Manuscripts and Editing
Smith, A. Mark, Transactions of the American Philosophical Society
In the previous edition of books 1-3 of Alhacen's De aspectibus I provided a detailed account of the available manuscripts and outlined my procedures for selecting particular representatives from among them for collation in the critical text. I concluded that the seventeen complete or virtually complete manuscripts could be broken into three family groups, the first consisting of six members (F, P1, Va, V2, L2, and S), the second of four (Er, C1, O, and M), and the last of seven (E, P3, P2, L3, C2, L1, and V1).1 I also concluded that the first family lies closest to the Urtext and, furthermore, that among its members F is closest to the family progenitor. F was therefore the logical candidate to represent this family in the critical text, but since it lacks a considerable portion of books 1 and 2,1 was forced to fall back on its nearest relative, P1. For the second family the choice was less clear, but I eventually decided on Er, thus bringing the total for collation to two. The third family was even more problematic, but I was finally drawn to E, P3, and L3 as the most suitable choices for collation. To the resulting list of five manuscripts I added S and C1, because both seemed to form inter-family links, S between the first and second families, C1 between the second and third. Altogether, then, I based my critical text on seven manuscripts-Pl, S, E, P3, L3, Er, and C1-using O for occasional cross checking when necessary.
Over the course of editing the text on the basis of these seven manuscripts, I was led to modify my initial conclusions somewhat.2 For one thing, it became clear to me that E and P3 are so close as to be virtually identical, the latter having most likely been copied directly from the former. It was therefore obvious that P3 added nothing of substance to the critical text. I also discovered that Pl, which I initially took to be the arch-representative of the first family, was less reliable as a textual witness than its distant relative S and, furthermore, that O, which I had marginalized somewhat in my initial evaluation, would have been preferable to Er as a representative of the second family. On the basis of these insights, I decided in retrospect that, were I to do it all over again, I would substitute O for Er and ignore P3. And that is precisely what I have done here in the critical edition of books 4 and 5, exchanging O for Er and dropping P3 from consideration. I have also added F to the mix, since it includes the entirety of books 4 and 5, thus bringing the revised list of manuscripts to be collated back to seven: F, P1, S, E, L3 , O, and C1. Sample pages from these seven manuscripts are reproduced on pp. cxxi-cxxvii below, each page containing the incipit of proposition 32, pp. 141 (Latin) and 446 (English) along with the relevant diagram, which is not included in the Saint-Omer manuscript (see p. lxxi).
As before, so now, the process of establishing the critical text has led me to reconsider my already-reconsidered assumptions about the selected family-representatives and their place in the manuscript-tradition. Central to this reconsideration is that the text shifts quite early in book 5 from narrative explanation to mathematical demonstration. In narrative explanation, of course, two different, sometimes even contradictory, readings can make perfect sense in a given context. Choosing the "right" reading is thus dictated more often than not by consensus of manuscripts (appropriately weighted for authority) rather than by the logic of the passage. In mathematical discourse, however, there is little or no ambiguity, so the right reading is dictated more often than not by logic, not consensus. In many cases, in fact, consensus is simply wrong. Accordingly, as the text of book 5 unfolded, it became increasingly clear to me that manuscripts F and P1 -were even less authoritative and reliable than I had expected and, conversely, that O was commensurately more so.
This re-evaluation of F, P1, and O in light of the critical text of book 5 raises questions about the authenticity of F and Pl as witnesses to the Urtext. …