Facsimile or Forgery? An Examination of America, Plates 4 and 9, Copy B
In 1927 the copy of Blake’s America we now know as copy L was sold as a “facsimile reprint” for £4, though it is quite genuine.1↤ 1 G.E. Bentley, Jr., Blake Books (Oxford: Oxford Univ. Press, 1977), p. 104. It is not unheard of for an original work of art to be sold as a facsimile, but it is certainly more common for a facsimile to be sold as an original. The authenticity of plates 4 and 9 of America copy B, in the Pierpont Morgan Library, has never been doubted in print. But in fact neither one is an original impression. Although both are facsimiles photomechanically reproduced on wove paper, very little about either image appears suspicious. Indeed, both are reproduced in The Blake Collection of Mrs. Landon K. Thorne.2↤ 2 G.E. Bentley, Jr. (New York: Pierpont Morgan Library, 1971), plates IX and XI. America, copy B, was given to the Morgan Library by Mrs. Thorne in 1973. For the complete provenance of copy B, see Thorne Catalogue, p. 26, or Blake Books, p. 100. The true technical and historical origins of plates 4 and 9 are not easily identified. What information we have is supplied only from the plate measurements and foliation, the binding, and the inscription on the front flyleaf of copy B. From these few facts, and with much caution, we can begin to speculate about who executed the plates, when, how, and why.
Plates 4 and 9 are printed on the same paper, the feel, weight, and texture of which are, from the verso, noticeably different from the other sheets in the book. The paper—which called to mind Murillo, an etching paper I had printed on before—and not the image, made me suspicious enough to check some measurements.3↤ 3 Thomas Lange, the assistant curator in charge of books and prints at the Pierpont Morgan Library in 1978, also noticed the difference in paper and suspected the two plates. Bentley seems to have been a bit puzzled by the difference too, but noted only that plate 9 was printed on “stiffer paper than the rest” (Thorne Catalogue, p. 26). As in most copies of America, the last four lines of plate 4 have been masked out. Unlike all other copies, however, the bottom platemark is only 2.4 cm. from Orc’s foot, instead of the usual 4.1 cm. If this bottom platemark were caused by the material used to mask out the four lines, then, because 2.4 cm. is not wide enough, half of line 3 and all of line 4 would have printed. If this were the platemark of an authentic impression, then Blake’s plate had been cut in about the middle of the third line, a possibility not yet to be ruled out since the plate could have been pulled posthumously and there is precedent for such plate tampering.4↤ 4 In the posthumous pull of “The Little Black Boy” (pl. 9), the woman’s bun was removed and her back narrowed with a burin. Songs of Innocence (U) contains such a pull, though it is recorded as an extra impression (Blake Books, p. 366). See “Posthumous Pulls in Songs of Innocence, copy U” (forthcoming). To make sure I was not looking at a late impression or posthumous pull, I also checked the distance in plate 4 from the top platemark to Orc’s toe: 19.5 cm. in copies A and L, but 20.4 cm. in copy B. In plate 9, the distance from the top platemark to the first line is 11.6 cm. in copy L, but 12.0 cm. in copy B; from the bottom of the text to the bottom platemark is 8.4 cm. in copies A and L, but 9.15 cm. in copy B. Because the relation of image to platemark did not check out, I measured the distance between parts within the image. For example, line 4 of plate 4 is 12.6 cm. long in copies A and L, but 13.0 cm. in copy B. Line 7 of plate 9 is 11.5 cm. long in copies A and L, but 11.9 in copy B.5↤ 5 These and a list of other measurements concerning plates 4 and 9 of copy B have been corroborated—and corrected—by Robert N. Essick, who has copies of both facsimiles in his collection, and who has been most generous in sharing information. He and G.E. Bentley, Jr. have read early drafts of this paper and have made valuable suggestions, many of which have been incorporated in this final version. More revealing than incorrect measurements within the image and between image and platemarks, however, are the plate measurements themselves. Plate 4 of copy B is 22.9 × 16.9 cm. vs. 23.8 × 16.6 cm., and plate 9 of copy B is 24.9 × 17.5 cm. vs. 23.5 × 16.8 cm.6↤ 6 Blake Books, p. 70. These and other differences are too great to have been the result of one paper shrinking less than the others. Damp printing paper shrinks only 1 to 2.5% of the sheet size,7↤ 7 Philip Gaskell, A New Introduction to Bibliography (Oxford: Oxford Univ. Press, 1972), p. 13. which could not, for example, account for the 1.4 cm. difference between the length of plate 9 of copy B (24.9 cm.) and of copy A (23.5 cm.).
Except for the platemarks, plates 4 and 9 are free of the embossment that is characteristic of a relief etching, or a process block. Instead, the ink lies flat on the paper, which is characteristic of lithography. The platemarks are themselves suspicious, as I’ll explain later, and do not positively identify the kind of lithography used. But whether these impressions were pulled from stone or metal plates, whether produced from a transfer or directly from a reverse negative, one thing is certain: they are photolithographs. The proportion of the parts to each other is correct, which indicates the work of a camera. But compared to the originals, the overall dimensions are slightly distorted, or elongated, which indicates either the slight distortion in the negative caused by the camera lens or in the contact between the transfer paper and the stone. In any event, with a photomechanically reproduced image, a good solid black ink (and not the thinner greyish black of tonal lithography), an impression that has been touched up in black water color (like the other prints in copy B), and platemarks, it is little wonder that these two plates have escaped detection.
America copy B was stabbed through two different sets of three holes. Plates 4 and 9, however, have no stab holes, though both plates are numbered. Copy B has two sets of begin page 217 |begin page 218 | begin page 220 | numbers: 1-16 and 1-18. Both sets are in pencil and start on plate 1, and neither set is by Blake or C.H. Tatham, the first owner of copy B.8↤ 8 On the verso of the frontispiece is the inscription: “From the author/C H Tatham Octr 7/1799.” Presumably this is Tatham’s hand; it is not Blake’s. In any case, the “7” and “9” do not resemble the “7” and “9” in either set of numbers. Blake did number two copies of America (A and M) 1-16, but he started on plate 3, not plate 1, and numbered, as was his custom, in the right hand corner of the plate in ink. Numbers 1-16 are written just below the platemark on the left hand side; numbers 1-18 are written at the top right corner of the leaf. Plates 4 and 9 are the two plates not in the 1-16 sequence, which indicates that 1-16 is the set of numbers first given to the plates.9↤ 9 The numbers of set I may not, however, have been the first attempt at pagination. Numbers 4, 5, 7, and 11 of set I are written over other numbers, or marks, that were erased; numbers 6 and 8 are very light, and a 7-like mark is written over the 3. Plates 4 and 9 are numbered in set II, and just below “4” and “9” is a “7”.10↤ 10 The “7” is the same size (.6 cm.) and in the same position (2 cm.) from the edge of the sheet, right under the number). In plate 4, the bottom of the number is partly erased, or smudged. There is a larger smudge mark just under the “7” on plate 9. This “7” may be a printer’s mark rather than a number. That plates 4 and 9 are not included in the first set of numbers, and that neither sheet has stab holes, suggest that copy B had for a time only sixteen plates to number and stitch.
America was “splendidly bound in Citron morocco” when it sold in 1878 with the library of A.G. Dew-Smith at Sotheby’s for £16.5 (lot 247) to John Pearson. According to the Sotheby catalogue, copy B sold in 1874, unbound, for £18. Being bound by 1878 would seem to imply that the facsimiles were executed before this date, but this is not necessarily the case, because plates 4 and 9 are not in the binding but are tipped in. They are glued to plates 5 and 10, are without gilding, and are slightly lower than the rest of the pages, noticeable only when looking along the foreedge. The facsimiles enter after, not before, the binding.11↤ 11 The pages were trimmed before insertion. The two leaves are slightly lower at the top, and less so at the bottom. They appear flush with the fore-edge, but are actually a fraction lower, noticeable when the fore-edge is riffled.In short, plates 4 and 9 were cut and trimmed separate from the other sheets.
The tipping-in of plates 4 and 9 indicates that one of the owners after 1878 could also have been responsible for their execution and insertion. Three of these owners had access to facsimilists and to models, and had reason enough to complete an incomplete copy. John Pearson, the bookseller, published the excellent Jerusalem facsimiles with 100 photolithographs in 1877, and was William Muir’s first agent, issuing four of Muir’s Blake facsimiles in 1884-85. Bernard Quaritch, who owned the book in 1890, was Muir’s agent in 1885-94, and issued a series of facsimiles by William Griggs from his own and other collections, including Poetical Sketches (1890) and The Book of Ahania (1892). He also owned another copy of America (R), which he lent Muir for his 1887 facsimile. Bernard Macgeorge, who owned the copy from 1892 to 1924, also knew Muir; Macgeorge lent Muir Europe copy A, which was then missing five plates, and which were “all supplied in facsimile by Muir.”12↤ 12 A Catalogue of the Library of Bernard Buchanan Macgeorge (privately printed, 1906), p. 8. Macgeorge also owned Songs copy A, which entered the British Museum in 192713↤ 13 Blake Books, p. 412. The only owner after 1878 whom we cannot suspect is Thomas Gaisford, the Greek scholar, and a major Blake collector (Thorne Catalogue, p. 14). Gaisford died in 1855, yet his bookplate is pasted on the inside of the front board, which is why Keynes and Bentley have assumed that America (B) was “acquired by Gaisford” after it was bound, and between Pearson, who bought it at Sotheby’s in 1878, and Quaritch, who bought it from Gaisford’s library, which sold at Sotheby’s 23 April 1890. We do not know who acquired America (B) for Gaisford’s library, when, from whom, or for how much. There are other Blake books with Gaisford’s bookplate, four of which may have been acquired before 1855 (Innocence [H], Europe [E], Urizen [C], Poetical Sketches [N]), and at least three, besides America (B), acquired after 1874 (Visions [I], Songs [M], Thel [C]). Apparently, someone familiar with the collector’s taste in books continued to build the library. with photolithographs of plates 51, 52, 53, and b.
Before eliminating suspects, we must examine a pencil inscription on the front flyleaf: ↤ 14 “America, a Prophecy, 18 designs, Folio, Lambeth, 1793, Sotheby’s, 1855, £ 2.7”—W.T. Lowndes, The Bibliographer’s Manual of English Literature (1857), I, 215.
Lowndes [pens?] 18 designs14Assuming “is only 18” refers back to “designs” and not to £18, the price the 1878 Sotheby catalogue says it sold for in 1874, the inscription can still be read in one of two ways, depending on whether “designs” refers to pictures or pages. In the Prospectus, Blake uses the word “design” to mean picture, not plate or page.15↤ 15 In the Prospectus, Blake says that America has 18 designs, but he also says that The Marriage of Heaven and Hell has 14 designs (counting top and bottom illustrations on one plate—probably plate 3—as separate designs), and not 27 plates, and that the Visions of the Daughters of Albion has 8 designs, not 11 plates or pages. In other words, Blake is using the word “design” to mean picture, counting illustrations on one plate which are independent of one another as separate designs. America has 18 pages and at least 18 designs, and therein lies the confusion. America, minus plates 4 and 9, would still have eighteen designs if, as Blake himself did with the Marriage, one were to separately count the top and bottom pictures on two plates as separate designs. By counting in this way, there would, indeed, be “some copies” with “2 more” designs, whereas in fact there are no copies with twenty plates. If, on the other hand, “designs” is referring to pages, then the inscription is referring to a copy in which the two facsimiles have already been inserted, a copy, in other words, with eighteen plates. In this second interpretation, the statement “There are 2 more in some copies” is probably a reference to Gilchrist’s description of America as “a folio of 20 pages.”16↤ 16 Life of Blake (London, 1863), I, 109. Gilchrist’s description of America as “a folio of 20 pages” is most likely derived from Richard Thomas’s description of America in Nollekens and His Times (London, 1828), II, 477, as having “18 plates, or twenty pages, including the frontispiece and titlepage.”
There are 2 more in some copies
but I believe these to be a supplementary number
& that the book is perfect & is only 18
The inscription is written on a flyleaf, thus written after the book is bound.17↤ 17 The inscription is on the verso of the front flyleaf, which is the conjugated part of the end paper pasted to the front board. Because Blake’s original “bindings” were simply wrappers of laid paper, and I do not know of any copy of an illuminated book which has its original wrappers plus a leather binding, I am assuming that the inscription, as well as Gaisford’s bookplate, are on material supplied by the binder, and thus were added to the book no earlier than 1874. The insertion of plates 4 and 9 also follows the binding, but may precede or follow the inscription. There are only two possible combinations of these three bibliographical facts: binding, inscription, insertion, or binding, insertion, inscription. If the inscription precedes the insertion, the inscription refers to a copy that still has only sixteen plates. If the inscription succeeds the insertion, the inscription refers to a book with eighteen plates. Neither combination rules out the possiblility of the inscription and / or insertion occuring after 1878—unless we can date the inscription.
The Sotheby catalogue entry for the 29 January 1878 auction and a comparison of the numbers in set II with those in the inscription make me believe that the inscription followed the insertion of the plates. The catalogue entry describes copy B as:
. . . presentation copy with author’s autograph inscription, splendidly bound in Citron morocco, ornamented with variegated leathers and gold tooling, e.g. by F. Bedford; two leaves said to be wanting, but Blake’s original prospectus says “America, a prophecy in illuminated printing, folio, with 18 designs, Lambeth, W. Blake, 1793.” This copy, unbound, sold for £18 in 1874.The cataloguer’s “two leaves said to be wanting” seems to refer to the assertion in the inscription that “There are 2 more in some copies.” If so, the inscription was already written by 1878. And his “but Blake’s original prospectus says” clearly implies that two leaves are not wanting, that this copy of America has 18 plates, the number that the prospectus says it should. The idea that the cataloguer is counting pages rather than illustrations is supported by internal evidence. As I mentioned, the two sets of plate-numbering are by different hands, but the “8” in the inscription—a very distinct —is, I believe, the same “8” in numbers “8” and “18” of the second set of plate-numberings.
If the inscription is pre-1878 and is written by the same person who renumbered the plates, then the two facsimiles begin page 221 | must have entered America copy B between 1874, when it is reported to have been unbound, and January 1878, by which time it was bound and in Dew-Smith’s library. Who bought copy B in 1874 is unknown; it may or may not have been Dew-Smith. Thus plates 4 and 9 could have been inserted either before or during Dew-Smith’s ownership.18↤ 18 It is not unlikely that Sotheby’s was given the information about the 1874 sale by Dew-Smith, and that he knew it because he was the one who bought America unbound in 1874 for £18. But the idea that Dew-Smith was the 1874 purchaser, i.e., the only owner between 1874 and 1878, cannot be proven. I know nothing about Dew-Smith except that he was “of Cambridge” (Bibliography, 1921), and that he also owned Songs copy J and Blake’s copy of Swedenborg’s The Wisdom of the Angels Concerning Divine Love and Divine Wisdom, both of which sold, along with America copy B, at Sotheby’s, 29 January 1878 (Blake Books, pp. 100, 417, 696). His copy of Night Thoughts printed on vellum and without plates sold at Sotheby’s 27-30 June 1906 (Blake Books, p. 644 n2). Dew-Smith is not mentioned in the Dictionary of National Biography. Knowing nothing of the man, I cannot ascertain his handwriting or his integrity. In the first case, someone other than Dew-Smith purchased copy B in 1874, had it bound, inserted plates 4 and 9, and, possibly, even wrote the inscription and renumbered the plates. In the second case, Dew-Smith bought the volume either bound with sixteen plates, or unbound in 1874 and had it bound, only to find out later that it was missing plates 4 and 9. He had the two plates made, inserted them, and described his copy as perfect. But it is also possible that Dew-Smith wrote the inscription and renumbered the pages in good faith, having bought a bound volume with plates 4 and 9 already inserted.
Reproductive Process Used to Make Plates 4 and 9
In the 1870s, photolithographs could be printed from finegrained limestones and from specially grained zinc plates. To use stone, the negative of the copy to be reproduced is placed in contact with paper coated with a light-sensitive gelatine film. The areas exposed to light become insoluble in water, so that when the sheet is covered with a transfer ink and soaked in a bath of water, all the soluble (unexposed) gelatine lifts off the paper, leaving only the inked image ready to be transferred (counterproofed) onto the prepared stone. To use zinc plates instead of stone, a reverse negative is exposed directly on a sensitized zinc plate, which, because it eliminates the intermediate step of transferring the image, is “much more likely” to retain “the finer qualities” of the original.19↤ 19 Alfred Seymour, Practical Lithography (London: Scott, Greenwood & Son, 1903) p. 100. Reverse negatives were first used in the 1870s with collotypes, and are necessary if the print is to be the same direction as the copy reproduced. Like a counterproof transfer, a reverse negative places the image in reverse on the plate so that the print is in the same direction as the original. The fine details of plates 4 and 9, including subtle traces of ink along the sides of the text, suggest—as do the platemarks—that these two prints were pulled from zinc plates. Lithography is a planographic method which, at least theoretically, should not leave platemarks. But because a press and a scraper bar are used, printing with too much pressure and with a bar wider than the plate will emboss the shape of the stone or metal plate into the print. The platemarks on plates 4 and 9 reveal the even edges of a metal plate, not the irregular edge of stone. But, as I mentioned, these marks are suspicious.
Robert N. Essick has in his collection what are possibly the proofs of plates 4 and 9. His two impressions were printed on larger sheets (40.3 × 28.5 cm. for plate 4, and 44.5 × 28.5 cm. for plate 9, vs. 36.3 × 26.1 cm.), on different paper (“thin, hard, ivory colored, machine made”),20↤ 20 Private correspondence. and in a dull black ink. Neither of his prints has any trace of side platemarks, while the side platemarks are the ones most noticeable in plates 4 and 9 of copy B. In plate 4, the bottom, top, and left platemarks are sharply defined, while the right side is slightly beveled. In plate 9, the plate lines forming the two bottom corners and top right corner intersect; the bottom platemark, for example, is 18.5 cm. long, while the distance between the platemarks is only 17.5 cm. In Essick’s plate 9, this bottom platemark is “about 20 cm. long.”21↤ 21 Private correspondence. In both plates 4 and 9 there are tiny ink dots slightly outside the right side platemarks. These ink dots seem to be from the border of the relief-etched plates. If so, they are part of the image the camera photographed and are thus on the litho plate or stone. Essick’s two prints, however, do not show these ink spots. This slight crossing of plate lines in copy B and the fact that the side platemarks are heavier while altogether absent in Essick’s two prints mean that the side platemarks in plates 4 and 9 of copy B are faked.
Such platemarks can be made with a stylus along a metal ruler with a bevel, or along a metal plate. The top and bottom platemarks, however, appear to have been produced by a metal plate in the process of printing, and they are the same distance from the image in Essick’s two prints as they are in copy B. Thus, plates 4 and 9 are either printed from zinc plates, or the same blank metal plates are registered exactly on both sets of impressions.22↤ 22 It seems that a blank plate or a stylus was used to make platemarks in another illuminated book. The leaves of Songs of Innocence and of Experience (J), which are inlaid into larger sheets, were either “provided with artificial plate-marks by putting each print, damped, in press with blank plate” (Keynes and Wolf, Census, p. 59) or “carefully indented at the join of the inner and outer leaves . . . to look like platemarks” (Blake Books, p. 416). Pulling a lithograph face down on a blank metal plate through a rolling press would give platemarks, but such a trick doesn’t explain why side platemarks are absent in Essick’s two prints, or how platemarks can intersect, or why the length of a platemark from the same plate is different in two impressions. Perhaps plates 4 and 9 were printed on a litho press using scraper bars of two different sizes, neither of which was as wide as the plates, and thus not able to force paper over the sides, but which came onto and off the plates with enough pressure to catch the top and bottom edges. The missing side platemarks, therefore, had to be added with a stylus. In any event, because lithographs from zinc and stone would ordinarily have been printed without platemarks, the platemarks, however they were made, were made to deceive, to make a flat image appear slightly embossed like a relief etching. The intention to deceive makes plates 4 and 9 forgeries, not facsimiles.
The Original Prints Reproduced
For there to be two photomechanically reproduced forgeries, two authentic prints must have been photographed. The bottom four lines of the fake plate 4 are missing, so the prototype plate 4 must have been masked. And because photolithographs of the period could reproduce line drawings or designs only in pure black and white (half-tone screens not yet having been perfected23↤ 23 “The half tones of a negative cannot be reproduced by this method, as they can with collotype.” H. T. Wood, Modern Methods of Illustrating Books, 4th ed. (London: Elliot Stock, 1898), pp. 64-65. Wood goes on to say: “For this reason the process is applicable for reproduction; it cannot produce a print from a negative taken direct from nature. What it can do is to reproduce in facsimile any picture or design that has been produced by any printing process whatever (in which tints are not employed), type, engraving, wood-cuts, or lithograph, also any line-drawing in ink, pencil, or chalk, anything in fact which already possesses the grain, line, or stipple required to hold the ink.” ), the two authentic prints photographed must have been uncolored. In addition to there being requisites for the models, both fake prints also have distinguishing, albeit, more subtle, marks. Plate 4 has an extra line on the lower right side of the leaf underneath the text, ink splatters on the right side of the tree, opposite line 8, and at the end of the last word, “foretold”; plate 9 is missing the two birds flying under the lowest tree branch, has an ink trail along the left of the tree trunk, a top branch that is lightly extended to the leaves, and ink splatters along the right side of the plate and along the bottom just under the grass.
Of the seven monochrome copies with a masked plate 4, America (F), in the British Museum since 1859, comes closest to having these peculiar markings.24↤ 24 I have checked copies D, E, F, H, and L; Robert N. Essick has checked copy I, and Roger S. Wieck (Houghton Library, Harvard) has checked copy C. Plate 4 (F) has both sets of ink splatters and an ink trace that could have begin page 222 | been transformed into an extra leaf line, and the two birds in plate 9 are very faint. On the other hand, plate 9 (F) has no ink trail, no completion of the top branch, and no ink splatters along the right side of the text. But just as the ink stain at the bottom of plate 9 (F), the “JY 59” museum accession number (1859), and the birds could have been rubbed off the plate so that they wouldn’t show in an impression, the light marks completing the branch in plate 9 (B) could have been added, and the ink traces accidently printed from the litho plate itself, possibly the result of it not being thoroughly sponged during inking. Because an image on stone or zinc could be altered in such subtle ways, absolute verification may not be possible, but it does seem probable that plates 4 and 9 of copy F were the prints photographed by or for the forgerer.
We do not know for sure who is responsible for executing and inserting the two forgeries into America copy B. The person who wrote the inscription and renumbered the plates is not necessarily the person who inserted the two prints. The inscription may be the hand of an honest owner warding off Gilchrist’s implication that his copy with only “18 designs” was incomplete. But, according to my reading of the evidence—and other readings may be possible—we do know when (between 1874 and 1878), how (photolithography), where (London), the probable model (F), and, I believe, why. Plates 4 and 9 are not facsimiles; they are not taken from a lost facsimile edition of America made in the 1870s. In short, they are not like Trianon Press facsimilies doctored up to fool the unsuspecting eye. Plates 4 and 9 were executed from the beginning with the intention to pass as originals and thereby secretly complete an otherwise incomplete copy. The plates have done well at both. Many professional bibliographers and bibliophiles have examined this copy of America. To their credit, plates 4 and 9 are not only extremely good forgeries, but, and I think this is equally important, as pages in a bound volume, they appear innocent by association. Had plates 4 and 9 been separate prints, it is more likely they would have been detected long ago. As it is, no other illuminated book, bound or loose, is known to contain a facsimile. Do we need to look harder? “Let the collector of prints be cautioned . . . to beware of buying copies for originals. Most of the works of the capital masters have been copied, and many of them so well, that if a person be not versed in prints, he may easily be deceived” (William Gilpin, An Essay on Prints, 4th ed., 1792).