ing off the last two letters, while it is given in full at Apollonia. In 11, 21, 13 Ancyra has ènì óvóparos, but at Apollonia the dative óvóμarı is given. Editors prefer the Ancyran reading for the Latin sub nomine. Elsewhere in the document the dative appears in Greek to correspond to the Latin ablative, where no preposition is used in either. We may then conclude that the significant omission of vaòv from both texts proves the Ancyran origin of the copy of Apollonia, for obviously the Ancyran inscription would not be derived from the Apollonian. The differences in readings may indicate that the Apollonian copy was not made from the stone at Ancyra, but from a manuscript, from which also the Ancyran copy was derived.

No Latin text has been found at Apollonia. No trace of a Greek text has been found at Antioch among all the numerous Latin fragments, and from this fact Robinson 11 concludes that a Greek translation was not set up. From these facts and from the general considerations advanced above in this paper it seems quite safe to exclude Antioch as the immediate source of the known copies. It remains that the inscription at Antioch is derived from the stone or from a manuscript at Ancyra, or that both Latin inscriptions are derived independently from Rome.

If a decision on the point is possible, it must depend primarily upon a study of the readings of the two inscriptions, though comparisons of the arrangement of the texts and even of the styles of letterings may be helpful. One would theoretically expect to find the texts very similar, even though neither inscription were a copy of the other, for they both were assuredly made within a very few years of the time of Augustus's death, when their original at Rome must have been quite complete and quite legible. This is not a case of copying from an age-old manuscript whose text may scarcely be legible. And striking likenesses are found. For example, the spelling fuere is given in Anc. 1, 4, 27, and in Ant., yet this short form of the perfect is apparently used in no other instance of any verb in either text. Clausum appears in Anc. II, 13, 44 and in Ant., though Anc. has claussum just two lines above, where the text of Ant. is unfortunately lost. In Anc. III, 15, 21 and in Ant. the spelling is paullo. The abbreviation Tib. is given in Anc. II, 8, 9,

11 AJPh., p. 2.

and in Ant., though elsewhere in the texts (Ant. has only one other instance) the spelling is Ti. In Anc. III, 15, 15 and in Ant. the text reads consul XII; and a similar case, consul XIII, is found in chapter 22 in both documents, but nowhere else in a number of passages, where the consular number is regularly spelled out in full. There are other correspondences, but none, I think, more striking than these. There are no apparent and peculiar errors or omissions in both texts, such as might indicate that one document was copied from the other. These textual likenesses, striking as they may be, can not prove copying, and may well have come from a common source, or even from the original at Rome.

In a number of passages there are different readings. Anc. IV, 21, 22 gives ad aede for which Ant. has ad aedem. Anc. III, 16, 27 has ad memoriam for which Ant. has ad memoria, plainly in error. But no such differences as these can demonstrate the relationship of the texts, and therefore other somewhat similar instances are not here mentioned. Only differences apparently significant will now be pointed out. Anc. II, 9, 17 gives collegia, while the corresponding passage in Ant. gives conlegia. In the one other passage in which the word appears in both, both have conlegio (chapter 22). The reading of Anc. 9 may be a slip. That of Ant., conlegia, was obviously not copied from Anc., and it is surely unlikely that the stonecutter would use by mistake an older spelling if he had collegia before him. In both documents there is a tendency, which is not consistently kept, to separate a prepositional compound from its verb, a practice found but seldom in Latin inscriptions, for normally such words were, of course, written as one. Anc. IV, 20, 13 gives profligata, while Ant. gives pro. fligata.12 It does not seem likely that the stone-cutter at Antioch would have made this strange separation if his copy had not shown it. In the same chapter both show praeter.misso. The readings above seem to indicate some independence in the document at Antioch, but can not be in themselves conclusive.

In dating, the consuls are often given in the usual ablative phrase. Of a total of some fifteen instances in Anc. about six are

12 Robinson, AJPh., p. 45, says that the simple preposition (where not used in a compound) does not occur in Ant. separated by a punctuation mark from the word it governs. It is found, however, in Ant. 1, 1, 9, and there are other instances.

392

"Res Gestae Divi Augusti"

in the common abbreviated form cos.13 In II, 8, 9 by the use of cos. at the end of the line the stone-cutter avoided dividing the word consulibus. In no other instance appears any good reason for a choice of the abbreviation. The word appears in the ablative in Ant. some ten times where the reading is sure, or reasonably so; and of these, three are probably in the abbreviated form. Apparently no requirements of space play a part in these abbreviations; but they correspond to abbreviations in Anc., that is to say, no abbreviations appear in Ant. where the full form is given in Anc. In one instance Ant. v, 16, 10 gives the word in full (as does Anc.) and to do so has to divide it, putting consu at the end of one line and libus at the beginning of the next. The abbreviated form would have nicely ended the expression of the date and the line. This appears to be a rather careful following of the copy. More significant is the fact that in the same chapter Ant. v, 16, 17 gives consulibus in full, where Anc. III, 16, 29 has the abbreviation cos. Again in the next chapter Ant. v, 17, 24 has the full form where Anc. III, 17, 36 has the abbreviated. These two cases go far to prove that the inscription at Antioch was not made from a copy of the stone at Ancyra. Would a stone-cutter expand the old and very common short abbreviation cos. into the much longer consulibus? Abbreviation is the rule in Roman inscriptions.

Similar to this apparently arbitrary use of the abbreviated and full forms of consulibus is the mixed use of sestertium in full and the sign for it, HS. Anc. chapter 15 has the sign three times; chapter 16 has the word spelled out twice; chapter 17 has one of each; chapter 21 has the sign. Ant. has preserved five instances that are sure or probable. In not one is the sign used. Twice, in chapters 17 and 21, the word is spelled out where in Anc. the sign is used. The sign is so common in inscriptions and so convenient for use that it is scarcely conceivable that the stone-cutter would have used the full form of the word if he had the sign in his copy. From these several words and forms the evidence derived seems adequate for the conclusion that the inscription at Antioch was not made from a copy of the stone at Ancyra.

18 In a few instances the word is supplied by editors where the stone is broken. Consul in singular and plural appears quite often in the nominative, and is not then abbreviated.

Other matters in the cutting of these inscriptions call for notice and explanation. I have already pointed out that in both prepositional compounds are spaced at times. Tall letters, especially T and L occur, while I is also tall at times to denote long I. Both documents use accents over long vowels, even in the headings; but, as Ramsay and Robinson have said, the use is not consistently the same in both. I think, however, that a closer likeness might appear if both texts were not so chipped and broken as they are. The Ancyran text in particular is badly weathered, and, since the accents were not cut as deeply as the letters, at times words can be read well enough whose accents may have been obliterated. For example, in chapter 19 Mommsen's edition and Diehl's put an accent over the u in Iunonis, while Ant. shows one clearly over the o. An examination of the word in Mommsen's facsimile might lead one to conclude that the mark over the u there is accidental, and to imagine at least traces of an accent over o.

In both the heading is cut in the style Scriptura Monumentalis, but with influence of the Scriptura Actuaria, seen especially in the M, and in the A, particularly in Anc. Also in the heading both disregard the joints of the stones, cutting letters right across them. The text in both is in the Scriptura Actuaria, with the letters in Anc. considerably larger and on the whole somewhat more regular, especially since parts of Ant. toward the end seem to have been carelessly, perhaps hastily, cut. A in Anc. tends to decoration a bit more than in Ant. Perhaps the lower lobe of B tends to be a shade more narrow in Anc. E is narrow and rather careless in both. G is in the capital and the actuaria form, the latter apparently more often used in Ant. R has a curved tail in both, and the lobe is sometimes open and sometimes closed. At the beginnings of lines and especially of chapters there is a tendency to use a larger letter and to make it a bit fancy. In both texts appears an unusual D, in which at the top a stroke overlaps the upright and extends far to the left, while the latter is quite normal at the bottom.1 V appears with a long flourish of the left stroke.16 The letterings of these texts seem more like each other than like any facsimile in the well known plates published by Diehl,

14 AJPh., pp. 25-26.

15

15 Mommsen's facsimile, pagina v; Robinson's paper, plate II.

10 Mommsen, pagina IV, and v; Robinson's paper, plate II.

Bruns, and Cagnat. Surely there were the same and explicit directions for styles and forms of letters given in the documents which the stone-cutters used. It is even possible that the same stonecutters were employed at both places. In such case the slight differences noted in the cutting would be accounted for by the presumably more careless work at Antioch, and by the fact that the larger size used at Ancyra permitted, and perhaps called for, more careful cutting, and allowed more ornamentation. At both places certainly the cutting shows the work of men experienced in the art of Roman lettering.

Correspondences in the arrangement of the texts in the two copies are still more striking. At Ancyra, as is well known, the text is divided into two parts, three columns, with the heading running across the top of all of them, to the left of the entrance to the shrine, and three, to the right. The heading tells us that at Rome the Res Gestae was cut on two bronze pillars, and Suetonius 17 speaks of bronze tables set up before the Mausoleum of Augustus. Certainly the Ancyran inscription was made from a copy of the inscription as set up at Rome-the heading so states.18 It has therefore been assumed, and with reason, that the arrangement on the stones at Ancyra imitated in general that on the bronzes at Rome.19 Kornemann points out that in the columns to the left are eighteen chapters, that is just over half, of the thirtyfive, and that the section to the right begins with the first line of a chapter. He has noted, too, that to the left there are 135 lines of the text, and to the right, 136, omitting the summary, and that these lines are grouped in the columns to the left, 46: 46: 43, and

17 Augustus, 101, 4.

19 The special attention paid in the heading to the facts that the tablets were two at Rome and were bronze, while, we know, at Ancyra the text is on two separate wall spaces, but of stone, seems to show a provincial point of view. See also below, p. 395. The fact that the Ancyran and Antioch inscriptions are evidently derived from the bronzes at Rome might be an indication that the central government did not send out copies to be used in the provinces, for it would scarcely need to have copies made from the bronzes. If anyone, however, sent from a province to Rome for a copy, he would naturally order it to be made from the inscription on the tomb; that would probably be the only way to secure it.

10 F. W. Shipley, Res Gestae Divi Augusti (in Loeb series, 1924), p. 333; E. Kornemann, Mausoleum und Tatenbericht des Augustus, 1921, p. 15.