t has long been thought, by the present authors among others, that Blake painted his plates using the standard à la poupée technique, adapted for his purposes. Compared to any alternatives, the method is direct, cost effective, and united with the art of painting (Essick, Printmaker 125-35; Friedman 16; Viscomi, Idea 119-28). Phillips does not believe this, but, citing Le Blon and Jackson as precedents, argues that Blake adapted the more complicated manner of printing and registering multiple plates by printing his own plates twice, once for the text in ink and again for the illustration in colors (95). It may seem that questions about printing technique in general and color printing in particular are of no real importance, but, as we argue below, using one or the other method significantly affects our ideas about Blake’s works and their conceptual implications. Phillips recognizes what is at stake, for he claims that by not recognizing the two-pull method we are grossly underestimating the “time and skill” Blake invested in color printing and misunderstanding his “intentions as a graphic artist” and his intended audience (95). On these issues Phillips says little beyond some general observations on Blake’s intended audience in his “Conclusion” (111-13). Nor does he develop further the effect of his theory on our understanding of Blake as artist, printmaker, theorist or poet.  Surprisingly, Phillips does not argue (let alone prove) that Blake’s visual effects in color printing were not possible with single-pull printing. In short, he does not directly consider (much less answer) the crucial question: Why divide the printing process into text (first pull) and illustration (second pull) to reunite them on paper if it was technically and aesthetically unnecessary to do so?

According to Phillips, Blake produced his color prints by inking the plate’s text areas, registering the paper to plate, printing and removing the paper, wiping the ink off the plate, adding colors, registering the paper exactly to the colored plate, printing and removing the twice-printed paper from the bed of the rolling press, and (presumably after drying) finishing it in water colors (95, 101, 107). To produce another print from the same copperplate, Blake would then begin the process anew by wiping the plate of its colors, inking the text areas, registering, printing, wiping the ink, adding colors, registering, and finally printing. Phillips claims that a significant part of his evidence for this labor-intensive method in which text is printed first and illustration second lies in the “Nurses Song” from Songs of Experience in Songs of Innocence and of Experience copy E (illus. 8). One can plainly see that this impression was indeed printed twice, as Essick and Viscomi separately recognized, but which they, according to Phillips, incorrectly identified as an individual aberration rather than as one of the most significant clues in revealing Blake’s color printing practice (Essick, Printmaker 127; Phillips 103; Viscomi, Idea 119). Phillips implies that this “Nurses Song” deviates from other color prints only in that, unlike them, it is misregistered, whereas all the other extant color prints were perfectly registered.

Phillips cites Le Blon as an example of multiple-plate printing to make the point that registering a plate onto a prior impression was possible (95-96). He states that for the three primary colors to be recombined into the original colors meant that “the precision of the registration had to be absolute” (96). From this statement, one might infer that Le Blon’s color prints show no signs of the second or third plate—that is, reveal no signs of their mode of production—but that one plate was registered on top of an impression from the other so precisely that all telltale tracks were covered. That, however, never happens.

Color prints produced with two or more plates or blocks—despite the plates being exactly the same size—always show signs of their production, usually to the naked eye but always under magnification or computer enhancement. We have yet to find a multi-plate (and hence multi-printed) color print that does not show evidence of at least slight misregistration at some point along its margins, usually at or near the corners. Such evidence generally appears in two forms: either as multiple platemarks and/or as a displacement of one color just outside another. For example, the top right corner of Le Blon’s Van Dyck Self Portrait (illus. 9) reveals one plate extending past the other. This effect is even clearer in the bottom right corner of Le Blon’s Narcissus (c. 1720s) (illus. 10). We see the same effect in all twenty of the prints in D’Agoty’s Myologie, including plate 3 (illus. 11), which were thought by contemporaries to be superior to Le Blon’s, and in all 53 of his smaller three-color mezzotints for Observations sur l’histoire naturelle, sur la physique et sur la peinture (1752-55), such as the Tortuise (illus. 12). Even the excellent two-color stipples of Louis Bonnet, such as Head of a Young Girl Turned toward the Left (1774) (illus. 13), reveal in their corners two platemarks, one slightly displaced from the other (illus. 14). The signs of production are also visible in the very best impressions of the mixed-method and pure chiaroscuro prints, including Kirkall’s Holy Family and Jackson’s Descent from the Cross, where the tonal blocks extend slightly past the key blocks (illus. 15-16). Even Jackson’s Venetian series—thought to be “without doubt the high point of chiaroscuro printing” (Friedman 6)—reveal their mode of production, as the corner of Holy Family and Four Saints, after Veronese (1739), demonstrates (illus. 17). In all of these illustrations, it is fairly easy to see the misregistrations.

Such subtle misregistrations are not signs of poor printing. They are to be expected, as printing manuals today acknowledge, regardless of the registration mechanism used, because damp printing paper stretches and shrinks in the course of printing the first and subsequent plates (Hayter 58, Romano and Ross 121, Dawson 100). Reviewing the various techniques used in his Atelier 17 for registering and printing multiple plates, Hayter, one of the greatest technicians of twentieth-century printmaking, states that “it is worthy of note that none of these methods is absolutely precise” and “examination of the edges of colour prints made by this system [i.e., multiple plates] will nearly always show some errors of registration between the different colours . . . ” (58). Slight misalignments, however, will not disrupt the visual logic and impact of the design; our eyes tend to make adjustments or “read” a slight fuzziness in an image as a pleasingly painterly style. The visual effect of multiple-plate color prints, in other words, was not dependent on absolute precision but on colors being overlaid one on top of the other. But the same eyes cannot be fooled when focused on the margins. “Absolute” (Phillips 96) accuracy in registration was impossible.

“Nurses Song” in the Experience section of the combined Songs of Innocence and of Experience copy E (illus. 8) clearly reveals its double printing. The next place to look for evidence of Blake printing his plates twice—on the grounds that modes of production can never completely conceal themselves, at least not to magnification and computers—is his other color prints, more than 650 of them. Given how poorly printed “Nurses Song” is, one would reasonably expect to find other examples of misalignment, albeit less overt. Yet not one of Blake’s other color prints reveals any sign of misregistration of the plate onto the impression previously printed in ink. Any suggestion that none exists because poorly printed impressions were thrown away ought to give one pause. Such a practice is refuted by “Nurses Song” and many of the other poorly printed impressions in Songs of Innocence and of Experience copy E and other illuminated books. It seems clear that Blake rarely threw away anything he printed that might be salvageable. He had little concern with the finer points of precision printing. If “Nurses Song” was acceptable (as its inclusion in a complete copy of the Songs of Innocence and of Experience sold to his major patron indicates), then any print less obviously misaligned would be too, including hairline misalignments not easily seen with the naked eye but visible under magnification. One would expect to see quite a few impressions looking like illustrations 18 and 19, where the text and designs, having been printed twice, are slightly out of register. It takes only a hairline misalignment of the second plate on top of an impression from the first—or on top of a prior impression from the same plate—to produce this out-of-focus effect. This is especially true with relief etchings like Blake’s, because the images are essentially in outline rather than tonal areas, which makes printing them twice analogous to double printing the key block in a chiaroscuro woodcut or mixed-method chiaroscuro. Even impressions that appear dead on, such as illustration 20, reveal, when magnified, the soft edges along the letters that evince a second printing (illus. 21). In impressions printed once, letters and other relief lines do not have any such ghosting (illus. 22), an absence characteristic of Blake’s color prints, as is demonstrated by details of “The Fly” and “Holy Thursday” (Experience) from Songs of Innocence and of Experience copies G and E (illus. 23, 63) respectively.

Phillips is correct to assume that Blake would have had to wipe the plate completely clean of ink before adding colors (95, 101), and then wipe the color off the plate before adding ink to pull a second impression. This is necessary to help disguise even the slightest errors in registration, for, as we have seen, if even minute traces of ink remain on the plate during its second printing and the registration is anything less than absolutely exact, then it will produce a slightly fuzzy impression. The same is true for the colors: if they are left on the plate, then they will be printed twice in the subsequent impression, once with the ink and once when colors are replenished. The slightest misregistration will show up. Masking techniques like these, however, work only to a point; the subtlest of misalignment may fall below the threshold of vision, but it can be detected with magnification and computer enhancement because relief lines or areas, even when devoid of ink or colors, still slightly emboss the paper around their edges. For example, the plate borders in “Nurses Song” in Songs of Innocence and of Experience copy F were wiped of ink but still embossed the paper (illus. 24). Such embossment is especially noticeable even without magnification in impressions color printed from both the relief plateaus and etched valleys of plates, such as those in Songs of Innocence and of Experience copies F, G, H, and T¹, The Book of Urizen copies A, C, D, E, F, and J, Visions of the Daughters of Albion copy F, and The Marriage of Heaven and Hell copy F, as is clearly evident in its plate 21 (illus. 25). If Blake printed his plates twice with pressure sufficient to print colors from the shallows, then the second printing, despite its carrying no ink, will reveal itself as a set of embossed lines around the printed lines (illus. 26). No embossments or haloes of this kind are found in Blake’s color prints.

Even wiping the plate of ink and colors between pulls cannot erase the signs of a second printing. Moreover, wiping oily ink usually leaves signs, as is evinced by the traces of ink on and along plate borders that Blake wiped of ink (illus. 27a-27b). There are hundreds of examples of such traces because Blake wiped the borders of nearly all illuminated prints produced by 1794 (see, for example, illus. 51, 53). In addition, wiping ink and colors for every pull is extremely wasteful in practice. Inking and printing pressure normal for relief can yield up to five useable prints from one inking in a dark color. Illustration 28, for example, is the third impression printed from one inking of a facsimile plate. Indeed, in the Tate Britain exhibition, the second pulls printed from facsimile plates were all more Blake-like than the first pulls, which were too dark. With lighter inks, like the yellow ochre used in Songs of Innocence and of Experience copy E, one can produce at least two acceptable impressions (illus. 71, 72, 73). The pigments, oil, and glues used to make inks and colors cost money, and so do rags used to wipe the plates clean. These unnecessary expenses and the time required to clean oily ink and glue-based colors from the copperplates between each impression make this method of color printing expensive and labor-intensive for no aesthetic gain, for it creates prints without any visual differences (other than the telltale signs of double printing) from those produced with single pulls at far less effort and cost.  

But one need not argue the point hypothetically about labor, time, money, and materials, or even about the astonishing absence of fuzzy impressions, ghost texts, and embossed haloes unavoidable in two-pull printing. To this negative evidence that argues against the two-pull hypothesis we can add a wealth of positive evidence that Blake did not wipe his plates of ink or color between pulls but continued to replenish the ink and colors. Printmakers are led by the physical properties of their materials to replenish ink instead of wiping and starting over again because ink transfers best once the plate is worked up. The repetition of inking accidentals and colors in sequentially pulled prints, such as the two proof impressions of The Book of Urizen plate 25, color printed but not finished in watercolors (illus. 29-30), or the finished impressions of plate 24 in copies F and C (illus. 31-32), demonstrates that Blake printed more than one impression from an inked plate and added ink and colors to a pre-existing base. To assume otherwise is to assume that repetition of colors and their placement was due to Blake trying to replicate the previous impression—i.e., reproducing a model—but, given the differences introduced, doing a very poor job of it. Clearly, it is more reasonable to conclude that the repetition of accidentals is due to Blake not wiping the plate clean between impressions than to conclude that he minutely copied irrelevant and even visually disruptive droplets and smudges of ink or color, using the prior impression as his model. The repetition of colors, and in some cases their diminishing intensity because Blake did not add more color for a second impression, lead to the same conclusion. Even the impression of “Nurses Song” that was printed twice, the very grounds for the two-pull hypothesis and for thinking that text and illustration were printed separately, shows two top plate borders in yellow ochre ink (illus. 33), which means that ink was printed with the colors and not wiped between pulls.

While one would expect to see fuzzy impressions and other signs of misregistration in two-pull printing, what one would not expect to see are perfectly clean fine white lines bordering the relief areas of prints color printed from both levels. For example, in illustrations 25, 29, 30, 31, 32, the fine white lines between the colors printed from the shallows and the ink printed from the relief surfaces are created by printing pressure that was insufficient to force the paper onto the escarpments between the etched valleys and the relief plateaus of the copperplate. Thus, the paper could not pick up any ink or color from those bordering escarpments. We see precisely the same effect in monochrome, ink-only prints, which no one doubts were printed in one pull, such as Europe copy H plates 1 and 4 (illus. 34a, 35). In these impressions, the inking dabber accidentally inked the shallows along with the relief areas, and both were printed simultaneously. The fine white lines between relief and recessed areas were created either by the dabber not depositing any ink on the escarpments or by the paper not creasing at an angle sharp enough to pick up any ink from those escarpments, in spite of relatively heavy printing pressure. The effect in plate 1 of Europe copy H (illus. 34a) become clearly evident when compared with an impression of the same plate which lacks the accidental deposits of ink in the etched shallows (illus. 34b).

The white line in the branches of plate 1 of The Book of Urizen copy D is most telling (illus. 36); here we can actually see Blake painting the plate, applying his green color on the inked relief lines and the green spilling over and touching the shallows on both sides of the line, creating white spaces between color and branches. If plates with colors from the shallows were printed twice, then the white line would be uniformly intersected with color. These white lines could not be perfectly aligned, even if registration of the plate was absolutely perfect, because the dampened paper, as Hayter and others have pointed out, would have minutely changed its shape while being passed through the press, even if printed with light pressure. This makes perfect registration of the white-line escarpments of the second pull impossible—and detection under magnification or computer enhancement possible.

Accidental flaws in one-pull printing can be mistaken as evidence of two-pull printing. That such accidentals appear in Blake’s monochrome impressions, unquestionably printed only once, should be sufficient warning against misinterpreting their mode of production. For example, the droplets of color in the margins of plate 24 in The Book of Urizen copies C and F (illus. 37-38), which may lead one to suspect the edge of a second plate, is an effect also present in monochrome impressions, such as America copy H plate 10 and Europe copy H plate 1 (illus. 39-40) that were assuredly printed just once. One-pull prints can even exhibit the slight fuzziness, so typical of multi-plate and multi-pull printing, at the margins between printed and unprinted surfaces because of slippage between paper and plate when run through the press. Color printing, particularly when done from the shallows as well as the relief areas, multiplies the chances for accidental deposits of ink and colors that do not contribute to the printed image, calligraphic or pictorial. Thus it should be no surprise that Blake’s color prints show, on average, more accidental effects than monochrome impressions.