among the descendants of the Chronicon. In general, however, both Moore and Kornemann accept Reinhold, though they have made no investigations into the question outside of the passages parallel to the Oxyrhynchus Epitome.

Moore's proofs that Oxyr.1 used the source of Obsequens, Cassiodorus, and Eutropius, are found on pp. 245 to 255 of his article, in which he discusses or refers to some seven different passages. In all Oxyr. shows a reasonably close relationship to these authors, but also to the Periochae Livii, Orosius, etc., wherever they treat of the same subjects. In other words, all the passages show indebtedness to the Epitome. In only one case does Moore find that his authors agree in a divergence from the Epitome tradition. I quote from p. 245 "Oxyr. 88, L. Marcio Censorino M. Manlio cos (=103, Manlio et Marc(i)o c[os]); Cassiodorus a. 605, L. Marcius et M. Manlius. The correct form in Per. Liv. 49, L. Marcio M' Manilio cos; Censorinus, de Die Nat. 17, II (Livy cited), Florus, 1, 31, 7, Orosius, 4, 22, 1, Appian, P. 75; 97; and Zonaras, 9, 26, all have the same, while Eutropius, 4, 10, in Mss P. and D., has Marco Mallio, though Manilio appears in the version of Paeonius." The trouble with this proof is that in many of the cases cited the correct form of the name is due to emendation. In the best Mss the name appears as follows: Censorinus, M. Manlio; Per. Liv. M. Manilio; Florus, Manilio; Orosius, § 1, M. Manilius § 7, Manlius; Eutropius, Marco Manilio (in the three oldest Mss); Appian, Μάρκον Μανίλιον; Zonaras, Μάρκος Μανί λιος. I add de Vir. Ill. 58, Tito Manlio; Velleius Paterculus, 1, 13, 1, M. Manlio. All descendants of Livy had Marcus for Manius, but Moore laid no stress on that variation. As regards the form Manlius, we must decide that it is only a Ms variation, liable to creep in anywhere. Further, Kornemann has removed all reason for discussing the passage by finding that Oxyr. really reads Man(i)lio in 1. 88 (i.e. there is space for the letter, though it is no longer distinguishable on the papyrus), but in l. 103, the reading is Manlio. I will Oxyr. Oxyrhynchus Epitome of Livy; the number following refers to the

1

line.

=

take up the question of language and minor points under Kornemann's article.

To sum up, Moore's paper presents welcome proof that Oxyr. was related to the Epitome of Livy, not to the entire Livy, but he gives no evidence on the question of Reinhold's Chronicon, except to throw doubt on the placing of Eutropius among its descendants.

We turn to Kornemann, who has openly accepted the Chronicon, yet without definite additional proofs. For in no case do his parallels show a different version for the Epitome, where there is special agreement between Oxyr., Cassiodorus, and Obsequens. Eutropius does not come in question at all. I discuss these few passages below in connection with many

more.

As Kornemann offers no decisive proofs for the Lost Chronicon, we turn to the considerations which induced him to accept it. These lie almost exclusively in the chronological character of Oxyr., Obsequens, and Cassiodorus. To this he adds (p. 70) that in certain cases kindred occurrences of two years are united in one, or such a union indicated for the common source by errors in dates. These he claims point to a source that was not chronological, so the original of the mistakes must go back of the Chronicon to the Epitome, which was thus not annalistic in its arrangement. He cites only two passages to illustrate, but a search of his commentary discloses eight such parallels:

(1) Oxyr. 17, Livy, 38, 36, 5, and 38, 28, 4;

(2) Oxyr. 44-45, Obsequens, 3, and Livy, 39, 54; (3) Oxyr. 49, Livy, 39, 20, 5, and Per. Liv. 39;

(4) Oxyr. 71-73, Livy, 40, 5-24, and Per. Liv. 40;

(5) Oxyr. 103-104 and Per. Liv. 49 ;

(6) Oxyr. 174, Per. Liv. 54, Eutropius, 4, 17, 1, and Orosius, 5,

4, 13-21;

(7) Oxyr. 202-205 and Per. Liv. 55;

(8) Oxyr. 213-214 and 216-217, Per. Liv. 55, and Orosius, 5, 4,

18.

Most of these cases are individual errors, but Oxyr. and Obsequens agree in error once, Oxyr. and Per. Liv. probably

twice. Both these errors and some of the others involve combination of events, both being given under the one which seems to the author more important. A few similar chronological errors had been previously noted (cf. Quellen-contamination, p. 44).

This is certainly interesting information, but as we are not positive that a non-annalistic Epitome occurred between Livy and these latest descendants, such data must be interpreted in the light of the information we now possess. Let us first note a few well-established points in regard to the Epitome:

(1) The consuls' names of Livy were given to later writers through the Epitome.

(2) These names were given in the ablative, therefore at the beginning of the years. This was proved by Mommsen by reference to Cassiodorus and Obsequens, but is supported by the other descendants; see further below.

(3) The Epitomator Livii combined material from other sources with his excerpts from Livy.

(4) The division into books was preserved in the Epitome. The mistakes in order noted by Kornemann and others find their adequate explanation in these facts, and themselves help to confirm this character of the Epitome. No one has yet explained how the Livian consular list could get to the later writers except through the Epitome, nor how the Epitome could transmit it intelligibly, unless it preserved the annalistic form. In spite of this, however, changes in order and time might creep in, owing to the fact that the author was combining material from other sources, was seeking to condense to the utmost, and above all was preserving the book division. This last had a tendency to make the author regard each book as a unit rather than each consulship. If we add to this the thorough acquaintance with Livy and the popular traditions of history possessed by the writer (cf. U. of M. Studies, I, p. 254), we shall be willing to admit that in writing his Epitome of each book, he would not have copied blindly from an unhandy roll, but trusted to a vigorous memory, quickened by repeated reading and perhaps aided by some few brief notes. The mistakes in chronology and

changes in order, and particularly the combination of related events, above noted, form a most convincing proof that such was his method of work.

More important as indication of an intermediate source is Kornemann's statement (p. 74) that of the few historical notices in Obsequens, three appear elsewhere only in Oxyr., though with different wording. This loses some of its force, when we note that in the brief parallel portion of Obsequens, there are twelve historical notices, and that of Kornemann's three, one (burning of the sacrarium Opis) was also a prodigy, and that another, Obsequens, 22, states the opposite from Oxyr. 167 in point of fact. There is left one historical notice found only in Oxyr. 132-134 and Obsequens, 20. Of the remaining historical notices in Obsequens, four are found in Oxyr. (one under different consuls) as well as in other epitomators, while six are not found in Oxyr. There is also a single notable agreement found only in Livy, 39, 22, 1, Cassiodorus, 568, and Oxyr. 42. These cases are most easily explained if there was a much abbreviated form of the Epitome used by Oxyr. and Obsequens, though they hardly prove the existence of such a work. Cassiodorus shows less clearly the same close connection.

The language of the Oxyrhynchus Epitome has been fully treated by Moore and Kornemann 1; they note the extreme brevity, often reduced to substantive and modifiers, excess of participles, and poverty of expression. The same characteristics occur in Per. Liv. I a and, in a less degree, in Obsequens. Moore even tries to discover the same tendencies in Cassiodorus. I fail to see them. Of the forty-three passages found in the Livian portion of Cassiodorus, twelve contain complex or compound sentences, and only seven show the pure participial construction, while some of these are known. to have been taken from Eutropius. The style of Cassiodorus is the same even in the last portion of his work, where he was supposedly original.

Of the special words or expressions found often in Oxyr., Per. Liv. I a, or Obsequens, the following are distinctive:

1 Cf. also Wölfflin, Archiv, XIV, p. 221 ff.

(1) devinco found rarely in Livy and only once in Per. Liv. (49);

(2) vexo, only once in Per. Liv. (32);

(3) prospere (dubie or varie) dimicatum (pugnatum); the Periochae have regularly combinations with male, dubio eventu, or feliciter and the active;

(4) re bene gesta; Per. Liv. has res prospere gestas as object.

The other words treated, caesus, subactus, and occisus, occur with special frequency in Per. Liv., usually with a copula. Clades accepta appears in Per. Liv. 67; also present participles in nom. sing. are fairly common, even those noted by Kornemann both occurring, decedens in Per. 30 and flens in Per. 89. Kornemann laid particular stress on the frequency of vastare and especially that it takes as object names of peoples in Oxyr. (not in Obsequens). Per. Liv. also uses the word frequently and once at least (Per. 47) with the name of a people as object. All these expressions point to the influence of the original Epitome, though favorite expressions doubtless increased in the later Mss or abridgments of the work (cf. U. of M. Studies, I, p. 188 ff.). Further we must admit that even these later historians sometimes show characteristics of their own time.

Omitting, however, everything questionable, we have still several agreements in language and style between Oxyr., Per. I a, and Obsequens, on the basis of which I believe we are warranted in assuming a fairly close relationship. This position is supported by the agreements between Oxyr. and Obsequens in historical statements. Cassiodorus does not seem to be as closely related, but this may be due to the character of his work.

In spite of the similarity in language Per. Liv. I a is not a fragment of the complete original of Oxyr. Kornemann (p. 78) has made this clear by noting the greater brevity of Per. I a. A still stronger point is that in spite of the extremely chronological character of Oxyr., in which the consuls' names project from the text, so that the years can be easily counted, Per. I a does not give the length of