articlebegin page 80 |
The Printings of Blake’s Dante Engravings
As readers of the annual sales reports in this journal may have noticed, complete sets of Blake’s engraved illustrations to Dante’s Inferno have consistently fetched over $30,000 at auctions during the last few years.*↤ I am indebted to Ruth Fine and Greg Jecmen of the National Gallery of Art, Washington, Arthur Freeman, and Thomas V. Lange for their help with the research reported in this essay. As a consequence of this high value, dealers have become increasingly interested in ways of ascribing a particular set of the seven plates to a specific press run since the earliest impressions would command the highest prices. In response to requests for information on this matter, as well as a desire to demonstrate how economic imperatives influence the study of printed images as much as their making, I began to study various sets of the Dante plates to correlate their physical characteristics with the known dates of printing. Since there are records of only three print runs in the nineteenth century, the problem seemed easily solvable. But the situation is far more complicated than I had anticipated, as the following report will demonstrate.
The documentary record of printings is less complete, and its interpretation by scholars less accurate, than one would wish. The first to offer any information on the subject were Bentley and Nurmi, who in their 1964 Blake Bibliography quote from pertinent documents in the Joan Linnell Ivimy collection. Since John Linnell had commissioned the Dante illustrations from Blake in 1824, and since the copperplates remained in the Linnell family’s possession until their sale to Lessing J. Rosenwald in 1937, these records have considerable authority. In a “List of John Linnell Senior’s Letters and Papers,” John Linnell, Jr., wrote that the Dante plates “have been printed at two dates, after a few proofs by Blake—J[ohn]. L[innell]sen. in 1 S[eries]—(at Bayswater[)] had India proofs taken (all disposed of)[.] soon after J. Ls death (J. L. jnr)—had 50 copies (india on drawing paper) printed by Holdgate.”1↤ 1 G. E. Bentley, Jr., and Martin K. Nurmi, A Blake Bibliography (Minneapolis: Univ. of Minnesota Press, 1964) 89. The interpolations in brackets are from Bentley and Nurmi. This statement provides crucial information: there were only two printings, laid India paper (or chine appliqué) was used for both, and Holdgate Bros. printed the second issue of 50 impressions “soon” after Linnell’s death in 1882. We are not informed, however, about the date, printer, and press run of the first printing.2↤ 2 The senior Linnell lived at 38 Porchester Terrace, Bayswater, from 1828 to 1851, and thus his son’s comment that the Dante plates were printed “at Bayswater” (i.e., while the family lived there) is not very helpful in fixing a date. Fortunately, other records in the Ivimy collection fill in these gaps. Two receipts from the plate printers Dixon & Ross, No. 4 St. James’s Place, indicate that £2.15s. was paid by Linnell on 26 September 1838 for “25 of each of 7 Plates Dante India,” and a further £1.10s. on 2 October 1838 for “95 Imprs. of 7 Plts. Dante India.”3↤ 3 Bentley and Nurmi 89. For information on Dixon & Ross (founded 1833) and its successor firm, see Iain Bain, “Thomas Ross & Son, Copper- and Steel-plate Printers since 1833,” Journal of the Printing Historical Society no. 2 (1966): 3-22, and Tony Dyson, “The Ross Records: 1833 to 1900,” Journal of the Printing Historical Society no. 12 (1977-78): 52-67. These receipts make it clear that there were actually two early press runs. That they were paid for only a week apart, and may have been performed even closer in time, accounts for the younger Linnell’s reference to nineteenth-century printings at only “two dates,” a habit continued by modern scholars but one that I will abandon here. As for the third printing, a draft of a letter from John Linnell, Jr., to the London book dealer Bernard Quaritch, dated 6 May 1892 states that the Dante “proofs” (i.e., impressions from the 1838 printings) had been “disposed of” and that the family was “about to obtain a few more copies of proofs—similar to the former ones.” Bentley and Nurmi reasonably conclude that “the second [i.e., third] commercial printing of the Dante plates was therefore about 1892” (89).
The letter actually sent by the younger John Linnell to Quaritch has never been located, but in October 1989 Arthur Freeman discovered a letter from another Linnell son, William, in the archives of Bernard Quaritch, Ltd. The references to both the Job and Dante engravings make this previously unrecorded letter worth printing in full:
May 18th. 92 The Avenue 76 Fulham Road SWbegin page 81 |
Dear Mr Quaritch
I waited a few days before sending you the five sets of Job you ordered—thinking I might send the Dante same time—but my brother James whom I saw yesterday tells me that there is a hitch in getting the Dante printed because the man whom my father employed & whom we wish to do it now is ill, though he is expected soon to recover— I told my brother that I would undertake to get it done safely & as well by some others—but in the country they are I think too conservative & slow coach—however I will have my man bring up the 10 sets of Job either tomorrow or Friday—I was about writing fathers to bring me 5 sets tomorrow but if they cannot look out the 10 in time he will bring them up Friday. With regard to the printing of the Dante it occurs to me that perhaps you might know who is the best copperplate printer in London and if you know that he is also thoroughly trustworthy (supposing we left the plates in his hand for a time) I think we might save time.—of course I should give the printer precise instructions as to the style of printing, etc.
With regard to the Papers which you alluded to in a previous letter—have you any suggestions to make as to the particular kind?
I rather think that Whatmans drawing paper is the nearest in quality and appearance to the old prints. I mean India on Whatman. So at present I cannot say any definite time when the Dante will be ready—but I will endeavour to get on as fast as I can make my brother James move on—he being Co Trustee we must go together.
Yours dearly and very truly
Several details in this letter are significant. Since the elder Linnell died in 1882, “writing father[’]s” for more Job impressions must refer to his home at Redstone Wood, Redhill, Surrey, which John Linnell, Jr., probably inherited as the eldest son. Surprisingly, there is no direct reference to John Jr. (born 1821), for the letter is written as though William (born 1826) and James (born 1823) had sole responsibility for the reprinting of the Dante plates.4↤ 4 All information about the Linnell family is taken from Alfred T. Story, The Life of John Linnell, 2 vols. (London: Bentley and Son, 1892), based in large part on the elder Linnell’s manuscript autobiography, and Katharine Crouan, John Linnell: A Centennial Exhibition (Cambridge: Cambridge UP, 1982). I have not been able to find the death dates of any of the Linnell brothers. Presumably John Jr. was still alive in 1892, if Bentley and Nurmi and Keynes (see note 5) are correct in ascribing the draft letter in the Ivimy collection to his hand. William’s request for advice on finding a printer for the plates opens up the possibility that Holdgate was chosen on Quaritch’s recommendation. In his draft letter of 6 May, John Jr. notes that the restrikes should be “similar” to the 1838 impressions. William’s letter stresses this desire to imitate the earlier printings and alerts us to the possibility that impressions on Whatman paper are from the c. 1892 printing.
The 1838 receipts, quoted above, are unambiguous as to the number of impressions (25) taken from “each of 7 Plates” in the first printing; but the record of the second printing, paid for5↤ 5 Keynes, “A Note on the Later History of the Dante Engravings,” Blake Studies: Essays on His Life and Work, 2nd ed. (Oxford: Clarendon Press, 1971) 228-29 (the first publication of this “Note”); Blake’s Illustrations of Dante (London: Trianon Press for the Blake Trust, 1978) . In the 1978 volume, Keynes states that “one hundred sets” were printed c. 1892 (), but gives no documentary evidence for this number, twice as large as the figure recorded by John Linnell, Jr. I assume that Keynes is simply mistaken on this point. Todd, in his essay on the twentieth-century restrikes (see note 13), repeats Keynes’s figure for the c. 1892 printing and claims that “170 sets” were printed in 1838 (p. 4 in the last version of Todd’s essay). Todd cites Bentley and Nurmi (88-89) as the authority for this latter number, but they say nothing about the number of impressions pulled in 1838 other than quoting the receipts. I take Todd’s “170” to be another error. This figure has been frequently repeated by dealers and auction houses and seems to have been widely accepted as an indisputable fact. But one detail in the 1838 receipts should give us pause. If 25 sets cost £2.15s. on 26 September, why would 95 cost only slightly more than half that amount six days later?
Bentley provides answers to the questions surrounding the 1838 press runs in his Blake Books of 1977. He prints therein the following transcription made by Iain Bain from the day-books (i.e., business records) of Dixon & Ross: ↤ 6 G. E. Bentley, Jr., Blake Books (Oxford: Clarendon Press, 1977) 545. Except for “[sheets]”, used here in the place of ditto marks in the original, the bracketed interpolations are Bentley’s.
Wednesday 26 September 1838This daybook apparently records the actual printing, as distinct from the billing dates, and thus the two press runs occurred only three days apart. As Bentley’s bracketed interjections indicate, these two printings yielded 38 sets (and four additional impressions), not the proverbial 120. His accurate figure is crucial in shaping expectations about extant suites of the engravings. If the larger number had been produced, then we could expect that most extant sets were printed in 1838 since only 50 were produced c. 1892. But we now know that well over half of the 88 documented nineteenth-century impressions on laid India were printed c. 1892. Given the additional 54 years in which the 1838 impressions could have been destroyed or lost, we can expect most extant suites to consist of c. 1892 impressions.
Mr Linnell 25 of each of 7 plts Dante Ind[ia]
87 1/2 Sh[ee]ts. of th[ic]k. pl[a]t[e]
Col[ombier] [i.e. 2 pulls per sheet]
25 [sheets] of India [7 pulls per sheet]
Saturday Septr 29th 1838
Mr Linnell 95 Impns of 7 plts Dante India [i.e. 13 4/7 sets]
47 Shts of Plt. Col. [2 pulls per sheet]
13 4/7 [sheets] of India [7 pulls per sheet]6
Both 1838 press runs recorded in the Dixon & Ross daybook used “Col[ombier]” paper. This apparent reference to the venerable French papermaker was probably the basis for Keynes’s assertion that all 1838 impressions are on “French Colombier paper.”7↤ 7 Blake’s Illustrations of Dante . In Blake Studies, Keynes calls the backing sheets for the laid India “(French) Colombier plate paper” (228). However, by the late eighteenth century, “Colombier” was regularly used in reference to a sheet size, not a specific manufacturer or country of origin.8↤ 8 See E. J. Labarre, A Dictionary of Paper and Paper-Making Terms (Amsterdam: Swets & Zeitlinger, 1937) 54. The Colombier produced by Whatman, whose paper Blake used for many of begin page 82 | his illuminated books and the Job engravings, measures 34 1/2 × 23 1/2 inches (87.6 × 59.7 cm.).9↤ 9 See Whatman’s mold sizes recorded in Thomas Balston, James Whatman Father & Son (London: Methuen, 1957) 61. Labarre 54 gives the same dimensions as the standard for Colombier. Further, the Dixon & Ross daybooks show that the firm acquired five quires of Colombiersize paper in 1835 from S. Tipper & Co., agents for Whatman.10↤ 10 Bain 7. Thus, there is no reason to assume that any 1838 impressions were printed on a paper of French manufacture. A French paper may have been used, but I have yet to find a Dante set, or any record of a set, definitely printed on such a paper.
Armed with the documentary record surveyed above, we can now turn to extant suites of the Dante engravings. Our concern should focus on the quality of the impressions, particularly as this shows various amounts of wear indicative of printing sequences, and the paper on which they are printed. Handwritten inscriptions and provenance information can also supplement the primary physical evidence.
We can begin by setting aside the easily-identifiable impressions pulled in this century. In 1953-55, Rosenwald had sets printed on heavy, dead-white wove paper with a surprisingly bold, pebble-grain surface. The plates had to be printed with considerable pressure in order to smooth the paper sufficiently to register fine lines. In a complete suite of these restrikes in the Huntington Library, San Marino, California, each sheet measures 35.5 × 50.5 cm. and is inscribed in pencil, lower right, “Impression taken from the copper plate in my collection 1953-4[.] Lessing J Rosenwald 4/19/55.” One sheet shows a watermark (“MADE IN ENGLAND [space of approx. 15 cm.] LINEN FIBRE”); another bears a few fragments of these letters. A set in the National Gallery of Victoria, Melbourne, has the watermark on four sheets, plus a countermark (“UNBLEACHED ARNOLD”) on the remaining three, and is inscribed on each plate, “Impression taken from the plate in my collection in 1955[.] Lessing J. Rosenwald 1/20/60.”11↤ 11 According to Ted Gott, “Eternity in an Hour’: The Prints of William Blake,” in Martin Butlin and Gott, William Blake in the Collection of the National Gallery of Victoria (Melbourne: National Gallery of Victoria, 1989) 187-88. I suspect, but cannot confirm, that all these restrikes bear some12↤ 12 I take the titles for each plate from those given the corresponding water colors in Martin Butlin, The Paintings and Drawings of William Blake (New Haven: Yale UP, 1981).
Rosenwald commissioned a further printing of 25 sets in 1968.13↤ 13 Information about the 1968 printing is taken from Ruthven Todd, “Blake’s Dante Plates,” Times Literary Supplement (29 August 1968): 928, and its two revisions: “Blake’s Dante Plates—Revised Version,” Book Collecting & Library Monthly no. 6 (October 1968): 164-71, and Blake’s Dante Plates, Reprinted, with Additions, from the October 1968 Issue of Book Collecting & Library Monthly (n.p., n.d.). All references to Todd are to the final version. The complete 1968 set in the National Gallery of Victoria is reproduced in Gott, figs. 50a-g. These were pulled by Harry Hoehn, who thoroughly cleaned the copperplates, on Japanese “Kochi,” a laid paper, with one set on “German copper-plate” (Todd 5). The Kochi paper (or “Koji,” the name of a papermaking district, not a specific manufacturer) is fairly thick but soft, with a much smoother surface than the pebble-grain paper used c. 1954. In a letter dated 29 July 1968 to Ruthven Todd, Rosenwald stated that he inscribed each impression in pencil, “Printed by Harry Hoehn 7/68 . . . Restrike from a plate in my collection . . . Lessing J. Rosenwald” (Todd 6). A 1968 suite on thick wove paper (apparently the “German copper-plate”) in the Rosenwald Collection, National Gallery of Art, Washington, lacks the reference to Hoehn but is inscribed on each plate, lower right just beneath the image, “A restrike from the copperplate in my collection August 1968[.] Lessing J Rosenwald” (see illus. 5). This set, with a sheet size approximately 36.5 × 45 cm., also bears a light pencil inscription lower in the right corner on all but plate 4, “printer’s proof 7/68,” written by Ruth Fine, Curator of the Rosenwald Collection. According to Gott (see note 11), the 1968 set in the National Gallery of Victoria begin page 83 |
In his essays, Todd is forthrightly critical of the c. 1954 impressions but praises the 1968 restrikes. He states that the latter are “more brilliant and show far more detail than any” others he had ever seen (6). Keynes is far more restrained, pointing out that “whether the result obtained [in 1968] with rather heavy inking (on dead-white Chinese [sic] paper) compares favourably with the first printing . . . is a matter of individual taste and judgement” (Blake’s Illustrations of Dante ). As a comparison between the nineteenth-century impressions of plate 3 (illus. 2 and 4) and a 1968 print (illus. 5) reproduced here will indicate, the heavy inking was only partially successful in bringing out the fine lines, such as those defining the mountain ridges and clouds top right, and had the unfortunate consequence of obscuring the distinctions between individual lines in densely engraved areas on the fighting devils and in the fiery lake lower left. This latter effect violates the linearist aesthetic implicit in the engraving style Blake deployed in his Dante plates. In my view, the 1968 restrikes are greatly inferior to all the nineteenth-century impressions I have seen. But no matter what one’s opinions may be about the two twentieth-century printings, they offer no problems in identification.
The very earliest impressions of the Dante plates can also be recognized easily. In his memorandum quoted earlier, John Linnell, Jr., mentions that “a few proofs” were pulled “by Blake” himself. In this context, the word “proofs” might mean nothing more than “impressions,” but at least some of them represent states earlier than those found in the 1838 and all subsequent printings. The identification of these earliest extant Dante prints rests simply on their states. I describe below the few proof-state impressions I have been able to locate.↤ 14 Bentley, Blake Books 546, records “proofs of each [Dante plate] in the BMPR,” but I have been able to locate only the three described here. The handlist of the British Museum collection in Blake Newsletter 5 (1972): 236, lists “Five Dante trial proofs” of pls. 2 (two impressions), 4, 6-7, accession nos. 1918.104.22.1683-77, but two of these (pl. 2, no. 274, and pl. 4, no. 275) are in their published states.
Pl. 1, “The Circle of the Lustful: Francesca da Rimini (‘The Whirlwind of Lovers’).” Princeton University Art Museum, Princeton, New Jersey. Lacking many lines of shading and radiance, the drypoint inscription in reverse letters lower right (“The Whirlwind of Lovers from Dantes Inferno Canto V”), the hands of the couple in the sphere upper right, and the inverted head between the whirlwind and the waves just left of the peninsula on which Dante lies. The heads of the two figures in the top left corner are re-cut in slightly different positions in the published state. Darkly but crudely inked. Reproduced Keynes, Blake’s Illustrations of Dante. Laurence Binyon, The Engraved Designs of William Blake (London: Ernest Benn, 1926), lists an even earlier state in which the “forms of the lovers are hardly more than outlines” and “there is a white patch in the river-bank under Virgil’s figure” (76). This was probably one of the unstated number of “proofs in early states” Binyon notes as then belonging “to Mrs. Sydney Morse” (76), but I have not been able to locate an impression of it. The other Morse proofs were probably those for pls. 2, 6-7 (described below), accessioned 13 July 1929 by the British Museum.14 There were no Dante plates in the auction of the Sydney Morse collection, Christie’s, 26 July 1929.
Pl. 2, “Ciampolo the Barrator Tormented by the Devils.” British Museum, Department of Prints and Drawings, accession no. 1922.214.171.1243. Some of the spiky wings around the devils on the right lack hatching strokes. A little more hatching appears on the upper thigh of the front-most seated figure on the right. This work appears to have been burnished away in the published state to create a highlight.
Pl. 4, “The Six-Footed Serpent Attacking Agnolo Brunelleschi.” Fitzwilliam Museum, Cambridge, bequeathed by T. H. Riches in 1935 and accessioned 1950. Printed in dark sepia and lacking hatching strokes on the serpent’s wings, the chest, stomach, and left hip of the man (Brunelleschi) in the serpent’s grasp, on the upper left thigh of the man on the right and on the stomach of his companion (second figure from the right), on the robe worn by Dante (left-most figure), on the ground beneath his begin page 84 | feet, lower left, and in the landscape background to the left of Dante’s right upper arm and to the right of Virgil’s hands. Brunelleschi’s left thigh bears some hatching that was burnished away in the published state to create a highlight. The spiky vegetation below and to the left of Brunelleschi’s right foot has been burnished away on its left side in the published state. The lower reaches of the rock above and to the left of this vegetation have also been removed. Reproduced Keynes, Blake’s Illustrations of Dante.
Pl. 6, “The Pit of Disease: The Falsifiers.” British Museum, Department of Prints and Drawings, accession no. 19126.96.36.1996 (illus. 1). Three figures in the lower left corner are incomplete; two are only scratched in outline. The figure on all fours just to the right and the supine figure above lack some shading strokes. The densely packed lines in the sky, just to the right of the arch of stony figures on the left, lack a few strokes. The faces of Dante and Virgil, upper left, are incomplete. This proof state may also lack some cloud lines top right, but the abraded surface of the print in this area leaves the matter indeterminable.
Pl. 7, “Dante Striking Against Bocca degli Abati.” Houghton Library, Harvard University, Cambridge, Massachusetts, bequeathed by Philip Hofer in 1984. Virgil’s left hand is raised to his face rather than to his chest (published state). Lacking many shading lines on the standing figures, particularly Virgil, the crosshatching on Abati’s right shoulder and a few lines around his mouth, and the tears on the cheeks of the old man on the left side of the column of figures on the right. Reproduced Keynes, Blake’s Illustrations of Dante. A later proof state in the British Museum, Department of Prints and Drawings, lacks only the finishing strokes defining the hands of the standing figures and a few lines on their clothing (accession no. 19188.8.131.527).
In Blake Studies, Keynes states that lot 179 in the Linnell sale at Christie’s, 15 March 1918, contained a “set of early proofs, probably those made by Blake himself” (229). Although Keynes may have inspected this lot (sold to Colnaghi for £26.5s.), he offers no evidence for this claim, which on the face of it seems unlikely. The auction catalogue describes this lot as “India Proof Impressions, before all letters, loosely inserted in an oblong folio vol., boards.” None of the traced working proofs are on laid India,
The next impressions to be pulled after Blake’s own working proofs are probably those printed directly on laid paper (not to be confused with India paper laid on wove) showing clear wire and chain lines, the latter approximately 3.7 cm. apart. Such a set was sold from the Doheny Memorial Library at Christie’s New York, 21 February 1989, lot 1713, plate 2 illustrated in the auction catalogue (sold to the New York dealer Donald Heald for $60,500). Part of a watermark, or countermark, “A & D,” is present in plate 3. The heavy foxing of most impressions in this set cannot mask the fact that these are superb impressions that justify the record price. All major engraved lines are dark, rich, and precise, while the drypoint sketching lines yet to be cut with the graver are delicately yet fully printed. Each plate in this suite reveals its superiority over all India-paper impressions I have seen. begin page 85 | Much of this excellence is the result of expert inking and wiping of the plates’ surfaces, but the quality of these impressions also indicates that the copperplates had not begun to show any effects of wear. The clarity with which each line is printed, even in densely-engraved passages, without any blurring of the boundaries between lines, suggests that the edges of each incision were still sharp and had not yet rounded to a gradual slope. This rounding can begin to occur after as few as ten impressions have been pulled from a copperplate. This initial wear is probably caused both by running the plates through the rolling press and by the inking process, for the craftsman must wipe the surface of an intaglio copperplate with dozens of strokes of his hand in preparation for taking a single impression.
As Thomas Lange has reminded me, another fine set on laid paper, also nastily foxed, is in the Pierpont Morgan Library, New York. A mixed group, with plate 1 on India laid on wove, lacking plate 4, and plates 2-3, 5-7 on laid paper, was sold in London at Christie’s, 29 June 1989, lot 13, plate 2 illustrated in the auction catalogue (£24,200 to the London print dealer William Weston). From this group I have seen only plate 3; but the auction catalogue reports that plates 6 and 7, both on the laid stock, show “part of a watermark [or countermark?] A & D.”15↤ 15 I am unable to identify this watermark, but it may have been produced by John Dickinson at the Apsley Mill he acquired in 1809—see W. A. Churchill, Watermarks in Paper (Amsterdam: Hertzberger & Co., 1925) 48. Weston tells me that these impressions were foxed badly enough to require careful cleaning; clearly, this laid paper has an inherent tendency to become stained in this way. Yet, in my view, they are the most desirable impressions of the published states.
The existence of the laid-paper prints is more than a little disconcerting. Blake used wove paper for almost all his illuminated books and original separate plates he printed himself.16↤ 16 The only exceptions known to me are copy L of The Marriage of Heaven and Hell (“A Song of Liberty” only, printed as a pamphlet of two leaves), the unique firststate impression of “Joseph of Arimathea Among the Rocks of Albion” (Fitzwilliam Museum), perhaps printed in the 1770s when Blake as still an apprentice, and the unique impression of “Charity” (British Museum). Linnell consistently favored India paper laid on wove and probably influenced Blake’s choice of those papers for some of his late intaglio graphics, including the “proof” issue of the Job illustrations in 1826. The documentary record of Dante printings makes no mention of laid-paper impressions. Indeed, that record would seem to exclude them, for all the press runs noted by John Linnell, Jr., and the Dixon & Ross daybook specify India paper. The laid-paper impressions suggest an unrecorded press run early in the history of the Dante plates. And if there was one such printing, might there have been others? I have no clear evidence of any further printings, but the mere possibility gives one pause before making bold statements about the total number of Dante impressions and their dates.
The quality of the laid-paper impressions indicates a printing date before the first India paper pulls of 26 September 1838. It is unlikely that they were printed by Blake himself. He left all seven Dante plates unfinished at his death in 1827, and there would have been no reason to print fine sets of the plates while there was any hope of developing them to a more complete state. Blake’s own working proofs, listed above, show areas of careless inking, as is often the case when a printmaker is taking impressions merely to check his progress. Perhaps the laid-paper impressions were taken by Dixon & Ross, or some other plate printer, just prior to the first recorded printing as a way of showing Linnell how expertly they could perform the task. This speculation, however, does not explain why at least two full sets, plus at least a third impression of plates 2-3, 5-7, were printed. The mysteries surrounding these beautiful laid-paper impressions only enhance their desirability.
We are now left with a variety of India-paper impressions, printed in two press runs in 1838 and one c. 1892—if we can still trust the completeness of the documentary record. How can these be sorted out and ascribed to the printing in which each was produced? Detailed comparisons of printing quality (never fully captured by photographs) and paper types (at best only hinted at by photos) are hindered by the wide dispersal of the Dante sets. The largest group I have found in one place is in the Huntington Library, but its four complete copies, plus one set in my collection, provide only a small sampling of the whole number extant. Visual memory, propped up by photos, must serve in such circumstances, in spite of their obvious limitations. The dimensions and thicknesses of backing papers can also be useful, but these too have their limits. Only untrimmed sheets can tell us what the original (and presumably uniform) sheet size was. The thickness of the same hand-made stock, even when produced in the same mold, can vary by measurable amounts. Even a single sheet can yield measurements that vary as much as 25%, depending on where one places the micrometer or paper gauge. All that follows should be understood in light of these caveats.
Among the dozen or more India-paper suites I have seen over the years, I have found a surprising uniformity in the quality of the impressions. The Linnell family was too successful, as far as my research is concerned, in replicating the 1838 impressions in 1892. There are of course small differences, but these can be accounted for by differences in inking of the sort that can occur within a single press run. Fortunately, one plate escapes this frustrating uniformity. Its special qualities require some background explanation.
Blake executed the Dante plates in a combination of drypoint, used mostly for preliminary sketching of outlines, and pure engraving without etching. In the latter technique, the tool is pushed through the metal to create V-shaped incisions. The metal removed takes the form of thin, curling wires. These are removed from each line as the engraver works over the plate. A drypoint needle creates a very different by-product. As the craftsman drags the needle across the metal surface, it leaves behind a “burr” on one or both sides of the furrow, much like the wake behind a begin page 86 | speedboat. This burr will catch and hold ink, even after the printer has wiped the surface of the metal clean, and its presence will register in impressions from the plate as a slight fuzziness or striations along the margins of drypoint lines. The burr, however, wears off the plate very quickly, and thus can serve as a useful (albeit approximate) index of the relative printing dates of two impressions if at least one shows some burr. That is, an impression with rich burr will probably be earlier than one with only a little, and this second impression earlier than one with none at all.
All the Dante plates show evidence of drypoint sketching, but only one shows any clear evidence of burr even in the early laid-paper impressions. Blake’s friend George Cumberland held that “the very early impressions of ancient [meaning early Italian] Prints in general” are not “the best, as they partake of the remainders of the Burrs, . . . but the best are those clear impressions which came early afterwards, when the Printer’s hand had entirely destroyed that Burr, and well polished the plate at the same time.”17↤ 17 Cumberland, An Essay on the Utility of Collecting the Best Works of the Ancient Engravers of the Italian School (London: Payne and Foss et al., 1827) 20. The graphic style of the Dante plates owes much to Italian Renaissance prints, particularly those by Andrea Mantegna and his circle.18↤ 18 See Robert N. Essick, William Blake Printmaker (Princeton: Princeton UP, 1980) 250-54. Cumberland showed Blake the manuscript of his Essay in November 1823 (G. E. Bentley, Jr., Blake Records [Oxford: Clarendon Press, 1969] 279). If Blake shared Cumberland’s opinions about such prints and made them a part of his own linearist, anti-chiaroscuro aesthetic, then he may have purposely removed burr as he worked. The one exception is plate 3, arguably the least developed in the series. Some impressions, including Heald’s on laid paper, show rich burr in the swirling lines lower right (illus. 2), on Dante and Virgil, in the landscape to the right, and on the seated figures center right (illus. 3). Just to the left of Dante and Virgil is a cluster of drypoint scratches that barely penetrated the copper except at their upper terminations. These are unrelated to the composition and are probably an unavoidable consequence of Blake having tested his drypoint needle. Like the burr, these shallow lines could be removed easily. If not erased, they would slowly wear off the plate during inking. Impressions without burr, and with only the slightest evidence of the scratched lines left of Virgil and Dante (illus. 4), are probably later than those showing both these features (illus. 2). The alterations over time in the condition of plate 3 are confirmed by the twentieth-century impressions, all of which show no evidence of either burr or the scratched lines (illus. 5), and by the present appearance of the copperplate (illus. 6).
The burr on plate 3 can serve as a rough indication of the relative chronology of impressions from that plate and, by extension, the chronology of its companions in the same suite—assuming that the suite is uniform as to paper, ink color, etc. Before applying this rule of thumb to the India-paper impressions, a cautionary note must be sounded. Burr records wear that occurs during inking and printing, not the gaps between print runs. All burr can be eliminated, gradually but completely, in the course of a single printing or retained through the early pulls of the next, no matter how many years pass between the two operations. It is highly unlikely for the very last fragments of burr to appear in the last impression of a press run, and thus for the first impression without burr to correspond precisely with the first pull of the next printing. Further, the wiping of the plate during the inking process will affect the amount of burr indicated in an impression. For example, the Weston impression of plate 3 on laid paper shows considerably less burr than the Heald impression on the same paper, even though both were very probably pulled in the same early press run. The direction in which the craftsman moves his hand for the final time over the burr areas will influence the amount of ink they retain. And we cannot assume that commercial plate printers of the last century took conscious measures to preserve evidence of burr or the test marks left of Dante and Virgil. Thus, slight variations in the amount of burr serve as a poor index of printing chronology. Only large-scale differences, such as those indicated by illus. 2 and 4, are very useful, and even then we must allow for the possibility that a few early pulls show no burr and a few later ones reveal its continued presence. We can have the most confidence in chronologies suggested by burr when supported by other types of evidence.
The Heald impression of plate 3 printed directly on laid paper shows the richest burr I have seen. But India-paper impressions laid on to a wove, hard surface, card-like paper, .46 mm. thick and without a watermark, show almost as much burr and reveal the test marks left of Dante and Virgil just as clearly. One such impression—the only one I have been able to compare directly with others—is in a complete set, with all plates on the same backing paper, in the Huntington Library (call no. 57438).19↤ 19 This suite is wrongly described as “probably the restrikes of ca. 1892” in Robert N. Essick, The Works of William Blake in the Huntington Collections (San Marino: Huntington Library, 1985) 209. All such sets are similarly (and wrongly) characterized in Sotheby’s New York auction catalogue of 9-10 November 1989, lot 168. This paper is so dense and thick that, when held to a backing light, it is completely opaque. In the Huntington set, bound in late nineteenth- or early twentieth-century full leather by Riviere & Son, the backing sheets measure 40.5 × 57.5 cm. (i.e., a little less than a half-sheet of Colombier) with all edges gilt. Another such set, formerly in the collection of Sir Geoffrey Keynes,20↤ 20 This set was not part of Keynes’s bequest to the Fitzwilliam Museum. According to J. J. Hall, Under-Librarian in the Rare Book Department, Cambridge University Library, the set cannot be located in the Keynes Collection there. is beautifully reproduced full-size in his Blake’s Illustrations of Dante. Its backing sheets measure 40.2×56.1 cm. Approximately the same dimensions are found in the only other thick-paper set I have located, now in the Kerrison Preston Blake Library, Westminster City Library, London.21↤ 21 Very well reproduced, with enlarged details from each plate, in David Bindman, assisted by Deirdre Toomey, The Complete Graphic Works of William Blake ([London]: Thames and Hudson, 1978) pls. 647-53. This may be the same as the set which Keynes describes as similar to his own and “until recently, in the possession of the Linnell family” (Blake’s Illustrations of Dante ). Both the Keynes and Preston sets are bound in green cloth with “BLAKE’S DANTE” stamped in gilt on the front cover and the letterpress Dante label, printed by Linnell for the 1838 issue, pasted to the inside front cover. This is probably the original binding in which such suites were issued.22↤ 22 This is also the conclusion stated independently about each of these copies in Kerrison Preston, Notes for a Catalogue of the Blake Library at the Georgian House Merstham (Cambridge: Golden Head Press, 1960) 10, and [Keynes], Bibliotheca Bibliographici: A Catalogue of the Library Formed by Geoffrey Keynes (London: Trianon Press, 1964) 65.
The conclusion, based on the presence of the burr, that these thick-paper begin page 87 | sets are early impressions can be refined further by documentary evidence that such prints were pulled in the first press run of 26 September 1838. One easily-overlooked detail in the Dixon & Ross daybook, quoted earlier, is particularly significant. The backing paper of the first printing is listed as “th[ic]k,” whereas no such adjective modifies the same “pl[a]t[e] Col[ombier]” used for the second printing (29 September). Keynes reports that the Dante set sold from the collection of the Marquis of Crewe at Sotheby’s, 8 May 1943, lot 312 (£68 to the London dealer Francis Edwards), was “similar” to his thick-paper suite (Blake’s Illustrations of Dante ) and was bound in the same green cloth stamped in gilt (Blake Studies 228). This Crewe copy, now untraced, bore an inscription on a flyleaf: “A few copies may be had of Mr. Chance, 28 London Street, Fitzroy Square, W, Artists Proofs £3.13.6. only 25 copies printed” (Bentley, Blake Books 545). The Preston set contains a clipping, pasted to the inside front cover, from an unidentified dealer’s catalogue that similarly claims that the impressions are “Artist’s Proofs, only 25 copies printed.” These statements further associate the prints on India paper laid on thick wove with the first press run. To call them “Artist’s Proofs” is certainly misleading, particularly since the artist had been dead for eleven years when they were printed, but the casual and self-serving use of the word “proof” is hardly unique in the annals of print selling (see my earlier comments on the 1918 Linnell sale). In this context, the term means, if it means anything at all, that these impressions are from the first press run. The connection with “Mr. Chance” and the number of “copies printed” are even more significant. In a letter dated 30 December 1856 (Ivimy collection), James H. Chance wrote as follows to his uncle, John Linnell: “I received the case quite safe with the 2[?] sets[?] of Dante & Two proofs of Emmaus[?] which are very beautiful (yesterday), and possibly I may be able to do something with them amongst my connexion. The Dantebegin page 88 | & Blake I have entirely on my own speculation being partial to them and if I can do any thing with them I shall be happy to enter into some agreements for 25 Copies. . . .”23↤ 23 Bentley, Blake Books 545. “Emmaus” very probably refers to Linnell’s line and mezzotint print of his own painting, “The Journey to Emmaus,” dated 17 June 1839 in the imprint. “Dante & Blake” is a bit odd; perhaps Chance meant “Dante by Blake.” Whether Chance carried out these plans is not known, but the inscription in the Crewe set indicates that he sold at least one of the so-called “Artist’s Proofs” printed on India laid on thick wove in the 26 September press run of 25 sets. Perhaps the Crewe and Preston sets are the two Chance notes he had already received.
It would be most convenient if we were now left with India-paper impressions on only two types of backing papers that could be ascribed confidently to the second (29 September 1838) and third (c. 1892) printings of the Dante plates. But the life of a chalcographer is rarely that simple, and what I have found instead is a bewildering variety of just off-white wove backing papers. All measure between .25 and .34 mm. in thickness and are only semi-opaque when held to a backing light. These characteristics are useful in distinguishing such prints from the thick-paper (.46mm.) impressions, but the range of only .09 mm. in thickness is insufficient for making clear discriminations among India-paper impressions on the thinner backing papers. Sheet sizes are of no help in separating out 1838 and c. 1892 impressions since no India-paper print I have measured is larger than the half-sheets (about 48×59.7 cm.) into which the Colombier-size paper was cut for both 1838 press runs. Here again, the success the Linnells had in matching the earlier prints in their c. 1892 restrikes thwarts the modern investigator.
Fortunately, some of the India-paper sets on the thinner backing papers can be identified on the basis of watermarks. I list below those I know of, with a few other particulars added.
1. “1822” followed at a considerable distance by “II [followed by an illegible letter fragment cut off],” with a countermark of “S & [cut off].” This paper is at the lower end of the range of thicknesses, is more pliable, and has a smoother and softer surface texture than all others listed here. A complete and uniform set is in the Huntington Library (call no. 57437), pl. 3 with the watermark fragment of letters, pls. 1, 3-7 with the date, and pls. 1, 4-7 with the countermark fragment. Sheets 44.5 × 59 cm., deckle edges right and bottom, minor foxing. A mixed group of nine impressions of the seven Dante plates in the Rosenwald Collection, National Gallery of Art, Washington, shows the “1822” watermark in pl. 2 only. Pl. 3 in the Huntington set shows considerable burr, roughly equivalent to that found in the impression reproduced here as illus. 2.
2. Fragment of an italic capital “N,” the remainder trimmed off. Found in pls. 1 and 3 of a complete (and apparently uniform) set in the Huntington Library (call no. 283403). Sheets 39.5×54.4 cm., one or two edges with deckle, the India paper (but not the backing paper) evenly browned on pl. 2. Pl. 3 shows no burr lower right and only very slight burr on Dante and Virgil. Only ghosts of Blake’s test scratches remain. The same amounts of burr and the scratches left of Dante and Virgil appear in a uniform set in my collection, sheets 39 × 54 cm., one or two deckle edges, the India paper of pls. 2 and 5 evenly browned, showing fragments of the “N” watermark on pls. 1, 5, and 6. A fragment of this watermark is reported to be in a set offered by Serendipity Books, Berkeley, California, December 1979, catalogue 39, item P 128, for $13,000 (untraced). The dealer’s catalogue claims that there is more burr on this set than in the set owned by Mrs. Landon K. Thorne (now in the Pierpont Morgan Library), which also contains fragments of the “N” watermark.
3. Letters, perhaps three or four, so poorly formed and densely packed together that they are illegible. Found in pls. 1 and 3 of a complete (and apparently uniform) set sold from the collection of The Garden Ltd., formed by Haven O’More, Sotheby’s New York, 10 November 1989, lot 168 ($41,250 to Donald Heald). Sheets 39.7 × 54 cm., no burr on pl. 3.
4. “J. Whatman Turkey Mills.” Variously reported in the following auction catalogues as present in complete sets: Sotheby’s London, 7 March 1985, lot 200, with a reproduction of pl. 3 showing no burr lower right but with fragments of the scratches left of Dante and Virgil as in illus. 4 (£23,100, untraced); Sotheby’s London, 27 June 1986, lot 747 (£24,200, untraced); Sotheby’s London, 27 June 1988, lot 168 (£24,200, untraced); Sotheby’s London, 27 June 1989, lot 226, sold from the “Property of a Member of the Linnell Family” (£38,500, untraced). Some of these may be resales of the same set, but it seems unlikely that they all represent only a single set.
The information recorded above allows us to draw a few conclusions. More than one backing paper must have been used in the second printing of 1838 and/or in the c. 1892 press run—assuming that there was no unrecorded printing in addition to the fine early pulls on laid paper (described earlier). It seems unlikely that the 1822 paper (no. 1, above) would have been used c. 1892, and even a little surprising that it would have been available in any quantity in 1838 (although the papermaker may have retained the 1822 mark in later years). Perhaps the uneconomical size of the second recorded printing of 1838—only 13 sets—and the four extra pulls (what commercial use would they have had?) were dictated by a limited amount of this 1822 paper. Of course this hypothesis assumes, rather rashly, that no other backing paper was used in the 1838 second printing. The presence of burr on the only impression located of plate 3 printed on this paper buttresses the conclusion that it was used in 1838.
All other backing papers listed here, nos. 2-4, are remarkably similar, at least to my eyes and fingers, as to color, texture, and density. William Linnell’s statement, in his recently discovered letter to Quaritch quoted earlier, that he preferred Whatman paper for the c. 1892 impressions does not insure that it was actually used; but this letter does lend support to the evidence of the burr (or lack thereof) on at least one impression of plate 3 indicating that sets on Whatman were pulled c. 1892. Circumstantial evidence, cited and interpreted (probably by Richard Godfrey) in Sotheby’s 27 June 1986 catalogue, further tips the scales toward 1892. The set sold in lot 747, with one backing sheet showing part of the Whatman watermark, was the suite Keynes acquired from “the last [nineteenth-century] printing of the plates” during a visit to the Linnell family home at Redhill.24↤ 24 Keynes, Blake’s Illustrations of Dante . I cannot confirm the claim in Sotheby’s catalogue that this visit was in 1912, but in The Gates of Memory (Oxford: Clarendon Press, 1981) 103, Keynes notes that he acquired a Dante set during one of his two visits to Redhill “before 1914.”begin page 89 |
Only two backing papers remain, nos. 2 and 3 above. The rule of thumb based on the amount of burr in plate 3, combined with the weight of numbers (only 13 complete sets produced in the second printing of 1838, but 50 in c. 1892), leads me to conclude that both types of paper signify the c. 1892 printing. However, the impression of plate 3 reproduced here, illus. 2-3, showing almost as much burr as the thick-paper pulls of the first printing, is on a backing sheet of unwatermarked paper which I cannot distinguish from my set with the “N” watermark.25↤ 25 A Dante set on laid India with very similar amounts of burr in pl. 3, now in a British private collection, was sold the property of D. R. Bollard (John Linnell’s great-grandson) at Sotheby’s London, 14 November 1980, lot 295, in poor (but probably repairable) condition (£5100). I have no information about its backing paper, but all seven plates are reproduced in The Print in England 1790-1930: Catalogue of an Exhibition First Shown at the Fitzwilliam Museum 12 March to 5 May 1985 (Cambridge: Fitzwilliam Museum, 1985) 54-60. I can only conclude from this that either a paper in addition to the “1822” variety was used in the second printing of 1838, or the copperplate of plate 3 retained considerable burr after 1838 and this show up in the first few pulls of c. 1892. If the latter is the case, as seems the more likely of the two possibilities, then the rule of burr cannot be used to discriminate in all cases between the second and third press runs.
In spite of its length, the foregoing is little better than an interim report. It concludes with the spectacle of a Blake nut imitating plate 10 of The Gates of Paradise (“Help! Help!”). It would be good to find more sets with the 1822 watermark, and more impressions of plate 3 with burr on any sort of backing paper. An impression on Whatman paper showing rich burr would force major revisions in the history concocted here. So let me offer my services as a Dante Clearinghouse and invite interested readers to send me information about Dante engravings they have inspected. Working together, we should at least be able to complicate matters further.