University of Virginia Library

At least three compositors produced the two volumes of Edmund Spenser's Faerie Queene printed in 1596 by Richard Field.[1] Within each volume two hands are clearly distinguished by different skeleton-formes, different ornamental boxes surrounding the canto "arguments," different type-cases, different spelling habits, and different stylistic preferences in such matters as capitalization in the "arguments" and the use of italic or roman fonts for certain catchwords. Although some links between type-cases prove that the two volumes were printed not simultaneously but consecutively, the volumes themselves differ from each other in skeleton-formes and ornamental boxes and display no two entirely parallel patterns of stylistic and spelling preferences. The evidence shows that the two pairs of compositors could not have been identical, but does not conclusively prove or disprove the possibility that one compositor may have worked on both volumes. I shall refer to the compositors as A1 and B1 (first volume) and A2 and B2 (second volume).

The two volumes of Field's 1596 Faerie Queene, printed for William Ponsonbie, have different provenance as well as distinct bibliographic identities. The first, commonly referred to as the "First Part" although not actually so designated on its title page, is substantially a page-for-page reprint of the volume containing Books I-III which John Wolfe printed, also for Ponsonbie, in 1590. Field's second volume, designated "The Second Part" on its title page, prints the new Books IV-VI from now-lost manuscript copy. Like Wolfe's 1590 edition, the two Parts of

1596 are quartos in eights, separately paged and quired. The collation of the First Part is 4°: A-Z8, Aa-Oo8. That of the Second Part is 4°: A-Z8, Aa-Ii8 Kk4. To distinguish the parallel series of signatures, I employ prefixed superscript numerals except in tables where the distinction can be made obvious by other means. I designate individual whole formes by the signature letter or letters followed by "I" or "II" (denoting the outer and inner sheets respectively) and the symbols "(o)" and "(i)" denoting the outer and inner formes of the sheet. Thus, for example, "2C II (o)" means the outer forme of the inner sheet of gathering C in the Second Part. Skeleton-formes I denote by letters a through h.

Eight different skeleton-formes can be identified in the two volumes. Four skeletons were employed in each Part. They appear in pairs, each pair turning up normally in alternate gatherings. Almost invariably, one skeleton of a pair was used to print and perfect the outer sheet of a gathering and the other to print and perfect its inner sheet. These four pairs of skeletons are the primary but not entirely sufficient means of distinguishing the compositors.

In the First Part, the two pairs of skeletons emerge slowly. Formes of gatherings 1A and 1B reveal a series of changing combinations of headlines; some of their running-titles turn up in each of the skeletons that eventually take shape. Skeletons a and b (used by Compositor A1) become recognizable in gathering 1C, attain constant sets of titles in gatherings 1E and 1G respectively, and appear in alternate gatherings thereafter. Skeleton a is found on both sides of outer sheets, b on both sides of inner ones. These skeletons are characterized by occasional diagonal switches in the relative positions of either the recto or verso headlines, a phenomenon that can be explained only if the compositor sometimes varied his usually fixed routine of stripping and reimposing. The other pair in the First Part, skeletons c and d (used by Compositor B1), are markedly more irregular. They become recognizable in gathering 1D and appear in alternate gatherings thereafter, but continue to show shifting combinations of titles and introductions of new titles. At times titles pass from one skeleton of the pair to the other. Although these shifts and migrations obscure their identities and indeed make it impossible to define them strictly as consistent skeletons throughout the volume, each is nearly always recognizable as the same skeleton on both sides of a given sheet. That is to say, sheets were printed and perfected with the same skeleton just as is true almost invariably of sheets printed and perfected in the other three pairs of skeletons. Skeletons c and d however, unlike the other pairs, frequently change between gatherings. Thus, for this pair, it is impossible to align one skeleton with outer and the other with inner sheets. To illustrate, a title with the error "QVEEENE" first appears on an inner sheet (page 1D3), switches to an

outer sheet on both sides (pages 1F1 and 1F2), reverts to an inner sheet (page 1H3), and subsequently appears corrected on various formes. Fortunately, in the present study no question arises which involves the order of printing formes in this pair of skeletons.

In the Second Part, all the skeleton-formes are easily identified. Skeletons e and f (Compositor A2) appear in gathering 2A and invariably thereafter show constant sets of running-titles. Skeletons g and h (Compositor B2) appear respectively on the outer sheet of gathering 2C (both formes) and on the inner forme of the inner sheet of gathering 2D. When first set up, skeleton h curiously duplicated on pages 2D6, 2G5, 2G6, and 2I5 the erroneous title "QVEEENE" found in the First Part. These errors are coincidental so far as the settings themselves are concerned, for the settings in the two Parts are distinct. After correction of this error, skeletons g and h have constant sets of running-titles, but the relative positions of the headlines are occasionally transposed when the forme was turned end-for-end between impositions. All four pairs of skeletons in the two volumes differ in the manner of their handling as well as in their physical identities.

Excluding the formes of gatherings 1A and 1B (where the skeletons cannot be identified) and grouping skeletons c-d together as a pair (since they cannot be consistently distinguished from each other throughout the volume), the following table indicates the formes imposed in each skeleton. Because the same skeleton nearly always appears on both sides of a sheet, only the sheets are discriminated in the table except in the few irregular instances.

                                           

First Part	Second Part
Skeletons	Skeletons	Skeletons	Skeletons
a	b	c-d	e	f	g	h
C I	C II	D I-II	A I	A II
E I	E II	F I-II	B I	B II
G I	G II	H I-II	C II		C I
I I	I II	K I-II			D I
L I	L II	M I-II	E I	E II	D II(o)	D II(i)
N I	N II	O I-II	F I	F II	G I	G II
P I	P II	Q I-II	H I	H II	I I	I II
R I	R II	S I-II	K I	K II	L I	L II
T I	T II	V I-II	M I	M II	N I	N II
X I	X II	Y I-II	O I	O II	P I	P II
Z I	Z II	Aa I-II	Q I	Q II	R I	R II
Bb I	Bb II	Cc I-II	T I	T II	S I	S II
Dd I	Dd II	Ee I-II	V I	V II	X I	X II
Ff I	Ff II	Gg I-II	Y I	Y II	Z I	Z II
Hh I	Hh II	Ii I-II	Aa I	Aa II	Bb I	Bb II
Kk I	Kk II	Ll I-II	Cc I	Cc II	Dd I	Dd II
Mm I	Mm II	Nn I-II	Ee I	Ee II	Ff I	Ff II
Oo I	Oo II		Gg I	Gg II	Hh I	Hh II
			Ii I	Ii II	Kk (i)	Kk (o)

With some exceptions which further evidence will elucidate, these four groups of formes represent the work of the various compositors.

Recurring impressions of identifiable type prove that within each volume the two compositors employed separate type-cases for most but perhaps not all fonts. In the italic used for the "arguments," some identifiable types are confined to one or another of the groups of formes produced by each compositor; but a few may occur in both groups of the same volume. The fineness of this type and the relative brevity of this part of the text make trustworthy evidence hard to find. But one instance at least is quite clear. The italic 'B' damaged in two places which appears in the "arguments" on pages 1F6, 1Cc6v, and 1Ee7v (skeletons c-d) is identical to that on page 1N8 (skeleton a). If the compositors began with different type-cases for this font, they did not keep them entirely separate. On the other hand, I have found no examples indicating a common type-case when the same italic font is used for names in the main body of the text. Several large display types at canto beginnings recur often enough to show that each group of formes is independent with respect to this font also.

In the main body of the text, including both the normal roman and the italic used for names, distinctive types are frequent and show that each group of formes in a volume reflects a different type-case. Two examples from each group will suffice for illustration. In formes imposed in skeletons a and b:

• (1) A roman 'B' showing a break in the vertical near the middle and a smudge within the upper loop: pages 1N2v, 1P6v, 1X6v, 1X7v, 1Hh8, 1Kk1, 1Mm7, 1Mm8, 1Oo7v.
• (2) A roman 's', its lower curve bent downward and separated from the tip: pages 1C3v, 1C6v, 1E3v, 1E6v, 1L2v, 1N1, 1N2, 1P1, 1T3, 1X1v, 1X4v, 1Bb4v, 1Dd4v, 1Dd7, 1Ff2v, 1Kk4v, 1Kk5v, 1Mm5v.

In formes imposed in skeletons c-d:

• (3) A swash italic 'M', its bottom serif at the right separated from the vertical and, beginning with the second impression, a break at the juncture of the right-hand slanting and vertical strokes: pages 1S1v, 1Y4v, 1Aa4, 1Ee5, 1Ii4, 1Ll4, 1Nn1v.
• (4) A swash italic 'M' showing a break in its left-hand slanting stroke: pages 1A2v, 1K1v, 1O2, 1Ll8, 1Nn6v.

In formes imposed in skeletons e and f:

• (5) A roman 'L', its horizontal broken near the juncture with the vertical: pages 2B6v, 2B8, 2C5, 2E4, 2F8v, 2H1, 2K4, 2O5v, 2Q3v, 2Ii6.
• (6) A roman double 'ff', the second letter broken above its cross-bar: pages 2A5, 2C6, 2E7v, 2F5v, 2O1v, 2Q5v, 2T6, 2V1v.

In formes imposed in skeletons g and h:

• (7) A roman 'g', its serif missing and the lower loop broken in three places: pages 2C1, 2C2, 2D6v, 2D7v, 2I3v, 2L5v, 2L6v, 2N5v, 2S2v, 2X4v, 2X5v, 2Bb4v, 2Dd4, 2Ff4.
• (8) A roman 'l', its bottom separated from the vertical: pages 2D3, 2D3v, 2I2v, 2L4v, 2X2, 2Bb3v, 2Bb6v, 2Dd2, 2Dd8v, 2Ff4, 2Hh5, 2Hh5v, 2Kk2, 2Kk2v.

Perhaps as many as eight distinctive types found in the first two gatherings of the First Part subsequently diverge into either group of formes: example (4) is an instance. In the Second Part, one type found first in a forme imposed in skeleton e next appears in skeleton g, probably the first and second times that skeleton was used:

• (9) A roman 'A' with a broken bar: pages 2B1v, then 2C8 and probably also 2C8v, 2D2.

I have discovered no other type in the main body of the text which occurs in both groups of formes within either volume.

Of the seven types I have found common to both volumes, three disappear for suspiciously long intervals between their last occurrences in the First Part and their first in the Second. One of these links the type-cases of Compositors A1 and A2:

• (10) A swash italic 'M', its lower right serif separated from the vertical and showing a break near the top of its left slanting stroke, similar to examples (3) and (4) but not the same: pages 1E6v (skeleton b); 2H4v, 2K5, 2O4, 2T3v, 2V2, 2Aa6v, 2Cc5, 2Ee8v, 2Gg4v, 2Ii4v (skeletons e-f).

The other two link the type-cases of Compositors B1 and B2:

• (11) A roman 'g' with a distorted lower loop: pages 1B4v, 1F6v, 1M1, 1M8, 1O3 (skeletons c-d); 2X2v, 2Z1, 2Bb4v (skeletons g-h).
• (12) A roman 'e', its tail distorted: pages 1A6, 1D1v, 1D8v, 1F2, 1F7, 1H8v, 1K7, 1M8, 1Aa7v, 1Cc8, 1Nn1 (skeletons c-d); 2P4v, 2Bb6, 2Dd8, 2Ff1v (skeletons g-h).

One of the remaining examples also links the type-cases of Compositors B1 and B2:

• (13) A roman 'w', its left-hand 'v' higher than the right and broken to the right of the point: pages 1F4, 1H1v, 1K1, 1K3v, 1M6v, 1O1, 1O7v, 1Q2v, 1Q3v, 1S2, 1V3, 1V3v, 1Y6v, 1Aa4, 1Aa7, 1Cc5, 1Cc6, 1Ee8, 1Gg3, 1Gg4, 1Ii4, 1Ii6v, 1Ll5v, 1Ll6v, 1Nn1v (skeletons c-d); 2D2v, 2D8, 2G1, 2G2, 2I1v, 2I7, 2P1v, 2P8v, 2R2v (skeletons g-h).

The other three examples contradictorily link the type-cases of Compositors B1 and A2:

• (14) An italic 'B', its lower loop broken at the bottom: pages 1Cc8v, 1Ee6v, 1Nn4v (skeletons c-d); 2A4, 2B5, 2E1, 2O7, 2Y8v (skeletons e-f).
• (15) A swash italic 'M' damaged across the tops of all but the right-hand vertical: pages 1A8v, 1S6v, 1Y4v, 1Ee1v, 1Gg1v, 1Ll4v, 1Nn5 (skeletons c-d); 2K7, 2O3, 2T8, 2V4, 2Gg6v, 2Ii1 (skeletons e-f).
• (16) A roman 'c' showing a dot above a downward bent tail: pages 1Ee1, 1Ll5 (skeletons c-d); 2A5v, 2B4v, 2O4v (skeletons e-f).

Four patterns of canto divisions paralleling the four pairs of skeleton-formes provide another means of distinguishing the compositors. Each pattern has several characteristics. The ornamental boxes enclosing the canto "arguments" are different repeated settings identifiable by the arrangement of the individual type-units composing them. Three patterns show duplicate boxes (employed because two cantos sometimes begin on the same forme) and the fourth has one box in four states. Both boxes used by Compositor A1 have an ovoid scroll design in obvious contrast to the floral scroll designs of all the others. The box in four states used normally by Compositor B2 is also easily recognizable by the position of its bottom border within rather than under the sides. Over a box appears the word "Canto" or "Cant." (followed by the appropriate number) which is usually a repeated setting associated with that box. Compositor A1 employs a normal 'C' in this heading with his first box, but a swash 'C' with his second box on its single occurrence. Compositors B1 and A2 employ swash forms consistently; B2 employs the normal. The "arguments" within the boxes provide a third identifying feature. These "arguments" are heptameter couplets printed as four lines, the even lines indented and not requiring initial capitals unless of course they begin with names or titles. In two instances words are split between lines (1.7 "great-/ly" and 2.6 "bro-/ther") so that capitalization would be quite awkward. Yet capitals occur in the second and fourth lines of some "arguments" both in 1590 and 1596. Each compositor in 1596 has a clear preference in this matter which is obvious from inspection and (in the First Part) comparison with his copy. Compositor A1 altered 1590 eight times to lower-case and preserved only four unnecessary capitals. Compositor B1 altered his copy four times to capitals; otherwise he merely reproduced 1590 with its preponderance of capitals. Disregarding for the moment two "arguments" of dubious provenance, Compositor A2 set only three

unnecessary capitals in thirty-three opportunities, never using two in the same "argument." Compositor B2, on the other hand, never set a lower-case initial in at least twenty-five chances. The final feature of the patterns of canto divisions is the font used for catchwords referring to new cantos when these begin at the top of a page. Compositors A1 and A2print them in roman capitals ("CANT.") but B1 and B2 in italic ("Cant.").

The pages on which features of each pattern appear are shown in the lists below. The various boxes and their accompanying headings are indicated by roman numerals followed by "a" or "b" to distinguish duplicate settings or by arabic subscripts to distinguish states. Initial words of the even lines of "arguments" are given to show each compositor's practice with respect to capitalization. It may be remarked that two unnecessary capitals in the "arguments" of Compositor A1 may have been introduced for the sake of symmetry when a name occurs in the other line, although this consideration does not seem to have influenced the other compositors.

First Part



• Compositor A1


• C1v Ia And, In
• C7v CANT.
• C8 Ib guides, doth
• E7 Ia fayre, and
• G5v Ia brings, and
• I4v CANT.
• I5 Ia to, the
• L6v CANT.
• L7 Ia betrouthed, her
• N7v CANT.
• N8 Ia the, striue
• P5v Ia and, by
• R1v Ia led, —ther
• T2v Ia Acrates, And
• X3v CANT.
• X4 Ia from, till
• Z6v Ia passing, and
• Bb4v CANT.
• Bb5 Ia faire, champions
• Dd6 Ia the, which
• Ff8v Ia three, and
• Hh8v Ia she, from
• Kk1v Ia like, is
• Oo1v Ia Chamber, Amoret



• Compositor B1


• A2v CANT.
• A3 IIa Foule, Doth
• B2v Cant.
• B3 IIa The, And
• D7 Cant.
• D7v IIa subdewes, his
• F6 IIa By, —ly
• H5 IIa The, Whom
• K6v Cant.
• K7 IIa two, most
• M6v Cant.
• M7 IIa The, With
• O6v IIa horse, Belphoebe
• Q4 IIa And, Attin
• S1 IIb Sunning, To
• V2v Cant.
• V3 IIa doth, knightes
• Y7 IIa besiege, Maleger
• Cc6v IIa describeth, in
• Ee7 Cant.
• Ee7v IIa Is, Long
• Gg8v IIb Of, With
• Ll1 IIb For, Both
• Ll8v IIa Malbecco, To
• Nn1v IIb findes, where

In the list below for the Second Part, five irregularities are marked by asterisks.

Second Part



• Compositor A2


• A2v CANT.
• A3 IIIa Duessa, Their
• B2v CANT.
• B3 IIIa Paridell, doth
• *C2v CANT.
• C3 IIIa Cambell, doth
• E1v IIIa of, doth
• E8v IIIa Doe, and
• F7v IIIa Belphebe, his
• H7v CANT.
• H8 IIIa Poeana, Prince
• K6v IIIa he, and
• M5v CANT.
• M6 IIIa Irenaes, his
• O2v IIIb where, in
• O8v IIIa two, and
• Q8v IIIb of, who
• T4v IIIb whom, and
• V3v CANT.
• V4 IIIb for, he
• Aa2v IIIb A, his
• Cc1 IIIb from, he
• Cc6v CANT.
• Cc7 IIIa till, with
• Ee3v IIIb doe, for
• Gg2v IIIb and, for
• Ii7 IIIb her, subdew



• Compositor B2


• D2v IV1 For, And
• G6v IV1 Sclaunder, And
• I5v Cant.
• I6 IV2 Of, And
• L6 IV2 In, And
• N2v IV3 Does, Does
• P7v Canto.
• P8 IV4 And, But
• *R6v IIIa Where, And
• S5 IV4 Free, Adicia
• *X1v IIIa Gerioneo, Belge
• *X6v Canto.
• *Y3 IV4 And, And
• Z3v IV4 A, Briana
• Bb1v IV4 Pursues, By
• Bb8v Canto.
• Dd4v Canto.
• Dd5 IV4 Of, For
• Ff2v Canto.
• Ff3 IV4 Quites, By
• Hh1 IV4 To, Into
• Hh7v IV4 Whilest, And

With four exceptions, all instances of the canto division patterns appear on pages falling in the corresponding groups of formes defined by the pairs of skeletons. The four exceptions are: (1) the roman catchword of Compositor A2 on page 2C2v occurs in skeleton g, on its first or second use; (2) the same compositor's box IIIa occurs on page 2R6v in skeleton h, where its "argument" contains two unnecessary capitals characteristic of B2 rather than A2; (3) the same box occurs again on page 2X1v in skeleton g, where proper names make the capitals inconclusive; and (4) box IV4 of Compositor B2, with two unnecessary capitals characteristic of him in its "argument," occurs on page 2Y3 in skeleton f. The fifth irregularity is that the catchword on page 2X6v, although in B2's normal italics and his skeleton, is erroneous: the predicted canto begins not on page 2X7 but on 2Y3. Explanations of these irregularities will be given in due course.

Spellings offer still another means of comparing the compositors. In the First Part, the habits of Compositors A1 and B1 can be ascertained quite precisely by observing their alterations and reproductions of the known copy. The spelling of 1590 is itself inconsistent, because (as there is evidence to prove) Wolfe like Field employed more than one compositor and they did not uniformly follow the manuscript of Books I-III. Since the division of 1590 between compositors does not parallel that of 1596, Compositors A1 and B1 each found in their copy the same variants in roughly the same proportions. If a compositor consistently rejects one variant offered by his copy in favor of another, he obviously exercises a genuine preference. If he tolerates more than one variant, especially if he sometimes alters his copy inconsistently, his lack of a real spelling preference is equally clear. Knowledge of his rejections and tolerations serves as a check on the significance of mere totals, which in themselves may be accidental. In citing statistics for Compositors A1 and B1, I use a system of notation which distinguishes between alterations and reproductions of copy and also indicates totals. Thus, for example, "71> farre = 88" means that in seventy-one instances the spelling of 1590 has been altered in 1596 to farre, resulting in a total of eighty-eight occurrences of this spelling (seventeen having been reproduced). If either symbol is given alone, all instances are alterations or reproductions. It is usually unnecessary to specify the spellings of 1590. Excluding a very few and very minor exceptions, only two variants are involved for most words, and the spellings of 1590 can be inferred from the symbols. Spellings in the first two gatherings of the First Part are not counted, of course, when there is a question of discriminating between Compositors A1 and B1, because the assignment of formes in those gatherings is uncertain.

Not knowing the copy for the Second Part, we have no certain check on the significance of total variants in the work of Compositors A2 and B2. But even if the manuscript was not Spenser's holograph—which it quite likely was—there is good reason to believe that it contained a high proportion of spellings known to be his.[2] Although known Spenserian spellings generally diminish between 1590 and the First Part, they often increase in the Second even for some words which both compositors almost certainly prefer to spell differently from Spenser. A particularly striking example is the fate of Spenser's medial -oo- in blood and bloody. In 1590, except in the first few gatherings,

Spenser's spelling occurs almost invariably. But in the First Part of 1596, 65> bloudy (-ie) = 70. The adjective is never spelled as Spenser spells it. When the noun appears in non-rhyming positions, 96> bloud = 101 and blood = 1. In rhymes, for the sake of conformity with wood, good and so forth, 1> blood =21 although 1> bloud = 2. Both compositors in the Second Part also prefer -ou- (which appears five times even in rhymes), but instances of -oo- increase in the adjective and in the noun when it does not form a rhyme. The adjective is bloud- 29 but blood- 9. In non-rhyming positions, the noun is bloud 42 but blood 8. Compositor B2 insists on -ou- somewhat more than does A2, but he contributes three -oo- spellings in the adjective and one in the non-rhyming noun. Other Spenserian spellings which similarly diminish in the First Part but increase in the Second are abrode, admyre, doe, farre, fynd (e), mynd (e); these will be cited in detail later. If we should suppose that Compositors A2 and B2 actually preferred the Spenserian forms of these words but found the non-Spenserian in their copy (or that they tolerated inconsistent copy), we should have to concede immediately that they could not be the same individuals as A1 and B1. But the more plausible explanation is that Spenserian spellings predominated in the manuscript. If this deduction is sound, we have a good clue to the significance of total variants in the Second Part when Spenser's spelling of a word is known.

Variant spellings of eighteen words are given below as they occur in the work of each compositor in 1596. Spenser's known preferences are marked by asterisks. Although his practice was not entirely consistent, his preferences for most of the words in this list are well-established.

                                     

Compositor A1				Compositor B1				Compositor A2		Compositor B2
	abroad	=	6	2>	abroad	=	10	*abrode	4	*abrode	8
	*abrode	=	4		*abrode	=	1	abroad	1	abroad	2
	*admyre	=	2	2>	admire	=	3	*admyre	9	admire	8
	admire	=	2					admire	6	*admyre	2
3>	find(e)	=	30	11>	find(e)	=	48	find(e)	34	find(e)	15
1>	*fynd(e)	=	3					*fynd(e)	9	*fynd(e)	8
4>	mind(e)	=	38	12>	mind(e)	=	43	mind(e)	46	mind(e)	28
1>	*mynd(e)	=	1		*mynd(e)	=	2	*mynd(e)	8	*mynd(e)	18
1>	alwaies	=	6	4>	alwayes	=	7	alwayes	5	alwaies	9
								alwaies	2	alwayes	1
4>	dayes	=	16	10>	dayes	=	24	dayes	18	daies	13
	daies	=	4					daies	3	dayes	5
1>	wayes	=	14	9>	wayes	=	16	wayes	9	waies	12
	waies	=	2					waies	2	wayes	1
25>	bl(ou)die	=	27	2>	*bl(ou)dy	=	31	bl(ou)die	17	*bl(ou)dy	14
	*bl(ou)dy	=	2	2>	bl(ou)die	=	3	*bl(ou)dy	6	bl(ou)die	1
17>	mightie	=	35	13>	*mighty	=	38	mightie	29	*mighty	26
1>	*mighty	=	3	2>	mightie	=	8	*mighty	3	mightie	1

                                 

Compositor A1				Compositor B1				Compositor A2		Compositor B2
14>	do	=	40	94>	do	=	116	*doe	124	*doe	104
6>	*doe	=	90					do	28	do	17
71>	*farre	=	88	3>	far	=	51	*farre	69	*farre	25
	far	=	3	1>	*farre	=	17	far	1	far	23
4>	*honour	=	20	5>	*honour	=	28	*honour	21	*honour	12
	honor	=	6					honor	7	honor	5
3>	horror	=	15	13>	*horrour	=	19	horror	5	*horrour	8
	*horrour	=	4					*horrour	3
5>	*litle	=	29	5>	little	=	15	*litle	48	little	25
5>	little	=	13		*litle	=	17	little	6	*litle	12
	*whiles	=	62	9>	*whiles	=	55	*whiles	23	*whiles	20
								whyles	9
3>	*whil(e)st	=	11	2>	*whil(e)st	=	14	*whil(e)st	20	*whil(e)st	26
	whylest	=	1					whylest	16	whylest	1
3>	whylome	=	14	6>	*whilome	=	15	whylome	18	*whilome	12
1>	*whilome	=	2		whylome	=	2	*whilome	3	whylome	5

A final set of variants is too complicated to summarize in a table. In 1590, usual spellings are suddein (e) 36 and suddeinly 15. Suddain(e) occurs twice and suddenly four times in the first two gatherings; in the corresponding gatherings of 1596, these spellings and three instances of suddein (e) are reproduced. In gathering 1C, the first known to be Compositor A1's, the usual spellings of 1590 are all changed: 3> suddaine and 1> suddainly. In the rest of his work, suddein (e) = 17 and suddeinly = 6, but 7> suddenly = 8. Compositor B1 in work known to be his invariably reproduces suddein (e) = 13 and suddeinly = 6. In the Second Part, Compositor A2's spellings are suddain 1, but sudden at least 9 and suddenly at least 8. On formes 2C I (i) and 2D I (i), among the first to be printed in skeleton g, the spellings are sudden 4 and suddenly 3. On forme 2D II (o), also printed in skeleton g, *sodein occurs once. This is Spenser's preferred spelling, although he also spells soddein. Compositor B2's invariable spellings in subsequent formes in skeletons g and h are sodaine 9 and sodainely 6.

Several of these variant spellings confirm the other evidence which distinguishes two compositors within each volume. In the First Part, Compositor A1 exhibits strong preferences for -ie and for *farre. Though less positively, B1 prefers *-y and far. Compositor B1's overwhelming rejection of *doe contrasts sharply with A1's nearly equal retention of this spelling. Compositor A1's alwaies, *litle, and whylome balance B1's alwayes, little, and *whilome. In the Second Part, regardless of what spellings appeared in the copy, it can scarcely be chance in every case that Compositor A2's majority spellings are *admyre, alwayes, dayes, wayes, bloudie, mightie, *litle, sudden, and whylome; whereas B2's are admire, alwaies, daies, waies, bloudy, *mighty, little,

sodaine, and *whilome. Neither compositor is consistently Spenserian. Whichever may have been more faithful to the copy in spelling any one of these words, the other generally departed from it. If we make the likely assumption that the manuscript spellings were Spenser's, A2 shows no clear preference in merely reproducing *farre; but then B2 shows a strong liking for far. Other examples which on this assumption can be added to the list of differences are A2's relatively stronger preference for find (e) and mind (e), honor and horror, whyles and whylest.

Comparisons between the two volumes do not yield equally positive results. On grounds of other evidence as well as spellings, the only pairs of compositors likely to be identical are A1-A2 and B1-B2. Several particular spellings such as -ie or -y and *litle or little suggest this alignment. Compositors A1 and A2 parallel each other not only in nearly every specific spelling but also in their general pattern of tolerating a relatively high proportion of minority variants. For some words, in fact, the real preference of both may have been not a particular spelling of their own but fidelity to the copy. Thus Compositor A1 generally reproduces either do or *doe, *litle or little, and inconsistently changes to both forms of both words. If his personal preference was actually for little, even though this is his minority spelling, so too very likely was A2's. All six of A2's minority spellings little occur within the first nine instances of the word in his work; after gathering 2E, he consistently spells *litle. In contrast, ten of Compositor B2's twelve minority spellings *litle occur within the first fourteen instances; he spells little consistently after gathering 2N except in two instances near the end of the volume. One or the other of these compositors began by expressing his personal preference but suppressed it in favor of adherence to the copy; if the copy spelled *litle, this compositor was A2. The evidence would permit a similar argument that both A1 and A2 may have preferred do, which they employ frequently, even though *doe is their majority form. Regardless of the force of these arguments, however, the tolerance of Compositors A1 and A2 is sufficiently marked that the few apparent discrepancies between them bear little weight. Only abroad- *abrode, alwaies-alwayes, and suddein-suddain fail to fit into a consistent pattern. But fidelity to the copy readily explains the first two exceptions. If the copy for the Second Part spelled *sod (d)ein, it is perhaps difficult to understand why a compositor who usually reproduced suddein (e) in the First Part should offer -ain in the Second. But Compositor A1 had also spelled suddain, and A2 uses it only once. The more telling fact is that both also spell sudden. No significant differences in spelling distinguish these compositors. Although similarities cannot

prove their identity, the parallels are sufficient in the absence of any clear negative evidence to raise the presumption that A1 and A2 were the same individual.

Compositors B1 and B2, however, can scarcely have been the same man. A compositor who ruthlessly altered all ninety-four instances of *doe in 1590 to do in the First Part is not likely to have introduced or reproduced *doe one hundred and four times in the Second. It is hardly any more plausible, in the face of Compositor B1's obviously strong bias toward alwayes, dayes, and wayes, to suppose that he would have tolerated the preponderance of alwaies, daies, and waies in the Second Part even if these forms appeared in the copy. Although Compositor B1 practically eliminated *fynd (e), *mynd (e), honor, horror, and whylome (which appear rather frequently in his copy), both the Spenserian spellings and the non-Spenserian -or and whylome show up with relatively high frequencies in the work of B2. When B1 merely reproduces suddein from his copy, he discloses no certain preference. But B2, who reproduced Spenser's *sodein the first time he encountered it and then settled on sodaine, does reject -ein.

With the evidence now set forth, it is possible to clarify the irregularities already noted in the two volumes and draw certain conclusions about their printing.

Formes and even individual pages in the first two gatherings of the First Part show signs of both Compositors A1 and B1. Most of the running-titles found in the skeletons diverge and reappear in either of the pairs which eventually emerge. Several distinctive types found scattered on six of the eight formes likewise recur later in either group of formes. Compositor B1's ornamental box or his italic canto catchwords appear on all four outer formes. Unfortunately, reliable spelling clues are rare. The only persuasive ones are eleven alterations of the copy to do, indicating Compositor B1, scattered on six pages of five formes. Since Compositor A1's spelling preferences are less definite, they cannot be recognized even if he also had a hand in the composition. On some of the same pages and formes, however, occur types which later turn up in the case of Compositor A1. Although the evidence is meager, it suggests that Compositor B1 may have set most of the type, perhaps under A1's supervision, but that both compositors cooperated in distributing. Commencing with gathering 1C, the compositors worked independently throughout the First Part.

If the Second Part was printed immediately after the First, it is difficult to explain why the skeletons and boxes were not continued in use for the second volume and indeed why the same pair of compositors

did not continue. But it is quite possible that Ponsonbie arranged for the reprinting of Books I-III in advance of Spenser's arrival from Ireland with the manuscript of the new Books IV-VI. We do not know exactly when he reached London. Ponsonbie entered the new books in the Stationers' Register on January 20, 1596, and it is reasonable to suppose, first, that Spenser would have brought so long and important a manuscript himself and, second, that he would have lost no time in arranging for its publication. On the other hand, Ponsonbie's dedicatory letter in Amoretti and Epithalamion implies that Spenser sent the manuscript of this volume to London and states that Ponsonbie published it "in his absence." The copy was entered in the Stationers' Register November 19, 1594, and the title page is dated 1595. In the same year, also probably while Spenser was still in Ireland, Ponsonbie published Colin Clouts Come Home Again. Since the only substantial revisions of the First Part of the Faerie Queene in 1596 were one added stanza (1.11.3) and five new stanzas substituted for the three originally concluding Book III, Spenser may well have sent ahead the necessary copy and entrusted his publisher with the reprinting. If the First Part was completed before Spenser arrived with the new manuscript, we have an explanation of why new skeletons and boxes were set up and why there are relatively few links—some apparently contradictory—between the type-cases employed in the two volumes.[3]

When printing did begin for the Second Part, Compositor A2 was apparently using the type-case used by B1 in the First Part. Distinctive types (14), (15), and (16) quite closely link the cases of B1-A2. On the other hand, three of the four types— (10), (11), and (12) —which seem to link A1-A2 and B1-B2 disappear for long intervals between their appearances in the two Parts and are probably only strays. The fourth, type (13), actually transferred to the case of B2 through the hands of A2 early in the Second Part. Compositor A2 produced gatherings 2A and 2B and sheet 2C II (both formes), using skeletons e and f. He also composed formes 2C I(i) and (o) and 2D I (i), imposing them in skeleton g. Type (9) is first found on forme 2B I (i) in skeleton e and

then recurs on forme 2C I (i) and probably on 2C I (o) and 2D I (i). The canto catchword on forme 2C I (o) is in A2's characteristic roman capitals. Three instances of his spelling sudden (ly) occur on forme 2C I (i) and four on 2D I (i). At this point, forme 2D I (i) was transferred to and stripped and distributed by B2. His box, with characteristic unnecessary capitals in the "argument," occurs on forme 2D I (o), the next imposed in skeleton g. Along with this skeleton, type (13) transferred from its appearance on forme 2D I (i) to 2D I (o) and subsquently continues in B2's case. He set up a new skeleton h, containing the misprint "QVEEENE" in one of its titles, and imposed in it forme 2D II (i), which shows the spelling *sodein. He completed gathering 2D by imposing forme 2D II (o) in skeleton g.

The order in which Compositor B2 began to produce formes has an important bearing on the irregularities later in the Second Part. Since the stanzaic form of the poem made casting-off the copy relatively easy, any order would have been possible. Actually, the way in which B2 began to work seems to have led him, starting with gathering 2G, into the sequence (alternating between skeletons) inner sheet outer forme, outer sheet outer forme, inner sheet inner forme, outer sheet inner forme. Errors in his headlines imply in five out of six instances that he imposed outer formes before inner ones, and the single exception is itself part of the chain of later irregularities.[4] If he imposed outer formes first, it can be deduced that inner sheets must conversely have preceded outer ones, for otherwise he could not have managed as long as he did with only one canto heading and box: the opposite sequence would have required canto beginnings on some consecutive formes. The sequence he did follow was probably a result of his poor workmanship on the first forme he produced, that is 2D II (i). This is the only forme in the 1596 edition known to exist in as many as three states and it contains nine variants besides the still-uncorrected "QVEEENE" in a title. It was composed by B2 simultaneously with A2's composition of 2D I (i), imposed in skeleton g. When skeleton g

was transferred to him, B2 used it for forme 2DI (o) and also—since skeleton h was still tied up during the correction of 2DII (i)—for 2DII (o). The use of g twice in succession led to the priority of inner sheets, though not inevitably, as the skeletons could have been switched between sheets. The sequence may be represented thus: If this sequence is projected it and it alone will be found to accommodate all the evidence of errors in headlines, recurring type, and repeated settings of canto headings up to gathering 2X.

This sequence accounts for the first irregular appearance of Compositor A2's box IIIa on B2's forme 2R II (o). Compositor B2's box IV was on the press in the immediately preceding forme 2P I (i), and rather than set up a new one he borrowed his colleague's. He reverted to his own box IV on 2S II (o).

Compositor B2's second use of box IIIa on forme 2X I (i) is another borrowing connected with the erroneous canto catchword on page 2X6v and the anomalous appearance of his own box IV and "argument" in skeleton f when the predicted canto does begin on page 2Y3. This "argument" for Canto 12, Book V, announces as a topic of the canto Artegal's rescue of Sir Burbon; but this episode has already been related in stanzas 44-65 of Canto 11. None of the explanations of this discrepancy hitherto offered has been entirely plausible on literary grounds or has taken into account all the relevant bibliographic evidence.[5]

A hypothesis which does fit and explain the facts is that Spenser inserted stanzas 37-43 of Canto 11 into the poem after some of the copy originally intended at this point was already in type. These stanzas relate Artegal's meeting with Sir Sergis. In one sense at least, Sir Sergis is clearly a newcomer to the poem. Although Artegal (and Spenser) declare in this passage that he attended Irena when she appealed to the Faerie Queene for aid against Grantorto, Irena actually seems to have been alone on that occasion (5.1.4) and Sir Sergis has not been mentioned earlier. He appears in Canto 11 to inform Artegal that

Irena's plight has become urgent while her champion has been distracted by other matters. Spenser may have inserted this passage to stress the urgency of protecting Ireland against Spanish Catholicism. Whatever the poet's motive, Compositor B2 did not expect these stanzas or those relating the rescue of Burbon. The stanzas about Sir Sergis now exactly occupy pages 2X7 and 2X7v. But B2 set his catchword "Canto." after stanza 36 at the foot of page 2X6v. He composed the "argument" for Canto 12, announcing the rescue of Burbon, and imposed it in his box IV with the expectation that it would appear on page 2X7.

On this view, in the original copy Canto 11 consisted of Prince Arthur's rescue of Lady Belge from Gerioneo (stanzas 1-35) and a final stanza (36) making a transition to the titular hero Artegal, who proceeds on his delayed quest to aid Irena. The narrative of Canto 12 began with the Burbon story. Let us put the closing and opening lines of these narrative portions together:

Excepting the parenthetical tag "the cause to know," these stanzas link together with perfect narrative and rhetorical consistency. The parenthetical words refer to the "rout of people" that (as the text now stands) caught the attention of Artegal and Sir Sergis in stanza 43. But the same phenomenon of "a rude rout" is given independently in stanza 44. There is no other connection with stanzas 37-43. If we suppose only some other tag than "the cause to know," Spenser could have written stanzas 36 and 44 consecutively. His practice may well have been to write narrative portions without stopping to consider canto divisions, "arguments," and reflective introductory stanzas. When he introduced a canto division between stanzas 36 and 44 (as I think he originally did), the close rhetorical connection between them was broken and the beginning of stanza 44 would have become abrupt unless the original introductory stanzas of Canto 12 supplied some immediate antecedent for "To which." The present second stanza of Canto 12 refers to both the Burbon and Gerioneo episodes as past and only the Grantorto episode as imminent. In other words, it fits the text we now have (unlike the "argument") and therefore must have been written or revised after the Burbon incident was shifted to Canto 11. All that need be assumed to reconstruct a smooth original text is a somewhat different version of this stanza and some other tag than "the cause to know" in the first line of stanza 44.[6]

The hypothesis that Spenser inserted the Sir Sergis passage during printing fits and explains the bibliographical evidence. If the sequence in which B2 was producing formes is projected into gathering 2X, it would be:

Having completed the first of these formes, 2X II (o), which contains the catchword, B2 began to compose forme 2X I (o), which would have contained the beginning of Canto 12. This forme was ready for imposition or nearly so, I believe, when Spenser decided to insert the Sir Sergis passage after stanza 36, that is on pages 2X7 (of this forme) and 2X7v. Most if not all the text on forme 2X I (o) could be salvaged as soon as the necessary re-arrangement of pages became clear. Pending this, B2 suspended work on the outer forme and proceeded to compose the inner forme 2X I (i). Pages 2X1v and 2X2 of this forme were unaffected and could have been composed before Spenser supplied the new copy. But for the opening of Canto 11 on page 2X1v Compositor B2 borrowed box IIIa, because his own box IV was already set up with the "argument" for Canto 12. As soon as new copy was at hand and it had been decided to shift the Burbon episode to Canto 11, forme 2X I (i) was completed and imposed. As an error in a headline shows, this inner forme exceptionally preceded 2X I (o). (If the new sequence for formes of outer sheets thus established is projected, it accommodates all the facts in the rest of his formes.) Forme 2X II (i), which was entirely unaffected by the textual changes, was produced, and then 2X I (o) was completed. Box IV, with the "argument" for Canto 12, was transferred to A2 for use on page 2Y3.

No reason appears for shifting the Burbon incident to Canto 11, unless it was to balance the canto lengths more nearly. Although Spenser seems to have made the necessary revision of the second introductory

stanza of Canto 12, the concurrent need to revise the "argument" was overlooked. The erroneous catchword may have been on the press and beyond the trouble of correcting when the shift was made. Minor revisions in the narrative of Canto 12 must have been attended to, for Sir Sergis is meant in the tag "that old knight" (5.12.6.1) and is once explicitly named (5.12.10.7).

All the significant irregularities in the printing of the 1596 Faerie Queene have now been accounted for. Each volume was produced by a pair of compositors, whose hands can be differentiated except in the first two gatherings of the First Part. The two pairs of compositors were not identical and may have been four individuals. The First Part may well have been printed in Spenser' absence, but the Second reveals a clear instance of his intervention during printing. These conclusions open the way for a more critical edition of the Faerie Queene than now exists.