to the right, 54: 54: 28. He thinks this a very awkward arrangement of the columns and assumes a division on the tables at Rome of 45: 45: 45, and 46: 45: 45, respectively. He assumes also that the summary was omitted at Rome, evidently assumes that the lines were individually identical at Rome and Ancyra, and finally assumes with Klaeyle that the first column left was made equal to the first on the right in number of lines by having over it the heading, Res Gestae Divi Augusti. This last assumption is a poor one for two reasons; first, in Roman inscriptions a heading is regularly in much larger lettering, which would cause his first column to appear longer than the second; again, in Roman inscriptions headings regularly run over several columns, if the inscription is cut in columns, and two or three columns on bronze or stone form a favorite grouping,20 as is the arrangement at Ancyra. Further, Kornemann's grouping for the original at Rome would throw quite into confusion a certain symmetry to be seen in the Ancyran text: columns 1 and 2 on the left start each with the first line of a chapter, column 3 does not; columns 4 and 5 (they are the 1st and 2nd on the right) start each with the first line of a chapter, column 6 does not. Also, column 3, the last on the left, has 43 lines, and column 6, the last on the right, has exactly that number, if we include the summary.

That summary has been rather unanimously 21 assigned to some provincial Greek at Ancyra, since Mommsen so decided, because it was poorly written, reckoned money by denarii, and referred to provincial cities. Of course it was not written by Augustus, as it is in the third person, but aside from that no compelling reason has been advanced for its provincial origin. Reckoning by denarii is to be found in the main document, though that by sesterces prevails. As for the reference to provincial cities, that is one short expression with no city named, while most of the summary concerns itself with a mention of buildings at Rome and expenditures for entertainments and for gifts to Roman senators, friends, the plebs, and the army-topics not obviously of special interest to

20 Bruns, Simulacra, pl. x, 12; XXI, 23; Diehl, Inscriptiones Latinae, 20 and 28.

21 Sandys, Latin Epigraphy, p. 275; W. Fairley, The Deeds of Augustus, p. 9; Shipley, op. cit., p. 403.

396

"Res Gestae Divi Augusti"

provincials. Hardy 22 calls attention to the fact that it would have been rather difficult for a person away in a province to get at some of the items of information given in the summary. Assuredly the summary may have originated at Rome and may have appeared on the bronze tables.

Now column 5 at Ancyra, as column 2, ends with the first line of a chapter, causing the following column to begin somewhat awkwardly with the second line of the chapter. The apparent purpose of the arrangement in each case is to make the column so ending agree exactly in the number of lines with the preceding column and to make the last column to the left and the last to the right equal to each other. Obviously the whole arrangement at Ancyra is the result of a very careful and skilful study of the text to be cut and is not at all awkward as Kornemann would have it.

We may now come to a study of the arrangement of the text at Antioch. Strangely enough, though by Robinson's study there. seems to have been a division into nine columns, column 8 ends exactly as does Anc. column 5, and so, of course, column 9 agrees exactly to the word in the beginning and almost exactly in the number of lines with Anc. column 6. Robinson 23 is troubled by this, since he has the idea that Ant. was not copied from Anc., but he has no explanation to offer. He thinks it an awkward division of the chapter, and it is very awkward if his arrangement of columns in plate VII B is approximately correct, for column 9 is separated by the width of an archway from column 8. It is much less awkward at Ancyra; and, if the Ancyran arrangement is that of the original bronze tables, it would have been even less awkward at Rome, as the two columns would have been on the same table. In facsimiles of bronze tables still existing, to which reference has been made, rather untidy runnings-over from one column to another are not at all uncommon.

Robinson has not seen the significance of another correspondence. Anc. column 3 ends with the end of chapter 18, just as Ant. column 5. This of course makes the beginning of Anc. column 4 the selfsame as the beginning of Ant. column 6. It is an important agreement, for Anc. column 4 is the first column of the section

22 E. G. Hardy, The Monumentum Ancyranum, pp. 163-164. 23 AJPh., p. 49.

of the document to the right. If Robinson's idea of the placing of the columns at Antioch is correct, columns 5 and 6 there were in a prominent position. 24 Naturally column 1 in both starts the texts. It is then clear that important divisions at Ancyra find corresponding divisions at Antioch.

There are other likenesses found in chapters beginning with lines of the same length, with exactly the same words and even the same letters, where words are divided at the ends of lines. Chapters 11, 17, 18, 22, 32, 34, and 35 agree in their first lines, Robinson's textual arrangement being accepted. Chapters 24 and 25 all but coincide in their first lines. Gaps in the text prevent a decision about some other chapters. I have an idea that the first line of Ant. column 6, chapter 19 may have exactly agreed with Anc. column 4, chapter 19, though Robinson's text does not give it so. Other correspondences were probably prevented by the different limitations of space at Antioch, as, for example, the heading is over only two columns instead of three as at Ancyra, and the text of these columns is considerably shorter to the line.

These corresponding arrangements in columns and lines are, of course, not by chance.25 An absolutely certain explanation is not possible. They do not come into the document at Antioch from a direct copying of the stone at Ancyra, as the textual study showed. Do they come as a result of the directions in separate manuscripts sent to each city from Rome? That is possible, and in that case we must have at Ancyra a very close copy of the bronzes at Rome, and at Antioch a copy made as nearly like the Roman as the spaces available would permit. But for general reasons given in the first part of this paper it seems most probable that one copy was sent out from Rome to Ancyra. If this is so, both the significant differences in the two texts and the remarkable likenesses in the arrangements of the documents on the stones may properly be explained. The text from Rome, or a workman's copy of it, was used by the stone-cutter at Ancyra. The same text from Rome, or a copy of it, was used at Antioch. And in the copy made for Antioch a description of the set-up of the text on the Ancyran

24 AJPh., pp. 23 and 44.

25 There may be other lines the same in both, but I have not looked for them throughout. It seems that not much effort was made to get lines exactly the same after the first lines, but see line 2 of chapter 32.

shrine was made. Or there was put in the copy for Antioch the directions for setting up the text which had been sent from Rome. Scholars seem to be agreed, as has been said, that there is some similarity between the arrangement at Ancyra and the original at Rome, which means that directions for the arrangement were sent with the manuscript from Rome, or that the manuscript itself was copied exactly line for line from the bronzes before Augustus's tomb. I think the similarities in arrangement of the two existing copies due to the manuscript that came from Rome. It does not seem probable that at Antioch, where they did not cut the inscription on the walls of a temple, and where they could not keep to the simple division into two sections of three columns each which is seen at Ancyra, the people in charge would have cared in any way to imitate the Ancyran disposition. They might feel an urge to approximate the arrangement of the original at Rome as closely as they could.

It is now possible to sugget a simple family-tree for our existing texts. The autograph of the Res Gestae is lost. The archetype was the copy on the bronzes at Rome. Of it a copy was made and sent to Ancyra, where a copy was made for the stone-cutters, and its copy is on the stones today. Another copy of the manuscript sent to Ancyra was prepared for use at Antioch, and, thanks to Ramsay and Robinson, the copy of it on stone is now known.

The copies on stone are, then, possibly three steps removed from the bronzes at Rome. But by a comparison of their readings we may arrive at a text only one remove from the inscription before Augustus's tomb, though we must remember that the Ancyran stone is not complete and that the text from Antioch is in fragments.

Statements in Cicero, Quintilian, and others make it clear that Roman authors were afraid of errors even in their first editions.26 Readings in our copies of the Res Gestae fully bear them out. Reference has been made to irregularities in the use of accents, the spacing of words compounded with prepositions, the long I, and the fairly frequent appearance of tall letters. At times paragraphing signs appear within chapters, or spaces are left to mark a division. The same elements doubtless appeared on the original

26 Cicero, Ad Q. F., ш, 5, 6; Quintilian's preface. Other interesting references are given by H. W. Johnston, Latin Manuscripts, pp. 32-33.

at Rome, and perhaps more consistently. In both copies the division into chapters is the same, and is made by the regular Roman method of slightly extending the first line of a chapter into the left margin of the column. We may be confident that we have precisely the original chapter-units.

Neither copy seems to have made any errors of omissions of words. There are differences in spellings and a few in word-order. Some of these have been given above. Others, though often slight, still help to restore the text of the original, and some have a bearing on theories that have been advanced on the basis of supposed variations of spelling in the inscription at Rome. Some of the noteworthy agreements will at least show readings of the Ancyran manuscript, supposedly a direct copy of the original.

Fuere (4, 27) in both, and only here with this short ending, is of the Ancyran manuscript.27 A numeral in letters (4, 27) in both, not spelled out as usual, is of Anc. MS. Similarly (15, 15) consul XII appears in both, and consul XIII (22, 38). A few similar instances can be found. Colonis, so spelled in the ablative plural for colonies, appears (15, 19) and (21, 30) in both, and the readings of Anc. are confirmed. Sexagenos, correct in Anc. (15, 20), seems to be by error sexagnos in Ant. Paullo (15, 21) is so spelled in both, and therefore was in the Anc. MS. Ad memoriam (16, 27) is by error ad memoria in Ant. Stipendis (16, 31), ablative plural, is the spelling of both. Praerant (17, 35) seems to be in Ant. prarant, by error. A noticeable agreement in accents appears (19, 7) in the phrase summá sacrá viá. Ad aede (21, 22) is given ad aedem in Ant. and correctly. Robinson has noted that it has been taken to be an old ablative.28 Quotiescumque is given by Ant. for quotienscumque (21, 28). Ant. is here in error, for regularly the inscriptions agree in using the n in the endings of the numeral adverb; e. g., milliens (16, 24). Anc. (22, 31) gives quinquens, which is given correctly quinquiens in Ant. The spelling spectaculum is sure in Ant., where editors have supplied spectaclum in Anc. (22, 34). They have done this because spectaclum appears in Anc. (23, 43), where the reading of Ant. is unfortu

27 The references are to chapter and column line of Mommsen's text. Ancyran manuscript is abbreviated Anc. MS.

38 AJPh., p. 43.