Sterling Dow (1903-1995), "Corinthiaca," Harvard Studies in Classical Philology
60 (1951) 81-100 (at 82):
If a lacuna is short, and if parts of the word or words to be filled in are preserved, and especially if the lacuna can be accurately measured, some certainty can be attained. But if as much as one whole substantive word is missing (more, I mean, than an obvious preposition, conjunction, or the like), restoration is equivalent to supposing that a modern scholar can possess such insight into the genius of the language, the spirit of the period, and the mind of the original author as to be able to compose verse identical with his. To say the least, such a supposition is difficult. I concede that erudition and sympathy can work marvels in the interpretation of texts transmitted without lacunae, and that "literary" emendations have occasionally been proved to be correct. But with this qualification fully admitted (and duly admired), the restoration of ancient verses, in the sense defined above, is a mere game; its results cannot be claimed as objective scholarship. Restorations of this kind have value, but only as illustrating what might have been written, as raising problems of usage and of grammar, and as proving that the readings and measurements given are not impossible. That is all.