University of Virginia Library

Compositor B, The Pavier Quartos, and Copy Spellings
by
William S. Kable

Printed books are the chief textual witnesses for the majority of Renaissance English literature. Lacking authorial manuscripts, our knowledge of the words from the author's pen comes to us through the medium of print. Modern bibliographical researches have only begun to explore fully the possibilities of reducing this distance between the author's pen and the extant printed text by investigating the process by which the author's words reached print. In this beginning, however, the printing house compositor has already emerged as a significant factor in the transformation of manuscript into printed text. Compositorial study concentrates on the identification and evaluation of the work of these men, and a comprehensive spelling analysis is one of the first necessities in any thoroughgoing compositor study.

I.

In June 1920, Thomas Satchell announced the discovery of two distinct spelling patterns which indicated the presence of two identifiable compositors involved in the setting of type for Macbeth in the First Folio Shakespeare. Satchell named these compositors A and B and outlined their significant spelling patterns based on his researches in Macbeth. Compositor A was recognized principally by the spellings doe, goe, and here; Compositor B, by do, go, heere.[1]

E. E. Willoughby extended Satchell's identification of A and B to other parts of the Folio and suggested the hypothesis that "since in the portion we have investigated there are many passages that are not characteristic of either of them [A and B], it seems probable that there was also another pair of compositors at work."[2]

Alice Walker attempted a start at assessing the quality of the work of A and B with special reference to I Henry IV, but little real progress was made at refining compositorial study in the Folio until Charlton Hinman undertook his exhaustive study of the many copies of the Folio at the Folger Shakespeare Library.[3] Hinman's use of type and case identification added a new dimension to compositor study in the Folio. In 1957 he was able to announce the discovery and isolation of a fifth, apprentice, Compositor E, whose work had clouded the picture in the Tragedies.[4] Until Hinman was able to demonstrate the presence of this fifth compositor on the basis of indisputable physical evidence, the similarity of E's most important spelling habits to those of B had caused investigators to assign E's work to B. The coincidence that do, go, and heere were both B's and E's preferred spellings of these words delayed the separation of their work until Hinman's new approach was applied to the problem.

Publication of Hinman's completed study of the printing of the Folio provides the first step toward a complete investigation of all five Jaggard compositors involved in setting type for the Folio.[5] His findings in respect to the three key words for compositor identification can be summarized as follows:

• Compositor A — doe, goe, here
• Compositor B — do, go, heere
• Compositor C — doe, goe, heere
• Compositor D — doe, goe (with tolerance for do and go found in copy), here
• Compositor E — do, go, heere (with early tolerance for copy spellings)

With the exception of E, who joined Jaggard's staff only in the later stages of work on the Folio, Compositors A, B, C and D make up a four-man staff which could be expected to carry out the composition in Jaggard's shop under normal conditions. Much work still remains to be done to produce an exact basis for identifying C and D in the Folio, and qualitative evaluation of all five compositors is needed. The above listing of do, go, here habits does, however, give the basic means of identifying those workmen who can be hypothesized to have made up Jaggard's compositorial staff during the setting of most of the Folio, and that listing can be used as a provisional limit to Jaggard's compositorial staff.[6]

The ten Shakespearean and pseudo-Shakespearean plays which make up the group called the Pavier quartos were stumbling blocks to bibliographical and textual researchers before the discovery and demonstration in 1908-1910 that they are all simple page-for-page reprints produced in Isaac Jaggard's shop for Thomas Pavier in 1619.[7] Several of these reprints had even been incorrectly identified as the true first editions of their texts on the basis of substantive variants. The following table lists the Pavier titles and the copy from which they were set. The printers' names are appended to the list of copy to show the diversity of copy which served for the Paviers.[8]

                     

Pavier Texts	Greg #	Copy	Printed by:
2 Henry VI	119(c)	Q1(1594)	Thomas Creede
3 Henry VI	138(c)	O1(1595)	Peter Short
Pericles	284(d)	Q3(1611)	Simon Stafford
A Yorkshire Tragedy	272(b)	Q1(1608)	Richard Bradock
The Merchant of Venice	172(b)	Q1(1600)	James Roberts
Merry Wives of Windsor	187(b)	Q1(1602)	Thomas Creede
King Lear	265(b)	Q1(1608)	Nicholas Okes
Henry V	165(c)	Q1(1600)	Thomas Creede
Sir John Oldcastle	166(b)	Q1(1600)	Valentine Simmes
Midsummer Night's Dream	170(b)	Q1(1600)	Richard Bradock(?)

After the initial argumentative flurry over the identification of the true nature of the Pavier quartos, Shakespearean bibliographical and textual study has in general ignored them or, at best, hastily dismissed them as simple reprints. It is true that there is little chance that the Paviers will produce any startlingly new substantive witness to the Shakespeare text, but since these ten reprints are all products of Jaggard's printing house just a few years before the production of the all-important Folio in that very shop, they offer an excellent opportunity

to see Jaggard's workmen dealing with dramatic texts. This opportunity is made all the more significant by the fact that all ten plays are reprints from earlier editions which are extant. The exact copy and the resulting Jaggard text can be compared side by side.

D. F. McKenzie pointed the way to the significance of the Pavier quartos in a preliminary study of "Compositor B's Role in The Merchant of Venice Q2 (1619)."[9] Although his spelling tests were based on the word list of Alice Walker, who in turn compiled her list before Compositor E had been distinguished from B, McKenzie's statistics capably demonstrated that Jaggard's Compositor B as recognized in the Folio had in fact set the entire Pavier quarto of The Merchant of Venice. It must be remembered that E, the only compositor who could easily be confused with B, did not enter the shop until 1622. McKenzie was then able to observe exactly what B did with his copy, Merchant Q1 (1600). His study was divided between the few substantive alterations which B gratuitously introduced into the text and a brief discussion of B's "alteration of accidentals."

McKenzie did not, however, suggest another important use which can be made of the Pavier reprints and the controls which exist for the study of these texts. In the past, compositorial identification has concentrated on the peculiarities of the workman, on the individual mark which he leaves on the copy he sets. Jaggard's Compositor B, for example, is known primarily for his strong spelling habits, do, go, and heere. Texts set by B are noted for the virtual invariability of these spellings. There is another aspect, however, to compositor study which has been neglected in the concentration on the features of a workman's habits which are his badge and serve to identify him from his fellows. After the bibliographer has determined with assurance that a given compositor did in fact set a given portion of text, the value of the identifying spellings is exhausted. A do, go, or heere in a Folio text may be of use in assigning that text to Compositor B, but its value ends there.

In many instances, it would be useful to be able to penetrate the layer of compositorial spellings and go behind the workman to the features of his copy. The study of a new class of words in relation to individual compositors may enable the textual investigator to do just that. These words are those for which the compositor's spelling treatment is recognizable but not so pronounced as to be called an invariable habit. For example, a do in a B text can represent any form of the

word (do, doe, doo) in his copy. B's strong habit virtually blocks out any trace of the copy spellings for this word. If, however, a new group of words can be discovered for which the compositor does not have an invariable habit, but rather a weak preference or even indifference, these words can then be used, not as identifying marks showing the presence of the compositor, but as a means of seeing through that compositor to his copy. If the compositor has no set preferential spelling for a given word but varies his spellings in some relation to the variations in his copy, the forms of that word in his texts will directly reflect the forms which were in the copy. If the compositor does have some preferential spelling for a given word but is influenced by his copy to set another form from time to time, these occasional occurrences of the non-preferential form(s) will again reveal copy spellings. Once a group of such words can be established for a given compositor, certain projections of expected copy spellings can then be made using the spellings in this new group. Such a projection could be of the greatest possible utility in situations where the copy is unknown, lost, or disputed.

The Pavier quartos provide just the controlled situation necessary to make it possible to discover whether this hypothetical group of words does in fact exist within a compositor's total spelling pattern and to establish a broader picture of the nature of compositorial spelling habits, including weak preferences and the cases of relative indifference. First, however, it is necessary to identify the compositor(s) at work in the Paviers.

The following tables represent the results of a count of the do, go, here spellings throughout the Paviers.

                 

	2H6	3H6	PER	YT	MV	MW	KL	H5	SJO	MND	Total
Do	45	47	76	15	101	55	98	48	89	102	676
Doe	2	0	11	1	12	1	4	0	1	4	36
	2H6	3H6	PER	YT	MV	MW	KL	H5	SJO	MND	Total
Go	56	16	10	3	41	29	37	20	38	21	271
Goe	5	3	11	5	16	23	21	1	3	17	105
	2H6	3H6	PER	YT	MV	MW	KL	H5	SJO	MND	Total
Heere	57	38	48	11	62	29	38	16	58	32	389
Here	22	11	21	3	15	24	33	8	25	29	191

Although the strength of do and the more or less strong preferences for go and heere seem to point to Compositor B, the real value of these tables lies in demonstrating the dangerous invalidity of overly simplified

statistical compilations of spelling evidence. Two important factors have been completely omitted from this set of tables, factors without which no really significant spelling analysis can be made of the Pavier quartos.

First, there is no recognition of the copy spellings which lie behind these statistics. For example, the evidence for do (101 occurrences) as opposed to doe (twelve occurrences) in The Merchant of Venice seems at first to conflict with Compositor B's almost invariable do habit. A closer look at this group of spellings with the additional evidence of the spellings of the copy text, however, makes an extremely strong case in favor of Compositor B's having set The Merchant. Of the twelve doe spellings in the Pavier, all twelve reproduce B's non-habitual doe as found in copy, Q1 (1600). Of the 101 do spellings in the Pavier, only five were found in copy and ninety-six represent gratuitous changes of doe to do on the part of the compositor. To state these facts yet another way, the compositor of Pavier Merchant found the word do spelled doe 108 times and do five times in his copy. He reproduced all five do spellings and changed ninety-six of the 108 doe to do leaving only twelve doe forms found in his copy. This additional evidence of the copy spellings overwhelmingly shows the strength of the do spelling in the habits of the compositor who set the Pavier Merchant even though the raw statistics point to a partial use of the doe spelling.

This distinction between the raw numerical totals in the first case and the weighted strength provided by the evidence of copy spellings is essentially similar to that between quantitative and qualitative evidence.[10] The quantities of spellings, do (101) and doe (twelve), did in fact point in the general direction of B's habit, but the quality of the evidence only becomes apparent after a review of the copy spellings reveals that ninety-six changes of doe to do are hidden in the raw statistics. Given the relative value of any one change over any one instance of following copy, the quality of the evidence makes the case for B one of certainty. Failure to consider the copy spellings greatly weakens and obscures the genuine evidence. From this example, an important principle governing the study of compositorial habits can be formulated. In compositorial analysis of texts for which the copy is known and available, all consideration of variant spellings must include an examination of the copy spellings.

The second factor which was omitted from the first group of tables is that of the influence of justification on a compositor's spellings. It

has long been recognized that, as McKerrow said, early compositors "had . . . a means of justifying the lines of type which is denied to modern compositors, namely, by varying the spelling of words."[11] If this is true, and one of the overall results of the examination of the Pavier spellings is to demonstrate graphically the truth of McKerrow's axiom, evidence of spellings in justified lines must be carefully separated from the evidence in short lines.[12]

In any given line, the possible influence of justification on the spellings is a matter for debate. The possibility of such influence must, however, cause one to segregate the spelling evidence which could be contaminated by the need for justification. Take, for example, the statistics for go/goe in King Lear. The simple counts record thirty-seven go and twenty-one goe spellings. When the possible effects of justification are taken into consideration, these statistics become go thirty-one times plus six justified occurrences and goe seven times plus fourteen justified occurrences. The original ratio of thirty-seven go to twenty-one goe spellings is rather questionable evidence for B's go as a preferential spelling. When the justified and, therefore, possibly deceptive spellings are taken away, the ratio of thirty-one go to seven goe becomes much more convincing. When this allowance for justification is combined with a consideration of the copy spellings, the total strength of the evidence emerges. Go was found ten times in Lear Q1 (1608), seven times in short lines and three times in justified lines. All ten of these go spellings are retained in the Pavier Lear. On the other hand, Lear Q1 contained thirty-one goe spellings in short lines and seventeen in justified lines. Of the thirty-one long spellings, twenty-four were changed to go. Apparently, therefore, the factor of justification caused the compositor's preference for the short go spelling to be obscured in the justified occurrences of this word. A second principle to govern spelling analysis is that the possible influence of justification on variant spellings must be acknowledged, and spellings in long lines must be separated from other spellings in any statistical compilation.

The following set of tables represents a restatement of the evidence for do, go, and here in the light of these two principles. Note that the words are recorded only in relation to copy spellings and that evidence in justified lines is separated from the general statistics. In these tables, J stands for justified, or at least long line, occurrences.

                                                           

Copy→
1619	2H6	3H6	PER	YT	MV	MW	KL	H5	SJO	MND	Totals
Do→Do	41	12	15	3	3	45	22	48	57	24	270
	J4		J6	J4	J2	J10	J5		J15	J5	J51
Do→	2								1		3
Doe			J2	J1		J1	J1				J5
Doe→		35	34	6	87		59		13	69	303
Do			J21	J2	J9		J12		J4	J4	J52
Doe→			2		8		1			1	12
Doe			J7		J4		J2			J3	J16
Copy→
1619	2H6	3H6	PER	YT	MV	MW	KL	H5	SJO	MND	Totals
Go→	49	13		3		15	7	18	20	4	129
Go	J3		J1			J4	J3		J10		J21
Go→	2	1			1	1					5
Goe			J1			J3			J2		J6
Goe→	4	3	5		34	9	24	2	5	16	102
Go			J4		J17	J1	J3		J3	J1	J19
Goe→	3	2	4	5	13	16	7	1		17	68
Goe			J6		J2	J3	J14		J1		J26
Copy→
1619	2H6	3H6	PER	YT	MV	MW	KL	H5	SJO	MND	Totals
Heere→	2	19	23	7	51	5	7	1	8	11	134
Heere			J7	J2	J5	J2			J2		J18
Heere→		3	2		6		1			1	13
Here		J1	J1	J1	J3	J2			J1		J9
Here→	51	19	14	1	6	19	25	15	42	19	211
Heere	J4		J4	J1		J3	J6		J6	J2	J26
Here→	19	7	16	2	6	21	18	8	17	23	137
Here	J3		J2			J1	J14		J7	J5	J32

Only in these complete tables which take the copy spellings and justification into consideration does the overwhelming evidence for Compositor B as the man who set type for the entire set of the Paviers become apparent. The already small total of thirty-six spellings in the first, overly simplified, tables becomes even more insignificant when it

can be observed that of these thirty-six occurrences of doe, twenty-eight (and of that twenty-eight, sixteen justified) reproduce copy spellings. In all ten plays, there are only three non-copy, non-justified doe spellings as opposed to 303 such do forms.

The go/goe evidence was not at all clear-cut in the first table but here is shown to be significantly in favor of Compositor B's go. Of the 105 goe spellings in the Paviers, ninety-four (and of that ninety-four, twenty-six justified) reproduce copy spellings. There are only five non-justified, non-copy goe spellings while there are 101 similar go forms.

The here/heere ratio was also not clear-cut in the simplified table. With the additional factors considered, however, 169 of the 191 here spellings turn out to derive from the copy used for the Paviers. Of that 169, thirty-two were in justified lines. The thirteen non-copy, non-justified here spellings must be compared with 211 similar heere forms.

The total number of B spellings gratuitously introduced into the Paviers is 615. In view of the token number of twenty-one changes in the opposite direction, the presence of B's hand throughout the Paviers is evident. It is also important to note that the few aberrant non-B changes are not significantly grouped, and in no instance do they suggest the presence of another compositor.

The examination of the do, go, here spellings in the Pavier quartos shows the special significance of change, that is, the altering of copy in contrast to the following of copy. The spelling pattern of go/goe in A Midsummer Night's Dream is an excellent example of the relative importance of spelling change. The raw evidence in the Paviers is almost evenly divided between twenty go and seventeen goe. The direction of the changes from copy which produced this result tells a far different story. Of the four go spellings in copy, there is not one instance of change to goe. Of the thirty-three goe spellings in copy, however, almost half, sixteen, were changed to go. On the surface, the sixteen changes in thirty-three occurrences of goe might appear to express a certain degree of compositorial indifference to the spelling of this word. When viewed as a part of the overall pattern of B's spellings and in relation to the four go forms, all of which remain unchanged, this change of sixteen out of thirty-three goe to go is far from expressing indifference. Rather, it indicates as strongly as possible the compositor's preference for the go form. A single change is of much more significance in assessing a compositor's preferences than is a single case of his following copy. The sixteen changes of goe to go tell a great deal more about the compositor's preference than do the seventeen instances of following copy.

This examination of do, go, here spellings in the Paviers has accomplished two things. First, it has shown the care which must be employed in avoiding over-simplification in the expression of statistical spelling evidence. Spelling analysis must be done in conjunction with an examination, where possible, of copy spellings, and the possible influence of justification on variant spellings must be considered at all times. Secondly, the introduction of these refinements into the analysis of do, go, here spellings has brought convincing strength to the demonstration of the hypothesis that Jaggard's Compositor B did in fact set the whole of the text of the Pavier quartos.

II.

Armed with this warning against over-simplification and, at the same time, strengthened by the assurance that B's hand is in evidence throughout the Paviers, I collated all of the Pavier texts against their copy and compiled word counts for over 500 individual words. This mass of statistical evidence was then analyzed in terms of orthographic groups which display significant spelling features. Although I do not wish to reproduce these tedious reports of observed statistics, I propose to give here a list of the groups of words I examined and one sample report in order to make the nature of my study known to those who may be interested in consulting it.[13]

The groups of spellings considered were as follows:

• 1. Internal -A-/-AU-
• 2. Initial DE-/DI-
• 3. Internal Consonants: -D-/-DD-, -R-/-RR-, -T-/-TT-
• 4. Final -D/-DE
• 5. Initial E-/I- before -N-/-M-
• 6. Internal -EA-/-EE-
• 7. Internal -EA-/-EI-
• 8. Internal -EE-/-IE-
• 9. Final -E/-EE
• 10. Final -F/-FE
• 11. Final -G
• 12. Internal -I-/-Y-
• 13. Final -IE/-Y
• 14. Final -K/-CK/-QUE
• 15. Final -L
• 16. Final -M
• 17. Final -N
• 18. Internal -O-/-OO-
• 19. Final -P
• 20. Final -R
• 21. Final -S/-SSE
• 22. Final -T
• 23. Final -W
• 24. Final -X
• 25. Reflexive Pronouns

One of the largest of these groups of words is that involving -ie/-y endings. Before one can attempt to generalize about the eighty-eight

words in this group for which there is sufficient evidence to permit valid statistical conclusions, they must be examined in sub-groups. First, those words ending in a vowel plus -ie or -y will be considered. After that, the words ending in a consonant plus -ie/-y will be examined in two groups defined by penultimate consonants.

All of the following words display B's extremely strong preference for final -y in conjunction with a preceding vowel:

         

away	enjoy	pay
betray	joy	pray
bewray	lay	say
boy	may	stay
day	nay	way

Without exception, B changes -ie and -ye forms of these words to -y and retains all of the -y forms found in copy.

The following words ending in -ie/-y after penultimate letters b, c, d, g, k, l, m, n, p, r, t, and v regularly display B's preference for -y endings:

                                                     

b: rugby	willingly	t: authority
c: fancy	m: enemy	beauty
mercy	n: any	charity
d: already	company	city
bloody	deny	cruelty
body	honey	duty
lady	many	fifty
needy	money	forty
ready	villany	guilty
speedy	p: happy	honesty
g: clergy	r: beggery	liberty
k: lucky	carry	loyalty
l: daily	contrary	majesty
dearly	country	mighty
early	cry	nativity
fly	dowry	necessity
heartily	every	petty
heavenly	fury	pity
holy	glory	pretty
kingly	history	quality
only	marry	safety
presently	merry	thirty
princely	misery	twenty
quickly	sorry	university
reply	tarry	v: envy
suddenly	very	heavy
truly	victory

Preferring final -y for all of the words on this list, B regularly changes the bulk of the -ie forms to -y and almost invariably reproduces the -y forms found in copy. A small and relatively regular percentage of -ie spellings slip through from copy into B's texts, but these carry-overs always occur in conjunction with B's definite preference for -y.

There are two consonants which do not fall into the above group and, in fact, display a distinctly contrasting pattern. These consonants are f and s. Defie stands in B's copy six times and all of these six occurrences are reproduced by B. B's copy spells satisfy with final -ie all five times it occurs. B follows copy for all of these forms. Signifie is so spelled in six (+J1) occurrences in copy. All of these forms are reproduced by B. All four (+J1) occurrences of busy are spelled busie both in copy and in B's texts. B reproduces the seven (+J1) curtesie spellings in copy, changes three (+J1) curtesie forms to courtesie, and reproduces three courtesie spellings. All seven easie spellings in copy are retained in B's texts. Heresie is reproduced from copy the three times the word occurs. One heresie form is set in place of copy's herisie. The seven occurrences of jealousy are spelled with final -ie both in copy and in B's texts. Lowsie is reproduced from copy three times and once changed to lowsy.

All of the words recorded for f and s plus -ie/-y display an uncanny pattern of strongly favoring final -ie in both copy and B's texts. It is likely that the anomalous -fie/-sie endings result from the use of the fi and long-s/i ligatures. Since all of the Pavier copy produced by various printers agrees with B's use of -fie/-sie in spite of the general -ie/-y treatment, the fi and si ligatures were apparently used with some consistency in order to make the supply of plain f's and s's in the case last as long as possible. The unanimity of the printers in following this procedure makes it appear to be a general convention for stretching the supply of type.

B displays, therefore, a special tolerance for final -ie after -f- and -s-. It is important to stress the word tolerance because this special feature of B's treatment of final -ie/-y cannot be termed a preference. Without being able to observe B's reaction to -fy and -sy in copy, we cannot equate his use of -fie/-sie with his demonstrable general preference for final -y as evidenced by the majority of the evidence given above. Indeed, the only change in that last group of -f- and -s- is that of one of the four lowsie forms to lowsy. Taken by itself, this shred of evidence points to the fact that B's preferential spelling was final -y but that his tolerance for ligatures with -ie was surprisingly strong. It must also be pointed out that B's tolerance for -fie and -sie is not simply a tendency to give in to copy spellings when they are weighted against his preferences

as were the -fie and -sie copy spellings against his general preference for -y. In the group which displays B's preference for final -y, many individual words had large proportions of -ie endings in copy but were consistently set by B with final -y. Happy was found in copy seventeen (+J1) to five in favor of -ie but becomes twenty-two (+J1) happy in B's texts. Victory, guilty, nativity, and thirty were all uniformly spelled with final -ie in copy but with final -y in B's texts.

To summarize B's spelling patterns for words ending in -ie/-y, B consistently prefers final -y after preceding vowels and consonants with the exception of f and s. Although his preference for final -y involves both the retention of almost all -y endings and the introduction of -y endings for the vast majority of -ie forms in his copy, a limited percentage of -ie forms is carried over from copy into B's texts. These few -ie forms serve as an indication of copy spellings. For the only exceptions to this general rule, -fie and -sie, B displays an almost absolute tolerance for the -ie endings which are consistently found in copy.

An examination of compositor B's group spellings has revealed both the dangers of and benefits to be derived from the use of analogy in the analysis of compositorial spellings. Although compositor identification in the Folio has generally been limited to the use of variant spellings of individual words, other compositor studies have employed classes of words such as that formed by all words with -ie/-y endings. Since investigators have employed these word classes in compositor identification, it is of interest to examine the nature and extent of analogy in the overall spelling pattern of a compositor.

There are several large groups of words which B treats in a unified manner. His consistent use of final -e after penultimate -f- is of little significance because the vast majority of copy spellings are also -fe. B's consistent preference for -esse forms of words varying between -es and -esse is an individual habit which extends to all of the similar words recorded in the Paviers. From the evidence of the Pavier spellings, a reasonably safe projection can be made from the uniformity of B's preference for -esse. It is, however, not demonstrable that the very next word examined will not for some special reason depart from the pattern and display the opposite tendency. The discovery of B's clearly defined preference for deere might lead one to expect that B would also prefer the yeere spelling. In fact, however, B's preference for yeare is equally strong as his use of deere. Word classes cannot, therefore, be formed by the use of simple analogy.

Although word classes cannot be automatically formed of analogical words, Compositor B's spellings are far from being patternless. Consistent patterns can in fact be observed for groups of similar words.

The key to the problem is that the word groups can be formed only after an examination of the individual words. For example, the real pattern of B's treatment of final -ie/-y becomes clear only after considering the individual words. Since the raw statistics indicate that B obviously prefers final -y for all but a handful of words, one's first reaction is to consider B's pattern a general preference for final-y with the tendency to accept a few aberrant -ie endings found in copy. These words might, therefore, be designated exceptions to the general rule and nothing more. When the individual words are analyzed, however, the so-called "exceptions" to the one rule are seen to behave according to a strict rule of their own. All of the words which display B's tolerance for final -ie have penultimate -f- or -s- and are probably governed by the fi and si ligatures. The exceptions to the hypothetical blanket rule are in fact governed by a separate, clearly defined, rule of their own. Only by reasoning from an examination of the individual words is it possible to define accurately the limits of B's pattern.

There are many similar situations which show the importance of building up word classes of a more limited sort. Internal -ai-/-ay- spellings appear to be mixed until it is observed that B's spellings split into an -ayle group and a more general -ai- group. In this case, the subgroup displaying -ay- is formed by words having analogous endings. The other group is not governed by any common feature other than the internal -ai-. Internal -r-/-rr- variation, however, falls into two groups, both of which are composed of analogous words.

In summary, therefore, Compositor B's spelling patterns indicate that analogy often governed his treatment of various spellings. At the same time, however, untested analogy cannot be employed to define linguistic units for compositor study. The exact nature and extent of these analogous groups must be determined only as a result of the study of separate words.

Although analogy must be employed with extreme caution, one important factor, aberration, should not be permitted to interfere with the careful formation of word groups and classes. The Pavier statistics confirm a widely held hypothesis that for some unknown reason a given compositor will on rare occasions contradict even his strongest habit. The point here is that these aberrations should not be allowed to obscure the valid evidence of spelling patterns.

In the examination of do, go, and here, for example, B was seen to go against his strong do habit and set doe three times in the course of the Paviers in instances when he was affected neither by the need for justification nor by a copy doe spelling. These three aberrations in

the course of ten plays and over 600 occurrences of the word must remain unexplained and be written off as human variability.

To take another example, a word first analyzed in the Paviers, the following statistics represent the word count for only. The first column represents copy spellings and the second the spelling of those words in the Paviers.

• ONLIE — ONELY 1
• ONLY — ONELY 12
• ONELY — ONELY 33+J2
• ONELY — ONLY 1

B's unique aberration in setting only for onely does not have any force against the formulation of the general rule that B's preferential spelling is clearly onely.

Although the existence of these aberrations should not be allowed to obscure clearly observable general patterns, the fact that single aberrant spellings occur calls for caution in placing weight on any one individual spelling. To paraphrase Fredson Bowers commenting on aberration in the printing process, the phenomenon of an individual spelling is narrow enough "to introduce the possibility that a recognized . . . [spelling preference] is not being revealed but instead that we are examining only an aberration — some human variation or failure that must be regarded as producing a sport."[14]

III.

The examination of the Pavier spellings has also produced valuable statistics on many individual words of interest. Spellings which have been proposed by previous investigators as characteristic B spellings are here reviewed in the light of the information provided by the statistics covering B's work in the Paviers. All of the words that have been proposed by Satchell, Willoughby, Cauthen,[15] and Walker[16] as aids in identifying B's work have been observed in the Paviers. None of these investigators considered B's habit spellings in direct relation to copy, and all four of them were working before Hinman's isolation of Compositor E. When reference is make to these spelling analyses,

therefore, it must be remembered that E spellings can have contaminated statistics for B. Hinman has commented briefly on the relation of spelling habits to copy.[17] Several other words are included in this section although they are not identifying features for the recognition of B's presence in a given text. These words have been chosen for their special value in indicating other features of B's total spelling pattern and new ways in which spellings can reveal evidence about the compositor's treatment of his copy.

1. been

As a part of his attempt to demonstrate that all of Folio Lear was set by B, Cauthen presented the following table as evidence that B's preferential spelling was bin.

       

A Spellings	B Spellings	Lear Spellings
beene (13)	beene (4)	beene (8)
bene (3)	bene (6)	bene (1)
bin (3)	bin (24)	bin (19)

When viewed in relation to copy, however, B's treatment of been cannot be defined in terms of a single preferential form.

These statistics show that B had no single preferential spelling of been. In fact, he actively employed, that is followed from copy and introduced against copy, three forms: bin, bene, and beene. Although B does not have a single positive preference, he consistently rejects the been and byn spellings which stand in copy. In the case of each of his acceptable spellings, B adopts about half the number of any given form found in copy. Although B's tolerance of bin forms found in copy is as high as that toward bene and beene, he introduces bin least frequently in replacing other forms. Between bene and beene, B apparently prefers bene, but the evidence is not strong. All in all, been is of little use in penetrating B's texts to his copy. The word serves, however, to show the possibility of a compositor's having several operative forms in his treatment of a given word, that is, various forms which he will both accept, reject, and introduce in place of others found in copy.

2. blood (bloody)

Compositor B displays a marked preference for -oo- forms of blood and bloody. Of the 103 (+J4) -ou- forms in copy, B changes sixty-three (+J4) to -oo-. At the same time, he retains sixty-one (+J5) of the sixty-six (+J5) -oo- spellings found in copy. It is interesting to note that in spite of his clear-cut preference for -oo-, Compositor B does allow four -ou- spellings in ten to come through into his texts. In all ten plays, there are only five abberant changes of -oo- to -ou- or about one in thirteen occurrences of the word.

3. brief

Cauthen's statistics suggest that B has a two to one preference for briefe over breefe. In the Paviers, only one justified -ee- form occurs in copy and this is followed by B. Of the twenty-one (+J3) -ie- forms in copy, B changes five to -ee- and retains sixteen (+J3) as -ie-. From this evidence, it would appear that any preference which these statistics point toward is for -ee- over -ie-. To complete the picture, it would be necessary to be able to observe B's treatment of several -ee- forms when found in copy.

4. choose

Alice Walker has suggested that B displays a preference for choose in contrast to A's chuse. In the limited evidence provided by the Paviers, B changes three of the five chuse forms in copy to choose. Of the three choose forms in copy, B retains two but sets one as chuse. Walker's suggestion that B favored choose is confirmed by the evidence in the Paviers.

5. cousin

Taking the -o-/-ou- variation as the operative feature, the evidence provided by the Paviers fails to confirm Willoughby's and Walker's suggestion that B preferred -o- forms over -ou- spellings. Of the nine (+J1) -ou- spellings in copy, B retains eight and changes only one (+J1). Of the eleven (+J3) -o- spellings in copy, B retains ten (+J3) and changes only one.

6. dear

The variation of -ea-/-ee- is the operative feature in dear. Of the sixty (+J12) -ea- forms found in copy, B changes twenty-three (+J6) to -ee-. He retains all thirty-seven (+J7) -ee- forms in copy. These statistics point clearly to his preference for deere.

7. devil

Compositor B expresses a strong preference for the di- form of devil. Of the eighteen (+J7) de- forms in copy, B changes seventeen (+J6) to di-. Only one (+J1) non-preferential de- spelling slips through from copy into B's text. All twenty-seven (+J9) di- forms in copy are retained by B.

8. forfeit

Forfeit provides another example, like been, of a situation where multiple alternative spellings [forfait(e), forfet, and forfeit(e)] were available to B. In this instance, he displays a double preference for forfet and forfeit(e) and completely rejects the third alternative. At the same time, he fails to display any single clear-cut preference for either of the two forms which he does use.

9. grant

Walker's suggestion that B preferred the grant form is strongly confirmed by the Pavier spellings. He changes every one of the eight (+J1) copy graunt spellings to grant and retains all twelve (+J2) copy grant forms.

10. grief

Walker included greefe in her list of B's preferred spellings. Of the thirty-eight (+J5) griefe spellings in copy, B changes seventeen (+J1) to greefe. Both greefe forms in copy are retained. The greefe preference is confirmed.

11. heart

B displays a marked preference for the -ea- form of heart. Of the ninety-two (+J13) hart forms in copy, B changes seventy-three (+J6) to -ea-. He retains 111 (+J21) of the 112 (+J24) -ea- forms found in copy. There is only one (+J3) anomalous change of -ea- to -a-. B's -ea- preference is clearly indicated.

12. heaven

Alice Walker observed that B preferred uncapitalized heaven up until Macbeth in the Folio. In the Paviers, B changes eight of the nine capitalized Heaven forms to lower case and reproduces thirty-six (+J2) uncapitalized forms found in copy.

13. honey

Alice Walker first observed B's preference for the hony spelling of this word. He changes all three honey spellings in copy to hony and reproduces the four short forms which stand in his copy. There is one anomalous change of hony to honny.

14. hour

Cauthen and Walker both indicate that B's preferred spelling was houre. Of the thirteen (+J2) -ow- spellings of hour in copy, B changes eleven (+J2) to -ou-. One hower and one houre-glasse spellings are reproduced by B from his copy. He retains all thirty-four (+J4) -ou- forms found in copy. The Pavier evidence indicates that B's preference for -ou- was quite strong.

15. "lushious"

The one occurrence of lushious in both copy and Pavier texts is included in this discussion because Alice Walker put great weight upon the coincidence of the lushious spelling in Q1 and Folio Othello. She used this coincidence as evidence that Folio Othello was derived from a marked copy of Q1 (1622). "It is ridiculous to suppose that anomalies like 'timerous' (I.i.76), 'lushious' (I.iii.346) and 'pudled' (III.iv.144) passed from a fair copy, via a prompt book, to a manuscript prepared by a book-keeper with (at best) only half an eye on the prompt-book, and so into Okes's quarto, thereby representing a common legacy in the quarto and Folio from some common ancestor."[18] If Miss Walker had checked the Concordance, she would have found two recorded uses of luscious, in the Othello passage in question (I.iii.354 — — Globe numbering) and in A Midsummer Night's Dream (II.i.251). Both Midsummer Q1 (1600) and the Pavier Q2 (1619) contain the lushious spelling (C2v in both Qq). Far from being anomalous, the lushious form is common to the compositors of Q1 Midsummer, Q1 Othello, and Jaggard's B. It is interesting to note that the Midsummer lushious is changed to luscious on N3v in the Folio. Hinman assigns this page to Compositor C.[19] The point of this example is that any common spelling, however "anomalous" it may appear, must be used with extreme caution in attempting to demonstrate direct derivation of one text from another.

16. madam

Alice Walker suggested that madam is a characteristic B spelling whereas madame is characteristic of Compositor A. Disregarding the

variation of -d-/-dd- and concentrating on -m/-me, the evidence of the Paviers is somewhat mixed. B's copy is slanted in favor of -m (fifty-two +J7) over -me (twenty-seven +J5). Of the twenty-seven (+J5) -me forms, B changes eleven (+J2) to -m and retains sixteen (+J3). On the other hand, a small proportion, seven (+J1) of fifty-two (+J7), -m copy spellings are changed in the opposite direction to -me. B's general preference is indeed for -m, but there are a number of aberrant changes of -m to -me.

17. mistress

Five variant spellings of mistress are found in the copy used for the Paviers: mistres (eight +J1), mistris (twenty-two +J10), mistresse (forty-seven +J11), mistrisse (one +J1), and misteris (thirty-five +J5). Of these five spellings, B prefers mistris and mistresse and rejects the other three. Although Alice Walker assigned mistresse to A and mistris to B as characteristic spellings, B's treatment of the word in the Paviers indicates that he employed both forms. On the surface, simple numerical superiority does favor the -is form, but this is a result of the numerical superiority of -is forms in copy.

18. power

Cauthen observed that B's texts contained both power (18) and powre (7).[20] In the Paviers, however, B changes the two powre forms in copy to power and retains all twenty-six of the power spellings which stand in his copy, thus indicating a preference for power.

19. prithee

Alice Walker noted B's constant use of the spelling prethee in the early sections of the Folio. In the Paviers, B normalized a total of twenty-six (+J6) occurrences of ten different spellings of prithee all to his preferential prethee. He follows one copy prithee, sets one justified copy prithe as prithee, and follows one copy prethe. B's habit in the Paviers coincides, therefore, with Walker's observation of his treatment of prithee in the early parts of the Folio.

20. proud

The Paviers confirm Walker's assignment of proud as B's preferential form. B changes the one copy prowd and one (+J1) proude to proud and retains twelve occurrences of the latter form from copy.

21. scarce

The evidence supplied by the Paviers confirms the observation that B had a strong preference for the -se form of scarce. He changes all eighteen (+J2) -ce forms in copy to -se and retains all five -se forms which stand in copy.

22. traitor

Willoughby and Walker both stress B's preference for traitor over traytor, and the Paviers confirm this preference. Of the twenty-five (+J4) traytor forms in copy, B changes twenty-five (+J1) to traitor. All thirty-five (+J3) -i- forms in copy are retained by B. The only one (+J1) traytor form in B's texts is derived from copy.

23. vile

Cauthen assigned the vile spelling to Compositor A and vilde to B. B changes all five vild spellings in copy to vilde and one copy vilde to vile. He reproduces the two (+J1) other vilde spellings found in copy.

24. wee'l (etc.)

Alice Walker suggested that B had a distinctive preference for wee'l (etc.) spellings as opposed to wee'le (etc.). The following discussion breaks these forms down by person.

The various compositors who set the texts used for Pavier copy and Compositor B display a surprising unanimity on the spelling of I'll. There are 124 (+J13) instances of ile in both copy and Pavier texts. One aberrant yle form found in copy is changed to ile. B's one deviation from the standard form occurs in a justified line in which he sets I'le for Ile.

He'll, she'll, we'll, and you'll occur in sufficient numbers to demonstrate B's strong preferences for final 'l forms of these words. Several variant factors are involved in these words. The use of the apostrophe, the ending (-l or -le) and the doubling (in the case of he'll, she'll and we'll) of the first e are all separate alternatives operative for these words. B's uniform preferences are for the doubling of the initial e, the use of the apostrophe, and the omission of the final -e.

Of the eighteen (+J2) times B encounters heele in copy, he changes eighteen (+J1) of these forms to his preferential hee'l. In the one justified instance, B changes heele to heel. In the two instances in which he encountered hee'le both are changed to hee'l. The one contraction introduced by B changes he will to hee'l. One aberrant hele in B's copy comes through as an unusual heele in his text.

Of the five sheele forms in copy, B changes four to shee'l and sets one as shee'll. He changes all four copy shee'le spellings to his preferential

shee'l. The one copy shee'l is reproduced by B. Of the two sheel forms in copy, B changes one to shee'l and carries the other over to his text.

We'll is a more complex word. B displays a preference for wee'l but is subject to the influence of copy and more prone to deviate from his strict preference. B changes twenty-nine (+J3) weele forms to wee'l. At the same time, he changes two justified weele forms to wee'll, fourteen (+J2) weele forms to weel, three weele forms to weell, one justified weele form to weel and retains eight (+J3) weele forms found in copy. If the three wele spellings in copy, B changes one to his apparent preference wee'l but the other two to weele. The one we'le form is changed to his preferred wee'l as are the eight (+J1) occurrences of wee'le. The two wee'l forms are reproduced from copy. Of the two (+J1) weel spellings in copy, one (+J1) is changed to B's preferential wee'l and one is retained. Both the one weell and the one we wil which are contracted by B become weel.

Compositor B displays a strong preference for you'l by changing twenty (+J3) youle spellings to this form. Only twice does he reproduce youle from copy and in one unusual instance sets youle in place of his preferred you'l. One other you'l in copy is retained, and the two you'le forms are shortened to you'l.

With the exception of the Ile which is uniformly spelled thusly, Compositor B displays a consistent spelling preference in favor of hee'l, wee'l, and you'l. The only variation comes in his tendency to depart occasionally from his strict habit with respect to wee'l.

One additional word can be considered with this group. B's treatment of contracted he is is as follows. Twice he sets hee's for copy he is. Of the eleven (+J4) hees in copy, B sets six (+J4) as hee's and five as he's. The one he's in copy is changed to hee's. Of the seventeen (+J3) hee's spellings in copy, B follows copy seventeen (+J2) times and changes one hee's to hee is in a justified line where these words stand at the very end of the line. B's preference is clearly for the hee's spelling.

25. widow

The Pavier evidence confirms Alice Walker's suggestion that B's preference is for widdow. Of the seven (+J1) -d- forms in copy, B changes three to -dd-. He retains eight (+J2) of the ten (+J2) -dd- forms found in copy. In addition, B drops the final -e found on one form in copy.

26. year

B's preference is clearly for the yeare spelling. He changes all seven (+J3) -ee- spellings in copy to -ea- and preserves twenty (+J4) of the twenty-one (+J4) -ea- forms found in copy.

27. young

Alice Walker assigns the yong spellings to B and young to A. The evidence in the Paviers shows that B's treatment of young does not display a marked preference for either form. Of the thirty-six (+J18) -ou- spellings in copy, B changes eight (+J6) to -o-. On the other hand, he changes eleven (+J6) of the twenty-eight (+J8) -o- spellings to -ou-. The number of changes of -o- to -ou- has a slight numerical superiority over -ou- to -o-, but no clear-cut habit can be observed.

It is worth noting the extent to which previous investigators dealing with Compositor B in the Folio and working without the ability to distinguish E from B as well as without reference to copy have determined B's preferential spellings as they are revealed in the Paviers. Four of the five test words proposed by Willoughby as means of identifying B are in fact strong B preferences.[21] Twenty-one of the forms proposed by Alice Walker as characteristic B spellings have been studied in the Paviers, and seventeen of the twenty-one are strong preferential spellings there. Two conclusions can be drawn from these results. First, the limited amount of text set by E and the closeness of his overall spelling pattern to that of B have not obscured B's spellings from previous investigators. Second, the fact that seventeen of B's Folio habits are confirmed in the Paviers indicates the extent to which set habits remained fixed. Although it will be necessary for later investigations to analyze B's Folio spellings fully, the preliminary indications are that the Paviers give an excellent idea of B's spellings. This is not to say, however, that any one individual B spelling as observed in the Paviers can be assumed to be his fixed preferential spelling. The general texture of B's spellings in the Paviers, however, should be reflected in his work in the Folio.

It is also important to note that the various words here discussed display all of the various logical possibilities for a compositor's treatment of variant spellings. B is seen to display strong preferences for one, two, and even three variant spellings of given words and can display indifference which appears neither to be influenced by a weak preference nor to depend on copy forms.

IV.

My examination of the Pavier spellings was undertaken as a result of the hypothesis that an analysis of compositorial spellings can provide not merely the basis for accurately identifying a compositor's work by his preferential spellings, but also the means of seeing through the workman to his copy. In many situations, it would be extremely advantageous, once the compositor has been identified, to penetrate the layer of his spellings and go behind the man to the copy from which he was setting his type. Compositorial identification has concentrated on the discovery of forms like do, go, and heere which are signs indicating that the text in question is B's work. The hypothesis in question stated, however, that if a new group of spellings can be discovered which contains not only these compositorial habits and preferences, but also spellings which reproduce copy spellings as well, there will be a means of seeing through that compositor to his copy. The examination of the Pavier spellings has produced the evidence to verify this hypothesis and has further shown that even the identifying badges, strong preferences, can be used to get at copy spellings.

There are various types of words which can produce information about copy. The ideal word is one that has two or more variant spellings which occur in mixed groups and are mechanically reproduced by the compositor in question. Such words which then directly depict the copy spelling at every occurrence turn out to be extremely rare. The only examples which are invariably reliable in the Paviers are art(20)/arte(2), brier(2)/bryer(2), marvell(4)/marveile(5+J1), and pallace(4)/pallas(1). All four of these words occur in the Paviers with limited frequency, but they serve to demonstrate that words can be found which will vary in direct relation to copy. Oh is the best example of a high frequency word which varies in extremely close relation to copy spelling. Of the 361 mixed O (244+J45) and Oh (177+J20) forms in B's copy, 341 reflect copy spellings. When one encounters an O or Oh spelling in B's Pavier texts, therefore, the odds are better than fifteen to one that B's spelling is the spelling which stood in copy. Words of this type which can be said to express compositorial indifference are unfortunately relatively scarce and are, therefore, only one contributing factor to the search for spellings which reflect copy. When they do occur, however, the evidential value is extremely strong.

At the other end of the scale are strong habits and preferences which, if viewed from a new perspective and not as identifying badges indicating a given compositor's presence, can contribute clues to copy spellings. Take, for example, do, the strongest of B's preferential spellings. In all of the Paviers, B violates his strong preference fifteen times

by setting unjustified doe spellings. Twelve of these fifteen spellings reflect copy spellings. When a doe form is encountered in B's work in contradiction to his known preference, therefore, the odds are four to one that the doe reflects a doe in copy. The same situation holds for goe and here. The odds are fifteen to one in the former case and about thirteen to one in the latter that these contradictions of B's preferred spellings had their origin in copy spellings.

Between the words expressing compositorial indifference and those which display a strong preference are all shades of a spectrum. Among these words are many which can be termed weak preference words. These words are those for which B displays some favorite spelling or spellings but neither a very strong habit nor indifference. Dear is an ideal example of a weak preference word. B clearly displays a preference for the deere spelling by following every one of the thirty-seven (+J7) deere spellings in copy and changing twenty-three (+J6) of the sixty (+J12) deare spellings in copy to deere. It is the thirty-seven (+J6) non-preferential deare spellings which B reproduces in his texts from his copy, however, which are of importance for copy study. B's preference was weak, and he allowed it to be subordinated to copy spellings in these numerous instances so that all of these deare spellings in his texts directly reflect copy spellings. We are, so to speak, able to see through B back to copy when we see a deare in his text. Dear is spoken of as an ideal example of a weak preference, copy-reflecting word because the complete absence of aberrant -ee- to -ea- changes makes the deare absolutely reliable as an indication of the copy spelling. Every single one of the thirty-seven (+J6) -ea- forms in B's texts directly reflects an -ea- form in copy.

There are, of course, many gradations of preferences ranging from words that approach the strong habits like do, go, and heere to those which fade into examples of compositorial indifference. In addition, many of these weak preference words are not like deare in being 100% reliable. Perfect reliability is not, however, a requisite for words to use in copy study. Take, for example, the weak preference word credit. B's demonstrable preference is for the credite spelling, but he sets credit seven (+J1) times. Of these seven times, six reflect six of the thirteen credit spellings in copy and one represents an aberrant change of the supposedly preferred credite to credit. This one aberrancy does not destroy the value of the word for the purpose of getting at copy since the odds are still six to one that credit in the Pavier text will reflect credit in copy.

The following compilation includes over 100 words which, within the Pavier texts, demonstrably reflect the characteristics of copy. Here,

therefore, is the verification of the hypothesis that a class of copy-reflecting words does exist. In the table, the second column lists B's observable preference or preferences, if any, for a given spelling of the word in question. The third column lists the forms which are members of the newly discovered class of copy-reflecting spellings. The numbers in parentheses after both forms give the number of times that these forms occur in B's work in the Paviers. A comparison of these figures will indicate roughly the place of that word in the spectum ranging from strong preference to indifference. The final column is a rough attempt to indicate the reliability of the form in column three as a reflection of copy. This percentage indicates the fraction of the instances recorded in column three which reflect copy. It must be noted that many of the copy-reflecting forms occur only infrequently, but the fact that a given copy-reflecting form occurs only once or twice does not by itself lessen the value of that form as an indication of copy. The total picture for each word must be examined. For example, alreadie occurs only once as a copy-reflecting form but this one occurrence is highly reliable when it is seen that B's preference for already is relatively strong and, therefore, any deviation from it is likely to be done under the influence of copy. B changed nine of the ten -ie forms in copy to -y and retained all three -y forms in copy. The one alreadie, therefore, that slipped through is, like the few doe, goe, and here forms which slipped through in contradiction of strong preferences, highly reliable as an indication of the copy spelling.

Words have been included in this list down to a 67% degree of reliability. This 67% figure in itself may not sound as strong as when it is expressed in the form of two-to-one odds that the copy-reflecting spelling does indicate the spelling in copy. This percentage is calculated only for the unjustified occurrences of the spellings in the third column so as to remove any possible contamination from the exigencies of justification. Since it is impossible to draw lines dividing the spectrum of copy-reflecting words into different groups, the percentage of reliability must be considered in conjunction with the number of times the preferential (column two) and copy-reflecting (column three) spellings occur in B's texts in order to evaluate the significance of a specific spelling as a form which reflects copy. Because of the dangers inherent in dealing with statistically small samples, even this rough indication of relative percentages of reliability has been omitted for words which do not occur at least ten times in the Paviers.

                                                                                                             

Word	B's Preference(s)	Spellings Reflecting Copy
aboard	aboard (5)	aboord (4+J1)	100%
act	acte (12+J3)	act (4)	100%
adieu	final -u (10)	final -w (2+J2)	100%
aid	medial -i- (12+J1)	medial -y- (3)	100%
alack	alacke (18)	alack (4)	75%
already	already (12)	alreadie (1)	100%
always	alwayes (7+J1)	alwaies (5+J1)	100%
and	and (General Rule)	& (4+J4)	100%
angel	angel (5+J1)	angell (5)	100%
answer	answer (43+J7)	final -re (12+J3)	83%
approach	approch (6+J1)	approach (5)	100%
art	_____	art (20)	100%
		arte (2)	100%
bar	barre (9)	bar (1)	100%
battle	battell (12+J4)	battle (1)	100%
		battaile (7+J1)	86%
been	_____	bin (18+J10)	72%
behind	behinde (17+J2)	behind (4+J3)	75%
bishop	byshop (19)	bishop (9)	100%
blind	blinde (14+J5)	blind (2+J2)	100%
blood	blood (97+J8)	bloud (32)	84%
bloody	bloody (27+J1)	medial -ou- (13)	100%
bow	bow (4)	bowe (1)
box	box (1)	boxe (2+J1)
briar	_____	bryer (2)
		brier (2)
brief	breefe (5)	briefe (12+J3)	100%
charity	charity (12+J3)	charitie (1)	100%
cheer	cheere (10+J2)	cheare (3)	100%
chief	cheefe (8+J1)	chiefe (1+J2)	100%
choice	choise (21+J1)	internal -y- (4)	100%
choose	choose (5)	chuse (3)
clear	cleere (9)	cleare (11+J1)	82%
coin	coine (4)	coyne (2)
command	command (17+J2)	commaund (1)	100%
courtesy	courtesie (6+J1)	curtesie (7+J1)	100%
cousin	_____	internal -ou- (9)	89%
		internal -o- (11+J4)	90%
credit	credite (8+J3)	credit (7+J1)	86%
crown	crowne (21+J3)	crown (1+J1)	100%
cruel	cruell (13+J1)	cruel (1)	100%
dear	deere (60+J13)	deare (37+J6)	100%
deed	deede (8+J2)	deed (14+J1)	71%
deny	deny (13+J3)	denie (1)	100%
devil	divell (44+J15)	devill (1+J1)	100%
do	do (573+J103)	doe (15+J21)	80%
duchess	final -esse (9+J3)	final -es (1+J1)	100%
duty	duty (14+J1)	dutie (1+J1)	100%
enemy	enemy (12+J1)	enemie (3)	100%
error	error (7+J1)	errour (1)
extreme	extreme (12)	extreame (2)	100%
fancy	fancy (4)	fancie (1)
forth	foorth (25+J3)	forth (16+J2)	87%
fury	fury (9)	furie (3)	100%

                                                                                                             

Word	B's Preference(s)	Spellings Reflecting Copy
girl	gyrle (6)	girle (1+J2)
go	go (231+J40)	goe (73+J32)	93%
grief	greefe (19+J1)	griefe (21+J4)	100%
grieve	greeve (19)	grieve (4+J3)	75%
guard	guard (13+J1)	gard (3)	100%
guess	guesse (5)	gesse (2)
hear	heare (19+J2)	hear (1)	100%
heart	heart (184+J27)	hart (20+J10)	95%
herald	herald (7)	herauld (5)	80%
here	heere (345+J44)	here (150+J41)	91%
heretic	hereticke (2)	heretike (6)
high	high (7+J2)	hi (y) e (9)	100%
hour	houre (45+J6)	internal -ow- (2)	100%
intreat	initial en- (17+J2)	initial in- (13+J1)	92%
	final -te (22+J2)	final -t (4+J1)	100%
jest	jeast (4)	jest (3)
keys	keyes (3)	keies (1)
kiss	kisse (15)	kis (2)	100%
liberty	liberty (4)	libertie (1)
loud	_____	lowd (4)	75%
		loud (7)	86%
mad	mad (13+J2)	madde (1+J1)	100%
madam	medial -d- (78+J12)	medial -dd- (1+J1)	100%
maid	internal -i- (25+J7)	internal -y- (2)	100%
majesty	majesty (47+J3)	majestie (2+J1)	100%
marvel	_____	marvell (4)	100%
		marvaile (5+J1)	100%
master	master (121+J19)	maister (4+J3)	100%
mighty	mighty (15+J3)	mightie (1)	100%
mind	minde (61+J10)	mind (4+J3)	100%
mock	mocke (9)	mock (1)	100%
month	month (6+J1)	moneth (2)
moral	morall (4)	morrall (1)
murder	_____	internal -d- (39+J3)	80%
		internal -th- (34+J3)	79%
near	neere (45+J4)	neare (2)	100%
need	neede (9+J2)	need (13+J4)	69%
noise	noise (5+J4)	noyse (3)	100%
oh	_____	o (244+J45)	99%
		oh (117+J20)	85%
old	olde (21+J18)	final -d (42+J16)	93%
palace	palace (3+J2)	pallace (4)	100%
		pallas (1)	100%
pity	internal -tt- (42+J7)	internal -t- (2+J1)	100%
poison	poyson (8)	poison (3)	67%
quick	quicke (6+J4)	quick (2)	100%
relieve	internal -ee- (11)	internal -ie- (1)	100%
run	run (16+J2)	runne (13+J7)	100%
sail	internal -y- (10+J2)	internal -i- (4+J1)	100%
seven	seven (9)	seaven (2)	100%
show(s)	medial -e- (75+J8)	medial -o- (13)	100%
singular:	final -w (60+J3)	final -we (2+J1)	100%
son	sonne (87+J14)	son (5+J3)	80%
sorry	internal -rr- (5)	internal -r- (3+J1)

                                     

Word	B's Preference(s)	Spellings Reflecting Copy
sudden	_____	internal -u- (4+J2)	100%
		internal -o- (11+J3)	90%
summer	summer (6+J2)	sommer (7+J1)	100%
sun	sunne (23+J1)	sun (5)	80%
together	together (12+J2)	togither (3)	100%
traitor	internal -i- (59+J4)	internal -y- (1+J1)	100%
truly	truely (7+J2)	truly (2+J1)	100%
try	try (6+J2)	trie (2)	100%
unless	unlesse (18)	unles (1)	100%
victory	victory (6)	victorie (2)
war	single r (5)	double rr (22)	100%
wax	wax (4+J2)	waxe (2+J5)	100%
we'll	wee'l (42+J5)	final -le (10+J3)	100%
wind	winde (34+J10)	wind (3)	100%
witness	witnesse (15+J4)	witnes (1)	100%
you'll	you'l (23+J3)	youle (3)	67%
young	_____	internal -o- (25+J8)	68%
		internal -ou- (39+J18)	72%

The forms in this table which are reflections of copy constitute the class of words hypothesized to exist at the beginning of this study. The large number of spellings which belong to this class and the consistently high degree of their reliability at reflecting copy spellings show that words which reflect copy are an important part of the overall pattern of B's spellings. Compositor B has more than simple preferential spellings; he has a complex pattern of strong preferences, weak preferences, tolerances, and even mechanical indifference. Time after time under the pressure of copy, he sets non-preferential spellings about which he was relatively indifferent and thereby allows his texts to transmit the evidence of spellings which lie behind them. No longer must the person investigating a B text look only for do and go and heere spellings. He will look for doe, goe, here, youle, wind and so on, which by long odds are indications of copy spellings. Using a large number of the forms which reflect copy spellings, we can now start to assemble from the evidence in B's texts a projection of the texture of copy spellings.

Until the Pavier statistics are carefully extended by word counts on control situations in the Folio, no thoroughgoing application of this new method for determining spelling features of unknown or disputed copy can be attempted. As a brief test of the validity of this theoretical technique, however, let us turn to the first text occurring in the 1623 Folio which was set by B from known copy and perform a formal experiment in hopes of indicating the validity of the technique. Such an experiment can offer a formal demonstration under laboratory-like conditions of this new method for determining the characteristics of

copy. From the Folio text, we can use the group of spellings reflecting copy to predict the spellings of words in the copy used for the Folio. Turning then to this copy, the validity or invalidity of the procedure will be apparent.

According to Hinman's study, Compositor B set the last six pages of Love's Labors Lost in the Folio (M4r-M6v).[22] Examining these six Folio pages, we find the following forms which occur in unjustified lines and should, on the basis of the results of the Pavier analysis, indicate copy spellings.

• alwaies 1
• bin 1
• deare 3
• griefe 1
• hart 3
• jest 5
• O 12
• show 3
• sodaine 1
• yong 1

These six Folio pages include, therefore, thirty-one spellings which, at odds ranging from two to one (yong) to virtual certainty (deare) should be reflections of spellings which stood in B's copy.

The experimental demonstration now depends on the confirmation or denial of the existence of these predicted phenomena by checking the control text, Love's Labors Lost Q1 (1598).[23] Checking these predicted spellings against the actual copy, we find that twenty-nine of the thirty-one or 93% of the projected spellings are in fact the spellings which stood in copy. Here is strong evidence that the hitherto unrecognized group of copy-reflecting spellings within B's total spelling system functions as an indication of copy. The strength of this experiment cannot be overemphasized. Starting with an hypothesis theoretically proposing the existence of a class of spellings which could be used as indications of copy spellings, a class of copy-reflecting spellings was constituted from the evidence derived from the control situation in the Paviers. The Folio sample was then examined for the presence of these newly discovered forms, and once they were found, yet another control confirmed the validity of the spellings themselves as reflections of copy and the validity of the technique of projecting copy spellings

from the evidence in B's texts. Here is a classic case in bibliography of hypothesis, prediction, and experimental verification.

It has now been demonstrated that a portion of a compositor's total spelling pattern is composed of spellings which directly reflect the spellings found in the compositor's copy. Although many investigators have conjectures about the origins and characteristics of copy used for various printed texts, the discovery and use of what are here called copy-reflecting spellings is the first procedure for determining features of copy. Many formerly vexing problems should be resolved by a thorough implementation of this method. After further investigation of spellings which reflect copy in Folio texts set by B from known copy, the road will be open to solve disputed problems like the possibility of the use of an annotated hybrid of Q1 (1608) and Q2 (1619) as copy for Folio Lear. Although authorities agree that the Folio Lear text was set from an annotated quarto, there is no agreement on the exact nature of copy. Using B's copy-reflecting spellings, one may be able to reliably project various features of the copy from B's share of the Folio text and then by inspection see which of the quartos matches the projection. In the case of texts set from MSS, these spellings which depend on copy spellings may provide the first substantial evidence concerning the specific physical features of copy. This new method, therefore, makes a start at enabling textual investigators to tackle one of the most vexing problems in establishing texts of English Renaissance drama, the nature of the lost MS copy behind extant printed texts. More specifically, when the technique has been applied to all of the workmen in the Shakespeare Folio, we may then possess an excellent tool for getting closer to the copy behind the texts of important Shakespearean plays that were printed for the first time in the Folio. The day may come when we can apply these techniques to texts set from Shakespearean foul papers and arrive at a far more intimate knowledge of the orthographical features of Shakespearean holograph than is at all possible today.

Notes