usually, however, a blank space has been left, often with a Latin version written over it in red or after it in black. The insertion of a Latin version of the Greek passages and the division into twelve books are features which the Arundel codex has in common with the Third Medicean, and these indications of kinship with that codex are confirmed by the evidence of the excerpts, which assigns to B a place somewhat similar to that of R1 in the Medicean group of the first class. While exhibiting unmistakable traces of its derivation from the archetype of Class I, it has suffered numerous 'corrections' from other sources. The following are some of the characteristic readings of Class I which it retains: 4, 35 actores; 18, 4 regione; 34, 8 ouio; 39, 29 carmensis; 47, 3 ultus est; 48, 3 ignominia; 48, 18 urbis; 49, 17 decimum; 52, 25 flamonium; 53, 35 tabulas; 57, 29 partimque; 89, 18 troiam; 114, 30 patria; 163, 28 defuit et ueterum; 183, 27 ueste; 227, 12 antiqua erat; 238, 11 aduerso rumore; 240, 20 uerum. Where the reading of the archetype was manifestly wrong or meaningless it has been corrected; thus we find 14, 16 aduentu suo (for sui); 21, 30 ad uinum (for ui); etc. In a few cases these 'corrections' may have been independent, but they are usually derived from some manuscript of the second class, as 4, 17 triumphalem uirum (for triumphalem); 21, 6 recentiora (for retiora); 22, 6 ecce — gallias, etc. In spite of these corrections and corruptions B maintains a distinct preponderance of agreement with A (47:38), and still more so with M (54:28). Where A and M3 part company B sides with A against M3 in a few cases: Much more numerous and more significant are the places in which B agrees with M3 against A. I have counted in my excerpts fifty-four such places, of which the following may serve as examples: 11, 23 aliis . . . additis M3 Bε 26, 32 moranti cuspide M3 B 33, 25 iulo M3 iulio B 33, 35 aureo M3 B 38, 14 minis gentis M B 39, 18 confiteri M3 B 40, 30 autem M3 B 42, 3 hac fraude M3 B ac fraude A 44, 31 de cleopatrae liberis M3 B 50, 28 prae se identidem M3 B 55, 23 uirilem togam M3 B 78, 9 ipse M3 B® ipse ei A 80, 21 in sinum eius signum rei p· quod M3 Bʻ rei p⚫ quam A 83, 11 rerumque missilium M3 B6 85, 16 quinque M3 Bo quinques A 93, 11 ignota M3 B¤ ignoti A 141, I omnem urbem M3 B 147, 13 magna ui M3 B 192, 4 male fracto M3 B 211, 16 erat M3 B laureo A alias ... addit A moranti se cuspide A ilio A eum A minos gentes A profiteri A de cleopatra liberis A urbem omnem A naui A rerumque missilia A male facto A tulerat A 217, 27 suae M3 B uenetae A 224, 7 incertum sponte an M3 B® incanum A 253, 9 canum M3B® in eius signum an A In orthography B is distinctly superior to the average mediaeval or renaissance manuscript. Not that there is any lack of the ordinary mistakes of copying; but its real errors of orthography are for the most part confined to those that arise from the confusion of c and tor i and y, or from the misplacement of the aspirate. Thus we find both acies and aties, otium and ocium; contio, concio, and conctio; sectius, suspictiones; dimictebat, actritus; prouintia (found in A also); circha, cathenati, cohercuerat, introhire; ymagine, tyberius, silla;1 and there are the usual distortions of proper names. But there are no barbarisms 1 See above, p. 7, footnote. like michi and nichil; and, on the other hand, we meet here and there with forms which we must recognize as survivals from the archetype.1 I have noted a few examples. The accusative plural of i-stems usually ends in -es, but often in -is, as 31, 19 pluris; 38, 14 gentis (in which the ending has been protected by the corrupt form of the adjective, minis); 39, 35 octobris; 205, 5 recusantis; 205, 6 flagitantis. In nouns and adjectives with stems in -io-, -ia-, a single --, for -ii-, is found in a few places, as 12, 1 stipendi; 211, 31 lecticaris; likewise in verb forms from perfect stems in -iui-, as 4, 13 redit; 14, 28 petissem; 42, 12 subisse. Compounds of iacio usually have -ii-, but -i- is not infrequent, as 26, 36 traiceret; 33, 13 deicerent; 33, 20 disicerent; 40, 11 obiciatur; 42, 18 adicit, etc. Prepositional prefixes are ordinarily assimilated, but the unmodified form often occurs, as 15, 28 adfirmaret; 18, 22 adlegit; 26, 12 obprobrium; 207, 9 adsimilis; 211, 16 adfuturi; 214, 22 adnexo, etc. There are a few instances of -qufor -cu-, as 4, 38 sequti; 207, 22 alloquturi. Most striking of all is the spelling of the temporal conjunction cum, which is regularly written quom, while the preposition is regularly written cum. The unfamiliar distinction has not been observed with entire consistency by the scribes, who have sometimes lapsed into cum for the conjunction and sometimes overdone the matter by writing the preposition quom. Thus in the Caligula and Claudius together, out of 79 occurrences of the conjunction (leaving out of the count a few cases where the text is corrupt), it is written quom 62 times, and cum 17 times; while the preposition, which occurs 43 times, appears as cum in all cases except four: 139, 21 quom exigua manu; 139, 37 quom equitatu; 148, 33 simul quom fratre; 153, 23 quom cetera turba. In this respect our manuscript bears some resemblance to the palimpsest of Fronto. The other forms of which examples are given in the foregoing paragraph are undoubtedly survivals of sound tradition. Is quom likewise 1 See Roth, Praef. xxxvi. Roth is no doubt right in accepting the more archaic forms found in A as authentic. Whether Suetonius used the archaic and the contemporary conventional forms of the same word indifferently is another question. The mixed spelling of A may well be due to the gradual encroachment of the conventional orthography of the scribes on the peculiar spelling of Suetonius (cf. Preud'homme, Troisième étude, pp. 20 f.), and B may represent a more advanced stage of the same process. traditional, or does it merely represent the unwarranted opinion of some mediaeval scholar? The testimony of our manuscript rather suggests than answers this question. Of this much, I think, we may rest assured: the form was not first introduced into B. It is plain that the writer of that manuscript had no well defined views on orthography; he copied what he found in his original. Back of this we cannot go. There is nothing intrinsically improbable in the supposition that Suetonius, like some of his contemporaries,1 distinguished the conjunction from the preposition by writing it quom, and that this spelling has disappeared in nearly all of our manuscripts; but the unsupported testimony of B cannot be said to carry us very far on the way to such a conclusion. The whole question of Suetonian orthography, however, deserves a more thorough investigation than it has yet received, and to that end I have set down the evidence of B for whatever it may be worth. The other manuscript which I examined, B', is a composite production both in its handwriting and in its text. It is written in six unequal instalments, apparently by as many different scribes; and the text is derived from two very different sources. As far as 97, 33, where the work of the second scribe ends, and perhaps a little farther, the text belongs to the Medicean group of the first class; while from this point to 230, 7 grauissima increpuit (Vesp. 8), where it breaks off incomplete, it belongs to the Florentine group of the second class. As the text stops at the end of a quire, we may suppose the copy was once complete and has suffered the loss of one quire. There has also been some disarrangement of the leaves: the order of foll. 28-34 should be 29, 28, 31, 30, 32, 34, 33. In the part of the text belonging to the first class the Greek passages are given, sometimes with a Latin version superscribed, as in V and M3; in the other part the spaces left for the Greek remain unfilled. The first part, like B and R1, is a much corrected text, which still retains a decided preponderance of agreement with the representative manuscripts of the first class, particularly with M3 (BA 44:24; B V 48:20; B' M3 56:12), as well as more specific evidence of its origin in numerous survivals of characteristic readings of that class. 1 Cf. Terentius Scaurus, VII, p. 28 K. The following may serve as examples: 4, 35 actores; 22, 27 aurum effutuisti; 31, 31 exta quondam; 32, 26 libris om.; 47, 3 ultus est; 48, 18 urbis; 49, 17 decimum; 52, 4 ut qui; 52, 25 flamonium; 53, 3 exempli; 58, 11 puerorum; 74, 28 tiburi; 85, 11 sextam; 85, 28 tribus; 86, 10 atque; 89, 18 troiam. Where A and M3 disagree, B' is found in pretty constant agreement with the latter. There are 42 such cases in my excerpts from this part of the text, in two of which B' differs from both of the older manuscripts. It sides with A in four places: 4, 30 desidere; 10, 6 licerentur; 10, 32 uatinia; 19, 22 exulabant. In the remaining 36 places it sides with M3 against A. I forbear to give examples, which would be mainly a repetition of the list given above1 for B. In 4, 5 regiae, Bì approaches V, but I have noticed no other indication of special resemblance to that manuscript. It clearly belongs, in this part of its text, to the Medicean group. In the other part (from 97, 33, or thereabouts, to the end) B is closely related to Vaticanus 1860 (Vo). In 64 excerpts in which both manuscripts are represented there are only seven cases of disagreement, and these are evidently due in every instance to unconscious error, with the possible exception of 194, 19 praefectorum (for peccatorum). On the other hand, B' faithfully follows Vo in all its peculiar readings, good and bad, such as 111, 27 cum coniugibus; 127, 18 quercina; 188, 38 perusurum; 192, 4 labefacto; 208, 19 addidit; 214, 19 claudusque ; 217, 27 ueterem, etc.; it reproduces the grosser interpolations of Vo, of which I gave some examples in my account of that manuscript;2 and finally it has the same transposition in the Galba, with the passages in the same order. From all this we may infer that B', in this part of its text, was copied directly either from V° or, more probably, from the common archetype of V° and V1. A word, in conclusion, about the value of the XV. century manuscripts, which have engaged a large share of our attention in these studies, and to which—for that reason, I suppose one or two of my 1 Page 9. 2 Vol. XII, p. 50. 3 See vol. XII, p. 48. The part of the Vespasian in which the great lacuna of |