the authority of M. L. Delisle, qui est si profondément versé dans la connaisance des manuscrits des bibliothèques de Paris et des départments. (b) The writing a bien les caractères de l'écriture de la fin du rii siècle. (c) T presents in its text a great improvement on M, and there was not enough scholarship at the end of the twelfth century to make these improvements by the exercise of conjecture. II. Thurot holds that T comes from the same archetype (A) as M, but is independent of M. He points especially to these passages to prove the independence of M and T:

a) Fam. iv. 6, 3:

maior mihi vatio mihi adferre nulla potest quam coniunctio consuctudinis sermonumque nostrorum M.

maius mihi solatium afferre ratio nulla potest, &c. T.

The usual medela of this passage is to read levatio for ratio, and adferri for adferre, omitting one mihi. It seems to me that T's reading is the work of an editor who saw ratio in ratio, and hence was forced to supply solatium as an object of adferre, and to correct to maius.

B) Fam. vi. 1, 6:

non debes . . . dubitare quin aut aliqua republica sis futurus qui esse debes, aut perdita non afflictiore condicione quam ceteri M.

Here for aliqua T gives recuperata, an obvious conjecture to supply a more regular antitheton to perdita, but a conjecture which materially impairs the force of the passage.

(7) Fam. iv. 12, 2:

.

Postumius . . . mihi nuntiavit, M. Marcellum . . . pugione

percussum esse . . . se a Marcello ad me missum esse, qui haec nuntiaret et rogaret uti medicos coegi M.

et rogaret utrum medicos ei mitterem. Itaque medicos coegi.

This seems to me really to point to a different origin for M and T. The scribe of M, through a common parablepsy, left out the words between the first and second medicos. Perhaps there is no stronger proof of independence between two mss than when one supplies a lacuna in the other under these circumstances. We can see how the words between medicos . . . medicos fell out in M, and it is immensely improbable that they should have been inserted erroneously or by conjecture in T.

I add three other places, which seem to show that T is independent of M, though Thurot does not use them for this pur

pose:

(a) Fam. iv. 4, 5:

de reliquis nihil melius ipso est, ceteri et cetera eiusmodi ut, &c. M.

de reliquis nihil melius ipso est Caesare, cetera, &c. T.

Whether Caesare is a gloss on ipso, which has crept into the text, or is the real reading of A which M has corrupted to ceteri et, at all events it is not a conjectural emendation of T.

(B) Fam. iv. 5, 4:

de imperio propter tanta deminutio facta est, M.

de imperio p. r. T.

The conjecture of Orelli, populi Romani, is generally accepted; p. r. is an abbreviation of propter; in common words such as prepositions, the scribes often wrote only the first and last letters. T preserved p. r. of A, which also stands for populi Romani.

(7) Fam. vi. 8, 1:

huic meae rogationi potius non responderent M.
huic meae rationi potius quam rationi T.

The reading of T gives a strong confirmation to the SchützWesenberg conjecture, efflagitationi potius quam rogationi.

The remarkable agreemeut between T and H has been referred to above, and it has been pointed out that they generally agree in an error against M.

The following passages bring out well the character of M and T respectively:

(a) Fam. v. 15, 2:

quod vinculum quas id est nostrae conjunctionis, M.

quod vinculum quasi est T.

The reading of M points to the certain conjecture of Wes.: quod viculum, quaeso, deest? The reading of T leads us away from it by a bad attempt to patch up the sense.

(b) v. 14, 2:

qua se levare M.

quas elevare T.

The reading of T is right, but the reading of M gives the right letters wrongly divided. Its very unintelligence is the guaranty of its good faith.

The same remark may be made on v. 12, 2, seiungere se quidem M: seiungeres equidem T.

hic tuae abesse urbe miraris in qua domus nihil delectare possit M. hic tu me abesse, &c. T.

But M points to the true reading, hic tu ea me abesse, &c. The word ea is absolutely necessary, to account for the subjunctive possit: cp. v. 17, 3, ea te republica carere in qua neminem . . . res ulla delectet.

(d) v. 1, 1:

(ď)

me desertum a quibus minime conveniebat M.

a quo quidem T.

The plural is thoroughly characteristic of the letters. It might be called the plural of caution.

Nere makes no sense, pointing, however, to ne vere, the probable reading of A; neque makes a sort of meaning, but points to nothing, and lulls the reader into false security.

(ƒ) v. 6, 13:

quidquid valebo . . . valebo tibi M.

quidquid valebo . . . conciliabo tibi T.

We have seen above, p. 62 (i), that the cognate acc. illustrated by quidquid valebo is very characteristic of the letters.

(g) i. 9, 24:

quod deque fratris negotio M.

quodque de fratris negotio T.

The true reading is quod de Q. Fratris negotio.

In the following places T is right, but probably through conjecture:

(a") ii. 8, 2:

quare da te homini complectetur mihi crede M.
quare da te homini complectendum mihi crede T.

G

Probably A had quare da te hominini complectendum. Mihi crede, &c.

(b") v. 21, 5:

tibique persuade praeter culpam et peccatum. . . homini accidere nihil posse quod sit honorabile aut pertimescendum M.

quod sit inhonorabile T.

T is probably right. The vulgate horribile is certainly wrong, as being stronger than the word aut pertimescendum, which follows. Inhonorabile is aπаž sip. So in v. 15, 2, T has permaxime.

T seems right; but the reading of A was not commonitorum, as Thurot suggests, but quod monitorum?

In i. 9, 17, the whole passage from idque non solum par pro pari, § 19, is transposed to ii. 10, 2, where it comes between qui mons and mihi cum Bibulo. This transposition is a strong proof that T is independent both of M and of H.

The net result of this examination seems to be that T is certainly independent of M, and on the whole presents a far more correct text. That it is of the twelfth century seems hardly to be established. As to the relative value of M, H, and T, as sources of knowledge of the letters, I have already indicated my opinion. I further refer the reader to what I have written on the subject of A and F in the next few pages. There I have explained fully what in my judgment makes a ms valuable; and what I have said to some extent applies to H and T as well as to A and F, except that we know so very much less about A and F, their very existence being problematical.

(4). Hofmann claims an independent place for P, a Codex Parisinus, including from Fam. i. to impediendi moram, Fam. viii. 8,6; and the same claim is made by some editors for one page of a Turin palimpsest, which includes Fam. vi. 9 and part of 10. Orelli, while classing the Wolfenbüttel ms with the other codices ultimately traceable to M, has remarked how desirable would be a thorough collation of the codez Guelferbytanus. R. Heine (Jahn's Jahrb., 1878, Seite 784) has examined the ms, and pronounces it to

belong to the fifteenth century, and to have no value independent of M.

(5). Very important in the criticism of the letters are the Editio Neapolitana (1474), and the editions of Victorius, published—one in Venice, 1536, another in Florence, 1558-as well as an edition preserved in the library at Zurich, of which the time and place of publication are unknown, the last leaf of the copy being lost. This is called A by Orelli, i.e. Editio Antiquissima, but must not be confounded with A, the supposed archetype of M, H, and T; nor with A, the Codex Antonianus, containing the letters to Atticus, Quintus, and Brutus, of which I shall have presently to treat. In this edition A will mean the Codex Antonianus. The other two are very seldom mentioned, and when they are mentioned each will be given its full title.

For the mss of the Commentariolum Petitionis (Ep. xii.), see Appendix C.

For the letters to Atticus, Quintus, and Brutus, we have the following authorities :

(6). M, the Medicean. This ms was discovered by Petrarch, perhaps at Verona, about 1345. The copy which we possess of it was probably, as I have already said, procured by Pasquino of Milan for Coluccio Salutato of Florence. In two letters of the present instalment, from the word reperire, Att. i. 18, 1 (Ep. xxiv.) to risus est et talis, nearly the last words of Att. i. 19 (Ep. xxv), we lose the guidance of M, some leaves of the ms having perished. But for Att. i. 19 we have the assistance of a Codex Poggianus in the Medicean Library, collated by Th. Mommsen.

(7). C. This is a name given to a ms of which we have no knowledge except from the marginal notes in Cratander's edition of 1528, which, however, show it to have been independent of M.

(8). W. Some leaves of a ms of these letters are preserved at Munich and others at Wurzburg (whence the leaves at both places are designated W): these contain portions of books xi. and xii. They coincide closely with the marginal readings in Cratander's edition, and are by some supposed to have formed a part of C.

(9). Z. The Codex Tornaesianus, now lost, our knowledge of which is derived from the notes of Lambinus and a few quotations by Turnebus.