II. A STUDY OF THE LEYDEN MS OF NONIUS MARCELLUS. Following the example of Mr. T. W. Allen and others, who have recently published careful studies of the chief MSS of Plato, Sophocles and Aristophanes, I will in this article attempt a detailed examination of the 'codex optimus' (L) of Nonius Marcellus, with the object of gleaning some information about the archetype and the history of the transmission of the text. The Leyden MS (Voss. Lat., fol. 73), of the ninth century, written in Caroline minuscules on 253 leaves, with two columns (each of 22 lines) to the page, comes from the Monastery of St. Martin at Tours and is one of the MSS selected by M. Delisle (Mém. Acad. Inscr. XXXII 29 sqq.) as a specimen of the calligraphy of Tours-that is to say, of the best work produced in the best scriptorium of all Europe. The care bestowed on this copy of the Compendiosa Doctrina of Nonius Marcellus is seen in the fact that the whole work, from beginning to end, has been revised by two correctors (L2 and L3), who have not been content with punctuation and emendation of the text and with correction of the spelling. In the division of words between the lines they have interfered whenever Priscian's rules of syllable-division were broken. Thus on fol. 11 r. i recta has been changed by L' to recta and on fol. 14 r. i pub|lica by L3 to publica. And even, a rare example of careful calligraphy, the correct division of consonant-groups in words at other parts of the line has been indicated by subscript commas throughout the volume, e. g. (fol. 1 r.) inhone,stis; di,ctis; indi,scretis; significatione; di,ctum ; o,mnibus; sene,clutem, etc. The scribes too have done their work well. All editors allow L'-i. e. the uncorrected transcript-to be the closest reproduction of the lost archetype. The only MS that can stand beside it is the Geneva MS (Gen.), which contains only book IV, and which belongs to the end of the ninth century. The consensus of L' and Gen.' gives us unmistakably the actual text of the original, with its barbarous spellings (e. g. 382 M(ercier) 24 Hecyra] hequira L'Gen.'; 241 M. 33-34 absinti... acerbum] absenthi... acervum L'Gen.; 246 M. 31 Zephyrumque] zeferumque L'Gen.') and illiterate word-division (e. g. 258 M. 7 satin astu] saginas tu), the unemended, or not fully emended, form in which the text of our author passed from the Dark Ages into the hands of Carolingian scholars. L' and L3 aim at adapting this to the standard of correct Latinity, but in so doing often suppress a genuine form or its trace, e. g. 443. 23 nomen habet] nominavet L', nominavit L corr. It is therefore to L' (and in book IV to Gen.' also) that we must look for light on the nature and composition of the archetype of our MSS. That all MSS of Nonius come from one archetype has long been recognized from their transposition of a passage of book IV (406 M. 12-409 M. 15) to near the beginning of book I (3 M. 13). This passage of book IV appears to have filled a single leaf of the archetype. The leaf became loose and dropped out, and, instead of being put back in its proper place, was slipped in after the first leaf of the whole work. We can thus estimate the size of a page of our archetype as about a page and a half of Mercier's edition. Now we get a clue to the size of the page of the immediate original of L from a mistake by the scribe of L at 379 M. 17. After the words iam tum religio there follow 380 M. 41 sqq. Verg. lib. XI multa dies etc. The most natural explanation of this mistake, a mistake not shared by the other MSS, is that the scribe had 'skipped' a page (or leaf) of his original. The amount omitted corresponds to what we have found to be the content of a page of the archetype. This suggests at least the possibility of our archetype, which had a leaf of book IV loose, having been also the immediate original from which L was transcribed. It is well known that the pagination of an archetype is often reproduced in a copy. Thus the Pithoean MS (P) of Juvenal is assumed to reproduce the pagination of the archetype of all the minuscule MSS, because XVI 60, the line immediately preceding the lacuna which characterized that archetype, is also in P the last line of the last page. This practice was found convenient when the task of transcription was distributed among several 1It is a strong argument for the genuineness of the passage discovered by Mr. Winstedt in a Bodleian MS (Class. Rev. XIII 201) that its content suits the theory of its absence from our MSS being due to the accidental loss of a leaf from the archetype and not to any doubt about the authenticity of the lines. scribes who worked simultaneously at different parts of the text. Suppose the original, of which a copy was desired, consisted of 80 pages (i. e. of 40 leaves or folia, i. e. of 5 quaternions), the first and fourth quaternions might be assigned to one monk, the second and fifth to another, the third to another; and the most certain way of ensuring that each transcriber should not find himself inconvenienced by having too much or too little parchment for his task, would be to make the three transcribe each and every page of their original exactly on one page of their transcript. This practice, more available for transcribers of poetry than of prose, has clearly not been followed by the scribe of L, if L was transcribed immediately from the archetype; for the transposed passage from book IV occupies in L not a single leaf, but three pages and one column, and the preceding part of book I, along with the index of contents and title-heading, takes up the same amount. Fol. 2 r. i ends pausimachomum IN and fol. 3 v. ii ends nascitur leat. Indeed, the calligraphic nature of L, with its large, uncramped, regular script, is inconsistent with a slavish reproduction of the form of, let us say, a Merovingian original. But that the transcription of the various parts of the original by the scribes of the Leyden MS was simultaneous there is some indication. Just before fol. 147 r., where a new hand appears, the writing of the concluding portion of the previous gathering is spaced out and straggling, so as to cover as much ground and leave as little of the page blank as possible. The Compendiosa Doctrina of Nonius is, in accordance with the fashion that prevailed in works of this class in ancient times, divided into twenty books. But several of these books (or rather chapters) are of very limited extent, and one (the fourth) is of exceptional length. If divided according to bulk, the work falls naturally into three parts, the first containing books I-III, the second, book IV; the third, books V-XX. And a division of this kind, possibly due to the mere breaking up of an archetype into these three sections, is traceable in our MSS; for some (e. g. the Geneva MS) contain only book IV, others (e. g. the Florence MS) only books I-III; while others that contain the whole are clearly transcripts from different originals in these three portions (e. g. the Harleian (H) is in books I-III a transcript of the Florence MS; in book IV, of the Geneva MS; in books V-XX, of some lost original), or even (as in the case of the Paris Nonius) are, in reality, mere accidental combinations of originally distinct MSS. It is one of the merits of the Leyden Codex that it is in all probability a whole transcript of a whole original. The transcription has been apportioned among three scribes and in this fashion: 1st scribe: foll. 1–94 r. (= pp. 1–170 M. 22 M. Tullius), 167 r.– 252 (= pp. 365 M. 18-557 M.). 2d scribe: foll. 94 v.-146 v. (= 170 M. 22-314 M. 14 multum). 3d scribe: foll. 147 r.—166 v. (= 314 M. 14—364 M. 18 conpertum est). In other words, the first scribe wrote book I and nearly the whole of book II (occupying some twelve quaternions), the second finished book II and wrote book III and the first half of book IV, the third wrote the third quarter of book IV; the first scribe then finished the volume. The lion's share of the work has thus been effected by the first scribe.' He made a separate numbering of the quaternions used by him in the second part of his task, but his numbers i, ii, iii, etc., have been altered later (by L'?), so that i becomes xxii, ii becomes xxiii, iii becomes xxiv, and so on. Near the end of the first quaternion in this second half of his labours, he omitted accidentally, through homoeoteleuton, a long passage of his original (379. 16 Virg.—380. 41 libro XI). To supply the deficiency a broad sheet (i. e. two leaves) was utilized, which had been discarded from some transcript of Priscian's Institutes, and which bore on one leaf the title-heading in gold letters: PRISCIANVS GRAMMATICVS | CVM OMNIS. On its other leaf the omitted passage of Nonius was written (in a new handwriting), and inserted in the gathering, so that the quaternion becomes a quinion, with 10 leaves instead of 8, the blank leaf being fol. 168 and the written leaf fol. 175 of our MS. I have been unable to ascertain whether this Priscian MS is still in existence. For the sake of completeness it may be as well to give here an account of the arrangement of our MS in quaternions, although this was a mere affair of the supply of material to the scribes, and does not throw light on the nature of the archetype. (I follow the account entered by the Leyden librarian on the fly-leaf) :Foll. 1-8 quaternio, 9-14 ternio, 'I think that 364 M. 18-366 M. 14 adfigebatur, occupying four columns, i. e. one leaf, is in the handwriting of the first scribe. But it may be in a fourth handwriting. Foll. 15-118 quaterniones, 119-129 quinio+1 fol., 130-137 quaternio, 138-146 quaternio, cui unum folium (139) additum est, scilicet ut textus congrueret cum sequentis quaternionis contextu, qui iam conscribi coeptus erat, 147-154 quaternio, 155-164 quinio, 165-166 duo folia, 177 sqq. quaterniones, More important for us is an examination of the procedure followed in the correction of the transcript. There were, as has been mentioned, two separate correctors whose services were given to our MS. In Prof. Lucian Mueller's critical apparatus they are merged under the symbol L'; and even corrections by the scribe himself are often included under this designation. Where the correction consists of a single stroke or dot or an erasure, it is often hardly possible to assign it with certainty to L'or L'or L3. But in the majority of cases we can distinguish fairly enough between the two correctors and keep their emendations separate from the mere correction by the scribe at the moment of transcription. The corrector, whom I call L', has left us a good specimen of his handwriting on fol. 181 r. and fol. 220 V. His revision was prior to that of L3, for L3 often confirms (by a dot or the like) the corrections of L2 (e. g. 298 M. 8 inplere L', implere L', confirmed by L3), and in the passage added by L' on fol. 188 r. ii (p. 405 M. 29) L3 changes ho to hoc. Both are later than the rubricator, as we see from fol. 178 v., where the word Aequales was written quale by the scribe, with space left for the initial. The rubricator has supplied an initial E (i. e. Equale); but L' has put A before this, so as to make the word Aequale. L3 has added a final s, producing Aequales. At 225 M. 32 L3 has stroked out the rubricated initial S of segetem and has replaced it with a small s. A good specimen of the handwriting of L' is seen in his lengthy addition in the upper margin of fol. 202 v. A corrector would, of course, make many emendations by his own impulse, but in the main would follow some text of the |