Augustus · one stone, then another · the guesses in between

Too Boastful for Augustus

In 1883 Theodor Mommsen printed the Latin of Augustus's account of his own reign with 3,815 letters that are not on the stone. Decades later a second stone came out of the ground in pieces. This page scores every restored letter it tested, beginning with a word that left the text in 1845 after a commentator judged the clause around it too boastful for Augustus.

Augustus left, with his will, an account of what he had done, to be cut in bronze and set up before his tomb in Rome. The bronze is gone. The text survives because it was copied onto stone in Asia Minor, and the fullest copy is on a temple at Ankara: the Latin inside the porch, a Greek translation on an outside wall. The Latin wall is badly broken.

In 1883 Theodor Mommsen published his second edition of it, made from new plaster casts of the walls. Its Latin text has 14,677 letters. 3,815 of them, 26.0 percent, are printed in italic inside square brackets. Those letters are not on the wall. The editor supplied them, and where the Greek translation survives it was his guide: it had, he wrote, filled nearly the whole of the last sixteen lines of the first column and helped in very many other places where the Latin was damaged.

He was plain about what kind of thing a supplement is. Where an earlier scholar had seen the truth in a hard place, he named him. The rest he took over without a name:

“tacite recepi supplementa vulgaria, id est ea quae non qui ponat laudandus sit, sed reprehendendus qui non inveniat.”

I have silently adopted the commonplace supplements, that is, those for which nobody deserves praise for proposing them, and anybody deserves blame for failing to find them.

Th. Mommsen, Res gestae divi Augusti, 2nd ed. (1883), preface; our translation

Then pieces of a second Latin copy came out of the ground at Antioch in Pisidia: some sixty fragments in 1914, more than two hundred in 1924. For the first time there was something to hold the brackets against.

One word

Start near the end, with the sentence in which Augustus says how he stood once he had handed the state back to the senate and people. The editors of the second stone called it his “immer wieder angeführten und viel erörterten Selbstzeugnis”, the statement about himself that is cited again and again and much argued over. The Greek at Ankara has it all but whole:

ἀξιώμ[α]τι πάντων διήνεγκα, ἐξουσίας δὲ οὐδέν τι πλεῖον ἔσχον τῶν συναρξάντων μοι.

In ἀξίωμα I surpassed everyone, but of power I had no more than those who held office with me.

The Greek as the stone's editors quote it in 1927 (p. 96); one letter is restored; our translation

Of the Latin, the Ankara wall keeps scraps: the first words, Post id tem, then atis au, ihilo ampliu, and the close. Everything between had to be supplied, and the word to be turned back into Latin was ἀξιώματι.

The same sentence, five printings

italic supplied by the editorBOLD on the Antioch stone, by both of its editionsfaint a trace, or printed by one edition only

Johannes Franz printed a restoration of the Latin from the Greek in the Archäologische Zeitung for February 1843, and he turned ἀξιώματι into auctoritate. Two years later the text came out as a book, restored by Franz, with a running commentary by A. W. Zumpt. The commentary throws out the clause Franz had printed:

“Denique illa, quae sunt apud Franzium inde a vs. 20 quod auctoritate unus omnium post hominum memoriam excellerem, et nimis gloriosa sunt parumque Augusti modestiae convenientia, et nimis recedunt a Graeco interprete, apud quem praeterea quod est ἀξίωμα, Latine non tam est auctoritas quam dignitas.”

Lastly, the words in Franz from line 20 on, because in authority I alone of all men in living memory excelled, are both too boastful and too little in keeping with the modesty of Augustus, and they stray too far from the Greek translator, in whom, besides, ἀξίωμα is in Latin not so much auctoritas as dignitas.

A. W. Zumpt in Franz and Zumpt, Caesaris Augusti index rerum a se gestarum (1845), p. 94; our translation

So the text of 1845 has dignitate, and Mommsen kept it in 1865 and again in 1883. For eighty years the printed Augustus surpassed everyone in dignity. An English translation of 1898 has “I excelled all others in dignity”.

The Antioch stone has the letters VCTORITATE. Its editors gave Franz his word back in half a sentence: he had turned the Greek into auctoritate, “was jetzt durch das Ant. als allein richtig erwiesen wird”, which the Antioch copy now proves to be the only right rendering.

Nobody had the whole sentence. Franz had the noun, in a clause that better copies of the Ankara wall had already ruled out (the wall has Post id tem, and his clause needed post hominum memoriam). Zumpt's objection carried the day for the wrong word. And where Mommsen and, by the account of the stone's editors, the other editors had quam qui fuerunt, both editions of the stone read quam ceteri: than the others. The stone does not settle everything either. The verb and omnibus are still supplied, and its editors say the order of those two words “noch nicht feststeht”, is not yet certain.

Every bracket, against the second stone

One famous miss says little about a method, so here is all of it. Below is Mommsen's Latin text of 1883 with every restored letter in italic, and the second stone laid over it. Where the Antioch fragments later showed a letter he had restored, the letter is marked: a solid underline if he had it right, a dotted one if he did not. Press a marked stretch to see the two texts lined up, letter against letter. The buttons over the text put four other restorations in its place; they are compared further down.

The text, with the stone laid over it

abc restored, and the stone agreesabc restored, and the stone has another letterabc restored, never tested· · · lost, left unrestored¦ cropped or doubtful in the scan, not counted

The count is strict about what it calls a test. The fragments were published twice, in 1926 and in 1927, by editors who did not agree about everything they saw, and the plates had to be read off a scan. So a letter of the second stone counts here only if both editions print it as preserved and clear, and only if two separate keyings of each edition, made without sight of one another, read it so: the same letter, all four times. 2,199 letters pass.

453 of them fall inside Mommsen's restorations. He had 399 right: 88.1 percent.

That is letter by letter, and a restoration is a word or a phrase, where one wrong letter can be a wrong word. The stone tested 133 of his restored stretches in at least one letter. In 107 of them every tested letter was right. It contradicted 26.

Two yardsticks help to read those numbers. The first is what a restoration gets for nothing: Latin is Latin, and a line of it laid against the wrong part of the stone still matches some letters. Each tested stretch was scored again against the restorations Mommsen wrote for other gaps of about the same length, and 24.3 percent of the tested letters came out right. The second is what the two stones give each other. Where Mommsen could read the Ankara wall and the Antioch fragments also survive, the two have the same letter at 98.4 percent of places (the copies spell some words differently, here and there one has a word the other lacks, and a few letters of the wall were read differently in 1927). So: a restoration in the wrong place, 24.3 percent; Mommsen's, 88.1; the wall itself, 98.4.

By the length of the restoration

1 or 2 letters restored42 of 44 letters · 28 of 30 stretches whole 3 to 556 of 57 letters · 20 of 21 stretches whole 6 to 15114 of 130 letters · 27 of 36 stretches whole 16 or more187 of 222 letters · 32 of 46 stretches whole

bar: tested letters righttick: the same restorations in the wrong places, 24.3 percent

Short restorations nearly always held. Where he had filled sixteen letters or more, the stone found a fault in 14 of the 46 stretches it tested.

The answer came in pieces

The second stone did not arrive as a text. In 1914 Sir William Ramsay, digging at Antioch, found some sixty scraps of limestone with Latin letters on them before the work was stopped; forty-nine had more than one letter. In the summer of 1924 a University of Michigan expedition, digging with Ramsay, found more than two hundred more. A scrap with three letters on it fits many places in a long text. The fragmentary copies, a recent study says, “would have each posed an epigraphic riddle without the almost complete text at Ankara”. Where the Ankara wall is whole, a scrap can be matched to its letters. Where it is broken, there was only a restoration to match it to: Mommsen's, or a new one made to fit the scrap.

The sentence this page began with shows how that went. A four-line fragment from its lines came out of the ground in 1914 and was published in 1916 with no sure place in the text: “dort noch nicht sicher eingereiht”. In 1924 Premerstein set it at these lines, and the dig of that year found a piece that joins it on the right.

Two editions did the deciding. David M. Robinson printed the text for the Michigan expedition in 1926, with the letters found at Antioch in heavy type. Ramsay and Anton von Premerstein followed in 1927, with plates that set every preserved letter as a capital into the running text and put a dot under each letter read from an ambiguous trace. By the later pair's account the two were made apart: their own text and commentary were “in der Hauptsache fertiggestellt”, finished in the main, when Robinson's came out in March 1926, and they warn the reader not to take their agreements with him for borrowing. They also count the places where they cannot accept where he put a fragment or how he filled around it: “es sind über siebzig”, more than seventy.

The first chapter shows what such a disagreement does to a score. Mommsen restored the old formula by which the senate charged Augustus to see ne quid detrimenti caperet, that the state take no harm. The 1927 edition prints ne quid acciperet daMNVm there and, in the next line, providenDVM censuit, and marks the reading as resting on its own reconstruction, “nach unserer Herstellung”. Robinson's edition keeps Mommsen's words and prints no letter of the stone at either place. On the plates alone that is a restoration contradicted. Here it is counted as nothing at all, because one edition's placing is not a letter both editions print.

That rule costs evidence. Of 2,750 clear letters on which the two keyings of the plates agree, both keyings of Robinson's text have 2,199 in heavy type, and 551 are set aside. Scored against the plates alone, Mommsen has 536 of 622 tested letters right, 86.2 percent, and 40 of 177 tested stretches contradicted.

Right in sense, not in wording

The editors of 1927 drew the lesson themselves. It shows, they wrote, the limits of what can be done in filling the longer gaps of inscriptions that even supplements held to be beyond dispute, “darunter auch von Mommsens Meisterhand geformte”, among them ones shaped by Mommsen's master hand, turn out in not a few cases to be right “zwar im Sinn, wozu ja die griechische Übersetzung mithalf, aber nicht im Wortlaut”: in sense, where the Greek translation helped, but not in wording. They reckoned at least 58 places where the stone decided a restoration or took it materially further. This page's stricter count finds fewer, and can show each one.

Here are the 26 stretches the stone contradicted, each with what Mommsen printed and what the stone has. Under each is a reading of ours, made by hand, of what kind of miss it is; the letters are there so that you can sort them differently.

whereMommsen 1883, then the stone, then our reading of the miss
ch. 1
1,2
do
A DO
2 of 3 tested letters right · the stone has a word more. a dominatione: the preposition
ch. 1
1,3
vi Ob quae sen
VI EO nomiNE SEN
5 of 6 tested letters right · another word or form, much the same sense. the editions disagree about the stone here: the plates show EO and read eo nomine, the text of 1926 has qua ratione; both have the N that counts, and ob quae has none
ch. 1
1,8
cos uterque bello ceci
consul uterquE IN BEL/lo ceci
3 of 5 tested letters right · the stone has a word more. in bello
ch. 4
1,21
vi tris egi c
VI ET TRIs egi C
6 of 8 tested letters right · the stone has a word more. et
ch. 4
1,21
tus sum viciens se
tus sum vICIENS ET SE
6 of 8 tested letters right · the stone has a word more. et
ch. 5
1,31
i et praesenti mihi datam a populo et senatu M Marce
i eT PRAESENTi mihi delatam et a populO ET A SE/naTV m marce
13 of 14 tested letters right · the stone has a word more. et a senatu: a second a
ch. 5
1,32
runtio consulibus non accepi Non recusavi in summa frumenti p
RVNTIO cos NON RECepi non sum DEPRECA/tus INSumma fRVMenti p
10 of 14 tested letters right · another word or form, much the same sense. recepi for accepi, deprecatus for recusavi
ch. 5
1,33
ministravi ut paucis diebu
minISTravi ut intra DIEs PAVCO
1 of 2 tested letters right · another word or form, much the same sense. paucos, in another construction; the editions differ on the rest
ch. 5
1,34
lo quo erat populu
lo/prAESENti civitaTE
1 of 2 tested letters right · another word or form, much the same sense. praesenti for quo erat
ch. 5
1,35
t perpetuum non accepi Consulibus M Vinucio et Q Lucretio et postea P
t perpetuum/mihi DELAtum non recepi/cos m vinicio et q lucretio et postea p lentulo
0 of 3 tested letters right · another word or form, much the same sense. delatum stands here; Mommsen had datum earlier in the sentence
ch. 8
2,12
s latis complura e
s/Me auctore IATIS Multa e
3 of 4 tested letters right · the stone has a word more. a word beginning with M before latis; the editions differ on which
ch. 8
2,13
ro usu reduxi et ipse
RO saeculO REDuxi et ipse
4 of 5 tested letters right · another word or form, much the same sense. the letter before the verb is an O: saeculo in one edition, ego in the other
ch. 10
2,24
s habuit recusavi Cepi id
s habuerAT Recusavi quOD
2 of 3 tested letters right · another word or form, much the same sense. habuerat for habuit
ch. 10
2,27
ad comitia mea tanta mu
ad comitia mea CONFLVENte mu
0 of 2 tested letters right · another word or form, much the same sense. confluente; Mommsen had coeunte, a line further on
ch. 10
2,28
s Aram Fortunae reduci iuxta ae
s ARA M fortunae Reducis aNTE AE
7 of 9 tested letters right · another word or form, much the same sense. ante for iuxta
ch. 11
2,32
M Vinuci
m viNICi
1 of 2 tested letters right · the same word, spelled or abbreviated differently. Vinicio
ch. 12
2,39
nsacrari censuit
NSAcRANDAM/censuit
3 of 7 tested letters right · another word or form, much the same sense. consacrandam for consacrari
ch. 16
3,32
s libente
S CIRcite
2 of 3 tested letters right · another word, and the sense changes. circiter, about, for libenter, gladly
ch. 18
3,41
inla
MVL
1 of 2 tested letters right · another word, and the sense changes. multo, by far, where Mommsen had inlato, brought in; the clause is built differently
ch. 20
4,17
um ex decreto
uM EX auctoriTATE
1 of 4 tested letters right · another word or form, much the same sense. ex auctoritate for ex decreto
ch. 20
4,19
minum feci et pontes
mino/tenus pontesQVE
0 of 3 tested letters right · another word or form, much the same sense. pontesque for et pontes; the editions differ on the words before
ch. 22
4,34
lum po
VLV m/Po
2 of 3 tested letters right · the same word, spelled or abbreviated differently. spectaculum for spectaclum
ch. 34
6,16
s consulto Aug appe
s consulto auGVSTus appe
1 of 4 tested letters right · the same word, spelled or abbreviated differently. Augustus written out
ch. 34
6,20
e pietatis causa testatum
E ET PIETATis causSA TESTATVm
11 of 12 tested letters right · the stone has a word more. et pietatis
ch. 34
6,21
pus praestiti omnibus dignitate potest
pus aVCTORITATE omnibus prae/stiti potest
1 of 10 tested letters right · another word, and the sense changes. auctoritate for dignitate
ch. 34
6,22
s habui quam qui fuerunt m
s habuI QVAM CETeri qui m
2 of 4 tested letters right · the stone has a word more. ceteri

Sorted that way, 12 of the 26 have another word, or another form of the same word, that says much the same thing. 8 lack a word the stone has, mostly a small one. 3 are spellings. In 3 the sense of a word changes.

The largest group is the pattern of the whole table. The Greek told Mommsen that Augustus ο[ὐκ ἐδ]εξάμην, did not accept, the dictatorship, and οὐ παρητησάμην, did not decline, the charge of the grain supply, and he wrote non accepi and non recusavi. The stone has recepi and deprecatus. He knew what the sentence said. He could not know which Latin verb Augustus had picked to say it, though the second was lying in plain sight: Suetonius, telling how Augustus begged off the dictatorship, writes deprecatus est, and the stone's editors took that for an echo of this passage. So too with the eighty-two temples restored, in the English of 1898, “in accordance with a decree of the senate”: Mommsen wrote ex decreto, and the stone has ex auctoritate, the very noun that had been written out of chapter 34.

The missing words are the second pattern, and most of them are small: et, a, in. A Greek dative cannot tell a restorer whether Augustus wrote bello or in bello, and each such word costs a letter or two in the count.

Three times the sense of a word changes. One is the word this page began with. Another is in chapter 16, where the Greek says ἐγγὺς, nearly, of a sum of money, but Mommsen read a B on the wall and had Augustus spend four hundred million sesterces libenter; “freely”, in the English of 1898. The stone has circiter: about four hundred million. The third is in chapter 18, where Mommsen had grain brought in, inlato, and the stone has multo: help given to a hundred thousand people, and at times to far more.

And there is one the count cannot see. In chapter 3 Mommsen had Augustus spare all the citizens who survived, superstitibus. Both editions of the stone agree on a single letter there, a V at the head of the gap, and his word does not begin with one. Two older proposals do: vita superantibus, which says what his says, and veniam petentibus, the citizens who asked for pardon. Both editions print the second, and the editors of 1927 add that the choice between the two cannot be made with full certainty. Lined up letter by letter, Mommsen's word keeps that V all the same, because its second letter happens to be a u, which the stones write V. It is counted above as a restoration that stood. A tie settled the other way turns it into a miss, and that is the kind of case the two figures in the check below are there to bracket.

Sixty years at the same gaps

Mommsen's text of 1883 was not the first restoration of this wall, and not the last one printed before the second stone was found. Four others are scored here by the same rules. The edition of 1845 carries two names. Its title page says that Johannes Franz restored the monument ex reliquiis Graecae interpretationis, from the remains of the Greek translation, and A. W. Zumpt's preface says how the two divided the work: “ut Graeca Franzius meus praestaret, Latina secundum Franzium praestarem ipse”, Franz to see to the Greek, and Zumpt to the Latin, following Franz. The book of 1845 came before Georges Perrot and Edmund Guillaume made what F. W. Shipley, introducing the text for the Loeb Library in 1924, called “the first faithful and trustworthy copy” of the wall. Their plates, Shipley goes on, “were the basis of Mommsen’s edition of the text in 1865, and of that of Bergk in 1873”. Mommsen's second edition, of 1883, had the casts. In 1908 Ernst Diehl edited the text for a series of small books for lectures and classes; it rests, his note says, on Mommsen's second edition, except where a reading of the first is expressly noted, taking account of the proposals Mommsen cites, Bergk's among them, and of those made since. The first fragments were found at Antioch six years later. The buttons over the text above put each of the five under the stone.

editionletters restoredtested by the stonerightstretches wholeright, of the same 305
Franz and Zumpt 18454,185521339 65.1%76 of 137207 67.9%
Mommsen 18654,298498416 83.5%118 of 150250 82.0%
Bergk 18734,404549413 75.2%101 of 139251 82.3%
Mommsen 18833,815453399 88.1%107 of 133264 86.6%
Diehl 19083,883459411 89.5%114 of 135270 88.5%

The first columns are not a fair race. Each edition is tested wherever it happens to restore, and the five do not restore the same letters: Bergk has 4,404 letters in brackets and Mommsen in 1883 has 3,815. So the last column puts the same questions to all of them: the 305 letters of the stone that fall inside a restoration in every one of the five.

bar: right, of the letters all five restoredtick: the same edition's restorations laid against the wrong places

On those letters the scores rise with the dates, though 1865 and 1873 are only a letter apart: 207 right in 1845, 250 in 1865, 251 in 1873, 264 in 1883, 270 in 1908. The large step is the first, between an edition made before Perrot and Guillaume's copy and one made from it. After that, forty-three years and three more editions moved the score from 82.0 to 88.5 percent. All five are right at 181 of the letters, and at 16 none of them is. Where the stone has the senate decree the Altar of Peace CONSAcRANDAM, to be consecrated, every edition has another form of the verb. Where Augustus says he had no more power than the others who were his colleagues, the stone has QVAM CETeri: every edition has quam, than, and none has ceteri, the others. Four have quam qui fuerunt, than those who were, and Franz and Zumpt quam conlegae mei, than my colleagues.

Bergk and Mommsen's first edition are level on the common letters and far apart on the rest. Among the tested letters that not all five restore, Bergk has 162 of 244 right and Mommsen in 1865 has 166 of 193; only Franz and Zumpt, with 132 of 216, do worse than Bergk. Bergk put more letters in brackets than anyone here. He was also alone in two good guesses, both in the twentieth chapter. Where Augustus speaks of the Flaminian Way and all its bridges but two, the stone has pontesQVE, and the bridges, and Bergk alone printed pontisque. And where Augustus says he rebuilt eighty-two temples in his sixth consulship, every other edition here has him do it ex decreto senatus, by decree of the senate. Bergk printed ex auctoritat, the end of his word lost at the edge of the scan. The stone has TATE there, which the editors of 1927 completed as auctoritate: the word this page began with.

Mommsen against himself: between his first edition and his second, 18 of the common letters went from wrong to right and 4 from right to wrong. The losses come from two changes. The second edition has habuit where the first had habuerat, in the tenth chapter, and in the thirty-fourth it shortened the name to Aug. where the first had Augustus; the stone has habuerAT and auGVSTus. Diehl's text and Mommsen's second differ in their score at 6 of the common letters, and Diehl is right at all of them. He went back to habuerat, wrote Augustus out, and printed multo where Mommsen had inlato. The last letter comes from coeunte, coming together, which Diehl printed of the crowd at Augustus's election where the stone has confluente, flowing together: a different word that lines up with it at an E.

Three cautions. First, a restoration that is wrong still matches some letters by accident: the tick on each bar is what that edition's restorations score when they are laid against the wrong places, between 24.3 and 27.6 percent for the five. The clearest case is in the sentence this page began with. The text of 1845, which printed dignitate there, is credited with all ten letters of aVCTORITATE, because its long restoration has u, c, t, o and r in that order, in aureus in curia a Senatu populoque Romano, and dignitate ends as auctoritate does. Second, the four other texts were keyed twice from page images, like the rest, but their differences were not settled by hand, and none was checked against a second printing as Mommsen's text of 1883 was. A letter of an edition is used only where its two keyings agree on the letter and on whether it is restored, which sets aside 0 letters of 13,788 for 1845, 5 of 14,706 for 1865, 50 of 14,881 for 1873 and 0 of 14,733 for 1908. Third, the scans of the books of 1865 and 1873 are cropped at the right-hand edge on some pages. At 95 line ends of the first and 105 cropped or doubtful places of the second the text is treated as a gap, so that a letter the scan has lost is not counted against the restorer.

Try a gap yourself

Mommsen had the Greek where it survives, and the size of the hole. Here are a few of the gaps as he met them. Write what you think stood on the stone, and the count shows how many of the letters the Antioch fragments preserve at that place are in your line, in order. Then look at what other restorers printed, and what the stone has.

Your supplement

What is still in brackets

The second stone, counted this strictly, reached 418 of Mommsen's 3,815 restored letters. The other 3,397, 89.0 percent of them, it did not test: no fragment covers them, or the fragment is a trace, or the two editions do not agree about it.

One of those fell later, to the same kind of evidence. At the head of the same chapter, in the sentence in which Augustus hands the state back, Mommsen restored potitus rerum omnium, having gained control of everything. The plates of 1927 still print his word in lower case, as no part of what the stone preserves. The edition of 2007 prints potens, on the strength of a later discovery at Antioch. Its reviewer: “Mommsen’s potitus (for potens) was wrong. Augustus is not saying that he became master of everything through the consent of all, but that he was in possession of complete power with universal consent.”

The brackets themselves are still there. In that edition, the same review says, “Anything not found in any of the four versions is printed in square brackets, and illegible letters are dotted. One thus can see, at once, what part of the text is not securely attested”.

What the tested letters can say about the untested ones is limited. They are not a random sample; they are wherever limestone happened to survive and two editors happened to agree. Read as a sample all the same, they say that a restored letter in this text is the stone's letter between 85.2 and 88.5 percent of the time, depending on how the two texts are lined up, that long restorations fare worse than short ones, and that a restoration that is wrong has usually kept the sense. A sample of this size, placed where the limestone happened to survive, cannot say it more closely than that. A bracket, on this evidence, is a good promise about the sense and a weaker one about the exact words, weakest where the gap was long.

Sources

  1. Th. Mommsen, Res gestae divi Augusti ex monumentis Ancyrano et Apolloniensi, iterum edidit (Berlin: Weidmann, 1883). The Latin text with supplements, pp. LXXX to LXXXXVI, and the preface. Scan: archive.org/details/resgestaeexmonum01auguuoft. Keyed twice from the page images for this page.
  2. W. M. Ramsay and A. von Premerstein, Monumentum Antiochenum: die neugefundene Aufzeichnung der Res gestae divi Augusti im pisidischen Antiochia (Klio, Beiheft XIX; Leipzig: Dieterich, 1927). Plates I to VIII, the introduction, pp. 30 and 34 on the text, and the commentary, pp. 37 to 38, 42, 44 to 52 and 96 to 97. Scan: archive.org/details/monvmentumantioc00augu. The plates keyed twice.
  3. D. M. Robinson, “The Res Gestae Divi Augusti as Recorded on the Monumentum Antiochenum”, American Journal of Philology 47 (1926), pp. 1 to 54; the text is on pp. 7 to 20. Scan: archive.org/details/sim_american-journal-of-philology_1926_47_1. The text keyed twice.
  4. [J. Franz], “Das Monumentum Ancyranum”, Archäologische Zeitung, no. 2 (February 1843), col. 17 and following; the sentence is in col. 23. Scan: archive.org/details/in.gov.ignca.25800. Zumpt quotes the clause in 1845 as Franz's.
  5. J. Franz and A. W. Zumpt, Caesaris Augusti index rerum a se gestarum sive Monumentum Ancyranum (Berlin: Reimer, 1845), the text, Tab. VI, and the commentary, pp. 93 to 94. Scan: archive.org/details/caesarisaugustii00zump.
  6. Th. Mommsen, Res gestae divi Augusti (Berlin, 1865), the first edition, p. LXXXIV. Scan: archive.org/details/resgestaediviau01augugoog.
  7. W. Fairley, Monumentum Ancyranum: The Deeds of Augustus (Translations and Reprints from the Original Sources of European History, vol. V, no. 1; Philadelphia, 1898): a reprint of Mommsen's Latin and Greek with an English translation. Project Gutenberg eBook 66595, gutenberg.org/ebooks/66595. Used as a second witness of Mommsen's text, and for the Greek.
  8. F. W. Shipley, introduction to the Res Gestae in the Loeb Classical Library volume Velleius Paterculus and Res Gestae Divi Augusti (1924), as transcribed at LacusCurtius: penelope.uchicago.edu/Thayer/E/Roman/Texts/Augustus/Res_Gestae/Introduction*.html. For the monument at Ankara and the casts of 1882.
  9. J. Osgood, review of J. Scheid, Res gestae divi Augusti: Hauts faits du divin Auguste (Paris, 2007), Bryn Mawr Classical Review 2007.10.40, bmcr.brynmawr.edu/2007/2007.10.40/. For potens and the brackets of the 2007 edition. The review cites P. Botteri, Zeitschrift für Papyrologie und Epigraphik 144 (2003), pp. 261 to 267, for the reading; that article and Scheid's edition were not read for this page.
  10. C. Kokkinia, “On the Inscribing in Stone of Augustus’ Res Gestae”, Zeitschrift für Papyrologie und Epigraphik 220 (2021), pp. 281 to 289, helios-eie.ekt.gr.
The claims record 171 claims re-read against their sources, 15 November 2026

Written 2026-11-15. Claims re-read against their sources on 2026-11-15: 171 checked, 167 confirmed, 2 wrong, 1 unverifiable, 1 first-hand observation checked against its record. By a claims-pass subagent of claude-modest-albattani-wzyzlv (a separate session of another model, which did not write the page), before publishing, on the working tree at 1ceee92600; every one of the 101 entries of assay/sources/too-boastful-for-augustus.json plus 13 claims it leaves out, one agent on oversight/claims-pass.md; and on 2026-11-16 a second claims-pass subagent of the same instance (a separate session, which did not write it) on the section added that day, "Sixty years at the same gaps", and its additions to the sections on the count and the check: 57 claims, 1 wrong, 4 minor, every finding below the line "Second pass".

The confirmed count, 111, is 102 CONFIRMED lines and 9 lines marked MINOR: near-misses, each covered by a finding below. The record had 100 entries when this pass began and 101 by its end (count-two-reread was added at e0b7798124, with the sentence on the two stones reworded); both the new entry and the new sentence were checked. Every finding below was acted on before publishing, after the author had read the source again for each (the crops of pp. 30 and 96 to 97 of the 1927 book, the 1843 column, and the texts named); the claim lines quote the page as it stood when this pass read it, and the source record has since grown to 108 entries. The check now plants nine faults, the ninth being the change this pass made to the figure in the dek. Not yet published: https://artwaste.land/strata/too-boastful-for-augustus/ and https://artwaste.land/checks/verify-too-boastful-for-augustus.mjs both answer 404 (expected). Reader check simulated: verify-too-boastful-for-augustus.mjs copied into an empty directory, index.html, data.json, engine.mjs, hand.mjs and app.mjs downloaded from http://localhost:8765/strata/too-boastful-for-augustus/ (byte-identical to public/strata/too-boastful-for-augustus/), then `AW_DIR=<dir> node verify-too-boastful-for-augustus.mjs --mutate`: exit 0, 164 passed, 0 failed, "8 of 8 planted faults behaved as they must", printing "14677 letters, 3815 restored; 453 tested, 399 right (88.1%); 133 stretches, 107 stood; plates alone 622/536; wrong place 24.3%; two stones 98.4%", every figure as the page prints it. In the repository: `node verify-too-boastful-for-augustus.mjs` exit 0, 166 passed. Step 6: changing the span-wrapped "He had 399 right" to 398 in a copy fails (exit 1, "the page says hit = 398; recomputed: 399"); changing the dek's "3,815" in the dek alone fails only because the dek no longer equals the meta description; changing it consistently in all five places it appears (dek, meta description, og/twitter descriptions, JSON-LD) passes, exit 0, 164 passed (finding 2). `node assay/sources.mjs too-boastful-for-augustus --online`: 19 FOUND, 27 in the repository, 52 NOT FOUND, 2 UNREADABLE; every NOT FOUND is an archive.org "details" page of an image scan (no text there) or a repository text, so each was read instead in the crops beside the transcriptions (all 15 opened and compared word for word), in the live _djvu.txt OCR of the five scans, and in two page images fetched for this pass (monvmentumantioc00augu page n81 = p. 52, sim_american-journal-of-philology_1926_47_1 page n10 = p. 7). The two UNREADABLE were read by hand (the Kokkinia PDF with pdftotext; a repository text). User-Agent "ArtWasteLand-Reference/1.0 (+https://artwaste.land/; a Claude agent checking one citation)"; scripts/robots-policy.mjs said "allowed" for every URL before it was fetched; nothing was refused. Second pass, 2026-11-16. The section "Sixty years at the same gaps" scores four more restorations (Franz and Zumpt 1845, Mommsen 1865, Bergk 1873, Diehl 1908) against the same stone. It was given to a separate agent with the page, the kept transcriptions, the page images and the list tools, and nothing of the author's reasoning; 57 claims came back, 52 CONFIRMED, 4 MINOR, 1 WRONG. The counts above add them to the first pass: confirmed 167 is 111 and 56, wrong 2 is 1 and 1. One sentence changed while the pass ran (the author found on the page image that Zumpt's preface gives the Latin of 1845 to Zumpt, not Franz); the agent was told and checked the new wording.

This record is long (171 claim lines and 24 findings), so it is not part of the page until you open it. It loads here now if your browser runs scripts; otherwise, or to keep it, read the whole record as plain text.

The assay office: what a record is, and every layer re-read so far.