Probably few scholars would now doubt that at least the bulk of the matter both of Servius' commentary on Virgil (S) and of the additions to it which were first printed by Daniel (D) descends from Donatus. The problem of these additions has been approached by a number of writers from different directions, and different lines of evidence have been found to converge on one conclusion. An important contribution to the discussion has recently been made by J. J. Savage in a thorough study of the multifarious scholia contained in the well-known Bernensis 165 and some other Virgil manuscripts.1 He shows that some of these scholia go back to a commentary which was fuller than that of Servius, and which moreover was related both to S and to D.2 Other fragments of such a commentary are to be found by those who have the patience to sift the rubbish of the Latin glossaries,3 and also, among other sources, in Isidore, who is all the more useful for his habit of verbal transcription.4 When material of this kind is found, there is a strong temptation to say ‘Here is Donatus,’ and one can hardly doubt that that is where much of it ultimately comes from; but the question how far we may assume that the words of Donatus are directly reproduced can hardly yet be regarded as satisfactorily answered. I have before ventured to doubt whether Donatus was the immediate source of D.5 Professor Rand and Mr. Savage think that he was, that the compiler of SD supplemented S with material which S had not reproduce from Donatus the statement to which S had objected.

In the Classical Review (1925, p. 118) I quoted, for Petronius 77. 6 ‘assem habeas assem valeas,’ a proverb (from Mayor on Juv. 3. 143, addenda) unnoticed as far as I know by other scholars—‘quantum habebis tantus eris; frange lunam et fac fortunam’—and suggested that we should invert and correct—‘frange lunam [et] fac fortunam; quantum habebis tanti eris’—thus getting an accentual trochaic tetra-meter, with rhyme in the first half (the type dealt with by Usener, Kl. Schriften 2. 256, and Lorenz, preface to his edition of the Pseudolus, p. 39), which could be added to the popular trochaics collected in Baehrens' Poet. Lat. Fragmenta.