For the sake of clearness the manuscripts are in this edition regularly grouped according to this classification, and in the critical notes each class is cited by itself. At the same time it must not be assumed that the manuscripts of each recension stand necessarily by themselves, and that no connexion is traceable between one class and another. On the contrary, we shall find that many errors in the text of the first recension appear also in some copies of the second, and even of the third. The process by which this was brought about is made clearer to us by the fact that we have an example of a manuscript which has passed from one group into another partly by erasure and partly by substitution of leaves, apparently made under the direction of the author. This is MS. Fairfax 3, which forms the basis of our text, and the handwriting of some of the substituted pages is one which may be recognized as belonging to the ‘scriptorium’ of the poet. The example is a suggestive one and serves to explain several things. It makes it easy to understand, for example, how the additional matter introduced into the second recension came to be omitted in the third. The author in this instance had before him a very fully revised and corrected copy of his first edition, and this by a certain amount of rewriting over erasure and by a substitution of leaves at the beginning and end of the poem was converted into a copy of what we call the third recension, which his scribes could use at once as an authoritative exemplar. The introduction of the additional passages in the fifth and seventh books could not have been effected without a process of recopying the whole book, which would have called for much additional labour of the nature of proof-reading on the part of the author, in order to secure its correctness. This argument would apply to a book which was intended to remain in the hands of the author, or rather of the scribes whom he employed, and to be used as an archetype from which copies were to be made. If a new book had to be specially prepared for presentation, the case would be different, and it might then be worth while to incorporate the additional passages with the fully revised and re-dedicated text, as we find was done in the case of the so-called Stafford MS. Another matter which can evidently be explained in the same way is the reappearance in some copies of the second recension of errors which belong to the first. In producing the originals of such manuscripts as these, partially revised copies of the first recension must have been used as the basis, and such errors as had not yet received correction appear in the new edition.