The relations of the group AdBTΛP₂ with the first recension and with one another are difficult to clear up satisfactorily. Broadly, it may be said that of these B represents an earlier type than the rest in regard to correction and Λ in regard to revision: that is to say, B retains a large number of first recension errors which do not appear in the rest (sharing some, however, with Λ), while at the same time, in cases where a line has been rewritten B almost regularly has the altered form, though with some exceptions in the first two books. On the other hand, though it often happens that Λ is free from original errors which appear in B, yet in many places where B has the revised form of text Λ gives us the original, in agreement with the earlier first recension type, while in others Λ agrees with B in giving the revised reading. Then again, there can be no doubt of the close connexion between B and T, but the agreement between them is not usually on those points in which B follows the first recension in error. It is as if they had been derived from the same archetype, but T (or a manuscript from which T was copied) sprang from it at a later stage than the original of B, when many of the errors noted in the first recension had been corrected, while the text of the book generally was allowed to remain as it was[AI]. Finally, the text of Ad approaches very near to a fully revised and corrected type. It very occasionally reproduces the earlier first recension, as if by accident, but seems never deliberately to give an ‘unrevised’ reading. It should be observed that from a point towards the end of the fifth book (about v. 6280) AdBT is a group which is very frequently found in special agreement, whereas before that point we usually find BT (or BTΛ) with Ad on the other side. Passing now to the third recension, which has the preface and epilogue as in Λ and S, but excludes the additional passages, we find it represented by eight manuscripts, with Fairfax 3 at their head. We have already seen that this manuscript was originally one of the first recension, and was altered by the author so as to substitute the new epilogue and the new preface. Besides these changes, fresh lines are in several places written over erasures, as i. 2713 f., iv. 1321 f., 1361 f., &c., the marginal date is erased at Prol. 331, and additions have been made to the marginal notes. All these alterations, as well as the points previously noted, in which F originally differed from the other copies of the first recension, are reproduced in the other MSS. of the third recension.