A few experiments further showed that one, and a very important, result of this collation would be to confirm the trustworthiness of 1633, to show that in places where modern editors had preferred the reading of some of the later editions, generally 1635 or 1669, the text of 1633 was not only intrinsically superior but had the support of tradition, i.e. of the majority of the manuscripts. If this were the case, then it was also possible that the traditional, manuscript text might afford corrections when 1633 had fallen into error. At the same time a very cursory examination of the manuscripts was sufficient to show that many of them afforded an infinitely more correct and intelligible text of those poems which were not published in 1633 than that contained in the printed editions. Another possible result of a wide collation of the manuscripts soon suggested itself, and that was the settlement of the canon of Donne's poems. One or two of the poems contained in the old editions had already been rejected by modern editors, and some of these on the strength of manuscript ascriptions. But on the one hand, no systematic attempt had been made to sift the poems, and on the other, experience has shown that nothing is more unsafe than to trust to the ascriptions of individual, unauthenticated manuscripts. Here again it seemed to the present editor that if any definite conclusion was to be obtained it must be by as wide a survey as possible, by the accumulation of evidence. No such conclusion might be attainable, but it was only thus that it could be sought. The outcome of the investigation thus instituted has been fully discussed in the article on the _Text and Canon of Donne's Poems_ in the second volume, and I shall not attempt to summarize it here. But it may be convenient for the student to have a quite brief statement of what it is that the notes in this volume profess to set forth. Their first aim is to give a complete account of the variant readings of the original editions of 1633, 1635, 1639, 1649-50-54 (the text in these three is identical), and 1669. This was the aim of the edition as originally planned, and though my opinion of the value of many of the variants of the later editions has undergone considerable abatement since I was able to study them in the light afforded by the manuscripts, I have endeavoured to complete my original scheme; and I trust it may be found that nothing more important has been overlooked than an occasional misprint in the later editions. But I know from the experience of examining the work of my precursors, and of revising my own work, that absolute correctness is almost unattainable. It has been an advantage to me in this part of the work to come after Mr. Chambers and the Grolier Club editors, but neither of these editions records changes of punctuation.