CORRECTORS AND CRITICS II
Ultimately, the purity of the surviving texts of the classics depended more on individual owners carefully checking copy against exemplar than on the odd scholar trying to “improve the text” by collating manuscripts. It is in any case far from clear that even the more scholarly readers of late antiquity had any clear concept of “improving the text” by collation. The prime example here is the Fronto palimpsest, with its margins full of variants added by the secondhand (C=Corrector). In a great many cases C explicitly designates them as variants rather than conjectures by the lemma i(n) a(lio) or i(n) a(liis), “in another copy,” “in other copies” (he seems to have drawn on at least three different copies, though not necessarily directly; most of his variants surely came from the margins of only one or two actual texts). In addition, he made many hundreds of corrections directly into the text, by erasure, expuncture, or interlinear insertion. Obviously, what we would like to know is on what basis he sometimes directly altered his text and sometimes just quoted variants in the margins.
To the scholar familiar with the modern apparatus criticus, it might seem obvious to identify the marginal variants as rejected alternatives. But a careful study by Zetzel has shown that the situation is more complex.1 Sometimes the variant is so obviously superior that it is hard to believe C could not see it; sometimes so obviously inferior that it is hard to see why he bothered to record it at all. The most plausible solution, one that would impose some degree of consistency on his choices, would be to identify the corrections made in the text as the result of collation against the original exemplar, and the marginal variants as the (later) product of collation against other manuscripts. This solution would incidentally underline the overriding importance of that routine first correction.
Divergences from the exemplar were automatically (and in most cases no doubt rightly) considered errors and corrected, while variants found elsewhere were simply recorded. It was the corrector’s obvious duty to reproduce his exemplar exactly, especially if that exemplar had been (to all appearances) authoritatively corrected. Any reading that differed from it could confidently be expunged and replaced, whereas
1. Zetzel 1980, 49–55; the standard text is now M. P. J. van den Hout’s Teubner (1988).