University of Virginia Library

Rickaby Cancellation: A New Species of Post-Impression Correction
by
B. N. Gerrard [*]

It is commonly accepted that the purpose of cancellation in printed works is to change the text in some way, perhaps to soften a possibly libellous view, or to correct, belatedly, an error in setting. The essence of cancellation is that it alters the text as originally set.

The intention of this paper is to report a new form of cancellation which has nothing to do with the text but rather with the quality of the engraved, illustrative matter. The London printer Thomas Rickaby practised this specialised form of cancellation at the end of the eighteenth century, and for this reason I term it 'Rickaby cancellation'. I define Rickaby cancellation as the cancellation of a leaf (or more) because the quality of an illustration, whether of itself or through bad press work, was below some arbitrary standard.

In 1793 Rickaby printed an edition of Virgil with variant readings and notes by the German editor Christian Gottlieb Heine.[1] The works and annotations were published as four quarto volumes. Since these would have been

very thick, part title-pages were supplied to allow the Virgil to appear in eight volumes, and it is normally found in this way. The set is generally admired, for it has many charming engraved vignettes, in the cameo style characteristic of the 1790's.[2]

An examination of the Virgil reveals a number of cancellantia, but these appear, at first, to be randomly distributed between copies, which is most improbable if textual changes have been made. The Virgil is printed on a good quality, heavy, wove paper watermarked 'J Whatman' once only on a shorter edge of each sheet. However, this mark occurs in two distinct places which are best explained by referring to the quarto leaves, rather than the original sheet. In those sheets, which we may describe as having watermark 1, the name, which must be read vertically, appears close to the foot of the outer edge of a leaf. With watermark 2 the name is in the middle of the outer edge of the leaf. With the exception of the preliminary leaves in vol. 1, these two types of paper are the only ones used in the Virgil, and there is never any overlap in the positions of the watermarks.[3] There is never a situation when an examiner is in doubt whether the example is of watermark 1 or 2. As there was only a single mark in the original sheet, so there is only one mark in each gathering of the finished book, and it will occur randomly in any one of the four leaves.

Traditionally the detection of cancellantia has been through the presence of stubs of cancellanda. But in the second half of the eighteenth century bookbinders became increasingly clever at disguising these stubs. This skill seems to have been practised especially upon more valuable books, and ordinary reading of the Virgil reveals no stubs at all, at least in the copies examined for the purposes of this paper. The only way in which cancellation can be detected in the Rickaby Virgil is by disruption of the watermark sequence; we must examine the individual leaves against a direct light.

Here we encounter a problem. Only one in four of the leaves of each gathering in its original state has a watermark. If we consider a single-leaf cancellation there is only a 1:4 chance that the leaf with the watermark will be removed. Similarly there is only a 1:4 chance that this leaf will be replaced

by another having the watermark, if the same paper stock is used. There is, therefore, a 1:16 chance that a watermarked leaf which is cancelled will be replaced by another watermarked leaf.

On the other hand, the chances of an unmarked leaf being cancelled are 3:4, and the replacement leaf is also likely to be unmarked 3:4 times. So that there is a 9:16 probability that an unmarked leaf, that has to be cancelled, will be replaced by another unmarked leaf. If the job of binding in cancellantia is done 'without trace' then 10:16 will not betray themselves by a disruption of the watermark sequence. Looked at the other way round we will find cancellantia through a disrupted sequence of watermarks in only 6:16 (or 37.5%) of all the instances where there are, in fact, cancellantia.

Could this low level of cancellantia revealing themselves through a disruption to the pattern of watermarks account for the apparently random distribution of cancellantia in copies of the edition? In part this could be the answer, but only in part. Certainly, without resorting to dissection of the book it would not be possible to identify absolutely the cancellantia in a particular copy. One of the copies examined for this paper was uncut and so enabled some extra gatherings, with normal watermark patterns, to be defined as having cancellantia, for lengths of leaves did not always coincide in what should have been conjugate pairs.[4]

In replicate examinations of the Virgil a number of cancellantia were noticed. It was also observed that:

• i. Cancellantia appeared to occur only in gatherings which had an engraved illustration.
• ii. Where checking was possible—that is because an extra watermark was now present in the gathering—the watermarks were always both type 1, or both type 2.

Examination of replicate copies did not seem to show the repetition of cancellation within the same gatherings that would be expected with an ordinary textural cancellation; but the apparent absence of consistent cancellantia might have been a function of detecting only 37.5% of the possible cancellantia. Nevertheless, where there was an obvious cancellans, it conformed to the two observations already given. One example was even found of three watermarks being observable in a single gathering, and all were of the type 1 kind.[5] Examination of more and more copies revealed that almost always the cancellantia were associated with gatherings in which there was also an engraving present. Furthermore, where an extra watermark was found in the gathering one of those watermarks occurs on a leaf with an engraving. For example, let us suppose that an engraving appears on the third leaf of a particular gathering. All we can say of the original gathering is that one of the four leaves will have the characteristic 'J Whatman' watermark in the type 1 or type 2 place. If there has been cancellation three results are possible in the case of trimmed copies of the Virgil with no visible stubs of cancellanda:

• 1. The gathering has a watermark in leaves 1, or 2, or 4, and, if there has been cancellation of an engraving on leaf 3, the event has not occurred with the addition of an extra watermark. Such a cancellation is at present undetectable.
• 2. The gathering has now no watermark. The absence of a watermark indicates that as the cancellandum had a watermark (one was present in the original sheet, and so in the original gathering) the gathering is a cancellans. The absence of a watermark from the gathering provides evidence of cancellation, but does not enable us to state which leaf is a cancellans: it might be any one of the four leaves in the gathering, as no leaf differs from any other.
• 3. The gathering contains two watermarks, which indicates cancellation. However, we now have some added evidence. Originally the gathering must have had a single watermark, and a cancellandum leaf without a mark has been replaced by a cancellans with a watermark. It therefore follows that the cancellans must be one of the two leaves with a watermark. It is found that one of the two watermarks always occurs in the third leaf of our hypothetical gathering, that is, the leaf with the engraved illustration. Where there are two watermarks in the gathering, both are found to be either type 1 or type 2; there is no occasion on which two watermarks are found in a gathering one of one type, and one of another. Replicate examination of the Virgil links the cancellation phenomena almost exclusively with engravings and permits no other explanation.

Why was such cancellation carried out? Rickaby's problems possibly lay with the fact that the plates used for the engravings were not new; they had been used previously for an edition of the Virgil printed by Caspar Fritsch in Leipzig during 1788 and 1789.[6] It a plate became too worn or damaged, it

might have been discarded. Because this could happen at different stages in the lives of individual plates, it is, moreover, conceivable that some books contain, or want, more illustrations than others. A second problem may have been that some illustrations were spoilt by bad press-work. Even in the book as issued, the vignettes occasionally cover some of the text,[7] and cancellantia, presumably containing the better state of a leaf, have the engraving misaligned with the text. Sheets with more than one engraving would have compounded the chances for error. To solve the problems he encountered, Rickaby seems to have cut up sheets with good engravings to provide cancellantia to replace the poor impressions in other sheets. With the recording by the ESTC of eighteenth-century printings still incomplete, it is difficult to know how many copies of the quarto Virgil still exist. At present some fifteen copies are known to ESTC and I have personally examined eleven of them. Bearing in mind that the same two papers have been used throughout the volumes, and allowing for the fact that, if the survey is large enough, we are able to detect only 37.5% of all cancellantia, then the results for each volume can be shown in tabular form (see Table 1).

Copies

• VMoU Monash University: *870.1 V497 A1/He.
• VU(1) Melbourne University: RBR.25D. Poynton.
• VU(2) Melbourne University: RBR.25D.
• BL(1) British Library: 56.f.5-12.
• BL(2) British Library: 687.k.2-9.
• Dm Marsh's Library, Dublin: W.1.17-21.
• BN Bibliotheque Nationale, Paris: Rés.m.Yc.435-442.
• GOT Universitätsbibliothek Göttingen: Auct.lat.II.7408.
• MH-H Houghton Library, Harvard: TYP705 93.868F.
• CtY Beinecke Library, Yale: Gnv6o. + A767C.
• CSmH Huntington Library, San Marino, California: 434 283.

Three comments are called for when consulting the Table:

• 1. Generally it is possible to state only that a particular gathering has a cancellans within it. This means that the ± signs used in the table represent a cancellation, within that gathering. The ± sign should not be taken to indicate absolute certainty that the cancellans is synonymous with the leaf bearing the engravings. Nevertheless, for easy reference the actual locations of all the engravings (throughout the volumes) are given. Thus there is a cancellans present in gathering k in vol. 1 in the copy now at Yale, and in all copies there is an engraving at k1v.

In all volumes there are instances where more than one engraving occurs in a gathering—in vol. 1, at D, M, 2P and 3C. In these gatherings it is not always possible to determine which leaf is a cancellans. These signs, like all other ± signs, indicate only that there is a cancellation within the gathering.

• 2. The amount of cancellation varies widely between copies. For example, in vol. 1 the BN copy has only one detectable cancellans, and the VMoU copy only two. On the other hand the BL(2) copy has six detectable cancellantia. Nevertheless, it is possible that these wide variations are a consequence of the 62.5% undetectable cancellantia.
• 3. In setting out Table 1 only the cancellantia are given which can be linked with cancelled engravings. There are others and, since these appear not to be Rickaby cancellantia, they must be analysed in detail. In vol. 1 the relevant gatherings, and the associated copies, are i (BL[1]); 1 (MH-H); O (BL[1]; BL[2]; Dm; CtY); X (VU[2]; BL[2]; Dm; MH-H; CtY); 2O (BL[2]; BN; MH-H). The opening gathering i 4 may have an ordinary or a Rickaby cancellans; it is simply not possible to tell.

Signature l in the Houghton copy has type 1 watermarks at l1 and l2. This is presumed to be an ordinary cancellation of matters textual, since there is no illustration within the gathering, and no place where an illustration might have been added. It may well occur in other copies and be undetected for the reasons already stated.

Signature O contains a cancellans in four copies. In the BL(2) copy there are watermarks at O2 and O3, one of which must be a cancellans. In all copies seen O3r has a large white space at its foot, and it is highly likely that there was once an engraving there, that it was not up to Rickaby's standards, and therefore that it was removed. It will be explained that at some stage in the book-production, some engravings, which were expected in the edition, were no longer included in the Virgil, and this is probably one of the lacunae. There is a chance that the leaf with the engraving will be found, in an as yet unexamined copy.

Signature X has a cancellans in five copies. X2 and X3 have watermarks in BL(2), X2 and X4 in Dm, X1 and X2 in CtY. X2 is the only common leaf here, and X2v has again considerable white space. Again this is presumably a Rickaby cancellation, and perhaps a cancellandum will be discovered.

Only signature 2O remains to be accounted for. In both the BN and MH-H copies, type 1 watermarks occur at 2O1 and 2O3. In all copies 2O3v has a large area of white space, and again this appears to be a lacuna where an engraving existed but was cancelled. The Rickaby cancellation is indirectly suggested by the apparent shortage of engravings in the final copies, but absolute proof depends, once more, upon the discovery of a cancellandum leaf with an engraving.

If we regard the existence of cancellation in vol. 2 in the same way as vol. 1, we are left with three anomalous cancellantia: b (BL[1]; BL[2]; CSmH); 2K (BL[2]; CSmH); 4G (CtY).

All three anomalous copies of gathering b lack watermarks. Nevertheless, there is a large white area at the foot of b2v, and this suggests Rickaby cancellation. In gathering 2K again both anomalous examples have no watermarks in the gathering and a white space suggestive of Rickaby cancellation is at the foot of 2K1v.

There are at least four more copies of the quarto Virgil which I have not examined. If Rickaby cancellation is sound as a principle, then two watermarks may be found in some of the second volumes, at b and 2K. Further, where two watermarks are found in these gatherings, one of them is likely to be at b2 and one at 2K1 to allow for leaves which held engravings, but which were cancelled. There is also a possibility that the cancellanda will be found.

In gathering 4G the CtY copy has type 2 watermarks in 4G3 and 4G4. One of these leaves must therefore be a cancellans. However, this is not a Rickaby cancellation, as the cancelled passage is in the middle of the Fifth Book of the Æneid.

The cancellations and their sites in vol. 3 in the same eleven copies are shown, except that the second part of the volume was not examined in the Göttingen copy, and here the anomalous cancellantia are at: M (CtY); 4F (VU[1]; BN).

The CtY copy has type 2 watermarks at M2 and M4. One of these leaves has to have been cancelled for textual reasons.

Gathering 4F in the VU (1) copy has no watermarks. But the BN copy has two type 1 watermarks, one at 4F3, the other at 4F4. It appears that the extra watermark at 4F3 could indicate a Rickaby cancellation as 4F3v has the lower half of the page blank.

There remains only vol. 4 to be considered. The volume now at Göttingen was not examined so that only ten copies were analysed. There are two anomalous cancellantia: 2A (BL[2]); 4A (MH-H).

Gathering 2A is split into two parts: the first two leaves conclude the first part of vol. 4, the second two leaves begin the second part. However, in the BL(2) copy there are no watermarks in the gathering and this may represent either a cancellation or simply that the two half-gatherings did not come from the same sheet.

Gathering 4A must contain a textual cancellation as it is in the middle of the index. In the MH-H copy there are no watermarks in the gathering, so that from this single anomalous copy it is not possible to suggest which leaf is a cancellans.

There are, then, only four examples of cancellation, in the copies which I examined, which appear to be textual:

• Vol. 1 MH-H copy has type 1 watermarks at l1 and l2.
• Vol. 2 CtY copy has type 2 watermarks at 4G3 and 4G4.
• Vol. 3 CtY copy has type 2 watermarks at M2 and M4.
• Vol. 4 MH-H copy has no watermark in gathering 4A.

At present nothing can be gleaned from gathering 4A in the Houghton copy, other than to state the obvious: that one or more leaves have been cancelled with the removal of the original watermark. This leaves only three other copies with apparently 'ordinary' cancellantia, and each example is unique. More copies, with cancellantia in the same gatherings, need to be examined before any conclusions can be drawn from these isolated examples. It appears strange that in the three cases of two watermarks within the gathering, identical paper has been used on each occasion, for, if a textual cancellans was required, printing must have been after the regular letterpress run, and would not necessarily have employed the paper which was used in the first instance. We might speculate that in only three cases the presence of identical paper is pure chance; it may also be significant that in each of these three cases the paper type could be identical because the watermark belongs to a pair of conjugate leaves: perhaps there has not been cancellation of a textual sort, but Rickaby bifoliar cancellation, perhaps because the letter press was poorly registered. Such speculation remains just that, without further evidence.

When all the watermarks were recorded, they reflected the statistics, in that many gatherings in which there is a cancellans (37.5% of all cancellations) had no watermark. There is a 1:4 chance that the leaf with the watermark has been removed and a 3:4 chance that it has been replaced by an unmarked leaf, that is 3:16 or 18.75%. In other words the analysis reveals only 37.5% of cancellantia, and half of these have no marks, whilst half have an extra watermark, in the gathering.

If the results of this rather limited analysis are tabulated they prove what is not visible to the bibliographer's prepared eye: which leaves—with an engraving—must contain the intrusive second watermarks in each gathering. Set out in Table 2 are the results of this analysis showing those gatherings with an engraving and in which two leaves with watermarks were noted. The leaf with the engraving is underlined.

The results show a relationship between an intrusive watermark and a leaf with either an engraving or a white area (where it is suggested that there was once an engraving). There are two anomalies in the table, both in vol. 1

of the Harvard copy. In gathering U, U1 has an engraving, but watermarks were found in leaves U2 and U4; in gathering 2A though there is an engraving at 2A3, watermarks are present in leaves 2A2 and 2A4. This suggests a bifoliar cancellation, or if it is a single leaf, a textual and not a Rickaby cancellation.

It will have been observed that records are given for only three of the four volumes. This is because only one copy shows any gathering with intrusive watermarks in vol. 4. This is the Harvard copy and in gathering n watermarks are present in n2, n3 and n4. Both n3 and n4 have engravings, which means that originally there was a watermark only at n2 within the gathering, but single-leaf cancellations at n3 and n4 have fortuitously brought two more watermarks in.

It is now possible to summarise cancellations in the Virgil: there are 69 engravings present in 54 gatherings, at least 38 of which have cancellantia within them, though once again it must be stressed that many cancellantia

will go undetected. In addition to these confirmed Rickaby cancellantia there are twelve cancellantia with no engravings. It seems reasonable to suppose that six of these are Rickaby cancellantia, where the printer, it is presumed, disposed of an engraving which was unsatisfactory. Thus we have perhaps only six cancellations of the normal textual sort and evidence that cancellation was practised to improve the appearance of engravings.

It is then my considered opinion that there appear once to have been 75 engravings adorning the Virgil, six more than are present in each of the copies examined for this paper. Confirmation of the originally greater number of vignettes is in fact given in the Critical Review 10 (1794): 303-304 where amongst the encomiums on the Virgil we read 'the plates (in number no less than seventy-five)'. Here surely is the strongest indication of Rickaby cancellation. Do any copies extant contain 75 engravings? Did any copy ever have that many? Did the reviewer base his remarks on advice from the booksellers or the printer? It is a very laborious process to count the engravings for whilst three may be found in close proximity, the next one may be found many, many pages on. It appears that the reviewer in the Critical Review relied on advice from an 'informed' source, or a very early source, possibly even a Proposals leaflet. Wherever this information came from, doubtless the reviewer passed it on in good faith, and in so doing confirmed the existence of six engravings which have not been seen, and Rickaby cancellation.

This 'improvement' of vignettes—rather than text—is the explanation for the observations noted. It means as a corollary that a careful description of the cancellantia within the work in the usual manner is pointless. There is little value in stating A-C4 (±C3) D-F4 (±F2) etc. when in another copy it may be A-B4 (±B2) C-D4 (±D1) E-G4 etc., even if we are fortunate enough to know that in these copies these leaves are definitely cancellantia. Such a formula is glib, as it stands, suggesting that the bibliographer is certain that particular leaves, and only these particular leaves, are cancellantia. The nature of Rickaby cancellation in the Virgil is such that some leaves which are cancelled will remain undetectable. This has important ramifications, for the phenomenon of Rickaby cancellation bedevils classical bibliographical description, though the possibility of multiple ideal copies had been suggested by Tanselle in SB 33 (1980): 18-53 (and perhaps especially p. 47). Now two quite different copies—as far as cancellantia are concerned—may be ideal, for these were the states which the printer released to the public. Since not all cancellantia can be identified in the Virgil with our present techniques, no certain description of any particular copy is possible. However, even with the statistical limitations of undetected cancellantia and the problems associated with wove paper, we may be able to offer a description which will, at least, alert the bibliographer to the need for care when examining an edition in which there are Rickaby cancellantia. If the bibliographer describes a quarto copy as A-C4 (±C1) D-E4 (±E2) . . ., then it is anticipated that there are possibly two loci for cancellantia, at C1 and E2, and absence

of the cancellantia suggests the presence of cancellanda. With Rickaby cancellantia the loci are not fixed throughout the edition, and not only do we expect to find variability between copies, but even what is known is still likely to leave about 60% of cancellantia undetected in the Virgil. I propose that in view of these factors the bibliographer should describe copies in a manner which will point up the differences between Rickaby cancellation and ordinary cancellation, and where the formula for two copies might be:

• A: A-C4 (±C1R) D-E4 (±E2R) . . .
• B: A-B4 (±B3R) C-D4 (±D4R) . . .

Here the copy is inextricably linked with the formula, and the bibliographer realises that copy A—possibly uniquely—has Rickaby cancellantia at C1 and E2 because the capital letter R is used as an index. Again the nexus is clearly shown in the second formula where copy B has Rickaby cancellantia at B3 and D4. In the case of the Virgil we might write, for example, of parts of vol. 2:

• GOT . . . 3I-3K4 (±3K1R) 3L-3X4
• MH-H . . . 3I4 (±3I4R) 3K4 (±3K1R) 3L-3X4 . . .
• CtY . . . 3I-3X4 (±3X2R) . . .
• CSmH . . . 3I-3X4 (±3X2R) . . .

—always providing that we could be certain which leaf was the cancellans in each copy. There is now an absolute nexus between copy and the Rickaby cancellantia within that copy. Of course, it may be that further examination of copies reveals information about the status of a copy that was previously examined, but the distinctive formula alerts the bibliographer to conditions governing the cancellations. Part of the importance of this novel style of describing the Rickaby cancellation status of particular copies may lie in future findings. If we consider the formulae advanced for the sections of vol. 2 described above, we see that the Yale and Huntington copies appear the same:

• CtY . . . 3I-3X4 (±3X2R) . . .
• CSmH . . . 3I-3X4 (±3X2R) . . .

However, bibliographical techniques may be developed which enable investigators to demonstrate cancellations at a number of other sites, cancellations which will clearly distinguish the Yale and Huntington copies.

Some, but not all, copies of the Virgil have gathering 2Yy4 at the close of vol. 4 detailing addenda and corrigenda. To a potential buyer of the works of Virgil this substantial list might have been somewhat discouraging, though it might not have bothered the buyer of the illustrated works of Virgil.

There are only fifteen copies of the Virgil known at present, and even with the completion of ESTC this number is unlikely to be doubled. It would have been interesting to ascertain whether Rickaby had intended to print more quarto copies than this, but so far it has not proved possible to trace any of his records. A great deal of paper had been wasted, as well as a great deal of ink and printers' time. What were needed now were positive reviews

to ensure successful sales of the Virgil. Rickaby himself printed the British Critic and it is tempting to think that he influenced the favorable reviewer of the Virgil in that work.[8] However, from a new price of £21 in boards the price of the Virgil tumbled. In Lowndes's Bibliographer's Manual, prices are quoted for the quarto ranging from the 16 guineas for lot 341 in the Steevens Sale in 1800 for a copy bound in russia, to £5-12-6d for lot 830 in Part 3 of the Sale of Sir Mark Masterman Sykes in 1824.[9]

Before leaving the Rickaby Virgil mention should be made of one other curious feature. The tail-piece on 3X2r of vol. 2 remains something of a puzzle. The engraving there has been clumsily sandwiched between the text and the direction line, with the plate mark overlapping two lines of text. Has any reader seen a possibly different state where the tail-piece is set beneath the text and the re-positioned direction line? This is where the engraving might be expected.

If applicable, this phenomenon is probably present elsewhere and suggests the initial examination of other illustrated works printed by Rickaby. Thomas Rickaby printed more than 60 works in the eighteenth century alone, but few are illustrated. I have examined all the other relevant illustrated works of Rickaby's from that century, but only one bears the characteristic stigma of Rickaby cancellation. This is the Hudibras edited by Treadway Russell Nash and printed in the same year as the Virgil, 1793.[10] The Hudibras is in a limited edition variously described as of one or two hundred copies.[11]

Again the format is quarto and whilst there are a few plates in the early part of vol. 1i, the interest lies in the engraved vignettes which are placed as headand tail-pieces to the cantos. At the time of the analysis 25 copies of the Hudibras had been catalogued by the ESTC, and I examined eighteen of these. The same two papers used in the Virgil are used for the Hudibras, and the results are set out in Table 3.

Table 3 shows the position of the engravings and cancellantia within vol. 1i-ii, and again the obvious question is, how many more cancellations are there in this volume? There is only one anomalous gathering with a cancellation: 4M, which has no vignette. There is cancellation within this gathering in six copies: Gerrard, CSmH, VU, GOT, MH-H and PPL. Again, like so many of the apparently anomalous cancellations in the Virgil, this is probably a site of Rickaby cancellation. 4M2r has the conclusion of Part III Canto III, and holds only eight lines of text—less than half the page is occupied. Once there was probably a vignette to conclude this piece. There are no other cancellantia in vol. 1, and this suggests that all the cancellantia in this volume are Rickaby cancellantia.

There remains vol. 2, that part of the whole book with a title page reading 'Notes, on Hudibras.', which were written by Nash.[12] Here, originally, there was printed a final gathering 3T4 in the volume. However, it appears that originally 3T4 had 3T1 the close of the index, and 3T2 an errata list. 3T3 was then a blank leaf, which if it is still found in the volume is usually at the close of vol. 2. 3T4 was the title-page to the second part of vol. 2. Sometimes the blank is bound in immediately before the inserted title page, but the

Copies

• Gerrard Author's own copy.
• BL(1) British Library: 77.1.9-11.
• Owo Worcester College, Oxford: N.11.6-8.
• Eastnor Eastnor Castle.
• E National Library of Scotland: Ak.1.8.
• CSmH Huntington Library, San Marino, California: 142396.
• CLU-C(1) Clark Library, Los Angeles: * fPR3338 H81 1793.
• VU University of Melbourne: BX: f 827.42H.n.
• BL(2) British Library: G. 11617-19.
• BL(3) British Library: 673:i.15-17.
• O Bodleian Library: fol.Δ.669.
• C Cambridge University Library: Eb.11.7.
• Ck King's College, Cambridge: J.60.1-3.
• Dt Trinity College, Dublin: O.pp.1-3.
• GOT Universitätsbibliothek, Göttingen: 4°Poet. Angl. 5685.
• MH-H Houghton Library, Harvard: * f EC 65.B 9782H. 1793.
• PPL Library Company of Philadelphia: * O Eng But Hudi 1793.
• CLU-C(2) Clark Library, Los Angeles: * fPR 3338 H81 1793a.

fact that these leaves are χ1 and 2χ1 rather than χ2 reveals their origin. The original gathering 3T4—the four leaves wherever they are placed in vol. 2—occasionally have two watermarks within them, notably, in this study, in the copies at Ck and GOT. Presumably this Rickaby cancellation represents dissatisfaction with the 'Bevereye' engraving on the title page to vol. 2ii.

In the Hudibras there are eighteen gatherings with engravings, and cancellantia can be identified in fourteen of these gatherings—and one other where there probably once was an engraving. There are no other cancellantia. It does not seem that there is a possibility of anything other than Rickaby cancellation. As with the Virgil, so originally with the Hudibras, not all the leaves with intrusive watermarks betraying Rickaby cancellations were recorded. However, the watermarks were recorded in ten copies, which had, in total, 35 gatherings with extra watermarks. One difference from the Virgil was immediately apparent: in every case the cancellation is bifoliar. This is confirmed by leaf lengths in the two uncut copies examined. For example in the Eastnor copy there are two gatherings with two watermarks: 2D and 2L. 2D1 and 2D2 have watermarks, but here the engraving is found on 2D4. Similarly 2L1 and 2L2 have watermarks but the engraving is on 2L4. In these gatherings Rickaby cancellation has caused a leaf containing a watermark to be introduced conjugate with the leaf bearing the engraving. Now it might be that the leaf conjugate with the engraving had to be cancelled, but this can be dismissed, by analogy with the Virgil, on two counts: the certainty of bifoliar cancellation in the uncut copies, already mentioned, and the regularity of cancellantia being associated with gatherings containing engravings.

Nash proceeded from the King's School, Worcester, in 1740 to Worcester College, Oxford, as a Cookes Scholar. Later he took orders and became an official of Worcester College before inheriting substantial estates, which caused him to resign his college positions at the end of 1758, move to Bevere and marry. It is possible that two of the copies have a direct link with Nash: those now at Worcester College, Oxford and Eastnor Castle.[13] Nash fathered only one daughter, and in 1785 this daughter, Margaret, married John Somers Cocks, eldest son of Baron Somers. Margaret Cocks became Lady Somers on the death of her father-in-law in 1806, and lived at Eastnor. It might be expected that a Nash-connected copy of Hudibras would be in the fine library now at Eastnor Castle. There is a red-inked copy at Eastnor, but though it has the bookplate of Margaret's husband, it does not have the armorial plate of Nash which is found in very many of the Eastnor books. Once more there is no definite link between the Eastnor copy and Nash himself.[14]

In the copies with red-inked engravings which were examined, the Gerrard, BL(1) and CSmH copies have all engravings in red ink—though, as noted, each of these copies possesses some Rickaby cancellations. However, the other four copies with engravings in red ink did not reach such a stage of perfection.

In the copy now at Owo the engravings on B2r in vol. 1i, and on the title page of vol. 2i are both in black ink. The Eastnor copy has vignettes in black in vol. 1 at e4r, B2r and 4R3v. In the other copies of the red-inked issue seen, those at E and CLU-C(1), the engraving at B2r in vol. 1i is again in black ink.

None of these gatherings, admittedly only seven in all, appears to possess a Rickaby cancellans, though this may reflect statistical probabilities. In the Eastnor copy, the Bevereye illustration on the title page of vol. 2 is badly mis-aligned, with the plate being skewed across the volume number beneath it.[15]

It has not yet been possible for me to examine other works with illustrations, except where they impinge on other work in progress. For this reason I can cite only one other case that I consider possibly a Rickaby cancellation. This is the bifoliar cancellation of 7Q1.4 in a copy of AI∧IANOΓ IIEPI ZΩΩN IΔIOTHTOΣ (London: 1744). It appears, on analogy with the Virgil and Hudibras, that the cancellation in one of the copies at the British Library

(54.g.9) was effected because the engraved illustration in the final leaf of the book was defective. This is the only copy of this edition that I know of with this leaf (or conjugate pair of leaves) cancelled, and there appears to be no textual difference from other copies.

Since there is no known archival material of Rickaby's substandard illustrations, it is possible only to conjecture what his standards may have been. All that can be said is that, in the two editions treated here, Rickaby's 'quality control standards' of 1793 found some illustrations to be unacceptable for publication. Rickaby cancellation applies to an illustration, rather than the text upon a leaf, for otherwise there is no distinction from any other cancellation. It is important to note that Rickaby cancellation is a term that I have coined purely from the source of the discovery; there is reason to suppose that Rickaby cancellations occur in other illustrated editions, not printed by, or associated in any way with, Thomas Rickaby. Furthermore, though this article describes the use of stock paper—that is, the same paper that is used throughout the edition—for Rickaby cancellations, there is no reason why this should be so for Rickaby cancellations of other editions. Indeed if an intrusive paper had been used for the cancellations described in this article, the determination of the specific cancellans leaf would have been much easier, just as the cancellantia are easily detected in Baskerville's 1757 Virgil. It is likely that Rickaby cancellations, in editions which have yet to be discovered, use previously printed sheets (or leaves) with, or without, engraved material; cancellation of the engraved material may be with a 'better' engraving than was originally present—whatever that may mean—or with a sheet, or part of a sheet, which had not been through the rolling press, but only the letter press.

Notes