which mark the 'Vatican' family of MSS. hereafter described. Perhaps the reading of 'L' in Inf. xxxiii. 54 may be similarly accounted for :- Infin che l'altro giorno sol nel mondo uscio, where probably "giorno," or perhaps "giorno 3," may have appeared as a marginal note. It is also just possible that the word giorno in the line before may have caught the copyist's eye. As it stands however it has the appearance of what is known in MSS. of the N. T. and elsewhere as a conflate reading' [see Westcott and Hort's Introduction &c., pp. 94 &c.], i.e. a reading formed by the combination of readings intended to be alternative. The metre of course in the case of Dante, though not always (as we have seen above in note 21) a sufficient safeguard, prevents the phenomenon of 'conflation' being at all common". 10. A small class perhaps remains, though its existence can never be more than problematical, in which changes may possibly have been introduced by the author himself in the way of revision, and so two varr. lectt. would both be genuine. In later times we are all familiar with this practice of authors, as well as with the fact that these 'second thoughts' are often 'not the best'. But there is so much uncertainty in this case whether Dante's work was published or put forth at all during his lifetime, or whether it was brought out afterwards"; to say 22 The following, however, seem (if the recorded variants are considered) to be instances of "conflation" : Inf. xii. 16. Lo savio mio Duca in ver lui (in ‘h'). " xvi. 14. xviii. 12. xxx. 18. disse, Ora aspetta, Disse a costor. La parte dove son rende figura sicura (!) (Bat. 55). Purg. xxi. 25. perchè colei (combined from perchè lei and per colei). Par. ii. 124. Riguarda omai a me. trenta tre fiate (found in '68'—trenta and tre being generally alternative readings). ben del ineterna (!) stampa. xxviii. 23. Alo al cinger. In Inf. i. 28 (q. v.) several instances will be found, in spite of redundant syllables, of the combination of alternative readings. Boccaccio's tradition, whatever it be worth, or whatever basis of fact it may represent, of the last thirteen Cantos of the Paradiso being missing for years, and the legendary method by which they were discovered, is well known. (See Ugo Foscolo's Discorso, p. 43, &c.) nothing of the uncertainty of what is meant by 'publication' (the phrase being in any case to some extent anachronistic and misleading), that the suggestion of such equally divided authority for different readings must remain a merely possible though not uninteresting speculation. 'Publication' probably then consisted in little more than forwarding a copy to a patron, or allowing a work to be read by friends. In any case facilities for constant retouching and altering would be greater than after 'publication' in our modern sense, and would therefore be likely to be much more common. Ugo Foscolo contends for its application (Disc. p. 424, &c.) to the well-known variants, 'moto' and 'mondo' in Inf. ii. 60. My friend Mr A. J. Butler, who has lately translated the Purg. and Par., has come to a similar conclusion from the phenomena of the variants in several places in those Cantiche. In this way, as he suggests, "two readings would be equally genuine, but representing a different stage in the polishing of the poem ". It should not be forgotten that the character and habits of copyists are widely different. Some were grossly ignorant and careless; others faithfully and mechanically reproduced what lay before them, including, it may be, obvious errors; others allowed themselves the utmost liberty of alterations and imaginary improvements". In estimating the critical authority of any given MS., secondary readings of this latter class should be carefully 'discounted', as the phrase is. They must as far as possible be kept distinct from those due to a corrupt and radically inferior original text; since a MS., though abounding in errors due to a careless or emendating copyist, if it give evidence of a good foundation-text, may yet 24 Again we may be allowed to quote Westcott and Hort as illustrating the application of the above principles in reference to another class of MSS. They speak (Introd. p. 35) of the necessity of "distinguishing the occasional errors of 'good' documents from the sound parts of their text "...." One MS. will transmit a substantially pure text disfigured by the blunders of a careless scribe, another will reproduce a deeply adulterated text with smooth faultlessness. It therefore becomes necessary in the case of important MSS. to observe and discriminate the classes of clerical [and we might add other] errors by which their proper texts are severally disguised" (p. 36). 25 See an instance of this in an account of the MS. denoted as '29'. be employed as an authority. Thus each MS. has, though in very different proportions, its family and its individual peculiarities. The problem in the case of each MS. is like 'a function of two variables', (a) the character of its exemplar text; and (8) the skill or capacity of its scribe, who often combines the functions of copyist, critic, and editor, as has been before observed (sup. p. viii.). Thus the Bodleian MS. 'I' affords an instance of a text unusually full of late and inferior alterations grafted upon a text fundamentally sound and good. The related MS. 'F' is another instance, but it is far more audacious in the number and freedom of its alterations. The MS. '' presents a somewhat similar case. Even 'L' appears to have a good foundation-text (occasionally in noticeable agreement with the important Egerton MS. [denoted by 'a'] of the British Museum), but it is disfigured and rendered nearly useless by the extraordinary carelessness and want of intelligence displayed by the scribe on almost every page. On the other hand 'H', though a late MS., is an example of a fairly good text, copied by a scrupulously careful scribe. The condition of a good foundation-text can only be ascertained by the examination of carefully selected test-passages, in which the alternative readings can be pronounced on recognized critical principles as original or derivative. This process we shall describe more at length presently. The condition of a text tampered with anew by the writer of any given MS., would be indicated by the relative number of rare or peculiar readings exhibited by the MS.26 A list of such 'peculiar readings' (so far as my knowledge extends) is for this reason given in the case of most of the MSS. that I have examined. 26 They are for instance exceptionally numerous in the MSS. F, 6, j, and others. It is remarkable how many purely personal and apparently unique varr. lectt. occur in almost all Mss. No doubt a wider collation would diminish the number so registered, but the results so far would lead one to believe that many would still remain in all cases. The rarity of such passages in some MSS. is of course one note of excellence, or at least of fidelity to the exemplar, whatever value that may have had. It is also very singular how extremely rare it is ever to find two MSS. that can be considered identical, even when we happen to know of more than one MS. having been written by the same scribe. Practically we have identity, I think, in the case of those marked ↓ and ∞, and 56 and 57, but I know of no others. (See further on this point Witte's Prol. p. xiv.) A more extended acquaintance with MSS. would probably shew that they are not all 'peculiar', but they are all, I think, at least uncommon, and in that case they have a singular value and interest of another kind, viz. as establishing a prima facie relationship between any two or more MSS. in which they are found: and thereby advancing us a step towards that most important, but at present rather distant, result, the classification of MSS. in 'families'. The common possession by two MSS. of only one or two striking and unusual variants has often proved in my experience very fruitful in suggesting further comparison by which sometimes an unmistakable relationship has been established. They are like what Bacon calls 'instances of the sign-post'. In this way it becomes important to register even mere blunders ('errori madornali' as Witte calls them), since their precise reproduction often affords, in the very proportion of their unreasonableness, convincing proof of connexion between the MSS. in which they are found. Thus then, to sum up, we have, as Francesco Palermo has tersely put it, "ignoranza, arbitrio, forza di dialetti," all in active operation upon the text for some two centuries. After that, it was subjected, with the invention of printing, to a fresh visitation of editors and grammatical purists. So that Vellutello, writing in 1544, complains that the text, both in MSS. and printed Editions, was 'incorrettissimo'; and declares that Dante, could he return to life, would be the first to recognize the necessity for its thorough reconstruction. This being so, by what means can we hope to recover the original text, or at least to make some approach to it? Surely by retracing, as far as possible, the various paths through which the errors have been introduced. But let us first clear the ground by shewing the futility of other methods that seem at first sight to offer a shorter, or more obvious means to this end, and by explaining what has already been attempted in these several directions. (1) It follows from what has been said that mere antiquity of a MS. is no guarantee of the purity of its text". Of It will be observed that the list of the best texts as selected by Dr Witte and as judged by my own tests, includes many of the 15th century MSS., and some even beyond the middle of that century. course there is, cæteris paribus, more chance of such purity as we draw nearer the fountain head, since every tributary stream has brought in some fresh elements of corruption. But, as we have already stated, a very short examination of the best and oldest MSS. in existence reveals readings which on critical principles are undoubtedly secondary and derivative. Therefore we cannot pin our faith to any one MS. now existing, however ancient; they are all echoes, some more, some less, distinct; not original voices. Our only hope therefore lies in a comparative study of MSS. At the same time, many editors have adopted the plan of reprinting the text of some one MS. which on various grounds they have exalted to preeminent authority, e.g. the Codice Bartoliniano of Udine (U), by Viviani"; the Codice Estense of Modena (M), by Montfaucon, Parenti and others; the Codice di Santa Croce (A), to some extent by Dionisi30; the Codice Vaticano, No. 3199 (B), by Fantoni", and so forth. I have said 'to some extent' when speaking of Dionisi, since while practically almost always falling back on the Codice di Santa Croce, he fully admits, as in the following passage (Anedd. v. p. 3), that no one MS. has preserved the true original, but that "they have all gone out of the way",-"Io cercava un Testo di buona mano e fedele che mi fosse d' esemplare alla ristampa della Divina Commedia da tutti i Letterati riconosciuta omai necessaria. In tutta Fiorenza non m' è riuscito di ritrovarlo, sicchè degno d' alcuna scusa mi sembra... il famoso Bastian de' Rossi, il quale volando (m' immagino) sovra le private e pubbliche Biblioteche, come la colomba di Noè sovra l' acque, nè trovando ove fermar il piede, si ridusse a posarsi su d' una Stampa," and of this he adds, "Qual ella sia, parole non ci appulcro." (2) It would seem superfluous (but for the warning of 28 See Witte, Prol. pp. xxxvii-xlii, and Foscolo, Discorso, pp. 16 &c. 115 &c. * See Witte, Prol. p. liii, and Foscolo, Discorso, p. 135. Parenti says of this MS., "che solo vale per cento." See Dion. Anedd. 11., Witte, Prol. p. xxxiv (init.). I venture to think that, in spite of its acknowledged excellence, Witte himself sometimes adopts an inferior reading in deference to it. 31 In his Ed. of 1820. See Batines, 1. p. 148, 11. p. 167. D. I. d |