The Printing of Lyrical Ballads, 1798

By D. F. FOXON

[Read before the Bibliographical Society 17 November 1953. Originally published in The Library vol. 9 no. 4 (1954) ; © D. F. Foxon and The Library; reproduced by permission.]

The history of the first edition of Lyrical Ballads might seem to be well established by now, and much of the work of Wise in his bibliography of Wordsworth (1916) and in Two Lake Poets (1927) has been generally accepted. But in view of some weaknesses in his account I have tried to approach the whole subject afresh from the point of view of the printer, and the results suggest some modifications in the traditional story.

The main points of this story are two: the cancellation of 'Lewti', and the change of publisher which involved a change of title-page. Wise's account of the first runs as follows:

At the last moment, when the sheets of the Lyrical Ballads had been printed off, and the book was ready for publication, it became apparent to one or both of the joint authors that the presence of one of the poems included in the series would not only afford a ready clue to the authorship of the Ballads, but would undoubtedly result in the whole number being attributed to the pen of one of them alone. Lewti ; Or, the Circassian's Love Chant had been published in The Morning Post, for April 13th 1798. Although it appeared over the nom de plume 'Nicias Erythraeus,' the poem was well known to have been the work of Coleridge. In order, therefore, that the anonymity of the Lyrical Ballads might be preserved it was necessary that this particular poem should disappear from its pages. Accordingly the three leaves numbered pp. 63-68 [Sigs. D8 and E1 and 2], together with the leaf carrying the table of Contents, were removed and were replaced by a new Contents leaf, and a new half-sheet of four leaves = 8 pages. The first seven of these pages carried The Nightingale, whilst the eighth remained blank.... The net result was the gain of one surplus leaf, or two surplus pages.[1]

The book was originally published with the imprint 'Bristol: printed by Biggs and Cottle, for T. N. Longman, Paternoster-Row, London. 1798.' (The relations between Biggs and Cottle change continuously from 1795 to 1802, but in this case it is clear that Cottle owned the copyright of the book and that though the two occur jointly as printers, the practical partner in that side of the business was Biggs.) But to quote Wise again:

Within a fortnight of the day of publication he [Cottle] sold the whole edition of Five Hundred Copies, less the few already distributed, to Messrs T. and A. Arch, of Gracechurch Street, London. By this firm the book was finally issued, after being furnished with a new title-page bearing their imprint.[2]

Finally, Wise describes the make-up of the book (excluding the cancel) as

Foolscap octavo.... The signatures are A to N (13 sheets, each 8 leaves) plus O (1 leaf), followed by an unsigned quarter-sheet of two leaves, the first of which has upon its recto the list of Errata, the reverse blank, whilst the second is occupied by the list of Books published for Joseph Cottle, Bristol, Mr. T. Longman, and Messrs. Lee and Hurst, Paternoster Row, London, the whole preceded by five unsigned leaves carrying the preliminary matter.[3]

So much for Wise's account of the book. Before I go any farther, however, I should make clear two physical characteristics of the book which will enter into my arguments. The first concerns the watermarks.

Lyrical Ballads, like many of the books produced by Biggs at this period, is printed on a wove foolscap paper bearing the watermark 'LLOYD 1795'. Paper of this type and period seems normally to carry the watermark at the bottom right-hand corner of the sheet, and when it is folded as an octavo the watermark appears at the foot of a pair of conjugate leaves in each gathering. In this case, 'LLO' of the watermark appears on one leaf and 'YD 1795' on the other. Which pair of leaves bears the watermark in any particular gathering depends, of course, on which way up the sheet is laid on the press; and in this instance the pressman does not seem to have worked through an ordered pile of sheets, for the position of the watermark varies in each gathering from copy to copy. On the other hand, as one might expect, there seems to have been a run of consistently arranged sheets from time to time, and certain gatherings show a preponderance of one or two positions of the watermark; but there is certainly not sufficient consistency to form a basis for argument.

The other characteristic is the incidence of point-holes. From the early days of printing, points have been placed on the tympan in order to pierce each sheet of paper placed on the press. When the sheet comes to be perfected the holes thus pierced fit again on the points, the paper takes up exactly the same position in the press, and correct register is ensured. By the eighteenth century it had become the almost universal practice in England to place the two points asymmetrically, that on the far side of the tympan from the pressman being closer to the edge than that on the near side. According to Moxon, the chief purpose of this was to save the pressman effort, in that he did not have to draw his body so far back from the press in order to fit the near hole over the point. To me it seems that the other function of this arrangement which Moxon mentions is more important. When the pressman perfects the sheet, he must (on the traditional methods of imposition) turn it over end to end to achieve the correct result. Now if the points were symmetrically placed on the tympan, sheets which had got turned round on the heap might be placed on the press the wrong way round and thus spoiled. When, however, the points are placed asymmetrically in this way, the sheet will only fit in the right position, and error is impossible.

Normally point-holes appear in the fold of a folio, at the head of a quarto, and along the fore-edge of an octavo. At the turn of the century, however, a number of small octavos appear with point-holes at the head of the page, for where the sheet is a small one it is possible to place it on the tympan at right angles to the normal. There are considerable advantages in doing this, for it means that the whole of the sheet can be covered by the platen and so printed with one pull instead of the two which are necessary for full-sized sheets on the presses of the period.[4] But when printing a foolscap octavo in this way, the sheet is still perfected by turning end to end; and the points must be symmetrically disposed in order to allow this. Thus when we come to examine an uncut or only lightly trimmed copy of Lyrical Ballads, we find that there are point-holes at the head of the leaves of the first half of each gathering, and the following diagram shows how the holes occur in relation to the leaves.

I might add that point-holes are one of the very few pieces of direct evidence that we have as to how a book was printed. It should, for example, be possible to identify the division of a book between different presses by measuring the distance between point-holes in each gathering, since the set of the points is unlikely to be identical on any two presses. But as point-holes occur at the fold, it is only with uncut or very lightly trimmed copies that it is possible to find them at all in an octavo or a quarto--in a folio, of course, they are usually hidden in the back fold. This is one of those occasions where the collector of books in original condition has a great advantage in bibliographical studies over the student who has access only to copies in libraries which have so often been cut down at some stage in their life.

Having dealt with these practical characteristics of the book's make-up we can turn to some of the problems in which they are involved. Wise when discussing the cancellation of 'Lewti', said 'the leaf carrying the table of Contents [was] removed and [was] replaced by a new Contents leaf'. Here my troubles first started when I examined his copy of the second issue in the original boards.[5] There is, of course, no difference between the first and second issue other than the change in the title-page; the body of the book is exactly the same and anything we prove from a copy of the second issue will be applicable to the first issue. Now this copy showed clearly that the preliminaries consist of a singleton title-leaf which is tipped in and a four-leaf gathering which contains three leaves of so-called Advertisement (which I shall refer to as the Preface) and a fourth leaf of Contents. The Table of Contents is, in fact, conjugate with the first leaf of the Preface. In what sense then was the Table of Contents cancelled? The problem is complicated by the fact that a copy which belonged to Southey, now in the British Museum,[6] and which contains 'Lewti' and the original Table of Contents as well as the cancels, has a note in Southey's handwriting, 'The Advertisement & the Circassian Love Chant in this volume were cancelled'. One might take this to mean that the whole of the preliminary gathering was cancelled and a new Preface and Table of Contents printed. That, of course, would explain the conjugacy. On the other hand, Southey must have obtained this special copy from his friend Cottle, and if there was an earlier state of the Preface, it is strange that this should not be included as well as the other cancellanda. Indeed, Southey says directly 'the advertisement in this volume [was] cancelled', and since he clearly meant his note to explain the make-up of his own copy, I feel confident that he made a slip of the pen and wrote 'advertisement' for 'Table of Contents': this has, in fact, been tacitly assumed by Hutchinson and later writers.[7]

If then the Preface was not reprinted, the fact that the corrected Table of Contents is conjugate with it must mean that the cancellation of 'Lewti' had been decided upon before these preliminaries were printed. The body of the book, on the other hand, must have been printed off or cancellation of the offending poem would have been unnecessary. But though we may say that the Table of Contents was not physically cancelled, since it is conjugate with the first leaf of the Preface, it is a fact that in this Southey copy and in two others we have an earlier state of the Table of Contents containing 'Lewti' listed among the other poems. The setting is the same as the normal Table of Contents except for the single line containing 'Lewti', but the paper is perhaps different from the rest of the book and in none of the three copies does this leaf bear a watermark. We cannot, I think, ignore Southey's statement that the Table of Contents was cancelled, but it looks to me as though the earlier state was never part of the completed book. We shall return to this problem later.

So much for the preliminaries. Now for the end of the book. Wise describes O1, which carries the last of the poems, as '(one leaf), followed by an unsigned quarter-sheet of 2 leaves (the first of which carries the list of Errata, whilst the second carries advertisements only)'.[8] This seems to me to ignore the facts of printing-house practice.

It is most unlikely that a printer would print off a single leaf before the book was completely set in type, when by waiting he might well be able to work it off with, say, the preliminaries. If indeed we add the title-leaf (which is a singleton) to the three leaves at the end, we have a respectable half-sheet, and I think a modern bibliographer would almost automatically assume that the four leaves were printed together. Wise's uncut copy gives the impression that O1 had been joined at the head to the leaf of errata which I consider to be 02; in particular, what is certainly a point-hole seems to be divided between these two leaves. There are, moreover, no watermarks inconsistent with this suggestion in the copies I have examined. It is then my view that the true collation for Lyrical Ballads is the following:

If we consider this formula, there is one point which it would be nice to establish. We have here three gatherings of four leaves, 2, , and O. Is there any evidence of how these were printed? This appears to be a matter of pure curiosity, I confess, but it is by pursuing these apparently unimportant matters that one sometimes finds a piece of evidence that may help one elsewhere. One would expect that a printer would print two of these half sheets together on a whole sheet, and then print the remaining four leaves as a separate half-sheet; if so, which is the odd half-sheet?

Watermarks give us no help since they appear at random in all copies. That is, one copy may have a watermark on all three half-sheets, another may have watermarks on none of them, or watermarks may appear on any combination of them. Clearly, if two of these half-sheets were printed together on a single sheet, the sheets were divided and stacked separately before the copies were gathered, and so any pattern was destroyed. The only evidence that is left to us is that of the point-holes. If the whole book was printed on the same press with the same setting of the tympan points, the point-holes will appear in approximately the same place in each gathering. Their position will depend slightly on how accurately each sheet is placed on the tympan and then on how accurately it is folded; but in the ten or so copies that I have checked, all the holes are found within very close limits--between 23 and 30 mm. from the back fold; the average is 27 mm. Now in 2 and the point-holes are similarly placed, but in O they lie between 45 and 50 mm. out. This evidence would suggest that 2 and were printed together on a single sheet on the same press as the rest of the book and that O was printed by itself. Unfortunately, the evidence for the printing of O is rather slight. The fourth leaf has been removed (perhaps for the title-page) and the leaf of advertisements which formed 03 was also very often removed when copies were bound. As a result I have seen only two copies which have the three leaves and have not been cut down so much as to destroy the point-holes; I have hearsay evidence of some others.

Now if O had been printed on half-sheets, we should expect to find two point-holes, one on each side, as shown in the diagram below:

If the sheet were folded so that the point-hole to the right appeared on leaf 2 (as we usually find), we should also expect to find a corresponding point-hole on 03, the leaf of advertisements; in the two copies I have seen no hole appears. The fact that there is but the single point-hole then suggests that these leaves were printed by half-sheet imposition. Why the placing of the point-holes is different here, one can only speculate; it suggests to me the use of a different press.

This leads to one more deduction. The following diagram illustrates the relation of the watermark to the point-hole in a number of copies:

The watermark appears with the date 1795 on 01. The date could appear on 02 if the sheet were laid on the press the other way up, and so it does appear on a couple of copies I have seen; similarly it might be on the other half-sheet. What is quite impossible is that the date should appear on the fourth leaf--that is, the title-page. Now the title of the London issue does sometimes bear the date as watermark, and this proves that it is a true cancel. This is of some importance since the setting of the London title-page, apart from the imprint, is identical with that of the Bristol title-page, and it might otherwise have been arguable that the change was made by stopping the press and making the alteration.

I must make one reservation here. I have followed the traditional view that the Bristol title-page was the first, and assuming that Lyrical Ballads was printed in isolation have argued as if it had been printed as 04. I must, however, record a disquieting fact. I have reports on eight Bristol title-pages, which were all I could trace. Not one of these bears any trace of a watermark; and, as with the first Table of Contents, I have a suspicion that the paper may be different from the rest of the book. As for the watermark, quite apart from the fact that the position shown above (with LLO on 04) occurs most frequently in the copies I have seen, one would expect by the laws of chance that one copy in four would show a watermark in the title-leaf. I have become cautious of small samples for a reason I shall mention later, but there is at least a possibility that these examples are typical and that the title-page we know as the Bristol title-page was not printed as O4. In that case what were these three leaves printed with? It is always possible, of course, that they were printed with some material for another book; the size and paper is one of Biggs's standard lines. But I believe we cannot rule out the possibility that there was a yet earlier title-page which has not survived and which proudly proclaimed Joseph Cottle as the publisher, as might be suggested by the advertisement on O3 which lists books 'published for Joseph Cottle, Bristol, Mr. T. Longman and Messrs. Lee and Hurst, Paternoster-Row, London'. We know that Cottle's financial difficulties were the reason for the sale of the edition to Arch, and I had previously suspected that the surviving Bristol title-page with Longman as nominal publisher represented a first but unsuccessful attempt to get the copies off his hands. One could therefore fit a yet earlier title-page with Cottle as publisher into what I believe to be the publishing history of the book.[9] There might then be some relationship between the unwatermarked title-leaf and the unwatermarked Table of Contents I have referred to. But this highly speculative subject must be reconsidered later.

There is one other curious fact which deserves some comment, namely the occurrence of watermarks in sig. E--the gathering of which the first two leaves were cancelled. Of the first half-dozen copies of Lyrical Ballads which I examined, five had no watermark whatever on the remaining leaves, 3 to 8. Since the whole book is printed on watermarked paper, the watermark should appear on a pair of leaves in each gathering; so this came as rather a shock. If one of the two cancelled leaves had carried the watermark one would still expect to find the matching half on either leaf 7 or 8.

Further exploration produced two copies which had a watermark on the conjugate pair 3 and 6 as well as half a watermark on leaf 8, and the other copies which I examined had half a watermark on either 7 or 8. Clearly chance had produced a far from representative sample on my first attempt. Where two watermarks occurred, one was in the inner and one in the outer half of the gathering.

It still remains to explain why two watermarks or none at all should appear, and suspicion falls on the method of cancellation since that is the only abnormal feature of the gathering. No stubs remain from leaves 1 and 2, and leaves 3 and 4 of an uncut copy have a neatly torn edge at the head, so we may presume that the sheet was folded before the first and second leaves were removed with a paper-knife. At first sight the obvious way of doing this is merely to separate leaves 1 and 2 from 3 and 4 at the head and then cut them out along the back fold. The effect of this is that the last two leaves are still attached at the head to the remainder of the gathering; this seems convenient since they cannot get lost in the collation and sewing. It also makes it impossible to explain the anomalous watermarks. If, however, we go on to consider the fate of 7 and 8 if this method were adopted, we find it is not so satisfactory as it might seem. The only way that these leaves can be secured in the book is by being tipped in. One can in theory do this by unfolding the pair at the bolt along the head which joins them to 5 and 6, applying paste along the margin of each, and replacing and sticking them down. The drawback to this, apart from the inconvenience to the binder, is that when the purchaser attempts to open the bolts which join them one to another and to 5 and 6, the paste may give way and the leaves come away in his hand. It would make the task much easier for the binder and it would remove this risk to the purchaser if the binder cut the bolts at the head and the fore-edge so that he had merely two separate leaves to tip in. But if he is going to separate these final leaves, he might as well divide the folded sheet in one operation; remove the outer half of the gathering; and then divide this into two halves, the first of which is discarded as the cancelled leaves. If in this way leaves 7 and 8 were at one stage severed from the remainder of the gathering before they were tipped in, the watermark problem becomes easier to solve. We can now argue that a pair of leaves not part of the same sheet as the centre four might have been bound up with them, and if it were the other part of each sheet which bore the watermark, the finished gathering would bear none. Is there any obvious way in which this muddle might be caused?

Unfortunately we know very little of the methods of collation and binding used at this period. Johnson's Typographia merely reproduces, with some additions, the story that Moxon had to tell, and there gathering and collating is related to unfolded sheets; at this period, however, when books were so frequently issued in boards, it seems possible that the modern practice of folding the sheets before gathering and collating them might have been introduced, and this would fit the situation here. On this hypothesis we can think of the binder folding the whole of each gathering en masse and stacking them in order on his bench. When he comes to gathering E, he folds it and then halves it, making a stack of E3, 4, 5, and 6 next to on his bench. He then removes leaves 1 and 2, the cancellanda, and makes another stack of leaves 7 and 8. After this is stacked sig. F, and so on. At a later stage E7 and 8 will be pasted to E3456, and if in the meantime the two piles have been in any way disarranged or if they get out of step, then the finished gatherings will be made up of two half-sheets with no relationship to one another and our curious phenomenon will be explained. It is worth mentioning that the book was put into boards in Bristol, and in close association with Biggs and Cottle at that, for Wise's copy of the London issue has end-papers of the same paper as that of the text. It is possible that Arch may have taken some copies in quires, but with the need to make the cancellation certain in order not to prejudice the book's anonymity it may well be that all copies were put into boards at Bristol.

So far we have dealt with the production of the ordinary Bristol and London copies of the book. There are, however, a number of variant copies of Lyrical Ballads, all but one of which have been discussed in Wise's Bibliography. Of these, the most notorious and the one which should be disposed of first is the copy in the Ashley Library[10] which has Sig. G1 'cancelled', in Wise's words; a cancel which is now accepted as a modern insertion. This leaf is reproduced together with the adjoining pages in Fig. 1, and the types are clearly quite different.

Wise describes this leaf as follows. 'After the book had been amended, and apparently after the sheets had been bound, yet another change was made. Signature G1 (pp. 97-98) was removed, and replaced by a cancel leaf. . . . A comparison of the two discloses the following variations. On p. 97 the second line was altered from Than fifty years of reason to Than years of toiling reason; and on p. 98 the title of the poem which commences there was condensed from Simon Lee, The old Huntsman, With an incident in which he was concerned, to Simon Lee, the Old Huntsman.'[11] This state of the text is not found again until the revised 1837 edition of Wordsworth's poems, and Wise had to postulate that the copy used for setting the second edition did not contain this cancellation, which was therefore overlooked until Wordsworth's final revisions--which is an ingenious theory but not a particularly plausible one. There is, in fact, no doubt that since this leaf, contrary to Wise's implications, appears to be unique, it is, in fact, a forgery. But a forgery by whom? I think we should consider this in some detail, because tempting though it is to assign it to Wise, there are some reasons for thinking that he may in this case have been the dupe rather than the perpetrator. There are several suggestions that I would like to make.

The first concerns the motive for the insertion. The copy in question is the London issue, but it is in the original boards, and in this state it is probably as much of a rarity as the Bristol issue. Now at one time this copy was in a pretty shaky condition; a number of leaves were loose and were tipped back in again, and one at least was completely detached, since it has a frayed and dirty margin which once projected from the body of the book. If then some leaves were once completely free, it seems quite likely that one of them was at some time lost for ever, and G1, the outside leaf of a gathering, is one of those most likely to have suffered this fate. We have here an obvious motive for printing a new leaf.

Now consider the text of the substitution. There is no subtlety about it that suggests the work of someone creating a bogus cancel. It looks to me much more like the work of a hack bookseller who finds an imperfect copy, has no other copy of the first edition to compare with it, and perhaps fails to realize any need for one, and who has the leaf set up from any edition of Wordsworth's poems which happens to be handy.

Another possible argument is this. According to De Ricci's Book Collector's Guide (in which Wise had a hand), the Wise copy was sold by Sotheby on 7 June 1907; it was Lot 41 and was bought for £39 by one Hopkins.[12] I have been unable to trace any bookseller by the name of Hopkins except one far distant in Glasgow in this year, and it is possible that 'Hopkins' represents the book being bought for Wise. It is most improbable that it would change hands at such a price without someone collating it with some thoroughness, and it therefore seems unlikely that there was no G1, either the original or the replacement, at the time it was sold. It seems even more unlikely that Wise would have lost a leaf of a Lyrical Ballads in boards. If then G1 was lost, it had probably been replaced before it reached the saleroom and Wise.

Wise did not call attention to this peculiarity in his bibliography of Wordsworth published in 1916 and it seems that he had not at this time noticed it. That he had this copy at the time is perhaps confirmed by the loving way in which he expatiates on the rarity of copies in boards. The discovery must have taken place, I think, about 1926 when he was preparing the appropriate volume of the Ashley Library Catalogue. By this time De Ricci had publicized his ownership of the copy and no doubt he had also shown it to his friends; it would have been embarrassing for him to admit that he had had a sophisticated copy for all this time without having been aware of it. This, I think, explains his motive--which may well have been unconscious-for arguing the genuineness of the leaf. That the argument could not possibly succeed if the facts were what we now know them to be is perhaps further evidence of his innocence.

Finally, as for the printing of the leaf itself, I have been unable to trace the types used. I hope that some member of the Bibliographical Society more learned than myself in typography may be able to do this, but to my eye at least the types are quite different from those used in the Wise forgeries; in particular the lower-case s is quite unmatched by anything I have found. And there I think we may leave that unfortunate episode.

The other variant which is represented in the Ashley Library[13] is a copy of the Bristol issue in which a leaf carrying Dr. Beddoes's 'Domiciliary Verses' is inserted before containing 'The Nightingale'. This leaf is also reproduced (Fig. 2 below). Wise says of the whole cancellation, 'to follow the events as they occurred is by no means a difficult task. After the removal of Lewti, and before The Nightingale was selected to replace it, these Domiciliary Verses were introduced with a view to filling the gap. Evidently, however, the incorporation of work by a third hand failed to commend itself to the authors of the Ballads; the Domiciliary Verses were rejected, and The Nightingale was made use of. . . . Apparently the specimen of the Domiciliary Verses which occupies a position in the present copy is the solitary survivor of this tentative effort at adjustment.'[14] Wise is here skating over many difficulties.

'Lewti' and 'The Nightingale' both start on page 63, though as each has a drop-head title the page is not numbered; the versos are numbered 64. The leaf with 'Domiciliary Verses' bears the number 63* on the verso. If this was the first attempt at filling the gap left by the excision of 'Lewti', we should expect its pages to be numbered [63] and 64; the asterisk, however, shows that when they were printed they were an additional insertion. It seems clear that while Cottle or Coleridge may at first have thought of these verses as a possible substitute, and may have asked Beddoes for permission to use them, they were not printed--though I suppose they may have been set in galley--until a substitute which filled the gap in pagination completely had been found. Thus in their present form, with this pagination, they were printed after 'The Nightingale' had been chosen. Why then were they printed at all? My theory would be that Cottle wanted to make some reparation to Beddoes for any false hopes of seeing his poem in print that may have been aroused, and that he printed off this additional single leaf with suitable pagination and inserted it in a copy for Beddoes; I think, in fact, that this may have been Beddoes's own copy.[15] Such a story would, I believe, fit well with Cottle's kindly but rather muddled character.

There is one point of interest which should be mentioned here, though it relates to the more general problem of the Bristol title-page. For some reason there is a great deal of offset throughout this copy between facing pages. In the body of the book this shows nothing of importance except that the original binder inserted the Beddoes leaf the wrong way round so that the 'Domiciliary Verses' began on the verso--no doubt the sight of 63* provoked the instinctive reaction that odd-numbered pages are rectos and he bound it in accordingly. The real interest concerns the end and the beginning of the book, and since they were printed in that order (as I shall show later) we can consider them in that order. The verso of O3 with the advertisements carries offset from the Bristol title-page and vice-versa; but the verso of the title-page carries not only offset from the first leaf of the Preface which now faces it, but by the inner margin the first word or two of several lines of the Table of Contents. Unfortunately the line which contained 'Lewti' or 'The Nightingale' is not one of them, so we cannot prove whether the title-page lay next to the earlier or later state of the Contents leaf; but since the later state forms the recto of 24, it is difficult to see how this could be responsible for the offset. It looks as though the book was originally folded and stacked with the Bristol title-page after 03, followed by the first Table of Contents; and that this was later removed and 2 substituted. If this is so, it suggests a connexion between the Bristol title-page and the first Table of Contents, and that if the Bristol title-page was not 04, 04 itself was removed before any offset could take place. The obvious explanation for the offset in this copy is that this was an advance proof, perhaps Cottle's own working copy, added to as new material was printed, since the sheets were apparently not given time to dry before being folded.

Of the other variant copies, two are associated with Southey. The British Museum copy[16] which I mentioned above contains both 'Lewti' and 'The Nightingale' and the first and second Contents pages. This is clearly a product of Southey's bibliographical interests and his friendship with Cottle. How this copy reached its present form would be of little importance if it were not that Wise had made some rather derogatory statements about it which have been often repeated. He wrote: 'A second set of the four cancelled leaves, which also belonged to Southey, is safely lodged in the library of the British Museum. They are bound up with a copy of the Lyrical Ballads in its completed form with The Nightingale taking the place of Lewti, but with the Bristol title-page. The cancelled Table of Contents is inserted between the Title-page and the Advertisement, and the three cancelled leaves with Lewti are inserted between The Nightingale and The Female Vagrant.'[17] This is a very tendentious statement, with its stress on insertion--as though these four leaves alone had belonged to Southey and had later been bound up with another copy of Lyrical Ballads. In fact, the whole book came direct from Southey's library in its present form, the Museum having paid 10s. for it at the Southey sale. It has, however, been rebound at least twice since it reached the Museum and is, in fact, about to be rebound again, so it has been possible to examine it piece by piece, but no useful evidence of conjugacy remains.

As far as the body of the book goes, one can get some evidence of what happened in the cancelled section by a study of the watermarks. They can be represented thus:

I have underlined the leaves which bear half-watermarks; the cancelled leaves D8 and E1 and 2 are enclosed in square brackets. The watermark on D8 is complementary to that on D1, and similarly the watermark on E2 matches up with that on E7. If these leaves had been inserted as Wise suggests, the odds would be very much against both of them matching up perfectly with the rest of these gatherings. What is upsetting is the extra half-watermark on E8, and though this has no apparent or obvious relation to the cancellation it does make one suspicious of the copy. This caused me a good deal of concern until the New York Public Library, having been asked about their Southey copy, expressed surprise at the curious fact that while their E1 bears half a watermark, there is no matching half on E8; E2 and E7, however, are conjugate. The obvious answer to the problem is that someone started to cancel 'Lewti' in the two copies; had second thoughts and stopped before he had finished dismembering the New York copy; and exchanged the two E8's in reconstituting the copies. Southey would be the likely culprit if both copies were his, but the original ownership of the New York copy is somewhat obscure. Since it is bound in cotton print it has been assumed that it was one of the books from Southey's 'Cottonian Library', and there seems little doubt that like them it was covered at Greta Hall sometime round 1820. But it is questionable whether this handicraft, carried on at Southey's home by family, friends, and relations, was limited to Southey's own books; there is, for example, in the British Museum a copy of Icelandic Poetry . . . translated into English verse by A. S. Cottle (Bristol, 1797) in a Greta Hall binding and with Wordsworth's signature. The New York copy bears a partially erased inscription on the front paste-down (added when the book was covered) which reads 'Martha Fricker / The Gift of / S. T. Coleridge'; it has not been possible to identify the handwriting. We cannot accordingly be sure that Southey owned this copy and so had the opportunity to make the change--the two copies may only have been together on the premises of Biggs and Cottle; but it seems clear that deliberate sophistication will hardly fit the facts. As for the first Contents leaf of the British Museum copy, it immediately follows the Bristol title-page, and the close association of these two rather puzzling leaves parallels the evidence of offset in the Beddoes copy. It is perhaps no accident that the two are together again.

We come finally to a pair of similar uncancelled copies which both appear to be imperfect. Of these, the first is the so-called Southey-Latymer copy, now in the New York Public Library, to which I have just referred. This has no preliminaries, except that the first Table of Contents is inserted after the first two leaves of the 'Ancient Mariner' which contain its half-title and argument. The body of the book is normal (except that 'Lewti' is not cancelled) and it ends with the usual three leaves of 0. The other copy, which is at Yale, has the Bristol title-page as well as the first Table of Contents, but no other preliminaries; the body of the book is the same as the New York copy except for the fact that G8 is missing. (This is merely an unfortunate accident so far as I can see.) Both libraries commented that the Contents leaves (and the title-leaf of the Yale copy) appeared to be of slightly different paper from the rest of the book, quite apart from the absence of watermarks.

Leaving that on one side for a moment and assuming that the New York copy has lost its title-leaf (its first gathering is rather loose), we should consider whether these copies represent a first state intended for publication;[18] in which case the Preface was an afterthought dating from about the same time as the cancellation of 'Lewti'. The difficulty involved in this hypothesis is again the nature of printing-house practice. These copies confirm what I argued above, namely that the preliminaries were printed together with the cancel , not with 0. Here we have O in its usual state, a first attempt at preliminaries, and no cancel; so O was printed before 2 and .

Now if the printer intended to produce the book in the form which these copies represent, how would he go about it? He needs two preliminary leaves for the title-page and the Contents; and two final leaves, O1 with text and 02 with errata. Surely, then, he would make up a half-sheet of the four and finish the job. But he does not; he prints a quite gratuitous leaf of advertisements on 03, as though he were filling up space. My own feeling is that the printer knew he would have more material coming for the preliminaries and finished off the text as best he could. It seems possible that, if there was a hold-up at this time (because the preface was not ready or because the need to cancel 'Lewti' was realized), Cottle may have prepared one or two advance copies with proofs of the Contents leaf as already set up--this was the one further piece of work that the compositor could do before more copy came to hand--and a trial title-page.[19] The New York copy may have been sent to Coleridge or to Southey--the latter has long been thought to have seen advance proofs and to have expressed his criticisms of them to Cottle. Who received the second extant copy, now at Yale, is something of a problem. In its present form it is in a mid-nineteenth-century binding together with an imperfect copy of an unidentified edition of Rogers's Poems (the title-leaf wanting) and the anonymous 1795 edition of Cottle's own poems. An early hand has rightly ascribed the latter to Cottle in a note on the title-page, but has also ascribed Lyrical Ballads to Southey. If the handwriting is that of its first owner, it is curious that someone who was honoured with an advance copy was not told the name of the author. It may be that Cottle, having received Southey's adverse criticisms at an early stage, asked for a second opinion from someone with no personal axe to grind. If so, it is rather ironic that this individual thought it was the work of Southy.[20]

There is one problem still outstanding; the status of the Bristol title-page and its relation to the first state of the Contents leaf. I have perhaps prejudged the issue in the references I have made above, but I will try to sum up the evidence. All eight copies of the Bristol title that I have traced bear no watermark.[21] The three copies of the first state of the Contents leaf bear no watermark. Three independent witnesses have, without prompting, suggested that the paper of these two leaves might be different from that used in the rest of the book. In the Yale copy these two leaves form the only preliminaries; in Southey's copy in the British Museum they are bound next to one another; in the Ashley copy the offset shows that they were once side by side. On the other hand, the normal London issue has the title printed on the same LLOYD paper used in the body of the book. The use of an odd paper suggests to me that these leaves were a stop-gap of some sort; the Contents leaf has all the marks of being a page-proof, and I am tempted to consider the title-leaf as an interim state as well. It seems possible that O4 was originally intended to be the title-leaf, with Cottle as publisher; why he should wish to withdraw is a matter to be considered later. For the moment I will attempt to codify the history of the book's printing.

(iv) Bristol title-page set up (or altered from 04) and proofs pulled (possibly contemporaneously with iii).
Advance copies issued with iii and iv as preliminaries.[22]

(v) 2 and set up together, proof-read, and printed.
Copies made up with Bristol title-page (? proofs as iv).

(vi) Arch title-pages altered from iv and printed.
Edition transferred to Arch in boards.

I do not wish to plunge too deeply into the problems of the publication of the Lyrical Ballads, even were it within my competence. On the other hand it is perhaps worth while to summarize what is known about their publication, and I am here largely indebted to Professor R. W. Daniel's article on the subject in M.L.R.,[23] though I do not accept all his conclusions. The evidence he brings forward may be summarized thus:

18 July 1798. Dorothy Wordsworth writes, 'William's poems are now in the press; they will be out in six weeks [i.e. by 29 Aug.].'

5 Sept. Southey writes to Taylor, 'Have you seen a volume of Lyrical Ballads,' and goes on to criticize it in terms similar to those of his later review.

13 Sept. Dorothy writes that the poems 'are printed but not published'.

[15 Sept.] William writes to Cottle on his return to England in 1799, 'The day before I left England I wrote to you to request that you would transfer your right to the Lyrical Ballads to Mr. Johnson [i.e. Coleridge's London publisher]'. Cottle comments, 'This I could not have done, had I been so disposed, as the engagement had been made with Mr. Arch'.

4 Oct. The Arch issue is advertised as published.

2 Nov. Southey's review appears in the Critical Review.

One point must first be made clear; the account Cottle gives in his Early Recollections (1837) and Reminiscences (1847) is quite misleading. For example, one of the reasons he gives for disposing of the poems to Arch is the adverse reviews the book received, but in fact Southey's review was by far the earliest, and this appeared a month after the book was issued by Arch. Cottle may have been confused by having seen Southey's review in advance; but the story has almost certainly been altered in order to conceal Cottle's financial troubles at this period. On the other hand, Cottle's statement quoted above concerning Wordsworth's request is probably reliable.

If we ignore Southey's letter of 5 September for the moment, the evidence suggests the following sequence. The book was intended to be published by the end of August, before Coleridge and the Wordsworths left Bristol for London and their tour to the Continent. Yet though the book was ready by early September there was a difficulty about getting it published, and Wordsworth in London found a publisher prepared to take it. Almost simultaneously Cottle made his agreement with Arch, and within three weeks the London issue was published. Professor Daniel points out that if the book was not published by the 13th and was transferred by the time that Wordsworth's letter reached Cottle, say the 17th, then the Bristol issue could only have been on sale for four days at the most. Indeed, it looks very much as though the Bristol issue was never published, but merely circulated to friends. Since Cottle had a large circle of friends in Bristol, and since Wordsworth and Coleridge presumably wanted a number of copies, it would not be difficult to visualize some twenty-five copies circulating in this way, of which some twelve are known.

Southey's place in the story is rather more difficult. It is clear that by 5 September he had made up his mind about the book and knew the terms in which he would review it; this is not surprising since he probably had an advance copy even if he had not, as Thomas Hutchinson suggested, 'got sight of the proof-sheets of Lyrical Ballads as they were preparing for press'.[24] Hutchinson suggests two consequences of this advance knowledge; that Southey was enabled to get his uncomplimentary review out well before the other notices, and that he advised Cottle to transfer the edition to another publisher before the review appeared so that he would not lose money if the book failed to sell. But if we take Southey's letter of 5 September at its face value, he was then under the impression that the book was on sale and that Taylor could have seen a copy. Since that could only have been a copy with the Bristol title-page which we know from Dorothy was unpublished, it appears that Southey was unaware of the delay in publication while another publisher was being found: therefore he had no hand in the transfer to Arch. This does not mean that Southey was completely out of touch with Cottle at this period; we know, in fact, that he received at least one special copy of the book. It merely suggests that Southey had not seen Cottle for a week or so and did not know the latest developments.

If we wish Southey to keep his traditional role of Cottle's adviser, we must fall back on the hypothesis that Cottle originally intended to publish it in his own name; in which case there was possibly a yet earlier title-page (printed as 04) in which Cottle appeared as publisher. If at this stage Southey had warned Cottle to dispose of the book, since he would do nothing to help his financial situation by publishing it, Cottle might well have hoped to sell the edition to Longman at a reasonable profit; he had previously done this with the second edition of Southey's Joan of Arc. This would explain the form of imprint 'printed by Biggs and Cottle for T. N. Longman' which one would not otherwise expect;[25] the normal form for Cottle's publications in 1798 is 'printed by Biggs & Cottle and sold by T. N. Longman'.[26] In this case we would have to argue that Longman agreed in principle to publish the book; that Cottle printed the appropriate title-page as a trial (not on the same paper as the text) and gave copies to his friends; and that at the eleventh hour the deal fell through.

The full story would then be on the following lines. The body of the book was printed by mid-August, and Southey warned Cottle that it would be a failure. Cottle offered it to Longman, and printed proofs of the Longman title-page. Then 'Lewti' was cancelled and the preliminaries printed; and copies were made up with the Longman title-page, since the Wordsworths were about to leave Bristol and wished to see the book completed. These copies were distributed to friends, when suddenly Longman had second thoughts--it is perhaps relevant that when he later acquired Cottle's copyrights, he offered nil for Lyrical Ballads. Southey knew nothing of this development when he wrote on 5 September; but the Wordsworths did, since Dorothy wrote in that sense on the 13th and William did what he could to find another publisher. By the time they sailed the agreement had been made with Arch and the book was duly published on 4 October.

This account is, of course, highly speculative; it is merely one attempt at stringing together the facts and some of the hypotheses into a self-consistent whole. I hope the biographers may find something in it that is not inconsistent with the personalities involved.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.