Evidence of interpolation in the text of Thucydides

The interpolation of glosses and other matter into the text of Thucydides is something sensed even by less attentive readers. The problem was exaggerated by Cobet, Herwerden and others but remains formidable. The dissertation analyzes the hard evidence for it, and the validity of the main ways in which we detect it. In Part A of this work, Interpolation in some MSS, I study those instances that seem certain or probable, because some MSS preserve the truth. Thus, Part A is organized with reference to the divisions of the stemma codicum (interpolation in individual MSS; in either main hyparchetype; in collation exemplars; and so on). Part B examines one by one the more difficult ways in which we sense Interpolation in the tradition itself. As presented roughly from what seem the least to the most heavily problematical, they are: (1) Evidence of inscriptions; (2) Scholia that preserve true readings; (3) Linguistic evidence; (4) Valla non vertit ; (5) Scholiasta non legit , (6) Evidence of papyri (hence, evidence of very early corruption); (7) Internal evidence such as tautology, anacolouthon. In Part C, Miscellaneous, these criteria are applied to various cruces in book 8. It seems that the chief criteria for deletion to which all editors have appealed are each treacherous. The hard evidence for interpolation, though real, is smaller than even cautious editors have assumed, and has clearer features. Beyond this, it is often profitable to give more careful thought to the stemma codicum; to Valla; to the original purpose (and to the lemmata, or their absence) of pertinent scholia; and above all to the mechanics and the exact point of the supposed gloss. For if it has none or if the point is vague or queer, probably nothing was interpolated, but we have some other kind of corruption. Usually, when editors bracket what seem relics of longer glosses, they are demonstrably wrong