same.

but in his manuscript these had been changed to suit the position occupied by each play in an alphabetical sequence. If the commentator was grossly deceived about these plays, we have a right to demand evidence that his information about the Phormio had a sounder basis. No such evidence. exists, for his apparent agreement with the Mss and the conventional chronology is due solely to the accident which makes this play, with the orthography Formio, the fourth in the alphabetical series. Whatever the numeral may originally have been, therefore, the result was bound to be the It is very doubtful, however, whether any numeral was included in the transfer of items from the Adelphoe to the Phormio. Since both the y and the & families of Mss have the word FACTA, but have no numeral, we may assume that the archetype had the same defect. Since the separation of these two recensions took place in early times, all authorities agreeing that it came in the period between the third and the fifth centuries, the loss of the numeral must have been very early. It may well have occurred before the alphabetical order of plays was made. If this was true, the author of the transfer of items from one play to another found none in the Adelphoe, and could have felt no prickings of conscience in arbitrarily adding one for the Phormio to suit his general plan.

It remains to consider the disposition of the items ousted from the Didascalia of the Phormio. Since the manuscript sources have a numeral for this play, it is obvious that in the transfer to another play a change in the numeral was necessary. No evidence, therefore, can be gained from this source. Nor can positive evidence be found in the tibiae, for it is clear that the praefationes to neither the Andria nor the Hecyra contain the tibiae dislodged from the Phormio. There remains but one play, the Hautontimorumenos, which had lost to the Eunuchus part of its Didascalia. The theory suggests itself that the deficiency was made good by inserting the corresponding items from the Phormio, with an arbitrary emendation of the numeral IIII to VI to suit the place of the Hautontimorumenos in the new order. If this was

done, the circle of transfers is complete. In the lack of a commentary on the Hautontimorumenos, direct evidence in confirmation of this suggestion is impossible, but indirect evidence presently to be cited indicates that it is correct.

The results of the foregoing discussion may briefly be summarized. The sources have been compared with reference to their agreement or disagreement on the modulator, the tibiae, the use of the word tota, and the numerals of the plays. In three praefationes, therefore, there are twelve points on which to compare Donatus with the Mss. Out of these twelve points Donatus gives no information on three, he differs from the Mss on five, and he apparently agrees with them on four. By the theory here proposed, four of the five points of difference are fully explained, and in the exception, the tibiae of the Eunuchus paraphrase, the difference is partly explained, and is otherwise easily accounted for. Again, by the theory of a transfer of items we can understand why Donatus omitted two of the three points of information which he fails to give. Here, too, the exception, the failure to use tota in the Eunuchus paraphrase, is in no way opposed to the theory. Lastly, three of the four points of apparent agreement between Donatus and the other sources on the same plays contain evidence neither for nor against the theory, since the information is identical in different plays. The fourth point is the numeral of the Phormio, which, as was shown above, is not an instance of real agreement, but is either an arbitrary emendation or, more likely, an arbitrary addition. In seven, therefore, of the twelve. points of comparison, the theory is fully supported by Donatus's information or by his failure to give any information, and it is not opposed by the other five. In other words, if we should transfer Donatus's paraphrases on the four points in question back to the plays to which they seem originally to have belonged, the commentator would differ from the Mss only in naming one set of pipes for the Hautontimorumenos, and in making the Adelphoe the fourth of the plays. Both exceptions have been fully accounted for. In every other point he would agree with the other

sources in the information he gives, and the reason for his failure to give certain information would be evident. The facts, therefore, completely support the theory advanced in this paper. For this reason I venture to believe that the parallel between Donatus and the manuscript sources on different plays is no mere coincidence and without significance. It rather points clearly to an actual transfer of portions of the Didascaliae of four plays in the manner I have indicated.

This conclusion finds support in a number of considerations general in character. First, the items which were transferred, except in one point, are always found together in the Didascaliae. The exception concerns the place of the Greek writer, who is usually named in the Mss after the tibiae and before the numerals. But the Bembinus is not consistent in this respect. On the Eunuchus and Phormio it seems to agree with the later codices, though the mutilated condition of the Phormio Didascalia may conceal the original order of items. But in the Hautontimorumenos and Adelphoe the Bembinus names the Greek writer immediately after the title, and this holds true also of the Ambrosianus of Plautus in the Didascalia of the Stichus. The proper place of the Greek writer among the other items is a point upon which scholars disagree, but the Ambrosianus and, in part, the Bembinus support the earlier position, and because of their age they have great weight. Moreover, the natural place for this item seems to be next after the title. Possibly the transfer to the later position was due to the common misconception or misunderstanding of the meaning of tota in the item on the tibiae. In the Hautontimorumenos, where tota could not be used, no such transfer had been made before the writing of the Bembinus. It is certainly possible, therefore, not to say probable, that the author of the changes in the Didascaliae, living as he did in the second or third century, found the Greek writers named after the titles in all the plays. If this was the case, the items transferred always stood together, and it is easy to see how a numeral when transferred to another play could have carried with it the adjacent items.

A second argument is found in the fact that it is on only three plays that Donatus differs widely from the Mss. With these the Hautontimorumenos would doubtless be included if we had the commentary on this play. Save on the Greek writer of the original, Donatus is in almost complete accord with the other sources on the Hecyra, and this exception is easily explained. In the case of the Andria no direct comparison is possible, but it is certain that there would be no disagreement on the number of the play. For the tibiae of the Andria the praefatio is our only source. It is an interesting fact that Donatus's disagreement with the other sources on the pipes of all the plays except the Hecyra has not led any one seriously to challenge his information about the Andria. If he gives its tibiae correctly, we may assume that on the Andria as well as the Hecyra he is trustworthy, just as he is untrustworthy on the other three (or four) plays. What is the reason for the difference? This fact has been observed by others, but no one has ever explained it. By the theory advanced in this paper it is easily accounted for. The numerals of the Andria and Hecyra mark the positions of these two plays in both the chronological and the alphabetical orders, hence both retained their original places when the new order of plays was made. For this reason their Didascaliae were not disturbed. With the remaining four plays the conditions were wholly different. In order to force agreement between the chronology and the alphabetical order, a transfer of numerals was made. In the Phormio this act was needless, for its numeral was already appropriate to both orders. But, as will presently appear, the position of this play in the manuscript followed by the author of the new order caused its numeral to be overlooked, so that it was included with the other three. In the transfer of the numerals, the author of the new order saw fit to include the adjacent items. For this act no satisfactory reason appears, but this is no valid objection to the theory. No one can deny that numerals have been changed on arbitrary grounds, an act which must be ascribed to the gross stupidity or the moral obliquity 1 See p. 154.

of the agent. Such a person would not hesitate to include other items with the numerals if this was suggested by convenience. It would increase his guilt only in degree, not in kind. He would not hesitate, therefore, if for any reason, such as the division into lines in the Didascaliae he was using, it seemed easier to transfer the numerals with the preceding items rather than alone.

Lastly, the transfer of items was not executed in haphazard fashion, but in conformity with a definite, though very simple plan. The freedom of the Andria and the Hecyra from such changes was due, as was explained above, to the fact that in the new or alphabetical arrangement these two plays retained the positions they had occupied in the original order. But it is easy to go further and to determine the positions originally occupied by the other four plays. The numerals alone seem to point to a manuscript like the Bembinus, in which the sequence was designed to be purely chronological. But since the numeral of the Phormio paraphrase is open to the gravest suspicion of having had no manuscript basis, the tibiae are a surer source of evidence, and these identify the original order of plays as that found in codices of the y family. To assist in making this point clear, the several ways in which the plays are arranged are given below:

ORDERS OF THE PLAYS OF TERENCE.

ence.

The fourth of the methods of arranging the plays, that of the d family, is given only for completeness and later referIn spite of the different positions it gives two plays, it betrays the same origin as the Donatus order. The latter, as is shown by the exemption of the Hecyra paraphrase from the scheme of transfers, is the original alphabetical order. In