of Carthage and Corinth (in the middle of the second century B.C.) to the activity of Lucullus in 69.  A like fate befell Books Seventy and Seventy-one at an early date.  The first twenty-one books and the last forty-five (save the two above noted) seem to have been extant in their original forms at least as late as the twelfth century.  Which end of the already syncopated composition was the first to go the way of all flesh (and parchment, too,) it would not be an easy matter to determine.  It is regarded by most scholars as certain that Ioannes Zonaras, who lived in the twelfth century, had the first twenty-one and the last forty-five for his epitomes.  Hultsch, to be sure, advances the opinion[1] that Books One to Twenty-one had by that time fallen into a condensed form, the only one accessible; but the majority of scholars are against him.  After Zonaras’s day both One to Twenty-one and Sixty-one to Eighty suffer the corruption of moth and of worm.

[Footnote 1:  Iahni Annales, vol. 141, p. 290 sqq.]

The world has, then, in this twentieth century, those entire books of Dio which have already been mentioned,—­Thirty-six to Sixty,—­and something more.  Let us first consider, accordingly, the condition in which this intact remnant has come down to the immediate present, and afterward the sources on which we have to depend for a knowledge of the lost portion.

There are eleven manuscripts for this torso of Roman History, taking their names from the library of final deposit, but they are not all, by any means, of equal value.  First come Mediceus A (referred to in this book as Ma), Vaticanus A, Parisinus A, and Venetus A (Va) of the first class; then Mediceus B of the second class; finally, Parisinus B, Escorialensis, Turinensis, Vaticanus B, and Venetus B, with the mongrel Vesontinus, which occupies a position in this group best designated, perhaps, as 2-1/2.

Vaticanus A has been copied from Mediceus A, and Parisinus A from Vaticanus A, so that they are practically one with their archetype.  Venetus A is of equal age and authority with Mediceus A. One can not now get back of these two codices.  There is none of remoter date for Dio save the parchment Cod.  Vat. 1288, containing most of Books Seventy-eight and Seventy-nine,—­a portion of the work for the moment not under discussion.  Coming to the second class, Mediceus B is a joint product of copying from the two principal MSS. just mentioned.  In the third class, Parisinus B is a copy of Mediceus B with a little at the opening taken from Mediceus A. This was the version selected as a guide by Robert Estienne in the first important edition of Dio ever published (A.D. 1548).  All the rest, Escorialensis, Turinensis, Vaticanus B, and Venetus B are mere offshoots of Parisinus B. The Vesontinus codex is derived partly from Venetus A and partly from some manuscript of the third class.