P. Cornelii Taciti Dialogus de Oratoribus. Mit Prolegomena, Text und Adnotatio Critica, exegetischem und kritischem Kommentar, Bibliographie, und Index Nominum et Rerum von ALFRED GUDEMAN. Zweite völlig neubearbeitete Auflage. Leipzig: B. G. Teubner, 1914. Pp. viii+528. M. 14 (16, bound). This completely remodeled edition of the Dialogus was issued at the beginning of July, 1914, on the eve of a far more momentous struggle than that which once divided the world of scholars over the authorship of the little work now adjudged beyond any reasonable doubt to the historian. The first edition, published in English in this country just twenty years earlier, has long been an indispensable tool to every Tacitus scholar. It will now be still more warmly welcomed as the mature product of intensive cultivation in a field in which these have not been lean years. Many opinions have been modified in response to criticism, but the main contentions of the first edition remain, strengthened by more exact statement and gradually accumulated evidence. Outwardly the book has not grown conspicuously in size, but a larger page and smaller type alone have kept growth within bounds. Text and critical notes, which formerly claimed 55 pages out of 585, now fill but 43 out of 528, while the commentary (allowing for the different page) has been increased by perhaps 60 of the new pages. In the Prolegomena there is a great gain in point of clear arrangement, and in added chapters. The question of authorship, time of writing, dramatic date of the dialogue, plan and technique of composition, the interlocutors, the two lacunae and the speakers of 36-40. 4, the fictitious character of the dialogue, literary sources, matters grammatical and rhetorical, the manuscripts-all these subjects are fully treated in the Prolegomena. The history of the controversy is given with much detail, and the arguments for Tacitean authorship marshaled with rare skill. It is safe to say that the ghost of anonymity or heteronymity conjured up by Beatus Rhenanus and Justus Lipsius has finally been laid. With regard to the date of composition, the newer view of Leo, Norden, and Wilamowitz, that the Dialogus was a tour de force in a style dictated by the subject-matter, and therefore may, or must, have been written after the Agricola and Germania, is critically examined and duly rejected, along with other arguments for a date after Domitian. The chapter on the dramatic date will probably win no less general assent, except in a passage (pp. 61 f.) concerning the magnus annus and Censorinus, where one may feel that the unknown quantities are more numerous than the equations available for their solution. The chapter on the two lacunae and the speakers of 35-40. 4 strongly defends the theory, already maintained in 1894, of a lacuna at the latter point, involving the loss of the conclusion of Secundus' speech and the beginning of that of Maternus. But much new matter is of course added, including a careful revision of the calculations printed in Classical Philology, VII (1912), 412–16. The summa summarum

is that that part of the dialogue between the long lacuna of 35 fin. and the shorter lacuna at 40. 4 would fill just four pages of the lost Hersfeldensis— lost except for eight leaves of the Agricola (cod. Aesinus), which furnish the basis for the computation. Of course opinions will differ as to the substantial value of calculations of this kind, in which Teubner lines are compared with those of a lost portion of a codex, that a lacuna, otherwise suspected on internal evidence, may gain external support also, if it can be made probable that precisely one leaf, beginning and ending with a complete sentence, was lost at this point.1

Prominence is deservedly given to Sabbadini's discovery (1901) of Decembrio's detailed description of the Hersfeldensis as he collated it in 1455. Decembrio counted 14 folia of the Dialogus up to the great lacuna, and 24 thereafter. In his addenda Gudeman supplements this (pp. 135 ff.) by a similar but less minute description found at the close of a manuscript of the philosophical works of Cicero and known since 1913. This is from an inventory made by Niccoli of manuscripts in convent and other libraries of Germany. The description goes back to 1425. It omits Decembrio's "Cornelii Taciti," and gives the folia of the Dialogus as 18. Thus Gudeman finds full confirmation of his theory of a lost leaf at 40. 4, arguing that the leaf was still extant and counted by Niccoli, but had been lost before Decembrio thirty years later counted but 14+2 folia. Some have no doubt scented a possible error in a Roman numeral, but to most scholars not already committed to another theory the assumption of a second lacuna, two pages in length, must now appear all but completely demonstrated.

Of the full description and comparison of the manuscripts it will suffice to say that as a result the value of A and B is somewhat lowered, and that the new stemma assigns E and V to a parent Z, a copy of the Hersfeldensis distinct from those (X and Y) from which the other manuscripts descend.

The text is much more conservative, ut fit, than in the first edition, many conjectures of the editor or his predecessors now appearing in the critical notes only, while an obelus in the text bids the reader seek enlightenment from below. In many cases Gudeman still favors the same conjecture, but does not admit it to the text. In not a few others readings of 1894 are silently discarded. Table I gives the changes which may be mentioned here.

1 Some curiosity on this matter prompted the reviewer to compare each successive group of four pages in the first Medici codex with Andresen's text of 1913, through the first book of the Annals, making careful allowance for all short lines and indented lines of Teubner as compared with solid text in the facsimile. With this correction it was found that, while four pages of this manuscript may mean only 86 Teubner lines, or as much as 101 lines, on the other hand 11 out of the 15 blocks of four pages examined run between 91 and 96-a very fair degree of uniformity, though perhaps not all that might be expected of a manuscript of the first order of calligraphy, with solid pages and no short lines, such as add another element of uncertainty in the Aesinus.

10, 18

10, 29

10, 30

10, 33

10, 34.

12, 13.

13. 3.

13, 21

13, 22

14, 4

14, 10

14, 12

15, 4.

15, 5

15, 12

16, 19

16, 29

17, 18.

18, 5

18, 15

18, 21

19, 2

19, 10

20, 2.

20, 6.

20, 10

21, 3

21, 28.

21, 34.

22, 11.

22, 13.

22, 19.

23. 4.

23, 20.

24, 9.

25, 12.

tet virides tteneri statim et rudes [et] virides statim et teneri [rudes]

In 7. 9, while the text reads si non in falio oritur, the former reading si non in alvo oritur is even more stoutly defended than before. No attempt is made to meet Baehrens' obvious objection that matris in alvo would be expected. The passages cited from Accius and from Pro Cluentio 34, in which a woman is in both cases the subject of the sentence, do not support in alvo without matris here. The same is true of De div. ii. 39, alvo contineret (not cited by Gudeman). And why, in the search for an antithesis to the äussere Vergünstigungen, should one reject the perfectly general in alio and dwell with exclusive emphasis upon the idea of heredity, as though application and experience had nothing to do with the case and would not be quite as likely to be mentioned here? The excellent authority for de alio, ex alio, and alio furnished by the Thesaurus may perhaps impress some who have been skeptical about in alio.

In 10. 18 the unhappy mox before summa adeptus is now dropped, but reluctantly, as we may infer from the commentary in which Gudeman still speaks of the "analogy" of Hist. ii. 82, egregios viros et mox summa adeptos. But in that case the historian is allowing himself an anticipation to which there can be no parallel in adeptus here. It is clear enough that the antithesis of levioribus must be found in summa, which then has no reference to fame or outward rewards-no more than Seneca's summa adeptum of the sage (Ep. 109. 1).

A glance at any part of the commentary suffices to show what a mass of material has been added in one form or another. Suppression has often made fresh insertions possible, or matter has been transferred to another context. The result is an encyclopedic commentary on the largest scale and of great value as a general work of reference, for which purpose it is made available by an adequate index. It is unnecessary here to cite instances of the learning accumulated oυpávcov oσov in these 326 solid pages. The most appreciative critic will sometimes feel that too much has been included, as when discarded readings are still more or less half-heartedly defended. Thus on pp. 308-9 imitatus esset (18. 5), though accepted in the text (in lieu of miratus esset, 1894), is found a hard nut to crack. Too much is made of a "direct contradiction" between Aper's own words at 22 init. ([Cicero] suorum temporum eloquentiam anteponebat) and the wish (18, §2) that Calvus,

Caelius, and even Cicero had not imitated the antiqui. For there Aper is speaking of general tendencies and everything in the large, here in all probability of incidental and perhaps rare expressions. If Seneca (Ap. Gell. xii. 2. 6ff.) could find fault with such Ennian words in Cicero as suaviloquens and breviloquentia, what certainty have we that Aper could not claim to find imitations of older orators in Cicero? The hearers might be horrified, as was Gellius, but the truth may still be on the side of the merciless critic. Cicero's acknowledged obligations to the old orators and Quintilian's recommendation that they be judiciously studied (x. 1. 40) become perfectly meaningless, if they did not have in mind at least occasional imitation. That Aper meant anything more than this does not appear; hence direkt falsch or stark übertrieben seem much overdrawn. To most readers miratus would seem to involve some degree of imitation and to be only a little less explicit than imitatus. But it is ungracious to look for rare flaws in so sound and admirable a piece of work.

The excellent Index nominum et rerum has already been mentioned. A copious Index locorum in the first edition has not been reproduced, no doubt for lack of space, since it would now be much larger.

FRANK GARDNER MOORE

COLUMBIA UNIVERSITY