At the risk of complicating an already highly complicated subject, I would like to develop a small point in the wonderfully thorough article by Mei-Ying Sung on the newly identified “Rosenbloom” set of Job prints. In paragraph 11 the author states that two of the proofs of this set are printed on laid rather than woven paper. These are the proofs for plates 17 and 20. By an extraordinary coincidence, if such it be, these are the two subjects, The Vision of Christ and Job and His Daughters, that, it now seems to be genuinely agreed, were added to the nineteen subjects painted in watercolor c. 1805–06 for Thomas Butts. When John Linnell commissioned Blake in 1821 to paint a new series based on copies of the Butts set, Blake added these two subjects. He then repeated them so that Butts would have the full set of twenty-one subjects. The additions are distinguished by, again, being on different paper from the rest of the Butts watercolors and also by being late in style and possibly completed by Blake’s widow.

Coming as part of one of the two early series of proofs (in the second of which Blake began to print the border designs), the two new subjects were probably added at this relatively late stage to those chosen for the original watercolors for Butts. In addition, it is possible to place this state of plate 20 at the end of a highly complex evolution of that subject (see Butlin #394, #550 20, #551 20, #555, #556, #557 42, and #557 49). Moreover, the placing of the two new subjects in the sketchbook in the Fitzwilliam Museum, Cambridge, dated 1823 by Blake, seems to show them to be integral to the whole scheme at that period in its evolution. The last few studies seem to have been drawn in this order: plate 17, a blank page, plate 18, plate 20 (the figures only), an alternative (not used) for plate 19, a blank page, plate 21, a blank page, plate 19, a blank page, plate 20 (the whole composition), and finally miscellaneous drawings and blank pages. There are tentative sketches for the borders in at least some of the earlier subjects (Butlin #557 7, 9, and 11). Similar rough sketches for the borders occur on six of the earlier, Linnell series of proofs, some close to the states as found in the Rosenbloom series, some not. Plates 17 and 20 in the Linnell series do not show any border designs, nor do the relevant sketches in the Fitzwilliam sketchbook. In the Rosenbloom series of proofs the border designs were more or less fully developed on the plates with, as Sung points out, an important early state of the title on plate 20.

Unfortunately the National Gallery of Art in Washington are too busy to examine the pre-publication proofs from the Linnell series to ascertain which of these are on what types of paper. However, Essick’s thorough discussion and catalogue of all the states of the Linnell proofs makes it clear that, with few exceptions, the earliest states were on laid paper; the exceptions were the proofs for plates 2, 4, 9, and 21, with no examples of the parallel states for plates 15 and 18. This seems to confirm that the printing of the Linnell and Rosenbloom proofs overlapped, with those on laid paper occasionally interspersed, it seems, with those on wove (in addition, a number of first states, and a few second and third states, are on oriental paper), and that the Fitzwilliam Museum sketchbook was being used by Blake at this same, crucial period.

All this suggests that the new subjects were probably not added until Blake began working on the engravings. But one must remember that artists, particularly visionary artists, do not always act in a logical manner.