University of Virginia Library

Recent studies have shown that the spellings of English Renaissance compositors are useful not only in identifying the parts of books typeset by these workmen but also in discovering what copy they used in setting the pages assigned to them.[1] Identification of compositors' shares in books by means of spelling rests chiefly on their characteristic preferences (that is, specifically their typical departures from the various spellings of their copy in favor of a single spelling), whereas identification of copy for a given text rests on their tolerances (that is, their habitual, though not necessarily constant, retentions of non-preferred copy spellings). The first kind of study is of course the more familiar and has generally been a preliminary step in inquiries into the faithfulness with which compositors transmitted the substantives of their copy.[2] Obviously it also must precede the second kind of study, which is less common but may nevertheless play an important role in solving problems in the texts of authors such as Shakespeare.

Indeed, much of the work done with compositorial spelling has quite naturally been concerned with the compositors of Shakespeare's plays—especially those workmen in William Jaggard's shop who set the First Folio, and particularly the notorious Compositor B. Although it reprints quarto texts of several plays with little or no reference to independent authorities,[3] the Folio provides the earliest and only

authoritative texts of about half of Shakespeare's extant dramatic works and also contains texts, having independence and authority in various degrees, of several other plays that appeared in quartos prior to 1623 (e.g., H5 or Oth.). The compositors of this important book have inevitably attracted the attention of students of Shakespeare's text, and because analysis has indicated that B alone set about half the Folio, he has become the object of considerable interest. Yet we still have no systematic study of his spelling habits in the Folio and have given very little attention to the question of his habitual tolerances. Since the nature of the printed copy for Compositor B's stints in seven plays of the First Folio is known, a study of his work in these plays should show what spellings (besides the traditional ones, such as do, go, and heere) can help identify B's share in Folio pages hitherto only tentatively assigned to a compositor and what spellings in his work can help identify the copy behind other Folio plays (for example, 2H4 or Ham.).

The nature of the Folio copy for these seven plays—Much Ado about Nothing, Love's Labour's Lost, A Midsummer Night's Dream, The Merchant of Venice, I Henry IV, Titus Andronicus, and Romeo and Juliet—seems to be one of the few questions about the printing of the Folio on which scholars are united. Another point on which nearly all scholars agree is B's share in the typesetting of these plays. Professor Charlton Hinman's assignments to B rest on fairly substantial evidence, and those few pages where the evidence is such that he has expressed doubt about B's precise role can be temporarily excluded from consideration without irreparable harm to the study of B's work in these plays.[4] Moreover, T. H. Howard-Hill's recent examination of all the Folio Comedies has brought to light several new kinds of evidence

that support Hinman's attribution to B of pages in Ado, LLL, MND, and MV.[5] When applied to IH4, Tit., and Rom., these new kinds of evidence confirm the conclusions about B's role in these plays that Hinman draws from the already substantial evidence of cases and spellings. (See the summary of this evidence appended to this article, and Howard-Hill's summary, pp. 90-99.) Only in sigs. f5v, f6, and ee3 is the spacing after medial commas not entirely consistent with B's usual practices, but these pages contain prose and other long lines which sometimes influenced B to set more unspaced commas (Howard-Hill, p. 67), and the other evidence shows that the compositor was B, rather than one of the other workmen who could have set these pages.

Not all spellings, of course, are equally reliable evidence of a compositor's share in a book or of the nature of the copy that he used in setting it: one reason some spellings are unreliable is that they were subject to various influences which often caused compositors to depart from their normal habits. For instance, there is no question that justification and rhyme at times affected the practices of some Renaissance compositors. Thus when investigating them, we must separate spellings in rhymes, long lines, and even nearly long lines from other evidence and use them, if at all, only when previous investigation has shown that such spellings are reliable evidence of the compositor's normal practices. This general rule applies not only to spellings in rhymes and long lines. Study of B's work indicates that in inflections (e.g., showes) and compounds (e.g., sicknesse) his normal habits of spelling endings (e.g., preferring final -w to -we or -cke to -ck) were sometimes affected by the addition of one or more letters. When drawing conclusions about his spelling of a particular ending, then, we must discount evidence in inflections and compounds, whereas when initial or medial spellings are the issue, such evidence becomes admissible because study shows that B's habits were otherwise unaffected by compounding or inflecting. In a similar way, most but not all words in stage-directions are good evidence of B's normal spelling habits. It is well known that in B's work the entrances are often carefully centered.[6] Such entrances are also generally symmetrical, because ordinarily B just as carefully set them in the form of an inverted pyramid. To do so, he sometimes deviated from his normal habits of spelling the words (usually by shortening one), and thus spellings in such stage-directions are suspect as evidence. Nevertheless, examination indicates that lines in other stage-directions set in a different manner

were not subject to this kind of justification. In these lines there is no compelling evidence that B spelled any differently when setting stage-directions than he did when setting dialogue, and thus such stage-directions contain reliable evidence of his normal spelling practices.[7]

Initially, then, we must separate spellings in stage-directions as well as those in rhymes, long lines, nearly long lines, inflections, and in compounds from other evidence and must use them in drawing conclusions about B's normal spelling habits only when examination has shown that they are legitimate evidence of his ordinary practices. The present study observes this kind of distinction and adopts the symbols NL, L, R, and SD to identify the number of spellings in nearly long lines, long lines, rhymes, and stage-directions, respectively.

Ultimately of greater moment, perhaps, than such specific issues in establishing reliable methods for the study of compositorial spelling is the more general question of compositorial uniformity. Of course, some kind of uniformity, revealed in certain spellings or other features of a text, is necessary merely to identify a compositor. The question is not, then, whether a given compositor is uniform in some of his habits (such as B's preferences for do, go, and heere), but to what degree such uniformity extends to his practices over a period of time and to his treatment of similar forms.[8] To have a sound basis for studying a compositor's particular habits, we need to discover something about his general constancy and consistency and to learn how much aberration should be expected in his work.

Thus, before we use the relatively few spellings of a specific word to define the stability of a compositor's practice of treating it, we should examine the larger question of his over-all constancy by finding a broad base of evidence that will minimize the effect of random cases of aberration on our conclusions. Such a broad base of evidence is available in a compositor's practices of spelling orthographically analogous words. An examination of these practices should allow us to study his consistency and constancy at the same time. In the case of Compositor B, a potentially ideal situation exists because of the printing

of the Pavier Quartos in the Jaggard shop in 1619. Jaggard seldom printed dramatic texts; hence these quartos provide a rare opportunity for seeing some of his staff at work on comparable material shortly before he undertook the First Folio. Moreover, William S. Kable's recent work on the Pavier Quartos includes a study of the treatment of orthographically analogous words that could enable us to determine to what extent B's generic habits in the Folio of 1621-1623 had already become established in 1619.[9]

Unfortunately, our ability to use these quartos is somewhat compromised by the current disagreement about the typesetting of them. Until recently it was generally thought that B was responsible for essentially all the work.[10] However, in the last several years two studies of the Paviers, by John F. Andrews and Peter W. M. Blayney, have independently attacked this view as it is represented in Kable's work.[11] Lamentably, these studies do not agree on the division of the Paviers between their hypothetical compositors nor on whether B was one of them, and in spite of bits of good evidence, neither view seems conclusive enough at present to be adopted as a whole. Now that the Paviers have become a matter of some dispute, it is likely that better evidence more rigorously treated will be necessary to settle it.

Nevertheless, some comparison of the generic habits exhibited in the Pavier Quartos with the implications of the spellings in the Folio pages set by B from known printed copy can help us discover whether or not the spellings in the Paviers are at all useful in studying B's Folio work. Since such a comparison deals in general habits and cumulative statistics, it cannot prove that one compositor, B, set all the spellings, or even all the pages, in the Paviers. Nor can examination of B's general spelling habits in the Folio and of the general habits of the compositor(s) of the Paviers prove that B's habit of spelling each and every word remained stable between 1619 and 1623. But it can deal with the main issues that must be settled before analysis of B's spelling of particular words in the Folio proceeds, though the results of such an examination can only be summarized here.[12]

For one thing, the recurrent agreement between the spelling patterns noted in B's Folio pages set from known printed copy and the patterns in the Pavier Quartos noted by Kable reveals how remarkably similar are the general habits of B in F1 and the habits of the compositor(s) in Pq. The inconsistencies as well as consistencies of the spelling habits of the compositor(s) of the Pavier Quartos are duplicated in B's Folio work, when sufficient evidence exists to define those habits. The only exception to this uniformity is the treatment of final -ie/-y after consonants other than -f and -s, but in this case B's own inconstancy in F1 suggests that the discrepancy between the Pavier spellings and some of the Folio spellings is to be expected and does not necessarily imply different compositors. Hence, if another compositor's work in the Paviers is to be distinguished from B's, it seems this distinction will have to be based mainly on other evidence (especially mechanical) such as Andrews seeks to exploit, rather than on the evidence of spellings.[13] Whether or not another compositor is in fact responsible for a large part of the Pavier Quartos has yet to be decided. But given the kind of uniformity of spelling patterns observed, it seems safe to conclude that on the whole Kable's findings have some relevance in a study of B's Folio work.[14]

Insofar as B's general practices of 1619 are often exhibited in the Pavier spellings, the generic habits examined show that on the whole B's general practices remained constant between the time he worked

on the Pavier Quartos and the time he set the Tragedy section of the Folio. This constancy is apparent not only in simple preferences but also in intricate habits, as when his practice of spelling endings is determined by penultimate letters or when his choice of -d or -de depends on whether -an-, -en-, or -in- precedes it.[15] In a number of complex cases, B's very inconsistency in treating the words within orthographical groups may be said to prove his constancy.

Within the context of this general constancy, we can briefly note the amount of random variation, or aberration, in B's work. Study of B's spelling in the pages considered here shows that even his handling of words for which he has a very definite preferred spelling is not absolutely uniform. In spelling words of high frequency like 'do', 'go', 'here', 'be', 'he',[16] and those with final -l/-ll, B sometimes sets a non-preferred spelling instead of the preferred spelling present in his copy. Various tests combine to indicate that at least 1%, and sometimes up to 6%, of the occurrences of words for which he has a definite spelling preference exhibit such aberration. Where random variation becomes inconstancy or indifference is no doubt a matter of definition. But there is reason to believe that aberration must be allowed for in analyzing B's spelling of practically all words, and that with fifteen or so occurrences of a word available for examination, we can begin to distinguish genuine random variation from variable practice that is due to indifference rather than to chance. In studying this compositor's work, we must take account of the fact that any one spelling may simply be evidence of his normal human variability; however, once this fact is taken into account, conclusions about his spelling habits drawn from a number of examples nevertheless remain possible, and accurate ones should in fact be more feasible.[17]

Another thing the analysis of B's treatment of orthographically similar words in F1 shows is that in his Folio work, as well as in the Pavier Quartos, orthographical analogy occasionally "governed [the] treatment of various spellings" (Kable, p. 48). But by no means does it show that all or even most spellings were so governed. There is no guarantee that a pattern ascertained to govern the spelling of some analogous words necessarily applies to the spelling of other such words. It may or not. Knowledge of a compositor's practice of treating the spellings of several analogous words may make us more aware of the possible pitfalls to be avoided in interpreting limited spelling evidence for another analogous word; it may even aid us in interpreting the limited evidence for such a word (e.g., 'entreat') when that evidence suggests the compositor's manner of spelling the word conforms to his habitual manner of spelling analogous words (e.g., preferring en- and tolerating in-). But such knowledge cannot by itself provide positive evidence that will help solve practical problems of compositor or copy identification. We must turn to B's spellings of specific words to find such evidence.

The purpose of the following remarks is to help identify and evaluate spellings in the Folio plays set by B from known printed copy that we can, in turn, use to study his work elsewhere in F1. In some cases this involves simply reporting the imperfect state of our knowledge of B's practices, pending the resolution of the current disagreement over the Pavier Quartos. The aim is not to review all the spelling evidence now available on the words discussed, but simply to reach conclusions, on the basis of the pertinent Folio evidence (supplemented by whatever Pavier evidence seems relevant), about the usefulness of the spellings of these words either in assigning disputed Folio pages to B or in identifying the copy behind Folio texts set by him.[18]

There are, then, two classes of spellings that are of interest here: those that in B's work possibly derive from his copy, and those that B possibly prefers, under normal conditions.[19] Naturally, it is more

likely that a compositor would develop a preferred spelling for words frequently seen, such as 'he' and 'me', than for words infrequently seen, such as 'battle' and 'courtesy'. Spellings that appear in B's work chiefly because they were in his copy, rather than because he personally favored them, often are spellings of words of low frequency; therefore, conclusions about them must often be based upon a relatively small number of occurrences and hence sometimes involve rather careful examinations of the evidence. However, it is by no means true that all words for which B often reproduces a copy spelling occur infrequently. For instance, he seems to have developed no clearly preferred spelling of the frequent 'oh' but generally to have followed his copy's spelling, and his practice of spelling 'been' includes both a spelling preference and a spelling that derives from his copy.

Most of the spellings discussed here have been examined previously, though not always extensively, and some are included here, despite the limited evidence, because of the attention they have received in the past. William S. Kable has presented simply statistical evaluations of most of the copy-derived spellings, and he as well as Willoughby, Walker, Hinman, I. B. Cauthen, Jr., J. F. Andrews, and P. W. M. Blayney have discussed many of the preferred spellings.[20] The present examination departs somewhat from these investigations by providing discussion of the usefulness not only of preferred spellings in identifying B's work but also of derived spellings in identifying his copy. Such discussion is necessary both because (as just mentioned) the quantity of spellings is often small and because (as the complex questions considered above show) more is needed in an evaluation of the quality of particular spellings as evidence than the presentation of raw statistics.[21]

Although the actual evaluations of the spellings take the form of discussion, the evidence analyzed appears in tables preceding each discussion.

This short-hand summary presents a concise view of the pertinent evidence gathered from various sources alluded to already.[22] Naturally this evidence simply represents the best that is now available; further study may well result in some refinement of the statistics cited here, though hopefully no major correction. Most of the evidence for the Paviers is, of course, Kable's.[23] When Andrews or Blayney presents corrections of Kable's figures, these corrections also appear, enclosed in round brackets. At one point (p. 194) Blayney offers corrections of what he calls the "greatest discrepancies" in Kable's statistics based on a collation of all the Pavier Quartos, but these totals ignore the distinction between long and short lines; in citing these figures I have adjusted them on the basis of Kable's statistics (i.e., by subtracting from Blayney's total the number of spellings Kable shows appearing in long lines) and followed them with a query to show that the precise statistics, and therefore their significance, are as yet uncertain. Finally, though the significance of some of the evidence from the Pavier Quartos is as yet unclear (either because of alleged inaccuracies in Kable's statistics or because he, Andrews, and Blayney are imprecise about spellings in rhymes, stage-directions, and sometimes inflections and compounds), the very fact that Blayney has checked Kable's statistics not only identifies those to be most distrusted, but also offers some assurance that much of Kable's other evidence is essentially correct, as experience with Andrews' totals suggests. Until agreement about the spellings in the Paviers is reached, conclusions based solely on Kable's statistics must be very tentative at best. But where a substantial number of spellings in F1 exhibit the same pattern as do those recorded by Kable, or where Andrews' or Blayney's corrections fit that pattern, there is reason to believe that these findings are pertinent to an examination of B's spelling habits from 1621 through 1623.

     

ART	Spellings	Pq	FI
	art > art	20	1 (sb.)	17 (vb.)
	arte > arte	2

B's tolerance for art in both the noun and verb seems indisputable: invariable retentions of this spelling when found in Folio copy parallel those in the Paviers noted by Kable. The theory that arte is also a spelling that B will generally reproduce rests only on the 2 instances of its retention in the Pavier Quartos. Both spellings are probably reliable clues to the features of copy, if Kable's statistics are essentially correct,[24] but since we cannot define exactly how B generally responds to copy's arte, it is remotely possible that art is a spelling slightly preferred rather than just generally tolerated. That arte is a preference seems unlikely. These words deserve delicate handling, but their forms are valuable witnesses to the characteristics of B's copy.

                       

BEEN	Spellings	Pq	FI
	beene > beene	23	4
	bin > beene	3	2
	bin > bin	13	3
	beene > bin	1
	been > beene	1	3
	bene > beene	12
	bene > bene	15
	beene > bene	20
	been > bene	3	5
	bin > bene	8	1
	bene > bin	3

For this word B prefers final -ne and tolerates bin. He is practically indifferent to beene and bene relative to one another. The 3 cases of Folio copy's bin reproduced verify the implication of the evidence of the Paviers that bin in B's work derives from copy's bin a large percentage of the time (about 80%, or 4-to-1 odds). Since in the Folio bin is never set by B in place of another spelling of copy, as it sometimes is in Pq, the reliability of this spelling in F1 as a clue to the spelling of copy is perhaps a little higher than the combined Pavier and Folio statistics would indicate, but there is no way to be sure about this. The spellings in the Folio and the Paviers (assuming Kable's statistics are essentially right) seem to corroborate one another in suggesting that B will follow copy's bin a little more than 50% of the time it appears under normal conditions.

             

BLOOD	Spellings	Pq	FI
	-oo- > -oo-	52+5L	17+1NL+1L
	-ou- > -oo-	45+3L	5
	-ou- > -ou-	27	1
	-oo- > -ou-	5
	-d > -d	118	20
	-de > -d	3

The forms blood and bloud remain, respectively, predictable preferred and copy-derived spellings in the Folio. In Pq, copy's -de always becomes -d, and approximately one-third of the -ou- copy spellings are reproduced.[25] In F1 only -d appears in known copy and B's work, and B retains only 1 of the 6 bloud spellings in his copy. The 5 anomalous alterations of Pavier copy's -oo- to -ou- are not surprising, since they yield a normal random-variation factor of 3.6%. If Kable's statistics are essentially correct, B's habit did not change materially between 1619 and 1623, although there is perhaps less chance in F1 that he will set bloud.

             

BLOODY	Spellings	Pq	FI
	-oo- > -oo-	9	5
	-ou- > -oo-	18+1L	1
	-ou- > -ou-	13
	-y > -y	23	5
	-ie > -y	15
	-ie > -ie		1

As in B's practice generally, the addition of the -ie/-y ending does not affect B's treatment of the internal vowels -oo-/-ou-. Here, too, about one-third of copy's -ou- spellings slip into the Paviers instead of being changed to -oo-, according to Kable's figures. The Folio evidence corroborates the Pavier evidence of B's preference for -oo-; but since the only bloudy in Folio copy is altered to bloody, for now we must simply assume that B will continue to reproduce copy's -ou- at times in F1 under normal conditions. The -ie/-y ending is subject to the same inconstancy that governs other words with this alternative.

       

BRIEF	Spellings	Pq	FI
	-ee- > -ee-	(1)
	-ie- > -ee-	5	1
	-ie- > -ie-	16 (15)	2

As in other analogous words, the internal -ie- in this word is one of the steadiest clues to the spelling of B's copy.[26] The retention in the Folio of 2 of the 3 briefe spellings and the change of 1 briefe to breefe corroborate the evidence of the Pavier Quartos that B prefers -ee- but will set -ie-, rather than -ee-, from copy's -ie- about 70% of the time under unexceptional circumstances. To date there are no examples of random variation for this word. We should note that -fe seems to be habitually tolerated spelling: B's known copy does not have -f, and hence there is no proof that he actually prefers -fe to -f (as Kable implies).

       

CHIEF	Spellings	Pq	FI
	-ee- > -ee-	2	1
	-ie- > -ee-	8
	-ie- > -ie-	2

Apparently conforming to his habit for other orthographically analogous words, B seems to prefer the -ee- spelling of this word and to reproduce the exclusive -fe spellings of it in his copy. Since the only Folio example is B's reproduction of the 1 cheefe of his copy, we cannot be certain about our conclusions, especially that -ie- in F1 will duplicate the -ie- of copy. The meagre evidence of the Paviers and the examples of analogous words encourage the belief that it will and that approximately 20% of -ie- copy spellings will be retained, but there is no conclusive evidence in F1 in favor of this inference.

             

CHOICE	Spellings	Pq	FI
	-i- > -i-	11	4
	-y- > -i-	10	1
	-y- > -y-	4
	-s- > -s-	14+1L	3
	-c- > -s-	9
	-c- > -c-	2	2

There are two sets of spelling alternatives for this word, and they appear to function independently of each other. The evidence of the Folio seems to confirm the implications of the evidence of the Paviers that B prefers -i- to -y-, but since the only -y- spelling in Folio copy is changed to -i-, the 4 instances of Pavier copy's -y- retained are the only evidence that -y- derives from copy. On the other hand, though Kable does not concern himself with -c-/-s- variation, his Pavier evidence suggests B's preference for -s- and his tolerance of -c-, and in F1 the 3 retentions of copy's -s- support

the first hypothesis, and the 2 retentions of copy's -c- confirm the second. Thus, if Kable's calculations are correct, -y- copy spellings should be retained by B about 28% of the time, as should -c- spellings; when either appears in B's work under normal conditions, the chances are excellent that it derives from his copy.

             

COUSIN	Spellings	Pq	FI
	-o- > -o-	9	2
	-ou- > -o-	1
	-ou- > -ou-	8
	-o- > -ou-	1
	-oo- > -o-	1	2
	-oo- > -ou-		7

Kable concludes that both the -ou- and -o- spellings of these words reliably duplicate the spelling of B's copy, but the truth is considerably more complex than this. Indeed, it is so complex that the evidence can hardly be rehearsed here.[27] The main complexity is introduced by the appearance in Folio copy of a proportionately larger number of -oo- spellings than Kable finds in Pavier copy. The implication of all the evidence so far available seems to be that B's -ou- will under normal conditions derive either from copy's -ou- or -oo-, that -o- in his work will derive either from copy's -o- or -oo-, and that whereas both -ou- and -o- can be used to decide between hypothetical copy having only these two variants, neither is useful where copy's -oo- is involved. B's spellings of this word are not totally useless, but in their use considerable care must be exercised, especially in view of the likelihood of random variation.

         

DEAR	Spellings	Pq	FI
	-ee- > -ee-	37+7L (38+3L)	3+2NL
	-ea- > -ee-	23+6L (10)	3+1L
	-ea- > -ea-	37+6L (41+2L)	6
	-ee- > -ea-	(16+2L)

The accuracy of Kable's statistics on this word has been questioned by both Andrews and Blayney;[28] until B's treatment of this word in the Pavier Quartos has been defined, final conclusions about his Folio practice are impossible. However, although they are insufficient to support final conclusions, the 15 spellings in the Folio are enough to outline a provisional view of B's handling of this word. The evidence in F1 suggests that B favors internal -ee- but often reproduces the -ea- spelling of his copy. Somewhere around 60% or 70% of the -ea- spellings (as in deare, dearer,

dearest, and dearely) occurring in his copy should be reproduced by B. What remains uncertain as a result of Andrews' and Blayney's findings is the extent of random variation in B's spelling of this word—how often he changes copy's -ee- to -ea- and therefore how reliable a clue to his copy's spelling is an -ea- in his work. The evidence in F1 suggests high reliability, but it may be deceiving because of its relatively limited quantity. If Kable's statistics were essentially correct, then an -ea- spelling of this word in B's Folio work under normal conditions (including long lines and stage-directions, but not rhymes) would be a good clue to the spelling of his copy. If, however, B's spellings vary randomly, then the word must be discarded in attempts to identify his unknown copy.

             

DEVIL	Spellings	Pq	FI
	di- > di-	27	3
	de- > di-	17	1
	de- > de-	1
	-ll- > -ll	18	4
	-l > -ll	10
	-ll > -l	2

The spelling divell appears to be one B prefers in the Folio, as Alice Walker has noted; however, the available spelling evidence of the Pavier Quartos suggests that, precisely speaking, B's preference is only for initial di- as opposed to de- (this preference determines the use of internal -e- after -v-), since -ll spellings in Pavier copy become -l and B may be genuinely indifferent to the -l/-ll alternative (the random variation for 'devil' is 6%). Thus the preferred spelling of 'devil' is initial di-, not divell, in the Pavier Quartos, and there is no evidence that B's habit changed by the time he set the Folio. Nor is there any evidence in the Folio to corroborate Kable's conclusion that the spelling devill in B's work derives from copy. Only 1 time is the devill of Pavier copy retained, and though Kable's conjecture is plausible, this is slender evidence on which to rear any conclusions.

         

DO	Spellings	Pq	FI
	do > do	270	41
	doe > do	303	47
	doe > doe	12
	do > doe	3

It is well known that do is B's preferred spelling. The only question is whether doe is a spelling that reflects B's copy, as Kable maintains. If the doe at MV, 28, is not B's but C's,[29] then only 1 (L) doe appears in B's Folio work, and it duplicates copy's doe in a long line; it appears to be a spelling influenced by justification (see 1H4, 991). All the other doe spellings in Folio copy are altered to do. That doe will often appear in B's Folio work seems unlikely (though some random variation must be expected), but, given the present meagre evidence, if it does appear in a short line, the chances seem good that it derives from copy.

     

DOOR	Spellings	Pq	FI
	-oo- > -oo-	19	1+3NLSD
	-o- > -oo-	9	1NL

Compositor B's practice of spelling this word remains constant in the Pavier Quartos and the Folio: he habitually sets the spelling doore. The 1(NL) and 3(NLSD) changes are primary evidence, because anticipatory justification would only have prompted B to set -o-, not -oo-. B's spelling of this word in stage-directions is probably as reliable a witness to his practice as his spelling in dialogue.

       

EMBRACE	Spellings	Pq	FI
	e- > e-	6	1NLSD
	i- > e-	2	1
	i- > i-	1	2

This word occurs in B's work in the Pavier Quartos and the Folio a total of 13 times; hence, conclusions about it must be tentative. The evidence indicates that B prefers initial em- but that he will follow about half of the im- spellings of his copy. These conclusions find support in B's generic habit of preferring em(n) - and tolerating im(n) -.

       

EMPLOY	Spellings	Pq	FI
	i- > e-	2	1
	i- > i-	3	1
	e- > i-	1

B's spelling habits for 'employ' seem to resemble those for 'embrace'. Since it appears a total of 8 times in the Pavier Quartos and the Folio, the conclusion that B prefers initial em- for this word but tolerates the im- of his copy must be tentative. The only employ appearing in Pavier copy is altered to imploy. This change may suggest that B did not set it or may

show his indifference to the alternative em- and im- spellings, but in view of the handling of em-/im- generally, that it is anomalous seems more likely. That B actually favors initial em- but reproduces roughly half of the im- spellings of his copy under normal conditions is the conclusion tentatively suggested by the evidence currently available.

             

ENOUGH	Spellings	Pq	FI
	e- > e-	24	5
	i- > e-	18
	y- > e-	1
	-gh > -gh	41	4+1NL
	-w > -w	2	1
	-gh > -w	1

For this word B prefers initial en- to in- and follows both final -gh and -w when they appear in copy. The alteration of all initial in- copy spellings to en- in the Pavier Quartos is sufficient evidence of B's preference for en- and finds support in the reproduction of all 24 en- forms in the Pavier Quartos and the Folio. Whereas 1 final -gh is changed to -w in the Pavier Quartos, B's tolerance for both these spellings is shown in F1 and suggested in the other Pq spellings. The chances seem good that -w and especially -gh spellings in B's Folio work derive from his copy.

                 

ENTREAT	Spellings	Pq	FI
	e- > e-	3 (2)
	i- > e-	13	2
	i- > i-	12 (11)	1
	e- > i-	1	1
	-te > -te	8
	-t > -te	14
	-t > -t	4	1
	-ts > -t		1

Kable concludes that the initial en- and the final -te are B's preferred spellings of this word and that initial in- and final -t spellings in B's work derive from his copy. The Folio spellings tend to confirm his conclusions about en-/in- variation. But the evidence is slender, and B's treatment of this word is more complex than Kable believes. The statistics reveal that B's practice of spelling this word is subject to some random variation—6% if we combine the Pavier and Folio spellings. B's preference for initial en- is suggested by the combined statistics and by the fact that he spells analogous words similarly. But it is not a strong preference: he reproduces 45% of the in- spellings of his copy, altering only 55% to en-. If we allow that the single example of Folio copy's en- altered to in- is just random variation, rather than evidence of his indifference in 1621-1623, then the chances are that 86% of the initial in- spellings appearing in B's Folio work under normal conditions derive from copy. However, his treatment of -t/-te in this word is unpredictable. His alleged preference for final -te is not confirmed or disproven by the Folio evidence; B is probably indifferent to this ending, as he is to -t/-te generally.

               

FORFEIT	Spellings	Pq	FI
	-e- > -e-	2
	-ei- > -e-	1
	-ei- > -ei-	2
	-e- > -ei-		1 (sb.)
	-ai- > -ei-	5	3+1L 3 (vb.)
	-ai- > -e-	2
	-ei- > -ey-		1

B's spelling habit for this word is complex. The one thing the Folio evidence confirms is that B invariably rejects the internal -ai- spellings of this word when they occur in his copy; apparently, he is indifferent to the -e- and -ei- spellings relative to one another. His alteration of the 1 forfeited of Folio copy to forfeyted, in accord with his equivocal attitude towards -ei-/-ey- spellings generally, merely complicates the picture. The -e-, -ei-, and -ey- spellings are all characteristics of B. He favors neither -t nor -te endings.

         

GO	Spellings	Pq	FI
	go > go	129	8
	goe > go	102	23+1SD
	goe > goe	68
	go > goe	5

As is well known, B prefers the spelling go. The question is whether when goe appears in his work it can be counted on to derive from his copy. The spelling goe appears in the Pavier Quartos 73 times in short lines, and of these 68 duplicate the spelling of copy and 5 replace copy's go (random variation is thus about 1.5%). The only goe spelling in B's Folio work from known printed copy is at the end of a long line (copy also has goe). Hence, it appears unlikely that goe will occur in a short line in B's Folio work, but if it does, the odds that it derives from copy under normal conditions would appear to be very good.

     

GRANT	Spellings	Pq	FI
	-a- > -a-	12
	-au- > -a-	8	2

Alice Walker's guess that B prefers grant to graunt is confirmed not only by the evidence of the Paviers but also by that of F1. He changes -au- spellings of his copy to -a- and reproduces his copy's -a- and -t- spellings invariably. The form grant is characteristic of B.

       

GRIEF	Spellings	Pq	FI
	-ee- > -ee-	2
	-ie- > -ee-	19	4
	-ie- > -ie-	21

In the Folio B prefers the spelling greefe to griefe. This habit also appears in the Pavier Quartos, but here we find about 53% of the griefe spellings in copy reproduced. The fact that in F1 he reproduces only 1 (L) griefe form in a long line may indicate that between 1619 and 1621 B's preference for -ee- had become stronger. However, the appearance of a stronger preference may simply be owing to the meagreness of the Folio evidence. Since not a single recorded instance of an internal -ee- copy spelling of this word is altered to -ie-, its conformity to the patterns in analogous words suggests that when an -ie- spelling occurs in B's Folio work, it almost certainly derives from copy under normal conditions.

         

GRIEVE	Spellings	Pq	FI
	-ee- > -ee	16	1
	-ie- > -ee-	5	2
	-ie- > -ie-	3	1
	-ee- > -ie-	1

This word follows a pattern almost identical to that of 'grief'. B prefers the -ee- spelling but almost half the time will set -ie- when his copy contains it in short lines. An anomalous -ie- spelling set from copy's -ee- may sometimes appear, but usually B's -ie- will reproduce his copy's.

           

HEAR	Spellings	Pq	FI
	-re > -re	19	22
	-r > -r	1
	-ea- > -ea-	17	25
	-e- > -ea-	4
	-ee- > -ea-	1

Kable's conclusion that heare is B's spelling preference and that hear is a derived spelling is imprecise, although it is true that B prefers the internal -ea- spelling to -e- and probably to -ee, as well as tolerates -re and -r. B invariably follows all the final -re copy spellings of this word in both the Pavier Quartos and the Folio. Whether he actually prefers the final -re enough to alter an -r spelling of copy to -re is another question. The only final -r spelling we have is in Pavier copy and is retained. Hence, no compelling evidence exists that B prefers the final -re spelling or that he will always follow hear when it appears in his copy.

             

HEART	Spellings	Pq	FI
	-ea- > -ea-	111	6
	-a- > -ea-	58	15
	-a- > -a-	24
	-ea- > -a-	1
	-t- > -t	152	15
	-te > -t	1

That B prefers the spelling heart is certain. In the Pavier Quartos, we find approximately 71% of the -a- spellings of copy altered to -ea- and approximately 29% of the -a- spellings that appear in copy reproduced. Had this practice continued in the Folio, at least 4 of the 15 -a- spellings in F1 copy would have been reproduced, but not a single one was. B simply retains copy hart all 4(R) times it appears in rhymes (matching, for instance, part) and changes the 1(R) heart of copy in a rhyme to hart. Indeed, many of the hart forms in the Pavier Quartos also occur in rhymes; my research is incomplete, but so far I find only 1 hart reproduced from Pavier copy's hart that is not in a rhyme. Hence, Kable's statistics appear to be somewhat misleading.

When the internal -a- spelling occurs in B's work in a short line and is not in a rhyme, the chances seem fair, given the present evidence, that it derives from an -a- spelling in copy. However, when an internal -a- spelling occurs in a rhyme in B's work and matches spellings such as part and dart, we cannot be certain whether -a- or -ea- was in B's copy.

         

HEAVEN	Spellings	Pq	FI
	h- > h-	36	10
	H- > h-	8
	H- > H-	1
	h- > H-		2

Alice Walker has distinguished A's use of the capitalized spelling Heaven from B's use of the heaven prior to Folio Mac., and she has remarked that both compositors capitalize the word from Mac. onward. Kable finds that B invariably follows the heaven spelling of Pavier copy and that he reduces to lower case 8 of the 9 Heaven forms of Pavier copy. In the Folio, B's copy for MND, MV, and LLL invariably has the lower-case spelling, and B follows all these forms. He retains 2 of the 3 heaven spellings of Q5 1H4 but alters the other 1 to Heaven, and he changes the 1 (+1L) heaven of copy to Heaven on sig. Gg1 of Rom. (set after Mac.). Hence although (as Walker notes) B prefers heaven at the beginning of F1 and Heaven at the end, his change of preference may have begun before the Tragedies were set.

         

HERE	Spellings	Pq	FI
	-ee- > -ee-	134 (133)	19+4R
	-e- > -ee-	126 (201)	27+1R
	-e- > -e-	137 (138)
	-ee- > -e-	13 (16)

B clearly prefers the spelling heere both in the Pavier Quartos and in the Folio. But whether here in his work is a reliable clue to the spelling of his copy is not as certain. The evidence of the Paviers suggests that it is. In spite of the fact that there may be minor inaccuracies in Kable's figures,[30] it seems that ordinarily here will not appear in B's Folio work under normal conditions and that, if it does appear, the chances are about 9 to 1 that it comes from copy.

       

HOUR	Spellings	Pq	FI
	-ou- > -ou-	34	15
	-ow- > -ou-	11	1
	-ow- > -ow-	2

B clearly prefers the internal -ou- spelling of this word. What is doubtful is the extent of his tolerance for -ow-. Since, on the evidence of the Paviers, B will retain only 14% of the -ow- forms of copy, at most, the fact that in F1 he alters the only -ow- of copy to -ou- need not imply that by the time he set F1 B no longer reproduced the -ow- spelling. The evidence of B's tolerance for the -ow- spelling is so meagre that we cannot determine whether he is likely to reproduce an occasional -ow- spelling of copy when setting the Folio.

         

JUDGMENT	Spellings	Pq	FI
	-ge- > -ge-	12	5
	-g- > -g-	1
	-ge- > -g-	1
	-e- > -ge-	1

Neither the Pavier nor the Folio evidence indicates that B simply prefers the spelling judgement (as Kable maintains). That B usually reproduces this form (the most frequent of Pavier and Folio copy) is clear. That he prefers judgement to judement is implied by his change of the 1 judement spelling in Pavier copy to judgement, but obviously this single piece of evidence does not warrant the broad conclusion that B has a blanket preference for judgement. Rather, the slender evidence indicates that B is usually inclined to follow his copy's spelling of this word and that—if he favors either judgement or judgment at all—he prefers judgment.

     

LABOR	Spellings	Pq	FI
	-ou- > -ou-	20	4
	-o- > -ou-	1

The evidence of B's preference—as distinguished from consistent toleration—for the internal -ou- spelling of this word is not very strong. The only evidence of this preference is his alteration of the 1 labor'd spelling of Pavier copy to labour'd. This alteration should not, of course, be discounted because it occurs in an inflectional form. Moreover, since no internal -o- spellings appear in either B's Pavier or Folio work, and since B retains all -ou- spellings of his Pavier and Folio copy, it is tempting to conclude that he actually prefers internal -ou-. But because the alteration of labor'd to labour'd could be mere random variation, such a conclusion is unjustified. All that is certain is that B is inclined to reproduce internal -ou- spellings of this word.

       

LION	Spellings	Pq	FI
	-y- > -y-	20	1+1SD
	-i- > -y-	5	1
	-y- > -i-	2	11

The peculiar nature of the evidence for this word in F1 makes conclusions about B's practice difficult. The evidence of the Paviers suggests a possible slight preference for lyon. But since most of the Folio evidence comes from the passage in MND where B apparently decided to use lion in the dialogue and lyon in speech-prefixes and stage-directions,[31] we cannot either confirm or deny Kable's conclusion that B prefers lyon. Elsewhere in F1, the 1 lion of Q5 1H4 is changed to lyon, but the 1 lyon of LLL copy becomes lion. Quite probably B is indifferent to the lion and lyon spellings.

       

LOSE	Spellings	Pq	FI
	-o- > -o-	6	2
	-oo- > -o-	18	2
	-oo- > -oo-	6	5

The 39 instances of this word in the Pavier Quartos and in B's Folio work from known printed copy indicate that B always follows his copy in its invariable use of final -se, occasionally reproduces the internal -oo- spelling if it appears in copy, and generally prefers the internal -o- spelling. The evidence for the Paviers is probably incomplete, but the spellings we now have indicate that B prefers internal -o- in both the Pavier Quartos and the Folio, and that when -oo- appears in his work under unexceptional

circumstances, it almost surely comes from copy.[32] What is not clear is how frequently we can expect B to follow his copy's -oo- spelling. In the Pavier Quartos he follows only 25% of the -oo- spellings of copy; in F1 he retains 5 out of 7—71%—of these forms. Perhaps his tolerance of the internal -oo- increased between the years 1619 and 1621; perhaps the meagre evidence of F1 is a bit misleading. In any case we should expect at least 25% of the -oo- spellings in Folio copy to be reproduced.

               

MADAM	Spellings	Pq	FI
	-d- > -d-	75	13
	-dd- > -d-	3
	-dd- > -dd-	1
	-m > -m	45 (42)	2
	-me > -m	11 (14)	8
	-me > -me	16 (14)	3
	-m > -me	7 (8)

B seems to prefer the internal -d- spelling of this word. The likelihood that he will reproduce the -dd- spelling of his copy is difficult to assess, and his practice of setting the final -m/-me spelling alternative is problematic.

Kable's conclusion that an internal -dd- spelling appearing in B's work derives from the spelling of copy is untrustworthy because it rests on only a single instance of retention of the -dd- copy spelling; the -dd- spelling does not occur in known Folio copy. In both the Pavier Quartos and the Folio, B reproduces all -d- copy spellings and alters most -dd- forms to -d-. His preference for the -d- spelling seems clear, but whether he was inclined regularly to follow a small proportion of the -dd- forms of his copy is uncertain.

B's attitude toward the alternative final -m and -me spellings is also unclear. Although he may favor the -m spelling while tolerating -me forms of his copy, if we hypothesize that -m is B's preferential spelling, the changes of Pavier copy's -m to -me yield a random-variation factor (9%) that is high enough to suggest that B was indifferent to the -m/-me alternative. The Folio spellings are too few to resolve the question. It seems that B was inclined to alter copy -me spellings to -m and to retain a large proportion of the -m forms of his copy. What is unclear is whether this is a result of indifference or of settled preference, and if the latter, whether his inclination to set the final -m became stronger between 1619 and 1621. Because this is unclear, we cannot know whether in F1 he will actually prefer the final -m and whether we can expect the final -me in his Folio work to derive from his copy.

       

MAID	Spellings	Pq	FI
	-i- > -i-	21	2 ('maid')	6 ('maiden')
	-y- > -i-	9		2
	-y- > -y-	2

Both the word 'maid' and the word 'maiden' should be discussed here. B prefers the internal -i- spelling of both. That the internal -y- spelling in his work necessarily reflects the spelling of copy is less certain. Any attempt to assess B's disposition to reproduce his copy's -y- is complicated by an apparent inconsistency in Kable's account. Twice in discussing 'maid' he mentions that B retains 2 -y copy spellings (pp. 28, 61); however, in his "word-count" (p. 99) no -y- spellings are listed as B's Pavier spellings of 'maid' (Kable ostensibly does not include statistics on 'maiden'). In my own (incomplete) count of the Pavier spellings of 'maiden', I find B following copy's internal -y- 2 times: once in setting maydens and once in setting mayden-head. Perhaps these are the 2 cases Kable refers to. In any event, when a -y- spelling occurs in B's Folio work under normal conditions, it probably derives from the copy spelling, for that B retains a very small proportion of -y- forms of his Pavier copy is clear, though that this disposition to tolerate -y- continues in F1 is not actually proven by the present evidence.

       

MASTER	Spellings	Pq		FI
	-a- > -a-	28	2 (sb.)	1 (vb.)
	-ai- > -a-	93 (114?)	3	1
	-ai- > -ai-	4	1

B prefers the spelling master but will occasionally set maister under the influence of this spelling in copy. This is the conclusion implied by the evidence of the Pavier Quartos and of the Folio about both the noun and verb. B will set a large proportion of -ai- spellings of copy as -a-, but when -ai- appears in his work under normal circumstances, it almost certainly derives from copy.

             

METHINKS	Spellings	Pq	FI
	me- > me-	18	1
	mee- > me-	2
	-kes > -kes	8	1
	-ks > -kes	8	1NL
	-ks > -ks	2
	-kes > -ks	2

B prefers the initial me- and final -kes spellings of this word, and he prefers to set either the hyphenated me-thinkes or the closed methinkes forms. He is intolerant of the mee- spelling, but he will occasionally set -ks and the spaced me thinkes under the influence of such forms in copy.

B's habits for this word appear to be stable. His preference for the me- spelling is revealed by his alteration to me- of all the mee- spellings in Pq

copy so far recorded and is suggested by his retention of all me- spellings of Pq copy and by his retention of the me- spelling of F1 copy. His preference for -kes is displayed in both the Paviers and the Folio spellings, including his alteration (not influenced by justification) of the 1 (NL) -ks spelling of Folio copy to -kes. The Pavier Quartos exhibit B following copy -ks 2 of the 10 times it appears and altering 2 of the 10 -kes copy spellings to -ks. There is no such disposition to set -ks displayed in F1, but this lack may be the result of the scarceness of Folio evidence.

Equally as important as B's spelling of this word is his treatment of its word-division. B prefers to unite the components of this compound: he closes up the spaced elements of his Pavier copy 9 (+2L) times and inserts a hyphen when his copy has a space 5 (+5L) times. He never separates the word when Pavier copy has it closed up, but he does reproduce the spaced compound 6 of the 15 (+2L) times it appears spaced in Pavier copy. In F1 he sets 1 (NL) me thinks of copy as methinkes and follows copy in setting me thinkes 1 other time. In setting 2 (L) spellings in long lines, he omits Folio copy's hyphen, leaving the word spaced; since the space seems to be narrower than the hyphen (although I cannot be sure that it is), these two spellings may be the result of justification. It seems safe, then, to infer from the available evidence (not complete for the Paviers[33]) that, when B separates this compound in a line not affected by justification, the spaced form probably comes from his copy.

       

MIND	Spellings	Pq	FI
	-de > -de	35	4
	-d > -de	26
	-d > -d	4

Kable finds that B prefers the spelling minde but that he will occasionally set the form mind under the influence of copy. The Folio statistics neither confirm nor deny these findings, but they tend to support them. Although the Folio evidence is slender, it seems to conform to the pattern exhibited by analogous words and clearly defined by the Pavier evidence. If Kable's statistics are right, when the spelling mind occurs under normal conditions in B's work, it will almost surely derive from copy. He will alter most but not all of the mind spellings of his copy to minde in short lines.

                 

MISTRESS	Spellings	Pq	FI
	-is > -is	20
	-esse > -is	6	1
	-esse > -esse	41	1
	-is > -esse	2
	-es > -is	6
	-es > -esse	2	1
	-eris > -ris	32
	other forms	4

In the Folio plays set from known printed copy, B continues to prefer mistris and mistresse to the other possible spellings of this word. Thus, as Kable properly notes, Alice Walker's conclusion that mistresse is A's spelling and mistris B's is imprecise. However, though B will set either, he seems slightly to prefer mistris to mistresse. For instance, at MND, 430, he alters Q2's mistresse (mistresse in Q1) to Mistris in F1. Although he is relatively tolerant of both spellings, it would not be surprising if B slightly favors mistris in Folio plays set from manuscript or annotated quarto.

         

MONTH	Spellings	Pq	FI
	-n- > -n-	10	8
	-ne- > -n-	4
	-ne- > -ne-	2
	-n- > -ne-	2	1

Kable concludes that B prefers the spelling month and that when moneth appears in his work it almost surely reproduces his copy's spelling. The Folio evidence does not confirm this conclusion, but it does suggest a pattern that was not identified by Kable but is present in the Pavier spellings. What is at issue is B's treatment of the internal -n-/-ne- alternative spellings of this word, and as has been indicated, B's habits for internal spellings are generally unaffected by inflection and compounding. Therefore, not only B's spelling of the singular but also his spelling of the plural is pertinent to defining his practice. The Pavier evidence for the singular inevitably leads to Kable's conclusion. However, in setting the plural, B not only retains many of the months forms of Pavier copy, but also changes 2 months spellings to moneths. If we argue that the internal -n- is B's preference, these last 2 changes yield a high random-variation factor of over 11%, which suggests that B's spelling of this word borders on indifference. It is, of course, possible that B spells the plural differently than the singular, but this random-variation factor is found in the Folio spellings of the singular. The random-variation factor of 11%—constant from the Paviers through the Folio—makes B's spelling practice for this word on the whole definable, but unpredictable in any one case. Given the present evidence, the odds are only 2 to 3 that an -ne- spelling in B's work duplicates on -ne- spelling in his copy.

         

MURDER	Spellings	Pq	FI
	-d- > -d-	31	1
	-th- > -d-	8 (9)
	-th- > -th-	26 (25)
	-d- > -th-	8 (7)	1

It appears that B is relatively indifferent to the alternative internal -d- and -th- spellings of this word. He prefers neither more than the other, and therefore the spelling of his copy exerts a strong influence upon his own spelling of the word. If B is responsible for the Pavier spellings,[34] then when the internal -d- occurs in his work in short lines the chances are about 4 to 1 that it duplicates the spelling of copy. The chances are almost the same that a -th- spelling in B's work under normal circumstances derives from the spelling of copy: 74% of the -th- copy spellings are reproduced by B. However, these observations rest almost wholly on Pavier evidence; hence conclusions about this word must remain tentative, at least until the controversy over the Paviers is settled.

             

NEAR	Spellings	Pq	FI
	-ee- > -ee-	33+4L	5+1L
	-ea- > -ee-	9	1L
	-ea- > -ea-	2	1
	-e- > -ee-	2
	-re > -re	39	2
	-r > -re	1

B's preferred spelling for this word is neere, and he sometimes reproduces the internal -ea- spelling when it appears in his copy. Although the evidence of his preference for the final -re spelling slim, the evidence that B prefers the internal -ee- spelling both to -ea- and to -e- is more substantial. This evidence is stronger for his preference of -ee- to -ea-, partly because, as has been shown, spellings in long lines are as valuable as those in short lines, since the letters a and e are both one-en thick.[35] While preferring the internal -ee- spelling to both -ea- and -e- and while probably favoring the final -re, B is led to set internal -ea- about 23% of the time it appears in his copy. If an -ea- spelling occurs in his work under normal conditions, it should almost certainly duplicate a copy spelling.

         

NEITHER	Spellings	Pq	FI
	-i- > -y-	3	1
	-i- > -i-	13	3
	-y- > -i-	1
	-e- > -ey-	1

From the limited evidence that is available it appears that B favors the spelling neyther but is very tolerant of neither when it occurs in copy. The Pavier statistics suggest that B retains roughly 80% of the neither spellings of his copy and changes the other 20% to neyther. The same practice is exhibited in B's Folio work from known printed copy. Unfortunately, no neyther spellings occur in Folio copy, and we cannot, therefore, view B treating the neyther of his copy. Hence, we cannot determine whether he actually prefers neyther or is indifferent to the -i-/-y- alternative. If neyther is supposed to be his preferred spelling, the single alteration of Pavier copy's neyther to neither yields a random-variation factor of 4.5%, which is high but not unreasonable. Although for the present it is not clear that B actually prefers the form neyther, it is clear that he reproduces 75% to 80% of the neither spellings of his copy and that when neither appears in his work in short lines, the odds are good that it duplicates his copy's form.

           

OH	Spellings	Pq	FI
	Oh > Oh	99+20L (109+13L)	3+1L
	O > Oh	18 (16)
	O > O	242+44L (190+19L)	43+4NL+16L
	Oh > O	2+1L (3)
	ô > O		2

Both B's O and Oh spellings are reliable clues to the spelling of his copy; if O appears in his work the chances are excellent that either O or ô was in his copy, and they are almost as good that an Oh spelling in his work derives from copy.

The evidence of both the Paviers and the Folio leads to these conclusions, but since the evidence of the Paviers is more ample than that of the Folio, we should examine its implications first. The spellings can be interpreted three ways: (1) B has no real personal preference for either spelling, and hence is usually influenced by his copy's spelling but also occasionally disregards it and sets Oh or O randomly; or (2) B prefers the spelling Oh to O, but because he infrequently sees Oh in his copy does not develop a strong preference and tends to tolerate the vast majority of non-preferred O copy spellings; or (3) two compositors, B and "G," both normally reproduce their copy's O and Oh spellings, but the former very slightly prefers O and the latter slightly prefers Oh (Andrews' theory, p. 381, which is damaged by the fact that "G" replaces an Oh of copy with O).

The Folio evidence affords no help in deciding between these hypotheses, but it does indicate that B's habit of spelling this word remained

constant over the years from 1619 through 1623. Folio copy, like Pavier copy, has predominantly O spellings, and B reproduces every single one. Only 3 (+1L) Oh spellings appear in Folio copy, but B retains all of these. Folio copy also has 2 ô spellings (I gather Kable did not distinguish between O and ô), and B sets these as O. With relatively few Oh spellings in B's copy, or his work for that matter, it is difficult to know whether or not he actually prefers Oh to O; were it concluded that he does, the factor of random variation for this word would be an encouraging 0.71%. Moreover, having no evidence from the Tragedy section of F1, we cannot be positive that B did not develop a preference for Oh late in the setting of the Folio. What is clear is that both spellings of this common word are rather reliable clues to the spellings of B's copy at least in the Comedy and History sections. The odds are at least 9 to 1 that an O spelling in B's work derives either from his copy's O or ô. The odds are 7 to 1 that an Oh spelling in B's work duplicates the spelling of his copy. Ordinarily, B will alter, on the average, only 1 out of every 7 O spellings of his copy to Oh; only 1 out of 10 Oh spellings in his copy will he alter to O. Finally, we may note that apparently the occurrence of this word in a long line rarely influenced B's spelling of it; usually the word is the first in a speech (either prose or verse) and hence is unlikely to be affected by justification.

     

PIECE	Spellings	Pq	FI
	-ee- > -ee-	12	2
	-ie- > -ee-	1	2

The change of copy's piece to peece in the Pavier Quartos is meagre evidence for the conclusion that B prefers the spelling peece. However, in F1, B alters 2 other piece copy spellings to peece. Thus, he alters to peece every piece spelling he encounters in copy and retains every peece form; although the evidence is still slender, the chances are that the spelling peece is indeed B's preference.

         

POWER	Spellings	Pq	FI
	-er > -er	19+2L	8+1L+1NLSD
	-re > -er	2
	-ers > -ers	6	1
	-ers > -'rs	1

The 2 alterations of copy's powre to power in the Paviers indicate that B favors the spelling power, even though Cauthen noted 18 power forms and 7 powre forms in Lr. (the latter may occur in E's pages). B's invariable retention of power in the Paviers and in the Folio tends to support this hypothesis. However, B's habit for the plural is still indeterminable. It is possible, but not proven, that B's apparent preference for power does not extend to the plural and that he might set pow'rs and powres as well as powers indiscriminately; it is also possible that the single -ers > -'rs in Pq is random variation.

                     

PRAY THEE	Spellings	Pq	FI
	prethee > prethee	4	3
	prithee > prethee	5
	prithee > prithee	1
	prethe > prethee	4
	prethe > prethe	1
	prythee > prethee	1
	prethy > prethee	4
	pray thee > prethee	3
	prethee > prythee		1L
	other forms	4

Alice Walker contends that B favors the spelling prethee until setting the late Histories, and the evidence of the Paviers and F1 supports this contention. The word occurs in B's Folio work from known printed copy only in IH4. There B retains prethee all 3 times it appears in copy, and this evidence, along with the Pavier evidence of a preference for prethee, substantiates Walker's conclusion. Moreover, the 1 (L) change of prethee to prythee also supports her statement that B begins in IH4 to prefer prythee, since justification could not mechanically affect B's spelling of this word, even though it occurs in a long line (990). However, the other 3 cases of prythee that Walker notes in early IH4 occur in pages that B may not have set; hence, they are not presently admissible as evidence of B's practice.

               

RUN	Spellings	Pq	FI
	-n > -n	4
	-nne > -n	12	5+1SD
	-nne > -nne	13
	-ns > -ns	2
	-nnes > -ns	1
	-nnes > -nnes	2
	-ns > -nnes	2

The evidence is excellent in support of the conclusion that B prefers the spelling run and that when runne appears in his work it derives from the spelling of his copy. If B, and not C, is the compositor of MV, 29,[36] then B reproduces 1 runne spelling of Folio copy in accordance with the disposition, evident in the Pavier Quartos, to tolerate runne occasionally. But even if this spelling is B's, he retains only 14% of the runne copy spellings when setting F1, whereas he retains over 50% in the Paviers. This decline in the proportion of runne spellings reproduced may be owing to a hardening of B's preference for run. Nevertheless, if Kable's statistics are essentially correct, when runne appears in B's Folio work under normal circumstances, it should duplicate the spelling of his copy. We cannot say whether the pattern observed in B's spelling of the singular extends to the plural. The evidence of Pq indicates that it does not; in F1 the plural occurs only in long lines.

             

SHOW	Spellings	Pq	FI
	-e- > -e-	53+7L	3+1L (sb.)	10+8L (vb.)
	-o- > -e-	22+1L	1L	5+1L
	-o- > -o-	13	1
	-w > -w	51	3	7
	-we > -w	11
	-wes > -wes	15	1	2

Because we cannot tell from Kable's statistics how many of the spellings that he lists occur in literal rhymes, precise conclusions about these words are presently out of the question. Nevertheless, it seems likely that B prefers the final -w and -wes spellings of both the noun and verb, that he favors the internal -e- spelling of both, and that when an internal -o- spelling appears in his work in unrhymed words (in long or short lines), it duplicates his copy's spelling. Kable's statistics (which ignore not only rhyme but the part of speech of the words) show B retaining all of the 60 the internal -e- spelling of both, and that when an internal -o- spelling ings, while altering the other 23 (22+1L) to internal -e-. In F1 the only show spelling appearing in B's work from known printed copy is in a rhyme (matching know). However, since inflection generally did not affect B's habits for internal spellings, his retention of 1 of the 4 showes forms in Folio copy verifies the hypothesis that he occasionally reproduces an -o- spelling of his copy. B prefers the internal -e- spelling, but—when literal rhyme is not involved—an -o- spelling in his work ordinarily derives from copy.

         

SON	Spellings	Pq	FI
	-nne > -nne	79	13
	-n > -nne	8
	-n > -n	4
	-nne > -n	1

B prefers the spelling sonne but will set son under the influence of copy. The 1 alteration of copy sonne to son yields a low random-variation factor of 1%. Although B prefers the spelling sonne, he will reproduce up to 30% of the son forms appearing in his copy. When the spelling son occurs in his work under normal conditions, the chances are good that it duplicates a copy spelling.

           

SUIT	Spellings	Pq	FI
	-ui- > -ui-	1	3+1R
	-u- > -ui-		1
	-u- > -u-	20	1
	-ors > -ors	1
	-ers > -ers	2	1

B is inclined to reproduce both the internal -ui- and the internal -u- copy spelling of this word; he also appears to follow the copy spellings sutors and suters. The 1(R) suite in F1 is good evidence of B's tendency to

reproduce this spelling because B ignores the opportunity to match sute with the rhyming mute. The single alteration of sute to suite may show us B developing a preference for internal -ui-, and it may be anomalous. But in any case, the evidence (as yet probably incomplete for the Paviers) indicates that whenever internal -u- appears in B's work in unexceptional circumstances, it almost certainly duplicates a copy spelling, and when internal -ui- appears, it probably duplicates a copy spelling. The evidence that B follows faithfully the spellings sutors and especially suters is slight but consistent.

       

SUMMER	Spellings	Pq	FI
	-u- > -u-	1	1
	-o- > -u-	5+2L	4
	-o- > -o-	7+1L

The combined Pavier and Folio evidence is not abundant, but it is consistent in suggesting that B prefers the internal -u- spelling of this word and occasionally tolerates the -o- spelling when it appears in his copy. If Kable's figures are right, B's inclination to tolerate internal -o- spellings of copy may have declined during the years 1619 through 1623, but when an internal -o- spelling appears in his work, it should duplicate a copy spelling, even in long lines.

       

TRAITOR	Spellings	Pq	FI
	-ai- > -ai-	35	1
	-ay- > -ai-	24	1
	-ay- > -ay-	1

B's preference for the internal -ai- of this word has long been known. Only 1 repetition of an -ay- spelling in Pavier copy exists to support Kable's inference that internal -ay- in B's work duplicates his copy's spelling, but in view of B's strong preference for internal -ai- generally, the inference is not unreasonable. Since a final -our spelling of this word occurs only 1 time in B's known copy, we cannot define his habit of spelling the ending, even though he alters this 1 -our to -or and reproduces all of the -or spellings of his copy. B's treatment of the final -or/-our spelling alternative for other words is known to be inconsistent and inconstant; hence, conclusions about his treatment of final -or and -our for this word are out of the question.

         

VILE	Spellings	Pq	FI
	-le > -le		3
	-lde > -le	1
	-lde > -lde	2	1
	-ld > -lde	5

B's practice of spelling 'vile' is still not perfectly clear. Cauthen assigned the spelling vilde to B and vile to A, and the evidence of the Paviers could be interpreted as supporting this notion, even though Kable understandably

remains noncommittal about its implications. If the meagre statistics show anything, it is that B prefers vilde to vild, tolerates both vilde and vile, and may even slightly prefer vile to vilde. The chances are that vilde in B's Folio work derives from the vilde of his copy or from vild, under unexceptional circumstances.

               

WIDOW	Spellings	Pq	FI
	-dd- > -dd-	8
	-d- > -dd-	3	1
	-d- > -d-	4
	-dd- > -d-	2
	-w > -w	12	1
	-we > -w	1
	-wes > -wes	3

The case for widdow as B's habitual preference is not so strong as Walker and Kable indicate. It is true that the change of Folio copy's 1 widow to Widdow supports the notion of B's preference for internal -dd- that is suggested by the 3 changes of Pavier copy's -d to -dd-. But the 2 alterations of Pavier copy's widdow to widow yield a high random-variation factor of over 11%, if widdow is considered the preferred spelling. Indeed, widdow may have been the form B favored, but the spellings that we have as evidence—those set from known printed copy—may be the product of indifference, not settled preference. The chances are that they are not, but for now B's spelling practices for this word must remain undetermined.

             

WIND	Spellings	Pq	FI
	-de > -de	14	3 (sb.)
	-d > -de	20	2
	-d > -d	3
	-des > -des	2	2
	-ds > -des	2
	-ds > -ds	2

The evidence indicates that B prefers the spelling winde and occasionally reproduces a wind spelling of copy.[37] It is not clear that this pattern extends to the plural—that B prefers windes but will set winds under the influence of copy. The conclusion regarding the singular finds support not only in the evidence for this word alone, but also in B's treatment of the analogous 'behind' and 'blind'. When the spelling wind occurs in B's work, it almost surely duplicates the spelling of copy. (These conclusions would not be materially altered by the attribution of MV, 26-28, to B, since there he retains Q1's winde and alters its wind to winde.) The evidence that this pattern extends to the plural is less convincing because there is less of it.

           

YEAR	Spelling	Pq	FI
	-ea- > -ea-	20+4L	4+2L
	-ee- > -ea-	7+3L	1+1NL+3L
	-ea- > -ee-	1
	-re > -re	11	1
	-res > -res	15	4

B prefers internal -ea- and perhaps final -re and -res when spelling this word. The evidence for the -ea- preference is substantial and rather uniform throughout Pq and F1. The spellings in long lines are primary evidence because justification could not materially affect -e-/-a- variation, since these letters were equally thick. The single case of -ea- > -ee- in Pq yields a normal random-variation factor of 2.9% and need not imply a different compositor. Incidentally, B's preference for -ea- in this word is inconsistent with his habits for other analogous words. Since copy has only -re and -res spellings, we cannot be sure that B actually prefers these forms, but his invariable retention of them might suggest that he does.

         

YOUNG	Spellings	Pq	FI
	-o- > -o-	17 (16)	1
	-ou- > -o-	8 (9)	6
	-ou- > -ou-	28 (30)	2
	-o- > -ou-	11

B's treatment of this word is a matter of some debate. Probably by the time he sets the Folio he prefers the spelling yong but sometimes tolerates young when it appears in his copy. Alice Walker assigned the yong spelling to B and young to A, but Kable finds that in the Pavier Quartos B is largely indifferent to the internal -o-/-ou- distinction. Andrews accounts, without total success,[38] for the discrepancy by assigning to "B" an -o- preference and to G an -ou- preference. If Kable is right, the indifference B displays in the Paviers seems to have been abandoned by the time he set F1; if Andrews is right, there is no conflict between the Pavier and Folio evidence. Thus, in any case, the Folio evidence, although it is slender, indicates that B prefers the internal -o- spelling of this word in F1. A corollary to this inference is that in B's Folio work the chances are better than in his Pavier work that an -ou- in a short line which he set duplicates the spelling of his copy.

This review of the evidence now at hand on B's spelling of selected words has shown that some spellings hitherto considered useful in identifying his work or his copy are unreliable, and it has clarified the status of others as clues to his work or his copy. But, more important, it has confirmed

the existence of a large group of words that, with various degrees of reliability, can serve as evidence that B set a given page or evidence of the spellings in the copy that he used. Not all words about which we now have evidence could be discussed here, but examination of them indicates that about 100 spellings are useful clues to the presence of B in pages of F1 not yet certainly assigned to a compositor, and that another 100 are important because in B's work they normally derive from his copy.

The study of Compositor B's spelling habits when setting from known printed copy enables us to identify both preferred and derived spellings in his Folio work. When carefully used with due regard for the kinds of considerations outlined above, these spellings promise to help us decide whether or not B set pages in F1 which as yet are only tentatively assigned to a compositor and to attack the crucial question of the nature of the copy behind certain Folio plays. And since B was perhaps the least conservative compositor of the Folio, such study suggests that similar examinations of the spelling practices of other Jaggard compositors may well achieve comparable results.