charles H. Beeson Saxon scripts; Lindsay describes* the form pi as characteristically Anglo-Saxon for post; he also calls attention^ to the use of the curious form pV for propter at Mayence, as evidenced by two Anglo- Saxon MSS originally from Mayence but now in the Vatican; he gives a third example from a Limoges MS. Now there are thirteen places where both families of our MSS have post while the sense requires propter and the only possible explanation is that this rare abbreviation for propter was found in Y and was misunderstood by the copyist of X, or some intervening MS. I count, of course, only the cases where both families are represented. For the greater part of the text we have only the testimony of the j3 family, which furnishes twenty-five additional examples. There is not much doubt, however, that most, if not all, of these errors were already in X. It is possible, therefore, that Y was written at Mayence and that France is not, as is generally assumed, the land that preserved Donatus for us.^ Further evidence that Y was written in the Insular script is found in the confusion of the letters n, p, and r. Only those errors are cited here which occur in all the MSS of both families and pre- sumably therefore existed in X; it is extremely improbable that X had the right reading in all these cases and that the errors were made independently by the scribes of a and 0: n for p: 318. 6 voluptatem] voluntatem; p for n: 82. 17 voluntate] voluptate; 87. 4 voluntas] voluptas; p for r: 110. 4 incertarum] inceptarum; 288. 10 rarum] parum; r for p: disceptationem] dissertationem} ' Notae Latinae, p. 193. ^ Op. cil.. p. 200. ^ This abbreviation regularly stands for praeter and was evidently so used in ii. 13. 18 where AV have incorrectly expanded it into propter. ' Wessner makes the Moguntinus a copy of the Carnotensis while Manitius, Geschichte der Lateinischen Lileratur des Mittelalters, p. 486, n. 5, thinks that the Moguntinus is derived from (he MS belonging to Lupus of Ferriferes. Mayence was an important Anglo-Saxon center; however, the presence of an Insular MS, especially a commentary or grammar, in Fleury or Chartres need not cause surprise. The archetype of the Mayence Insular MS of the fourth decade of Livy apparently came from Chartres (Traube, Palaeographiscke Forschungen, IV, 20) . ' Of course the confusion of these letters can take place in other scripts; any letter can be mistaken for another no matter how unlike they may be; one word may be contused with another. The interchange of voluptas and voluntas is common, due to the confusion of the words, which were often associated; cf. the addition made in the insular MS Bale F. iii. 15. a, f. 29' at the end of De proprietate sermonum: "Inter voluntatem et volumtatem et volumptatem et voluptatem hoc interest quod voluntas DoNATus' "Commentary on Terence" 291 There is practically no evidence as to the character of the script of X. We cannot depend on abbreviations, because any Insular forms we assume for X may have been copied from Y. They show only that X or Y was Insular. The MSS from which X descended contained a bewildering variety of abbreviations which caused great confusion throughout the entire tradition. Some of these were undoubtedly in the uncial archetype but most of them may be regarded as Insular in char- acter, whether by adoption or invention. The Insular scribes tended to use many abbreviations, especially in grammatical texts, com- mentaries, etc. The early MSS of Donatus seem to have been written while the abbreviating systems were still in a state of flux. Thus pt, with one or two abbreviating strokes, meant not only propter and post, as stated above, but also potest and praeter — all Insular except the last, and that occurs in our MSS. This explains why we find praeter for potest (four times) and for post (three times). Pro for post (30. 18) may point to an abbreviation p with suprascript 0, common in the Insular script; postquam for proquam (441. 2 V) may be due to the same abbreviation. We find the same uncertainty in regard to the abbreviation s- (or s,); it is found in Insular MSS for sic, sed, secundum and scilicet and is often mistaken for si, espe- cially in the Insular script on account of the ligature (cf. 193. 15 where A has off ens- for offensi). So we find in our MSS si = sed; sic = si, sed, secundum; scilicet = si, sic, sed; in all there are about thirty cases of this sort. Similarly id caused trouble. It is found for idem once or twice and for id est several times; in ii. 439. 3 RC have idem, has id est and V has vel (i.e. ul; the loop of d was mis- taken for the second stroke of u; ii. 270. 16 we find id est for vel in X; the confusion of these letters and of the propter-post abbreviations is responsible for the reading postulare for propter id 127. 17 (Y must have been written in a compact script). Wessner here reads ideo; it is pos- sible that we have here I'cf = ideo, of which Lindsay gives an example* dei est volumtas diabuli voluptas mali desiderii." In the same way the presence of Insular abbreviations does not necessarily mean that they were copied from an Insular MS. A continental scribe writing Caroline minuscule might use them occasionally. Later many of them passed into common use. ' Notae Latinae, p. 110. 292 Charles H. Beeson from a Cologne MS (Irish scribe), but it may be that the usual abbreviation (ido) was misread. Nam seems to have been abbreviated with the initial letter as well as non; there are about twenty cases where they are confused. The usual confusion between ergo and igitur occurs — there are about a dozen cases. We find dicit abbreviated in several ways in our text: by sus- pension, d, or (f; the former caused such errors as quid or quod for qui dicit (358. 2) and the latter has been expanded into de nine times, seven times in the archetype of all the MSS ; d, or another suspension die, which occurs a few times, is responsible for the reading dum (i.e. dU) which occurs eight times; do for dicit, which occurs twice, is a mistake for de or for dc (syllabic suspension); I have observed no case of dl for dicit but the many mistakes in writing dicitur (df) for dicit point to such an abbreviation (contraction) unless we assume an abbreviation by suspension as the cause. There are also many cases of dicitur for dixit. The Insular abbreviations that perhaps caused the most trouble in our text are the quam, quia, quod series; the abbreviations for these words consist of q with the tail bisected by strokes that often differ very httle from one another, especially when carelessly made; they were frequently mistaken by the Insular scribes themselves and still more frequently by scribes not famihar with that script. They were also confused with the continental abbreviation for qui (Q) and apparently with a rare Insular abbreviation for que, which also involved quae in the confusion. In all there are about a hundred passages where one of the series quam, quia, quod, qui, que (quae) has been mistaken for another and in many of these passages the scribes have expanded the abbreviation in three different ways. A few errors ah-eady existed in X, e.g. 19. 19 quod] quam A quia TFCV; 302. 19 quam] quam quod BTCV (an interiinear correction was copied in the text along with the original error) ; errors in all the MSS in that part of the text represented only by the /3 family may have originated in /3 or in X, e.g. 475. 14 quam] quia /3; ii. 205. 10 quod] quam CV; ii. 390. 17 quia] qui V quam RCO. The confusion of quando for quomodo in X (290. 1) was probably due to an Insular abbreviation; quando was not ordinarily abbreviated except in Insular. DoNATus' "Commentary on Terence" 293 To prove that X was an Insular MS we must depend on the confusion of n,p,r in AB as against the MSS of the /? family or vice versa; we are thus limited to that part of the text represented by both A and B (about a seventh part of the whole). Errors in A alone or in B alone do not count since their archetype (a), as will be shown presently, was an Insular MS and the error could have been made by the scribe of A or B as well as by the scribe of a. Now there are no cases of this sort in AB and only two in /3; 122. 3 volun- tatem] A, voluptatem TCV; 8. 5 sarcinarum] A, sati {y FV) rarum TFV. The evidence therefore is quite insufficient to warrant any conclusion. The /S family furnishes a number of examples from that part of the text not contained in A or B but we cannot tell whether the mistakes were made by the scribe of X or of ^; e.g. ii. 33. 13 aptius] artius V arcHus C; ii. 64. 11 quae sit ars] quesit as C, queritans V; there are a half-dozen cases where voluptas and voluntas are confused. Wessner has shown that A and B are not independent copies of X but are derived from a copy (a) of that MS. There can be no doubt that this copy was written in the Insular script. I shall first give the evidence furnished by that part of the text contained in A and B. Here again abbreviations are of no value but if we find A or B repeatedly confusing n, p, and r the only explanation is that in the archetype these letters were written in a script that made these confusions possible, and that script must be the Insular. We may have a confusion of n for p in 197. 5 supra] sim' A; the Insular abbreviation for supra is sup, which was corrupted to the meaningless sun; this in turn was corrected, consciously or unconsciously, to sim'; n for r: 128. 2 efficaciorem] efficationem A, corrected by A^, to efficatiorem; 134. 10 rede] node A; 150. 17 gratia] gna A {gratia is commonly abbreviated gfa); 176. 18 adierunt] adie non A (the verb was written adier — r is regular for -runt — r was mistaken for n and n was expanded into non; the words were evidently not separated in the archetype) ; a probable case is 174. 9 expelit oratorie] expenatorie A (the lemma was abbreviated, as often, to exp- and o was mistaken for e — a common error); possibly also 28. 21 ara] ami A; r for n: 134. 6 genere] gr A (evidently an arbitrary abbreviation stood in the archetype, gn) ; 199. 13 anacoluthon] anacoluthor A; 215. 3 inm.] in B, 294 Charles H. Beeson ira A (a probably read in, A mistook r for n and added a to make a Latin word). The confusion of I and b occurs with especial frequency in Insular MSS owing to the peculiar curve to the left at the bottom of the shaft in both letters; there is one example in A 240. 10 celerem] ceberem. The evidence from B that a was Insular is meager. We find r mistaken for n 139. 12 miratur] minatur B; r was apparently mis- taken for p 135. 17 convenire] conve- nup- B. The following errors are found in A where B is lacking; here we cannot be sure that they were not already in a ; n for r: 35. 3 Terentio] tenent-; 60. 4 ralionem] natu; 100. 1 res] ines; 112. 7 ingralum] ingnatum; ii. 19. 18 ratione] natione; r for n: 25. 21 Rinthonica] rinthorica; 31. 7 Serranae] serrarq; 36. 10 minoris] miroris, with minoris above the line; 93. 14 consentivam] consercinam; 256. 8 acceptione] accepto rone (probably written in the archetype acceptiore, with on above or); A has confused r and y twice; 9. 13 maceror] maceroy and 253. 17 asyndetos] asrndetos; this confusion is not uncommon in Insular MSS, but it also occurs in ordinary minuscule of a pronounced cursive character. Examples of the confusion of the abbreviations for quam, quia and quod are numerous; e.g. quam for quod 89. 21 (A); gwod for quam 124. 9 (A, quam A^); 127. 3 (B); 172. 2 (B); 255. 13, twice, (B); ii. 285. 3 (B); ii. 298. 4 (B); ii. 308. 19 (B); quod for quia ii. 270. 3 (B); ii. 294. 23 (B); ii. 301. 20 (B); ii. 308. 18 (B) ; quam for quia 231. 16 (A). There are a number of cases where que is confused with quam, quia or quod: 25. 8 que] q; quia A (evidently quia was written originally, q; was written above as a correction, and later copied into the text) ; 94. 19 and 95. 1 where A has the Insular abbreviation for quod; 155. 10 quoqu^] g Q A; 134. 9 nusquam] nusq; A; 240. 17 quicquam] quoquam A C^, quoque BTC; ii. 307. 11 quod] quia C, que B ; ii. 310. 4 quod] que BC. It is possible that the non- Insular abbreviation for qui (q) is responsible for the following errors: qui for quia'' 23. 8 (A); 53. 15 (AV); 189. 3 (A, quod T); 202. 6 (A, quid T, quod C) ; 239. 17 (BCV) ; ii. 311. 16 (B/3) ; qui for quod 104. 23 (A,pTC); 147.5(A); ii. 309. 14 (B) ; gwod for gm 99. 8 (A) ; 243.22 (B); 252. 8 (A); 333. 14 (B, p TC; p and q are confused several ' 1 he cases where quia is followed by a are not included. DoNATus' "Commentary on Terence" 295 times in our MSS); que for qui 170. 4 (A); 172. 17 (A); ii. 317. 14 (B/3). Of course it is not necessary to assume that all these errors are due to abbreviations; the carelessness of the scribe is without doubt often responsible and the text of a commentary (like that of a grammar) is especially liable to change. On the other hand, this multiplicity of abbreviations does not appear impossible if we bear in mind that several generations of MSS are involved, that more than one scribe was probably active on each one of them and that some, at least, of the MSS were carelessly and hastily written. The number of cases where the Insular abbreviations for haec and hoc were confused is surprisingly small — less than a half-dozen. The mistake of homo for hoc may be due to the fact that Insular scribes often abbreviated homo {ho). The lemma hem is transcribed several times as hoc est or /laec est (i.e. he, with the m-stroke over e mistaken as two abbreviating strokes), e.g. 184. 19 (AV); 192. 14 (A); 204. 15 (A, twice); 229. 9 (A, hie TC). The Insular abbreviation for huiiis (hs) is responsible for a number of errors. It was probably in X and passed into o and /3; in 95. 20 A expanded it into haec sunt (i is the regular abbreviation for sunt in A), V expanded it correctly, but TC have has; 102. 16 A repeated his mistake, V misunderstood the abbreviation but V corrected it, while TC have hes; 108. 22 C has hes; 115. 16 T has hec; 115. 1 A has haec sunt, V^ again corrected the error of V, TC have huius. The error in A 192. 15 ante] ea n is due to the Insular abbreviation for ante (an). This abbreviation may account for the reading of V 174. 14 apud (expunged) ante for ante (ap is the Insular abbreviation for apud; n was mistaken for p; the correction was written above the line and later copied in the line along with the error) ; C has in. Another Insular compendium {a = aut) caused the error aul for an (a with an n-stroke above it) in 206. 10 (BTC) and 289. 6 (B). These words are confused eight times in our MSS. The words tamen and tantum were originally abbreviated only in the Insular script; the compendium tfh was used for both and this naturally caused confusion; 174. 4 B has tm for tamen but in 245. 9 he writes tantum for tamen. The (late) Insular abbreviation for usque (us) occurs once in B (240. 17) and the Insular abbreviation for vi (m) once (223. 3). 296 Charles H. Beeson The practice of Insular scribes of putting an apex over certain monosyllabic words sometimes caused confusion, as the apex was often undistinguishable from an abbreviating stroke or an m- (or n-) stroke; e.g. we find sic for sic in A (73. 3), in T (240. 11) and C (203. 9) but sic is also the common abbreviation for sicut; this explains why in 163. 13 B has sic, the correct reading, while ATCV have sicut; X and a probably had sic with an apex, which A and mistook for an abbreviating stroke; 171. 2 V wrote sicut for sic which V^ corrected; the reverse process took place in 60. 10 where sic ut is right; A has sic and TC have sic. The error in 59. 8 where A has nc { = nunc) for sic is probably to be explained in the same way; the ligature si, which is peculiar to the Insular script, often resembles a high-backed, round-shouldered n, the apex was again mistaken for an abbreviating stroke; the same error occurs in ii. 276. 1 in B. The confusion of letters is reversed in 225. 18 where C has esiphasis for enphasis {exifasis T) and in 233. 13 where A has occulta sis for occuUans, and probably in 173. 9 where A has necafis for notans; n looked like the si-ligature but the eleventh-century scribe was familiar only with the yi-ligature which is often hard to distinguish from the Insular form and so copied fi; the confusion of o and t for e and c occurs several times in A. The apex may be responsible for such errors as an for o and of hinc for hie, of which there are some half-dozen cases; it may have contributed to the error contra (ctra) for o terra (d Ira) in A (207. 3); in this case the apex was combined with the abbreviating stroke; in 120. 10 A actually has an apex over o in o iurgantis (for obiurgantis). There are a dozen cases where A omits i after letters which admit an i-subscript; the practice of writing i (and a) in ligature below the line, especially characteristic of the Insular script, often caused the letter to be omitted; in one case A wrote minimi (41. 8) with three i's subscript; this was probably so written in a. Another peculiarity of Insular scribes, reflected in A about fifty times, is the omission of one of two double consonants or the doubling of single consonants, e.g. apareo, comorata, erat {=-errat), teritus, percuro, teror, sumus, amisis, none, opido, ociso, etc., and gerrit, rapperet, commis, etc. These errors are rare in B, as the scribe was probably the scholar who revised and condensed the text for his DoNATus' "Commentary on Terence" 297 own use; he eliminated most of the misspellings. Many such errors sifted down through the /3 family into C, but the Italian scholars would naturally correct most of them. The archetype did not have the words separated as the following errors show: 30. 12 poeta sed\ poetas A (s. = sed); 44. 16 positi sunt] positis; A (s = sunt); 121. 5 scripta sunt] scriptas A; 177. 5 pauds] pauci sunt A; 211. 10 dies] die s A; 91. 12 non (i.e. n) Oedipus] noie (i.e. nomine) dippus A; 98. 16 ut Lesbia] utiles' ina A. Many errors were caused by the fact that the letters were compactly written, e.g. 36. 15 initium (with initial I-longa) butium A; 63. 1 interea] hie ea (i.e. from Inl ea); the confusion of hi for In is a characteristic Insular symptom. Other errors are: id for ul, ul for id, ol and cl for d, di for ch, du for hi (e.g. duo for hie in A, 136. 2; this is not, then, an Insular symptom as Lejay assumed [see p. 289, n. 3]; the error could occur in ordinary minuscule) ; li for b, iq for up, etc. Turning now to the /3 family we find, according to Wessner, that it divides into two groups, one represented by V, the other by the lost Camotensis. Accepting for the moment this scheme as the correct one, it can be shown that /3 was an Insular MS. I cite here only the evidence of V. We find p for n 199. 6 where V has volup- tatem for voluntatem and probably 487. 19 where V^ reads voluntarium, with n in an erasure; n for r: possibly 245. 16 cur] eum V, with cur written above by V^; ii. 74. 18 rursus] nutus V; r for n: 431. 1 Donacem] doracem V, so also 435. 7 and 448. 4; ii. 469. 12 ante] arte V; r for p: 439. 8 torporem CV^, terrorem V; ii. 20. 11 aptius] artius V. Insular abbreviations again caused trouble; e.g. 125. 13 tantum] V^ tamen V; ii. 48. 18 hodie] autem V; the word was written as a suspension as is often the case in a lemma, and h was mistaken as the Insular abbreviation for autem {h, with a suspension stroke). In 120. 11 V has qn for quoniam, a form that Lindsay finds in the older continental specimens of the Anglo-Saxon script' — among them a MS written in the script of Fulda, the type of Anglo-Saxon script that prevailed at Mayence; ii. 280. 15 V has quoniam for qvxindo. Lindsay^ cites qh for quando from an old Mayence codex Vat. lat. 1447; ii. 478. 14 V has quin for quando, RCO have quoniam; evi- ' Notae Latinae, p. 263. ' Op. cit., p. 221. 298 Charles H. Beeson dently j3 has qn; ii. 276. 7 V has the Insular abbreviation for id est ([•). Two Insular abbreviations are probably involved in the error 475. 21 where all the MSS (a lacking) have unde for bene (un = unde, bn = bene^); the same error occurs in V (ii. 472. 1); un for bene is found in T (159. 12); this form caused the error of uestri i = un) for unde in T (317. 2). What do we know about the archetype of V ? We have proved that it was Insular. We know that it was an old MS; in the margin of V, opposite Hecyra v. 3. 1, where there is a gap in the text, the scribe has written "deletum propter vetustatem." This sounds suspiciously like two marginal notes found in R, a copy of the Mogun- tinus ("consumptae erant Utterae ob vetustatem" and "deletae sunt htterae") but V cannot be a copy of the Moguntinus. There are a thousand errors in the Moguntinus not found in V. Sabbadini,* assuming that a certain gap in V was caused by the omission of a single line, argued for a small script with many abbreviations. He dates the archetype in the twelfth or thirteenth century, that being the period when condensation reached its maximum. But the Insular script is capable of condensation equally great; the Insular scribes, more than any others, practiced the art of economizing time and money — and parchment cost money and abbreviations saved time and space. The archetype of V, like that of AB, was probably written in a compact script with many abbreviations. We come now to the script of the Camotensis. According to Wessner's stemma we must depend upon the readings of the Mogun- tinus and the descendants of Aurispa's copy of the Camotensis; the latter, however, have suffered so much from interpolation and crossing that it is impossible to reconstruct their archetype with any degree of certainty. If the following errors are peculiar to TC, as they probably are (Wessner does not give the readings of the deteriores here), we may assume that the Camotensis was an Insular MS: 149. 4 u.c.a.p.] ucar. TC; 204. 2 non eveniet] pene {pone C) veniet TC; 260. 16 gnatam] gratam C, grata T; 230. 11 g.t.p.r.s.] c.{t. T) r.n.n.s. TC; 283. 16 fontem] fortem TC. In any case these errors ' Lindsay, Notae Latinae, p. 26, cites one example of this syllabic suspension for bene from an eighth-century Anglo-Saxon MS, Boulogne 63-64. 2 Studi ital., II, 45. DoNATus' "Commentary on Terence" 299 show that the Moguntinus was a copy, direct or indirect, of an Insular MS. The evidence of in two passages might also indicate that the Camotensis was an Insular MS; ii. 481. 5 repandum] ore pardum O; in ii. 437. 19 RC have cum citatione for concitatiore (OV); since R and C are independent copies of the Moguntinus the latter MS must have had cum citatione and, if it was a copy of the Camo- tensis, the latter must have been in the Insular script and the error must have been made by the copyist of the Moguntinus. The error could not have been in the Camotensis since Aurispa's copy of this MS (represented by O) had the right reading. The position of O, however, is very uncertain; its affinity with RC is very strong; in fact Wessner later' derives it from the Moguntinus. In the rearrange- ment of the stemma which I shall suggest there is no difficulty in prov- ing that the Camotensis was an Insular MS and was copied from an Insular MS. One old MS remains to be discussed, the lost Moguntinus. Here we are on firmer ground. Candido in his letter accompanying the copy of the Phormio which he made from the Moguntinus for Pizzol- passo (see above, p. 285) writes feelingly of the difficulties of his task. His statements might be discounted as due to a desire to magnify his service or to excuse his errors — ^Italian scribes were likely enough to exaggerate the difficulties caused by the script — but there are two expressions in his letter that make it clear that he is dealing with realities. He refers to " barbariem quandam veteris scripturae" and "cariem vetusti operis," which must refer, not to the text, as Dziatzko assumed, but to the MS and the script. We have to do, therefore, with an old MS; it was in bad condition and the script was unusual. If Candido had used a technical term he would have called the script "Langobardica" (i.e. "barbarous"). That the script was difficult is proved by the mmierous errors in RTC; that the MS was old and in bad condition is shown by the marginal notes in R, opposite gaps in the text ("consumptae erant fitterae ob vetustatem" and "deletae sunt litterae"). The fact that these omitted passages are found in C, another descendant of the Moguntinus, need mean only that R was a less painstaking scribe or less skilful; in fact the corrector of R did add passages that the • Berliner PhUologische Wochenschrift, 1906, 797. 300 Charles H. Beeson copyist omitted. According to Wessner's stemma three copies were made of the Moguntinus, represented by R, TCF, and Aurispa's transcript. The last is of Httle use, either from a critical or paleo- graphical standpoint, since the MSS derived from it are badly inter- polated; R contains only the Phormio; we are therefore compelled to rely almost entirely upon the readings of TC (the readings of F are given by Wessner only for the Vita and the Tractatus de comoedia) for the reconstruction of the Moguntinus. The picture we thus get of the Moguntinus is greatly obscured, since we see it through the medium of intervening fifteenth-century archetype of TC. An examination, however, of the cases where T and C disagree will show that their archetype could not have been a fifteenth-century MS; we find the same sort of evidence with which we have grown familiar in examining the archetypes of AB and of V — and more of it. The number of cases where n, p, and r are confused is striking; e.g. n for p: 255. 9 voluptati] volunlati T; p for r: 21. 5 Rinthonicas] pinthorijcas {rij corrected from ris) C, rynthonims Q2; 449. 18 retice- mus]2 Pee ticemus T; p for n: 126. 12 nunc iam] N2, nuptia T, nunciata C; 241. 15 omnium] impiusum (s expunged) C, enim T (this may have been corrupt in the Moguntinus) ; 305. 19 voluntatis] voluptatis T (the same error in T 308. 4 and 400. 11); possibly 416. 14 diminu- tivum] ditopanatium C (in the archetype of C the reading was probably ditopatium, with na above pa as a correction); n for r: 122. 15 ratione] nomine C {ratione was abbreviated roe and was mistaken for noe, the common abbreviation for nomine); ii. 432. 12 may belong here, rationem] nom (from rorn) RC, numero V, vim 0; 133. 5 cur] C^ cum C, cui T; 201. 12 qvurto] qnto T; 232. 8 officiorum] offensionem TC, officionem T (in margin); 274. 5 possessorem] poxessionem T; 279. 22 diversorium] discensorium T; 376. 10 littera] Ira VC, na T; 404. 12 cur] cm T; r for n: 25. 21 Rinthonica] sinthorica T; 74. 1 peccationis] peccatoris T; 110. 11 i.n.a.p.] n.r.a.p. C; 117. 11 con- ventionem] conventorem T; possibly 352. 21 bene] di, corrected to dr T {bn = bene; cf. ii. 420. 1 where R has bh for dicitur). The following errors occur in that part of the text where T is lacking, but it is safe to assume that most of them existed in the archetype: n for p: ii. 231. 9 voluptatis] voluntatis C, so also ii. 297. 21 ; p for n: ii. 222. 23 inertia V, ineptiam C; n for r: ii. 324. 16 quantopere] DoNATUs' "Commentary on Terence" 301 quanta pene C; ii. 483. 12 corde] CV condere RO; r for n: 462. 24 cur] V, cum C; ii. 11. 11 retentum] retortum CV^; p for r: ii. 360. 4 ret] :pei C. This mass of evidence proves two things: (1) that T and C are not derived from the same fifteenth-century archetype — no renais- sance script could have caused so many errors of the type hsted above; we have here two descendants of the Moguntinus instead of one. The agreement of TC, therefore, represents a text that has escaped the revising hand of the Humanists and brings us six cen- turies nearer to the uncial archetype. (2) The Moguntinus was written in the Insular script. Neither T nor C is a direct copy of the Moguntinus. This is proved for C by the fact that it has errors in common with F where T preserves the right reading; for T one error furnishes sufficient proof, 17. 17 smis for sententiis; snis would not occur in a ninth- century MS. The Moguntinus, therefore, was written in the eighth and ninth centuries and not in the twelfth as Sabbadini and Wessner assumed. They based their conclusion on the confusion of et and in and of et and quia. There is indeed an extraordinary number of cases of these confusions in T and C (nearly fifty) but the same errors are found in other MSS (BVRO) and three or four times in X, the archetype of all our MSS. We have shown that of Wessner's three archetypes of the Italian MSS two were Insular and one a copy of an Insular codex. But were there really three archetypes? Wessner in his first classifica- tion* derived V from the Moguntinus, and the deteriores, with one or two exceptions, from the Carnotensis. He made the Moguntinus and the Carnotensis independent, each of the other. His later grouping was determined largely by two lacunae, a large one in the Hecyra^ involving the loss of several quaternions, a small one in the Phormio^ involving the loss of one, or possibly, two folios, and by the fact that in the better MSS the schoha for Phormio ii. 3 are arranged in two series, while in the deteriores this double version has been revised into a single composite one. His starting-point was the passage in Valla's letter (see above, p. 286) where it is ' Rheinisches Museum, LII, 95. ^ iii. 5. 8— v. 1. 2. ' ii. 1. 4—9. 302 Charles H. Beeson stated that in the Carnotensis the Hecyra was "non Integra" and the Phormio was "cum defectu." According to his theory the Moguntinus was copied from the Carnotensis' which at that time had only the Phormio lacuna; later the Carnotensis suffered a further loss of leaves in the Hecyra and both of these lacunae passed into Aurispa's apograph. It was natural enough, since RCO actually have a lacuna in the Phormio, to identify the "defectus" in the Carnotensis with the Phormio lacuna in the Moguntinus. But this theory is not without its difficulties. In the first place it forced Wessner to assume that there was another archetype, entirely inde- pendent of the Carnotensis and the Moguntinus, from which V descended, for V alone among the meliores does not have the Phormio lacuna.^ In the second place it forced him to assume that the Phormio lacuna in Aurispa's apograph was immediately filled in, because with two exceptions all the MSS of this family lack the lacuna. The two exceptions are Oa, which as I have said above (p. 299) show a strong affinity for RC, copies of the Moguntinus, and which Wessner himself later groups with the Moguntinus family. If we remove these two MSS therefore there is no need to assume that the Carnotensis and Aurispa's apograph ever had the Phormio lacuna; in that case we can make V a copy, direct or indirect, of Aurispa's apograph and so eliminate the third archetype. What was then the "defectus" in the Carnotensis? It is per- haps significant that while Valla's informant reports that the Hecyra is "non Integra" he speaks only of a "defectus" in the Phormio, as if there were a difference in kind, though it could refer, of course, to a small lacuna. Sabbadini suggested' that it might refer to the gradual thinning out of the scholia in the latter half of the Phormio; Wessner at first accepted this explanation^ only to reject it in his 1 There is no proof whatever that the Moguntinus was a copy of the Carnotensis. The situation could have been met equally well by making the two MSS gemelli, assigning the Phormio lacuna to the archetype and putting the Hecyra lacuna in the Carnotensis. ^ V does not have the Hecyra lacuna either but this fact has no bearing on the argument. As Wessner has clearly shown, V belongs to the meliores only as far as Ad. ii. 3. After this point it was copied for the most part from a MS belonging to the LcN family which did not have the lacuna. ' Studi ital., II, 53. * Rheinisches Museum, LII, 83. DoNATus' "Commentary on Terence" 303 edition. 1 It is possible, however, that the "defectus" may refer to the double recension in the Phormio; here we find the following notes in V: "error in exemplar!. A," "verte usque ad tale signum," "haec praecedunt ea quae praeeesserant ante in taU signo " "duppli- cata sunt et tamen aliter dici videntur quam antea dicta sunt." It is quite clear that V is following here one of the meliores though in general he follows a MS of the poorer class after Ad. ii. 3 (see footnote, p. 302) — only the meliores have the double recension. These marginal notes may be the work of a fifteenth-century scholar (A = Aurispa?) but such notes are common enough in ninth- century MSS. I am inclined to believe that these or similar marginal notes were in the Carnotensis and furnish the explanation of the "defectus." They are more likely to have attracted attention than the dropping out of a single folio. The assumption that there were more than two archetypes is therefore not only unnecessary but it is also improbable. The correspondence of the Humanists shows how keen they were in the search for new MSS, how close their contact was with one another, how rapidly the news spread when a new MS was brought to light, and how eager they were to correct their copies. This was espe- cially true in the case of Donatus, as Aurispa's correspondence shows. Is it Hkely that another old MS of Donatus — the archetype of V was an Insular codex — one that filled a gap existing in all the other MSS, could be brought into Italy and used, without any mention being made of it, brought not merely to Italy but to Florence, the center of Donatus' study, and more than that, brought into the same script&rium that produced a half-dozen other MSS of Donatus ?* Wessner^ advanced much the same kind of argument against Sab- badini's assumption that there were ten independent copies of Donatus in Italy.* ■ Praef . p. xxvii. ''According to Wessner, Praef. p. xxx, VOacdLN were written in the famous scriptorium, of Vespasiano. ' Rheinisches Museum, LII, 73. * Ambrosianus L 53 sup., a miscellaneous MS of the fifteenth century, discovered by Sabbadini (Studi ital., XI, 185) contains at the end the Vita of Donatus, the treatise on tragedy and comedy, and the Preface to the Andria. It belonged to 304 Charles H. Beeson In considering the improbability that there were three arche- types in Italy it is well to bear in mind the circumstances in which the Latin texts that are preserved only in MSS of the fourteenth and fifteenth centuries have been transmitted to us. Almost without exception they owe their preservation to the fact that during this period a single MS was discovered and copied. I give the list: Asconius, all our MSS from three copies of the St. Gall codex, dis- covered by Poggio and his friends; Catullus, from the lost Verona; Cicero, Brutus, from the lost codex of Lodi which is also the source of the integri of the De orator e and the Orator; Pro Roscio Amerino and Pro Murena from Poggio's lost Cluniaeensis; also Pro Roscio comoedo, both Pro Rabirio's from a lost codex of Poggio; Nepotianus, one MS; Panegyrid, three copies of a lost Moguntinus, the Italian MSS from Aurispa's copy; Petronius, Cena, one MS; Priapeia, all from a lost ninth-century codex; Pelagonius, one MS; Rutilius Namatianus, two sixteenth-century copies of a lost Bobbio codex; Silius Italicus, Punica, all from Poggio's copy; Statins, Silvae, all from Poggio's copy of the St. Gall MS; Tacitus, minor works, all from Hersfeldensis, of which one quaternion still survives; Tibullus, all from one MS which appeared in Italy in the fourteenth century. There are two families of MSS of Cicero's Pro Quindio and appar- ently two for the letters to Atticus, Brutus, and Quintus, but the date of the archetype is uncertain. The existence of two Cicero MSS in Italy would not cause surprise, but the only library in Italy where one might expect to find an Insular codex of Donatus was Bobbio whose treasures, curiously enough, escaped the notice of the Pizzolpasso, who apparently brought it from France. There is no indication that the MS was ever used by the Humanists; we may doubt whether Pizzolpasso recognized the fragment as belonging to the Commentum. If he obtained it while he was bishop of Aquis in Gascony (1422-23) as Sabbadini asserts (Storia e critica di testi latini, p. 121), ten years before the Moguntinus was discovered, he had no means of identifying the text for there is no heading, except a late one, in the MS. The text of the Ambrosianus is independent of the Italian tradition, being closer to A than any other MS. Some parts of the text are written twice, due, according to Sabbadini, to the difficult script of the archetype. We may have here again the old story — the difficult Insular script. In one version the scribe wrote scipiore, in the other scipione; one has cum, the other em (from the Insular abbreviation c =cum) ; one has lelio, the other bello, corrected to lelio; we also find appollinantes for appollinares and mrmicum for mimicum (i.e. mymicum). There are many cases of single consonants for double and vice versa. There is not evidence enough, however, to warrant a definite conclusion. DoNATus' "Commentary on Terence" 305 Humanists until the close of the fifteenth century. Fortune was, therefore, lavish when she bestowed two old MSS of Donatus on the Humanists; to have given them a third would have been an act of unparalleled generosity. To sum up: The archetype of the Commentum was written in uncials with the lemmata in rustic capitals; at least two MSS inter- vened between the original and the immediate archetype of our MSS (X); one of these was probably in Caroline minuscule, the other in a compact Insular script with many abbreviations; the script of X is uncertain. Two copies were made of X, a and ^; both of them were Insular. The Camotensis and the Moguntinus are copies of /3; both again Insular. V is a copy, direct or indirect, of Aurispa's apograph of the Camotensis. T descended independently from the Moguntinus, with at least one copy intervening, and is not derived from the same fifteenth-century MS as C (F). (and a) are probably descendants of the Moguntinus. University op Chicago