The manuscripts fall into three main groups (1) D, H49, Lec. These with a portion of 1633 come from a common source. (2) A18, N, TCC, TCD. These also come from a single stream and some parts of 1633 follow them. L74 is closely connected with them, at least in parts. (3) A25, B, Cy, JC, O'F, P, S, S96, W. These cannot be traced in their entirety to a single head, but in certain groups of poems they tend to follow a common tradition which may or may not be that of one or other of the first two groups. Of the Elegies, for example, A25, JC, O'F and W transcribe twelve in the same order and with much the same text. Again, B, O'F, S96, and W have taken the Holy Sonnets from a common source, but O'F has corrected or altered its readings by a reference to a manuscript resembling D, H49, Lec, while W has a more correct version than the others of the common tradition, and three sonnets which none of these include. Generally, whenever B, O'F, S96, and W derive from the same source, W is much the most reliable witness.
Indeed, our first two groups and W have the appearance of being derived from some authoritative source, from manuscripts in the possession of members of Donne's circle. All the others suggest, by the headings they give to occasional poems, their misunderstanding of the true character of some poems, their erroneous ascriptions of poems, that they are the work of amateurs to whom Donne was not known, or who belonged to a generation that knew Donne as a divine, only vaguely as a wit.
These being the materials at our command, the question is, how are we to use them to secure as accurate a text as possible of Donne's poems, to get back as close as may be to what the poet wrote himself. The answer is fairly obvious, though it could not be so until some effort had been made to survey the manuscript material as a whole.
Of the three most recent editors—the first to attempt to obtain a true text—of Donne's poems, each has pursued a different plan. The late Dr. Grosart[36] proceeded on a principle which makes it exceedingly difficult to determine accurately what is the source of, or authority for, any particular reading he adopted. He printed now from one manuscript, now from another, but corrected the errors of the manuscript by one or other of the editions, most often by that of 1669. He made no estimate of the relative value of either manuscripts or editions, nor used them in any systematic fashion.
The Grolier Club edition[37] was constructed on a different principle. For all those poems which 1633 contains, that edition was accepted as the basis; for other poems, the first edition, whichever that might be. The text of 1633 is reproduced very closely, even when the editor leans to the acceptance of a later reading as correct. Only one or two corrections are actually incorporated in the text. But the punctuation has been freely altered throughout, and no record of these changes is preserved in the textual notes even when they affect the sense. In more than one instance the words of 1633 are retained in this edition but are made to convey a different meaning from that which they bear in the original.
The edition of Donne's poems prepared by Mr. E. K. Chambers[38] for the Muses Library was not based, like Dr. Grosart's, on a casual use of individual manuscripts and editions, nor like the Grolier Club edition on a rigid adherence to the first edition, but on an eclectic use of all the seventeenth-century editions, supplemented by an occasional reference to one or other of the manuscript collections, either at first hand or through Dr. Grosart.
Of these three methods, that of the Grolier Club editor is, there can be no doubt, the soundest. The edition of 1633 comes to us, indeed, with no a priori authority. It was not published, or (like the sermons) prepared for the press[39] by the author; nor (as in the case of the first folio edition of Shakespeare's plays) was it issued by the author's executors.
But if we apply to 1633 the a posteriori tests described by Dr. Moore in his work on the textual criticism of Dante's Divina Commedia, if we select a number of test passages, passages where the editions vary, but where one reading can be clearly shown to be intrinsically the more probable, by certain definite tests,[40] we shall find that 1633 is, taken all over, far and away superior to any other single edition, and, I may add at once, to any single manuscript.
Moreover, any careful examination of the later editions, of their variations from 1633, and of the text of the poems which they print for the first time, shows clearly that some method more trustworthy than individual preference must be found if we are to distinguish between those of their variations which have, and those which have not, some authority behind them; those which are derived from a fresh reference to manuscript sources, and those which are due to carelessness, to misunderstanding, or to unwarrantable emendation. Apart from some such sifting, an edition of Donne based, like Mr. Chambers', on an eclectic use of the editions is exactly in the same position as would be an edition of Shakespeare based on an eclectic use of the Folios, helped out by a quite occasional and quite eclectic reference to a quarto. A plain reprint of 1633 like Alford's (of such poems as he publishes) has fewer serious errors than an eclectic text.
It is here that the manuscripts come to our aid. To take, indeed, any single manuscript, as Dr. Grosart did, and select this or that reading from it as seems to you good, is not a justifiable procedure. This is simply to add to the editions one more possible source of error. There is no single manuscript which could with any security be substituted for 1633. Our analysis of that edition has made it appear probable that a manuscript resembling D, H49, Lec was the source of a large part of its text. But it would be very rash to prefer D, H49, Lec as a whole to 1633.[41] It corrects some errors in that edition; it has others of its own. Even W, which has a completer version of some poems than 1633, in these poems makes some mistakes which 1633 avoids.