Robbins put P in the position of Π in this last stemma, but on the assumption that it did not contain the indices. That is not true of Π.
[Further consideration of the external history of P, Π, and B] Still further evidence is supplied by the external history of our manuscripts. B was at Beauvais at the end of the twelfth or the beginning of the thirteenth century, as we have seen.[43] Whatever the uncertainties as to its origin, any palaeographer would agree that it could hardly have been written before the middle of the ninth century or after the middle of the tenth. It was undoubtedly produced in France, as was F, its sister manuscript. The presumption is that Π1, the copy intervening between Π and B, was also French, and that Π was in France when the copy was made from it. Merrill, for what reason I fail to see, suggested that the original of BF might be “Lombardic,” written in North Italy.[44] An extraneous origin of this sort must be proved from the character of the errors, such as spellings and the false resolution of abbreviations, made by BF. If no such signs can be adduced, it is natural to suppose that Π1 was of the same nationality and general tendencies as its copies B and F. This consideration helps out the possible evidence furnished by the scribbling in a hand of the Carolingian variety on fol. 53v;[45] we may now be more confident that it is French rather than Italian. But whatever the history of our book in the early Middle Ages, in the fifteenth century it was surely near Meaux, which is not far from Paris—about as far to the east as Beauvais is to the north. Now, granted for a moment that the last of our stemmata is correct, X, from which Π and B descend, being earlier than Π, must have been a manuscript in majuscules, written in Italy, since that is unquestionably the provenience of Π. There were, then, by this supposition, two ancient majuscule manuscripts of the Letters, most closely related in text—veritable twins, indeed—that travelled from Italy to France. One (X1) had arrived in the early Middle Ages and is the parent of B and F; the other (Π) was probably there in the early Middle Ages, and surely was there in the fifteenth century. We can not deny this possibility, but, on the principle melius est per unum fieri quam per plura, we must not adopt it unless driven to it. The history of the transmission of Classical texts in the Carolingian period is against such a supposition.[46] Not many books of the age and quality of Π were floating about in France in the ninth century. There is nothing in the evidence presented by Π and B that drives us to assume the presence of two such codices. There is nothing in this evidence that does not fit the simpler supposition that BF descend directly from Π. The burden of proof would appear to rest on those who assert the contrary. Π, therefore, if the ancestor of B, contained at least as much as we find today in B. Some ancestor of B had all ten books. Aldus, whose text is closely related to BF, got all ten books from a very ancient manuscript that came down from Paris. Our simpler stemma indicates the presence of one rather than more than one such manuscript in the vicinity of Paris in the ninth or the tenth century and again in the fifteenth. This line of argument, which presents not a mathematically absolute demonstration but at least a highly probable concatenation of facts and deductions, warrants the assumption, to be used at any rate as a working hypothesis, that Π is a fragment of the lost Parisinus which contained all the books of Pliny’s Letters.
Our stemma, then, becomes,
P (the whole manuscript), of which Π is a part.
[Evidence from the portions of BF outside the text of Π] We may corroborate this reasoning by evidence drawn from the portions of BF outside the text of Π. We note, above all, a number of omissions in BF that indicate the length of line in some manuscript from which they descend. This length of line is precisely what we find in Π. Our fragment has lines containing from 23 to 33 letters, very rarely 23, 24, or 33, and most frequently from 27 to 30, the average being 28.4. These figures tally closely with those given by Professor A. C. Clark[47] for the Vindobonensis of Livy, a codex not far removed in date from Π. Supposing that Π is a typical section of P—and after Professor Clark’s studies[48] we may more confidently assume that it is—P had the same length of line. The important cases of omission are as follows:
32, 19 atque etiam invisus virtutibus fuerat evasit, reliquit incolumen optimum atque] etiam—atque om. BF. P would have the abbreviation for bus in virtutibus and for que in atque. There would thus be in all 61 letters and dots, or two lines, arranged about as follows:
| ATQ· | |
| ETIAMINUISUSUIRTUTIB·FUERATEUA | (30) |
| SITRELIQUITINCOLUMEMOPTIMUMATQ· | (31) |
The scribe could easily catch at the second ATQ· after writing the first. It will be at once objected that the repeated ATQ· might have occasioned the mistake, whatever the length of the line. Thus in 82, 2 (aegrotabat Caecina Paetus, maritus eius, aegrotabat] Caecina—aegrotabat om. BF), the omitted portion comprises 34 letters—a bit too long, perhaps, for a line of P. The following instances, however, can not be thus disposed of.
94, 10 alia quamquam dignitate propemodum paria] quamquam—paria (32 letters) om. BF. Cetera and paria, to be sure, offer a mild case of homoioteleuta, but not powerful enough to occasion an omission unless the words happened to stand at the ends of lines, as they might well have done in P. As the line occurs near the beginning of a letter, we may verify our conjecture by plotting the opening lines. The address, as in Π, would occupy a line. Then, allowing for contractions in rebus (18) and quoque (19) and reading cum (Class I) for quod (18), cetera (Class I) for alia (20), we can arrange the 236 letters in 8 lines, with an average of 29.5 letters in a line.