"Press variants in Q2 Hamlet" by Gabriel Egan*

Because of stop-press correction on certain formes and the retention of sheets showing the uncorrected state(s), the 7 surviving exemplars of Q2 Hamlet are not identical. When John Dover Wilson collated Q2 in the 1930s, only 6 of these 7 exemplars were known (the Polish copy was undiscovered), and with this new exemplar and an altogether more careful collation Ann Thompson and Neil Taylor have, for their Arden3 edition, added 8 variants to the 18 variants found by Wilson (Wilson 1934, 123-24; Shakespeare 2006, 479n1, 524-5). Thompson and Taylor's abbreviated labels for the exemplars (followed in this essay) are different from Wilson's, so for the sake of readers comparing their work the following mapping is given:

Dev[onshire] = Huntington Library, so here HN
Huth = Beinecke Library Yale, so here Y²
Folg[er] = Folger Library, so here F
B[ritish] M[useum] = British Library, so here L
Cap[ell] = Trinity College Library Cambridge, so here C²
Grim[ston] = the earl of Verulam's copy at the Bodleian, so here VER
Unknown to Wilson = University of Wroclaw, so here Wro

    Of the 26 variants, 10 are clustered on a single forme, N(outer), which exists in three states, listed in Table One according to which of the 7 exemplars contains each combination of variants and with the order of correction running down the page. Thus the uncorrected (or earliest-known) state is witnessed in exemplars F, HN, Y², Wro, the first corrected state is uniquely witnessed in exemplar L, and the last (known) corrected state is witnessed in exemnplars C² and VER.

Table One

F, HN, Y2, Wro	thirtie	pall	sellingly	dosie	yaw	neither in	too't	reponsiue	be hangers	sir
L	thereby	fall	sellingly	dazzie	raw	neither, in	doo't	reponsiue	be might hangers	so sir
C2, VER	thereby	fall	fellingly	dazzie	raw	neither, in	doo't	responsiue	be might hangers	so sir

 

Fredson Bowers's analysis of the reuse of headlines in skeleton formes showed that it is highly likely that Q2 Hamlet was set by two compositors, each working almost exclusively on his own sheets and providing type to each of two presses (Bowers 1953, 19; Bowers 1953-4, 79-80; Bowers 1955; Bowers 1956). Bowers presented his evidence using the post-war convention of assigning an upper-case roman numeral to each headline and listing the pages it topped (Table Two), but the same data are here also presented using the conventions developed by Peter W. M. Blayney and G. Thomas Tanselle (Table Three) that allow the patterns to be seen more clearly.

 

Table Two

I = a	B2v C1v D1v  F1v  I3v  N3v	IX = n	E1v  G1v  H1v  K1v  L3v  M3v
II = d	B2r  C2r  D2r  F4r  I4r	X = m	E2r  G2r  H2r  K4r  L4r  M2r
III = e	B1v  C3v  D3v  F3v  I1v  N1v	XI = p	E3v  G3v  H3v  K3v  L1v  M1v
IV = f	B4r  C4r  D4r  F2r  I2r  N2r	XII = o	E4r  G4r  H4r  K2r  L2r  N4r
V = b	B3r  C1r  D1r  F1r  I1r  M1r  O1r	XIII = i	E1r  G3r  H3r  K3r  L3r  N3r
VI = c	B4v  C2v  D4v  F2v  I4v  N2v	XIV = j	E2v  G4v  H2v  K4v  L4v  N4v  O1v
VII = h	[B1r]  C3r  D3r  F3r  I3r  M4r  N1r	XV = k	E3r  G1r  H1r  K1r  L1r  M3r  O2r
VIII = g	B3v  C4v  D2v  F4v  I2v  M4v	XVI = l	E4v  G2v  H4v  K2v  L2v  M2v

 

Table Three

Sheet		Inner forme				Outer forme
		1r	2v	3r	4v	2r	1v	4r	3v
B		--	a	b	c	d	e	f	g
C		b	c	h	g	d	a	f	e
D		b	g	h	c	d	a	f	e
	E	i	j	k	l	m	n	o	p
F		b	c	h	g	f	a	d	e
	G	k	l	i	j	m	n	o	p
	H	k	j	i	l	m	n	o	p
I		b	g	h	c	f	e	d	a
	K	k	l	i	j	o	n	m	p
	L	k	l	i	j	o	p	m	n
	M	b	l	k	g	m	p	h	n
N		h	c	i	j	f	e	o	a

 

It is clear that 4 skeleton formes were set up, each containing 4 headlines (thus 16 headlines, a-p). The skeletons were for the most part consistently used in two pairs, containing headlines a-h and i-p. The pair containing headlines a-h was used for sheets B, C, D, F and I and the pair containing headlines i-p was used for sheets E, G, H, K and L, while sheets M and N break this pattern. (The half-sheet O+A, containing the last 3 pages of the book and the title-page, has been omitted because half of its 8 pages had no running-titles.) Within each pair of skeletons, one was for the most part consistently used for the outer and the other for the inner forme. It is possible for one compositor working with one press to construct and use skeletons in this way, but there is no reason to do so and it would put him to unnecessary trouble. If we suppose two compositors and two presses, however, the skeletons would have provided a convenient orderliness to the processes of composition, imposition, and distribution. The advantage for one man of working with two skeletons is that it allows him to impose and lock up in the chase the forme next required by the press before the press has finished machining the current forme.

    John Russell Brown's examination of two distinctive spelling habits in Q2 Hamlet corroborated Bowers's findings (Brown 1955), as did W. Craig Ferguson's discovery that two distinct roman typefaces are used in the book in the same pattern of divided labour suggested by Bowers's analysis of headline reuse and Brown's spelling analysis (Ferguson 1989, 15). Adrian Weiss's scathing dismissal of Ferguson's book did not reject this discovery, and elsewhere he confirmed it (Weiss 1989; Weiss 1991). Further confirmation was provided by Eric Rasmussen's analysis of the reappearance of distinctively damaged type throughout the book (Rasmussen 2008). These studies all point the same way: two compositors using distinct sets of type divided the work, one (compositor X) taking sheets B, C, D, F, I, N and setting them in Lyon-a type using his own pair of skeleton formes, and the other (compositor Y) taking sheets E, G, H, K, L and M and setting them in Lyon-b type using his own pair of skeleton formes.

    None of this evidence can establish the order that the formes went through the press, since any arbitrary order is possible once we accept that printers often worked on several books at once rather than racing to complete each one before turning to the next, and that they did this to regulate the work-flow of the whole printshop (McKenzie 1969; Weiss 1999). However, the rational sequence of working alphabetically through the sheets of this book and maintaining a consistent pattern of alternating the sides (inner and outer) printed first makes the best sense of the evidence of type and headline reuse. The only alternative sequence that fits the evidence equally well is reverse alphabetical order, from the end of the book to the beginning. Working that way, printers might easily paint themselves into a corner, since in the event of miscalculation of the length of a book it is harder to extend the beginning than the end. Although extra leafs or gatherings could in principle be added anywhere in a book, Q2 Hamlet has an unmovable head-title before the first lines of the play that was presumably intended from the outset. The trouble taken to make attractive beginnings to books suggests that publishers expected prospective buyers to examine them more closely than the ends. We may assume, then, an alphabetical progression of sheets and alternation of sides. This granted, the pattern of headline reuse might explain the clustering of variants, 10 out of 26, on a single forme, N(outer).

    Bowers showed that each compositor kept to his own 8 headlines until near the end of the work. Compositor X imposed two of the pages of the outer forme of his last full sheet, N3^r and N4^v, using headlines i and j that he borrowed from compositor Y, who used them on L3^r and L4^v. Similarly, compositor Y imposed two of the pages of the outer forme of his last full sheet, M1^r and M4^v, using headlines b and g that he borrowed from compositor X, who used them on I1^r and I2^v. (The movements are shown in boldface in Table Three.) I have called this 'borrowing' on the assumption that the compositors worked simultaneously and swapped headlines. If the compositors were not working simultaneously, the pattern of headline reuse requires that towards the end of his job, one of the compositors made up and used an extra pair of headlines that he did not need, and that this extra headline was then used by the other compositor for most of his work, only to be abandoned near the end of his job in favour of a different one of the first compositor's headlines. If the swapping of headlines by two compositors working simultaneously is accepted, it does not matter just which of them did this first. The important point is that they invited trouble since they were in fact setting to slightly different measures.

    Fresh examination of 4 of the 7 exemplars shows that the two compositors' sticks were not set to the same width. On pages where there are two or more lines set as prose, the width of the stick can be measured because type was justified by the insertion of spaces between words already set, or by changing spellings and abbreviations, rather than, as was done with verse, by adding spaces to the end of the line. 36 pages of Q2 Hamlet have such measurable passages, 12 of them set by compositor X and 24 by compositor Y. Appendix One shows the width of the stick (the measure) for each of these 36 pages in each of 4 exemplars. The readings fluctuate around 98 millimeters, but statistical analysis shows a small yet significant difference between their readings: compositor Y set his pages around 0.75 millimeters wider than compositor X. Necessarily the headlines varied by this amount too although they cannot be measured directly as they begin and end with spaces. When compositor X borrowed one of compositor Y's headlines, he ought to have either removed a thin space from this headline to match the page of type, or else add a thin spacing shim, a reglet, along a vertical edge of the page of type to widen it to match the headline.

    On compositor X's forme N(outer), the pages set with compositor Y's headlines are N3^r and N4^v. If we allow images of an impression taken from the forme to stand for the forme itself (with reflections in both axes as necessary), Picture One represents how N(outer) was imposed in what is supposed the uncorrected state as witnessed in the Folger, Huntington, Yale, and Wroclaw exemplars. The red boxes indicate how much of the type present in this state had to be adjusted to make the 10 known corrections in the two later states: 8 witnessed in the unique British Library copy and a further 2 corrections witnessed, together with those previous 8, in the Cambridge and Oxford exemplars. The adjustments marked here include not only the altered readings, but also other type noticeably shifted in making them. Yet more type may have been temporarily disturbed to make these corrections but then reinserted into the forme in precisely its previous location and so not detectable from the printed books. As can be seen from Picture One, the most extensive adjustment was made in the type just underneath headline i on N3^r, one of the headlines borrowed from compositor Y and hence about 0.75mm too wide for compositor X's page of type.

    Whether the chases of early Jacobean printers had fixed crossbars, as later ones did, is not known; McKerrow guessed they had at least one (McKerrow 1927, 15n2). Crossbars would confine the effects of imperfectly locked up type to within the space they enclosed (half a forme for one crossbar, one quarto page for two crossbars), whereas if moveable furniture did the work of crossbars then an oversize headline topping one page might affect other pages held in placed by the same furniture. As well as N3^r (which shows press corrections), the other page on this forme imposed with a borrowed headline is N4^v--topped by headline j from L4^v--and it shows no press corrections in the extant exemplars. The same problem of a mismatch between headline and type-page would have occurred when compositor Y borrowed compositor X's headlines: on M1^r headline b from I1^r was used, and on M4^v headline g from I2^v was used. In these cases, the headline would have been too narrow and any problem of looseness would be confined to the headline. There are no extant variants on these pages. The last two swaps were compositor X's imposition of N4^r (a page showing one variant) using headline o from L2^r and compositor Y's imposition of M4^r (no extant variants) using headline h from N1^r.

    The commonest type of accident in these circumstances is that loose sorts stick to the leather balls used to smear the ink onto the type and are thus lifted out of the press. The obvious remedy would be simply to reinsert the sorts back into the forme, unlocking the forme if necessary and tightening it more than usual to prevent recurrence of the problem. However it may be significant that compared to press correction elsewhere in the book, the changes on N(outer) are more extensive, more difficult to make, and largely lacking in obvious motivation (Egan 2009). The preceding press corrections witnessed in the extant exemplars involved just one or two lines in the forme, the lines are always short (in the sense that there are spaces between the last word on the line and the end of the line), and the changes fixed literal errors of some importance. Typical examples are "Showe me the step and thorny way to heauen" becoming "Showe me the steepe and thorny way to heauen" (C3^v) and "Your Officres" becoming "Your Officers" (L1^r). Because of the spaces at the ends of the lines, the changes on the preceding formes were easily accommodated by increasing or decreasing that end-of-line spacing as needed and the disturbance of other words was minimal or avoided altogether. By contrast, 7 of the 10 corrections on forme N(outer) occurred in full lines, several required extensive alteration to the rest of the line, and one required alteration in 4 lines.

    For the sake of argument, let us consider these changes using Wilson's inferred order of correction. On N1^r thirtie > thereby and pall > fall were executed with no disturbance of surrounding type. On N2^v sellingly > fellingly was executed with no disturbance of type, but dosie > dazzie required the removal of spaces from three inter-word gaps earlier in the line. On the same page, to insert the comma to execute neither in > neither, in spaces were removed from two inter-word gaps later in the line, and in executing too't > doo't the whole of the three-word line was shifted. Still on N2^v, the change reponsiue > responsiue required the previous word to be shifted also. On N3^r, the change of sir > so sir required removal of space between the speech prefix and the first word of the line, and from within two inter-word gaps later in the line. The most extensive alterations were needed for be hangers > be might hangers near the top of N3^r, 4 lines below the too-wide headline i that compositor X borrowed from compositor Y. Earlier in this line a small inter-word space was removed but this made little difference. To get the new word might into the line the last word on the line, then, and its terminal comma were moved to the second line, which line had then to lose the last 6 letters of its last word, assignes, plus its terminal comma, which were moved to the third line (the initial s of signes being changed to a long-s because now heading a line), which had then to lose its last letters of bet a-, which were moved to the fourth line (the word-breaking hyphen being removed as no longer needed), where the adjustments could stop because the line was not full and spaces could be taken from its end. It is significant that in this adjustment of 4 lines, 3 runs of words seem to have been moved as unbroken units, for there is no sign of adjustment within them: "but on, six Barbry horses against six French swords their as", "and three liberall conceited carriages, that's the French", and "gainst the Danish, why is this all you call it?". As far as one can tell, the smaller units of type that had to be moved around these longer runs also underwent no internal readjustment, only repositioning as units. The resetting seems, then, to have involved the orderly removal or shifting along of small and large groups of sorts and their replacement in new positions. If there was an accident of the press here, it did not involve the extensive pieing of type that may occur when furniture fails under pressure.

    Conclusion

    Nothing in the resetting of type on N(outer) rules out correction for the usual purpose of improving the goodness of what was printed, judged by any of the usual criteria of 'fidelity to copy', 'linguistic plausibility', and 'attractiveness of presentation'. But the coincidence of unusual clustering of changes on one forme with a break in the pattern of headline reuse on that forme provides a plausible mechanical explanation for the extensive resetting of type on N(outer). That the changes on this forme are relatively undermotivated compare to those elsewhere in Q2 Hamlet gives further cause for suspicion. The largest single alteration (involving resetting 4 lines of type on N3^r) occurs where an oversized headline (likely to make the type underneath it loose) was borrowed, quite possibly for the first time in the job, from the other compositor. If we assume no accident then this difficult correction was made simply to improve the reading, in which case it is hard to understand why the correction itself was bungled: what was intended was presumably be hangers > might be hangers not be hangers > be might hangers as the three UK exemplars read. It is easier to suppose that an accident, short of extensive pieing, affected these lines and that it caused the page to receive disproportionate attention and resetting. The consequences for editors are a matter considered elsewhere (Egan 2009).

 

    Appendix One. The printer's measure in 4 of the 7 exemplars of Q2 Hamlet

Page by X	Page by Y	Its width in C2	Its width in C2	Its width in VER	Its width in VER	Its width in L	Its width in L	Its width in HN	Its width in HN
F1 F1v F2 F2v F3 F3v F4 F4v N2 N2v N3 N3v Mean SD SD²/n	E4 G3 G3v G4 H1 H1v H2v H3 H3v H4 K1v K2 K2v K4 K4v L2 L2v L3 L3v M1v M2 M2v M3 M3v	98 99 98 98 98 98.5 98 98.5 97 98 97.5 97.5 98 0.522233 0.0227273	98.5         99 99 98.5 98.5 99 98.5 98.5 98.5 98.5 99 99 99 99 99 98.5 98 98.5 98.5 99 99 99 98.5 99 98.729167 0.294115 0.0036043	98 99 98 98 98 98.5 98 98.5 97.5 98 97.5 98 98.0833333 0.41742355 0.0145202	98.5 98.5 98 98.5 98.5 99 99 99 99 99 99 99 99 99 99 98.5 98.5 98.5 99 99 99 99 99 99 98.8125 0.28788962 0.00345335	97.5 98.5 97.5 98.5 98 98.5 98 99 97 98 97.5 97.5 97.9583333 0.58225008 0.02825126	98.5 98.5 98.5 98.5 99 99 98.5 99 98.5 98.5 99 99 99 99 99 98 98 98 98.5 98.5 98.5 98.5 98.5 98.5 98.604167 0.3290027 0.0045101	98 98.5 98 99 98.5 99 98 99 98 98.5 98 98.5 98.41666667 0.417423555 0.014520202	99 99 99 99 99.5 99.5 99.5 99.5 99.5 99 99.5 99.5 99.5 99.5 99.5 99.5 99.5 99.5 99.5 99 99 99 99 99.5 99.3125 0.247267678 0.002547554
Difference in means		0.729166667		0.729166667		0.645833333		0.895833333
Variance		0.02633159		0.017973553		0.032761377		0.017067756
SD of difference in means		0.162270115		0.134065482		0.181001042		0.130643624
95% confidence lower limit		0.41111724		0.466398322		0.29107129		0.639771831
95% confidence upper limit		1.047216093		0.991935011		1.000595376		1.151894835
mid-point		0.729166667		0.729166667		0.645833333		0.895833333
average of midpoints		0.75

 

Legend and notes
Type-page widths are in millimeters
SD = standard deviation
n = number of data points (12 for compositor X, 24 for compositor Y)
Variance = here, the variance of the difference in means, thus the sum of the SD²/n values for each compositor
SD of difference in means = square root of the variance of the difference in means
95% confidence lower limit = difference in means minus 1.96 times SD of difference in means
95% confidence upper limit = difference in means plus 1.96 times SD of difference in means
mid-point = half-way between 95% confidence lower limit and 95% confidence upper limit

 

Appendix Two. Statistical Analysis

    Regarding the printer's measure used by compositors X and Y, we can say with some confidence that they were different and by how much. We can think of the 12 readings for compositor X as a sample from a wider body of measurements that we could not take (because he set prose on only 12 pages) and this wider body of measurements would have a mean value that we do not know. Likewise for compositor Y, although we have a larger sample, 24 readings, from the wider body of measurements with an unknown mean. We are interested in the difference between the two unknown means, and can use the statistic called 'the difference in the sample means' to comment upon it. In Appendix One, the numerical means of the sample for each book are given: this is simply the sum of the readings divided by the number of readings, 12 for compositor X and 24 for compositor Y. An expression of how widely or narrowly the readings are spread around the mean is called the 'standard deviation'. This is calculated by squaring each reading's difference from the mean, summing these squares and then dividing that sum by the number of readings, and finally taking the square root of this quotient.

    Once we have the standard deviations for the sample readings, these can be used to calculate a pair of numbers, a lower limit and an upper limit, for which we can say to an arbitrary level of confidence that the mean of the unknown distribution readings (that is, the actual width of compositor X or compositor Y's composing stick) falls within those limits. Obviously we cannot say with 100% confidence what this unknown mean is, but we can say with very nearly 100% confidence that for both men it lies between 50 millimeters and 150 millimeters. The lower the confidence level, the narrower the span between the lower and upper limits, and a typically useful value for the confidence level is 95%. The formulas giving the lower and upper limits for a confidence level of 95% are:

Lower limit = Y_mean - X_mean - (1.96 x √ Comp X's SD²/n + Comp Y's SD²/n  ) 

Upper limit = Y_mean - X_mean + (1.96 x √ Comp X's SD²/n + Comp Y's SD²/n  ) 

where X_mean is the average of the compositor X readings, Y_mean is the average of the compositor Y readings and n is the number of readings in each man's sample.

The results of the calculations for each exemplar are in Appendix One. Of course, the compositor's measure did not change between exemplars. The raw data vary because each sheet of hand-made paper would have absorbed a different amount of water when wetted for printing, would have shrunk by a different amount when dried (and during storage over the ensuing centuries in different locations), and because the depth of ink applied before each pull would vary, as would the pressure exerted by each pull and hence the depth that the type bit into the paper. Also, there is human error in measuring by eye. The readings were taken by placing a measuring rule on the page to press it flat and recording the full distance from the first sign of ink in the first letter of the line to the last sign of ink in the last letter on the line, ignoring where necessary letters with kerns extending beyond the body of the type. Where different lines on a page produced different readings, the readings for the page were averaged. The Huntington exemplar values for both compositors are consistently higher (by just under half a millimeter) than those for the Capell, Verulam, and British Library exemplars, which might reflect a permanent expansion upon washing. The British Library reckons its exemplar has probably not been washed (Goff 2009). To combine the evidence from the four exemplars, we may take the average of the four mid-points lying half way between the lower and upper limit for each exemplar. This produces the result that compositor Y's stick was around three-quarters of a millimeter wider than compositor X's stick.

* I am grateful to Ian Gadd for his comments on an early version of this paper, especially regarding its scope and the presentation of data, to John Jowett for discussion of the likely regularity in the work of compositors and pressmen, and to my sister Mary Egan for explaining the mathematical procedures for statistical analysis of the raw data. I would also like to thank the librarians at Trinity College Cambridge, the Bodleian Library Oxford, the British Library London (especially Moira Goff), and the Huntington Library California (especially Holly Moore and Stephen Tabor), for their kind offices in providing access to exemplars of Q2 Hamlet, and the owner of the copy deposited at Oxford, the earl of Verulam, for permitting use of it.

Works Cited

Bowers, Fredson. 1953-4. "Shakespeare's Text and Bibliographical Method." Studies in Bibliography 6. 71-91.

Bowers, Fredson. 1953. "A Definitive Text of Shakespeare: Problems and Methods." Studies in Shakespeare. Edited by Arthur D. Matthews and Clark M. Emery. University of Miami Publications in English and American Literature. 1. Coral Gables FLA. University of Miami Press. 11-29.

Bowers, Fredson. 1955. "The Printing of Hamlet, Q2." Studies in Bibliography 7. 41-50.

Bowers, Fredson. 1956. "Addendum: The Printing of Hamlet Q2." Studies in Bibliography 8. 267-69.

Brown, John Russell. 1955. "The Compositors of Hamlet Q2 and The Merchant of Venice." Studies in Bibliography 7. 17-40.

Egan, Gabriel. 2009. 'Press Correction and the Shakespearian Editor': A Paper Delivered at the Conference 'The Future of Shakespeare's Text' at Loyola University Chicago on 7 November.

Ferguson, W. Craig. 1989. Pica Roman Type in Elizabethan England. Aldershot. Scolar Press.

Goff, Moira. 2009. 'Expansion of Paper in Q2 Hamlet': Personal Correspondence to the Author.

McKenzie, D. F. 1969. "Printers of the Mind: Some Notes on Bibliographical Theories and Printing-house Practices." Studies in Bibliography 22. 1-75.

McKerrow, Ronald B. 1927. An Introduction to Bibliography for Literary Students. Oxford. Clarendon Press.

Rasmussen, Eric. 2008. 'Blindness and Insight: Damaged Type, Damaged Eyes, and Q2 Hamlet': Internet Shakespeare Editions, University of Victoria, Http//internetshakespeare.uvic.ca/Library/Criticism/textualstudies/erasmussen.html:.

Shakespeare, William. 2006. Hamlet. Ed. Ann Thompson and Neil Taylor. The Arden Shakespeare. London. Thomson Learning.

Weiss, Adrian. 1989. "Review of W. Craig Ferguson Pica Roman Type in Elizabethan England (Aldershot: Scholar, 1989)." Papers of the Bibliographical Society of America 83. 539-46.

Weiss, Adrian. 1991. "Bibliographical Methods for Identifying Unknown Printers in Elizabethan/Jacobean Books." Studies in Bibliography 44. 183-228.

Weiss, Adrian. 1999. "A 'Fill-in' Job: The Textual Crux and Interrupted Printing in Thomas Middleton's The Triumphs of Honor and Virtue (1622)." Papers of the Bibliographical Society of America 83. 53-73.

Wilson, John Dover. 1934. The Manuscript of Shakespeare's Hamlet and the Problems of its Transmission: An Essay in Critical Bibliography. Vol. 1: The Texts of 1605 and 1623. 2 vols. Shakespeare's Problems. 4. Cambridge. Cambridge University Press.