seeking for the basis of this order, the arrangement found in codices may be disregarded. Turning to the Bembinus and the y codices, to discover which of these two methods of arrangement served as the basis of the new order, it is necessary only to see how the transfers of items were made. The Andria is first in all orders, and since its numeral was appropriate to this position, it retained its place in the new order. The same was true of the Hecyra in the fifth place. The Adelphoe was advanced to the second position in the alphabetical order, and to make it appear to have been the second also in order of composition, the numeral from the Eunuchus with the adjacent items was inserted in its Didascalia. Similarly the Eunuchus was made the third play, and its Didascalia was filled out with items from the Hautontimorumenos. In both the Adelphoe and the Eunuchus, the Didascaliae received items from the plays which occupied corresponding positions in the manuscript serving as the basis of the new order. Thus far this manuscript had the order of plays found both in the Bembinus and in the y family. The fourth play determines which of these orders was used. If a manuscript like the Bembinus had been used by the author of the transfers, he would have made no changes in the Phormio, which in the Bembinus already occupies the fourth place. When we find, therefore, that the Phormio in the new order received items from the Adelphoe, we can hardly doubt that the latter play was fourth in the original order. But this is the position it occupies in the y codices. The Hautontimorumenos, as the sixth play in the new arrangement, had its Didascalia completed by the insertion of items from the Phormio, the sixth play in the original. Since the Adelphoe in the fourth place and the Phormio in the sixth are found only in y codices, it seems clear that a member of this family was the basis upon which the new order was made. The author of the new order must have regarded the arrangement of plays in his manuscript as chronological, as indeed it is intended to be in all but two plays. His failure to note the exception in the Phormio and Adelphoe was most natural, especially if the Didascalia of the latter play had no numeral,

leaving the Didascalia of the Phormio as his only means of discovering the truth. Since the Phormio was sixth in his manuscript, he might easily at first have overlooked its numeral. After finding it, or even if he observed it at first, he may have doubted its authenticity.

All these peculiar conditions attaching to the groups of items give evidence completely in harmony with that derived from a comparison of the separate items in the different sources. I maintain, therefore, that the manuscript followed by Donatus had suffered a series of interchanges in the Didascaliae of four plays. To explain these the theory of chance is utterly untenable. A conscious hand, working in accord with a definite plan, must be recognized.

The conclusion stated above and the facts brought out in the discussion have an important bearing on a number of interesting questions. First, from this source we get new light on the work of Donatus. It is necessary to acquit the commentator of all responsibility for the changes in the Didascaliae, a charge often brought against him because he happens to preserve them. With a single exception these changes are due to one person, in all probability the person who first reduced the plays to an alphabetical order. After this order was established there was no reason for such extensive and radical emendations. That Donatus was not the author of the new order and the consequent changes in the Didascaliae is shown by the fact that he sometimes questioned the information given by his manuscript. He did not indeed reject the numeral SECVNDA in the Didascalia of the Adelphoe, but it is clear that he doubted its accuracy. It has been suggested that he found no numeral in the Didascalia of this play, or that he had another source of information, but both suggestions are needless. The plan followed in altering the Didascaliae to suit the new order of plays required that every play should have a numeral. Without any other source of information Donatus had the best of reasons for suspecting the numeral of the Adelphoe, and would have been very stupid if he had not seen that this play, brought out at the second attempted performance of the

Hecyra, could not have been the second in order of production, if the Hecyra was the fifth. Unable to change the numeral of the Adelphoe alone, and unequal to the task of correcting the chronology of all the plays, yet he clearly expressed his suspicion in the words Hanc dicunt ex Terentianis secundo loco actam. The same scruple appears in every reference to the Greek writer of the Hecyra. In the praefatio to this play he gives information in harmony with a number of passages in the commentary in which Apollodorus is mentioned as the author of the Greek original. Why, then, does he speak with hesitation both in the praefatio and in the Auctarium to the life? In all probability he found the play assigned to Menander in the Didascalia, as it is in the Bembinus. Forced to choose between the Didascalia and the sources of the commentary, he rejected the former, but with a feeling of doubt which appears every time he mentions this subject. It seems certain, therefore, that the new order of plays and the changes in the Didascaliae were prior to Donatus. The commentator was so unfortunate as to use a manuscript of this sort, and through his adherence to it was led into gross blunders on many points. This is particularly true of the tibiae, which, as given by his manuscript, he had no reason to doubt. As a consequence, his attempt to characterize the different kinds of tibiae according to the plays in which they were used was marked by inevitable errors. For these he should not be too severely censured, for the material at his disposal precluded any accurate results. In general, judging from the paraphrases alone, Donatus was not a man of much learning, but he used faithfully the material at his command, and deviated from his sources with misgivings, and only when no other course was possible. He surely does not deserve the oft-repeated charge that he made reckless and arbitrary changes because of preconceived ideas.

1 As Fischer, De Terentio . . . quaestiones selectae, p. 18 f., remarks, a further reflection of Donatus's hesitation on this point appears in the praefatio to the Phormio. The words manifestum est indicate Donatus's feeling of relief at finding a play assigned to Apollodorus both in the Didascalia and in citations of the Greek original by older commentators.

If the changes in the Didascaliae were made before Donatus's time, we can scarcely doubt his authorship of the paraphrases. However widely the information given by him varies from the other sources, it is so completely explained by the theory of a transfer of items as to forbid the thought that its origin was due to different persons, or to some one later than Donatus. For many years it has been the fashion to deny that the commentator wrote parts, or in some cases, any part, of certain works which have come down under his name. This view may be justified in the first of the two treatises, that usually entitled de Fabula, which preserves strong evidence of its origin, in part at least, from Euanthius. It is possible also, or even probable, that the commentary, conceded to be in its present form a compilation of two sets of excerpts from the original Donatus commentary, may have matter added by the excerptors, the compilers, and possibly others. But it is easy for one to go too far in refusing to credit Donatus with other works besides that entitled de Fabula, or in applying the theory of a compilation of two sets of excerpts. There is no adequate reason for doubting that Donatus wrote at least the essential and substantial portions of most of the works under his name. Even if it is difficult to distinguish the true work of Donatus from that of others in the commentary, this is not necessarily true of the other works. Both of these remarks apply especially to the life and the Auctarium. They stand in the same relation to the commentary on Terence as the life of Virgil did to the commentary on this author. Without any evidence against them, we must suppose that both are in substantially the form in which they were left by Donatus. The same I believe to be true of the praefationes. Rabbow, it is true, does try to show that those of the Eunuchus and Adelphoe had a different source from those of the other three plays, holding, apparently, that one set of excerpts was the source of the two praefationes, the other of the remaining three. This theory is based in part on Donatus's variations from the Mss in the paraphrases. But in these, as I have shown, the Phormio must be included with the Eunu

chus and Adelphoe, not with the Andria and Hecyra. If this is done, Rabbow's theory breaks down. Whatever the vicissitudes through which the praefationes have passed, there is no evidence of any serious changes. It is clear that none were made in the paraphrases. Even the many instances of language in the praefationes similar to that in the treatise de Comoedia, the life, or the Auctarium are not necessarily due to insertion from these sources by later hands. More probably Donatus wrote the language in question in both places. The best example of this is perhaps the references to the Greek author of the Hecyra explained above, but there is no reason to doubt that most of the other instances are to be accounted for in the same way. A similar parallel between the praefationes of the Eunuchus and Adelphoe and the Ars of Donatus is pointed out by Smutny.1

A second result of the conclusion reached in this paper is the removal of nearly all uncertainty about the text of the Didascaliae. The evidence of Donatus, if rightly viewed, so far from differing from the other sources, really confirms them. It seems certain that the latest common ancestor of all the sources made the Adelphoe the sixth, not the second of the plays in order of composition. In view of the hesitation with which Donatus refers to this point, it is strange that so many scholars have accepted his statement and have wasted their ingenuity in attempting to defend it. Incidentally it has also been made clear, I think, that the archetype ascribed the Greek original of the Hecyra to Menander, not to Apollodorus. This, of course, does not prove the Mss right and Donatus wrong. More probably Donatus gives the correct information, though he rejected the Didascalia and gave the preference to the sources of his commentary. The naming of Menander in the archetype may have been due to an early copyist, who, under the influence of the preceding plays, mechanically wrote the name of the same Greek author. Lastly, the variations in the tibiae, which have always baffled explanation, are now fully accounted for. It is true that scarcely any one since the publication of 1 Diss. Phil. Vindob. VI, p. 104 f.