3. Computer Applications

Electronic texts serve a very wide readership at a low cost. They belong to a worldwide virtual library in which every book exists in an unlimited number of loanable copies. The research uses of electronic texts are less well known, but text-analysis software can generate lists of textual variants between any two or more early editions, or between any early edition of a text and a modern commercial edition of that text. Electronic transcriptions can also be used to generate interactive concordances and word- and phrase-frequency lists. With these indexes, scholars can analyze word usage by text, date, speaker, printer, signature, forme, or indeed any other taggable characteristic of the text. Such flexible transformations of the original text can assist in thematic and bibliographical studies.

Thematic Analysis

What is SHAKE-SPEARE'S SONNETS about? Critics and biographers have answered this question differently for two centuries. This edition chooses to draw on electronic materials and tools not available readily to a paper-edition reader so as to show how they can help recover Shakespeare's English, from which we are estranged by a gap of nearly four centuries. Two excerpts from the sonnets are analyzed: the publisher's dedication, and neglected sonnet 105, in which Shakespeare tells us what he believes the poems mean. Electronic sources like the Early Modern English Dictionaries Database (EMEDD), help identify lost or now archaic senses of still-existing words as well as anachronistic meanings, ones that emerged only after Shakespeare's death. Electronic tools like TACT also help determine whether the second excerpt, sonnet 105, is faithful to the sonnets as a group.

The Dedication

The title of the book, "SHAKE-SPEARES SONNETS," is unusual. Thomas Thorpe, the publisher who is probably responsible for the title-page, does not call it "Sonnets and A Louers Complaint, by W. Shakespeare." Thorpe places Shakespeare's surname directly in the title, as if his reputation and authorship were so well-known that his first name was unnecessary, and as if the term "Sonnets" applies to everything in the quarto. As far as the maker of the title-page was concerned, "A Louer's Complaint"--at several hundred lines--belonged with the 154 sonnets with which the book begins. The word "sonnet" in this period need not characterize only 14-line poems but was on occasion used for a longer narrative poem (Bailey 1980). "The Louers Complaint," written in 47 seven-line stanzas, also resembles a series of 14-line poems.

Editors and critics are sharply divided on the meaning of the 12-line dedication by Thomas Thorpe ("T. T.").






            T. T.

Disagreement hinges on three terms that name persons: "BEGETTER", "Mr.W.H.", and "ADVENTVRER." Does Thomas Thorpe speak for himself and thank the person who "got" the sonnets for him, or does he address the person who was Shakespeare's inspiration and was addressed by many of the sonnets? The answer to that question turns on how we understand the word "BEGETTER": is it the one who created the poem, the one who created them in the mind of the poet, or the one who obtained them for Thorpe? (If "BEGETTER" means the poet himself, then "Mr.W.H." must be a typographical error for "Mr.W.S.") Finally, who is the "THE.WELL-WISHING. /ADVENTVRER", the one who wishes all happiness and immortality to the mysterious "BEGETTER"?

Nine of the 225,000 word-entries in the bilingual dictionaries and monolingual dictionaries forming the EMEDD have the word "begetter" in them. Five lexicographers (Thomas Thomas, John Florio, John Minsheu, Randle Cotgrave, and Thomas Blount), in works published from 1587 to 1656, use it as equivalent to Latin, Italian, Spanish, and French terms. The word appears in two senses. Eight entries place it among the following loosely synonymous English expressions: "engenderer," "creator," "maker," "father," "one from whence came the offspring or first original," "breeder," "progenitor," and "beginner." One entry only, Minsheu's "Conciliador, m. a winner, a procurer, a begetter, a reconciler", associates the word with terms for a solicitor or agent or factor or obtainer. It is very unlikely that Thorpe would have used "BEGETTER" not in the first sense but only in the second sense because the second is minor and rare, and appears to be derived from the more common first sense.

Five entries in the EMEDD by John Florio (1598), John Minsheu (1599), and Randle Cotgrave (1611) use the noun "ADVENTVRER" synonymously with "a hazarder", "a venturous fellow, a hardie man, one that putteth all vpon haphazard", and "one that freely and without compulsion, or charge goes to the warres; also, a free-booter, or boot-haler." The role of risk-taker seems best suited to the publisher Thorpe, who signs the dedication, than to anyone else. EMEDD entries for the phrasal verb "set forth" as "to publish." substantiate this. Thomas Thomas explains Latin "Edititius" as "Published, named, set forth, or to be set forth" in 1587; and Randle Cotgrave describes French "Publier" as "To publish, diuulge, manifest, proclayme, noyse abroad, lay open, set forth, make common, or knowne" in 1611. Thorpe risked his own money in printing a book of verse.

Early Modern English usage, then, allows the following paraphrase of the dedication: Thomas Thorpe, who ventures funds to set forth the book, wishes that the happiness and immortality promised by the poet to a Mr. W. H., who alone begot the poems in the writer. Mr. W. H. is the "onlie begetter of these unsuing sonnets": that is, he is the occasion of them.

If we ask, "who is Mr. W. H.?" we go beyond what the EMEDD can supply. Still, it is reasonable to think that Thorpe knew the answer to our question, and that he thought the buyers of the book would in all likelihood recognize the name too and take a greater interest in the quarto because of this identification. Reading the minimum information from this dedication, and doubting none of it (that is, emending none of the words because we believe they are wrong), E. K. Chambers and Samuel Schoenbaum, among the most conservative of Shakespeare's biographers, make Mr. W. H. out to be William Herbert, third Earl of Pembroke. He is one of the dedicatees of the Shakespeare's first folio. Young Herbert was being groomed by his family for the approval of Elizabeth I, who was looking for an heir, but Herbert as heir to the earldom had some failings. Although "exceedingly beloved of all men" (DNB 678), he refused to marry for some years, preferring to undertake affairs. The case for young Herbert as Shakespeare's young man suggests itself, if readers work towards an interpretation that does not emend the text.

Schoenbaum has one reservation: "it is doubtful that the obsequious Thorpe would insolently address a noble lord by the unhonorific `Mr.'" (219). In this period, however, the term "Mr" was an honorific, sometimes used to identify the heir to an estate. The sonnets are not about an earl, but about an earl-to-be, someone who could be called "Mr." If it was this earl who in 1609 was addressed in this dedication, would he not have been complimented that Thorpe's dedication desired his youth to live forever? A wish that Mr. W. H. have all happiness and immortality need not be insolent. As well, Thorpe might not have formally dedicated the quarto to the mature earl so as to avoid implying that Herbert contributed towards the publication of the book. Patronage is unlikely, given Thorpe's reference to an adventurer's risks.

Sonnet 105

The poems appear to fall into three groups. Although the poet's (or his muse's) voice utters them all and describes himself, the youth, the lady "colored ill" whom they shared (40-42, 133), and a rival poet who drew the youth's affections away from Shakespeare for a time, the first 154 sonnets form two sequences, once we segment them by person addressed. The first group addresses either the young man or the poet himself (1-126), and the second either the dark lady or the poet himself (127-54; cf. 144, 152). If we look at the order of the sonnets, these sequences cover the same chronological events twice. Shakespeare alludes to the affair between the youth and the lady mid-way through the first sequence (40-42), but the sonnets addressed to her by the poet--assuming that she is the same person--are bundled after this first sequence about the young man. The third member of this group is A Louers Complaint. This consists of 47 7-line stanzas, that is, 23 and a half 14-line "sonnets." Because Shakespeare speaks, not in his own person (or in a persona whom we are clearly intended to regard as himself), about people who live in a country that still allows women to become nuns, Complaint is not set in Renaissance England. It markedly contrasts with the two sequences of 14-line poems. Can computer-aided analysis of the content of Shakespeare's verse indicate why two such sequences are followed by a long narrative poem?

To begin, consider Shakespeare's own summary of what at least the first two-thirds of the short sonnets are about. Consider what sonnet 105 says.

Therefore my ver{s}e to con{{s}t}ancie con{fi}n'de,
One thing expre{{s}{s}i}ng,leaues out di{ff}erence.
Faire,kinde,and true,is all my argument,
Faire,kinde and true,varrying to other words,
And in this change is my inuention {s}pent,
Three theams in one,which wondrous {s}cope a{ff}ords.

Shakespeare says that his verse expresses only one thing in three themes, that each manifests itself dominantly in a single word, and that his invention consists of varying these three words with other, synonymous words. Shakespeare seems to adopt a formalist, text-centered approach. A theme manifests itself in repetition and variation of a few keywords. Shakespeare confidently says that his verse "leaues out di{ff}erence." If the sonnets written up to the 105th mean one thing only and do so transparently and unambiguously, he might have been nonplused by the subsequent literature disagreeing with him. His meaning is hardly transparent to us. On the other hand, Shakespeare can hardly be dismissed as unqualified to interpret his own writings.

Shakespeare is not writing a critical essay in sonnet 105, what we might call a theme. His language belongs historically to the study of logic, not to literary criticism, as we can see from the EMEDD. Its entries show that, if he uses the words "theme" and "argument" in contemporary senses, they mean the same thing: a logical proposition or argument.

Tesme: m. A theame, argument, position, proposition. (Randle Cotgrave, 1611)

THEME: Theme. A sentence or argument whereupon one speaketh. (John Bullokar, 1616)

Shakespeare's own syntactic variations already imply this synonymy. "Faire,kinde,and true" are equally all his "argument" and the "Three theams in one": thus clearly both the argument and the themes are one thing. Other early dictionaries take this explication further. They show that any argument (and thus also any theme) consists of at least one "proposition" in a logical syllogism or other proof.
THEME: a Theame. Proposition. (Henry Cockeram, 1623)

Syllogisme. An argument consisting of three parts, whereby something is necessarily proued, as thus; Euery vertue is honourable: Patience is a vertue; Therefore Patience is honourable. The first part of a Syllogisme is called the Proposition or Maior; the second, the Assumption or Minor; and the third, the Conclusion. (Henry Cockeram, 1623)

Syllogismus, mi. m. g. Quint. A perfect argument of three parts, inferring a necessarie conclusion: a syllogisme. (Thomas Thomas, 1587)

Note how Cockeram explains the term "theme" as a "proposition," and the term "proposition" in turn as the first part of a three-part "argument" called a syllogism. This proposition also functions as what was then called a logical definition. When Shakespeare says that, in expressing one thing, his verse "leaues out di{ff}erence", he is using the terminology of definitions that describe a thing, not a word (the notion of a lexical definition is a century in the future). The thing represented by "difference" belongs to logical definition, as Sir Thomas Elyot, the author of the first Latin-English dictionary, says:
Diffinitio, onis, a diffinition, or declaration of the nature qualitie or propre sygnification of a thynge by generaltie, specialtie, and difference. (Thomas Elyot, 1538)
By using the language of logical definition, Shakespeare situates his verse alongside encyclopedias, herbals, and scientific treatises. All are about things in the world; and the sonnets too focus on Shakespeare's love, both his feelings and the young man who is their object.

When Shakespeare says that his "inuention" consists of changing "Faire,kinde and true" into other words, he might be thought to imply that he thinks his verse is only about language, not about things in the world. However, note how Thomas Thomas translates the Latin word "Topice, es, f. g."

Invention or finding out of arguments: the arte of Inuention: a part of Logicke noting the places of inuention.
The context of sonnet 105 associates "invention," not with imagination, but with the making of what were understood, by Shakespeare's contemporaries, to be equivalent logical propositions. In sonnet 105 this three-in-one proposition defines his love as something in the world.

Critics have been quick to suspect playful irony in this poem--Shakespeare disclaiming idolatry while becoming increasingly guilty of it, for example, through his use of Trinitarian imagery ("the three in one")--but the poem is entirely truthful in what it says about the role these logical themes play in the other poems. Text-analysis software such as TACT verifies that Shakespeare repeats the three words "Faire,kinde,and true" elsewhere and varies them with other words. (An introduction to the use of TACT for literary analysis appears in Using TACT with Electronic Texts: A Guide to Text-Analysis Computing Tools, Version 2.1 for PC-DOS and MS-DOS.) The collocations generator Collgen, part of the TACT system, gives a starting point.

Collgen lists all repeating phrases within the quarto. These lists appear in appendixes 1 and 2. There are 1,742 repeating fixed phrases in the 14-line sonnets, and 83 in the Complaint. Three factors define a distinct fixed phrase: length, frequency, and span. The longest repeating phrases are called maximal phrases (e.g., "for I haue sworne thee faire", labelled "M" in Table 1), but they may contain subphrases that occur more frequently and that are called subordinate phrases (e.g., "for I haue sworne", labelled "S" in Table 3). Both types are distinct, the first because of its greater length, and the second because of its higher frequency. By defining the length in words beyond which a fixed phrase cannot go (ten words in this analysis), the span sometimes causes very long repeating phrases to appear as a series of shorter ones. One such phrase appears below.

Most phrases consist of function or closed-class words, such as articles, prepositions, and auxiliary verbs, combined with one content or open-class word. Only 30 of these repeating phrases from the 156 14-line poems, found below in Table 3, contain two or more content words, i.e., lexical nouns, adjectives, adverbs, and verbs. Aside from the couplet that identically closes sonnets 36 and 96, which has four content words, Shakespeare's three-themes-in-one, "Faire,kinde,and true," is the repeating phrase with the most content words (three). This describes the beloved, the subject of three of the four most repeated fixed phrases in this list, "deare loue", "sweet loue", and "sweet selfe."

Freq.Fixed PhraseType Locations
2antique penM 19.10; 106.7
2ayde my verseM 79-1-2; 86.8
2blacke nightM 27.12; 73.7
2but doe not so I loue thee in such sort M36.13-14; 96.13-14
as thou being mine mine is thy good report
2day by dayM 75.13; 117.4
2deare friendM 30.13; 111.13
2deare heartM 95.13; 139.6
5deare loueM 39.6; 72.3; 122.10; 124.1; 151.14
2doe thy worstM 19.13; 92.1
2face sweetM 93.10; 127.6
3faire kinde and true M105.9, 10, 13
3for I haue sworneS 147.13; 152.9; 152.13
2for I haue sworne thee faire M147.13; 152.12
2loues eyeM 148.8-9
2loues fireM 154.10, 14
3mens eyesM 20.8; 29.1; 81.8
2mine eye and heartM 46.1; 47.1
2sinfull earthM 146.1-2
5sweet loueM 29.13; 56.1; 76.9; 79.5; 93.10
5sweet selfeS 1.8; 4.10; 114.6; 126.4; 151.4
2sweet theefeM 35.14; 99.2
5ten timesM 6.8-10; 37.14; 38.9
2the better part of me M39.2; 74.8
2the wide worldM 19.7; 107.2
2thoughts of loueM 39.11; 47.8
4thy sweet selfeM 1.8; 4.10; 126.4; 151.4
2thy will and willM 135.1; 136.2
4true loueM 40.3; 61.11; 72.9; 107.3
2true sightM 148.2; 150.3
2two louesM 36.5; 144.1

Table 3: Repeated Fixed Phrases with at Least Two Content Words in the Sonnets

Shakespeare's three thematic words in the phrase "Faire, kinde,and true" appear in 69 of the first 154 sonnets. Usebase, the TACT interactive concordance program, shows that they collocate with one another strongly (that is, with high z-scores). The words "kinde" (nine occurrences) and "truly" (five occurrences) collocate with "fair", respectively, four and two times within a span of five words on either side. The words "faire" (48 occurrences) and "true" (39 occurrences) co-occur with "kind", respectively, six and three times. The word "kinde" co-occurs with "truth" three of nine times. These numbers indicate that the three words make up a highly associated, cohesive network, "Three theams in one" (as Shakespeare says). See Appendix 3 for the data.

In sonnet 105 Shakespeare does not tell us which words he chooses to vary these three words, but the following passages in other sonnets suggest the words "beauty", "bright", "gracious", "sweet", and "worth" collocate with the three theme words. By co-occurring with them, these additional words suggest themselves as alternates.

FRom faire{{s}t} creatures we de{{s}i}re increa{s}e,
That thereby beauties <f pi>Ro{s}e<f pr> might neuer die, (1.1-2)
Be as thy pre{s}ence is gracious and kind,
Or to thy {s}elfe at lea{{s}t} kind harted proue, (10.11-12)
But from thine eies my knowledge I deriue,
And con{{s}t}ant {{s}t}ars in them I read {s}uch art
As truth and beautie {{s}h}al together thriue
If from thy {s}elfe,to {{s}t}ore thou would{{s}t} conuert:
Or el{s}e of thee this I progno{{s}t}icate,
Thy end is Truthes and Beauties doome and date. (14.9-14)
Neither in inward worth nor outward faire
Can make you liue your {s}elfe in eies of men, (16.11-12)
Til what{s}oeuer {{s}t}ar that guides my mouing,
Points on me gratiou{{s}l}y with faire a{s}pe{ct}, (26.9-10)
So I , made lame by Fortunes deare{{s}t} {s}pight
Take all my comfort of thy worth and truth. (37.3-4)
OH how much more doth beautie beautious {s}eeme,
By that {s}weet ornament which truth doth giue,
The Ro{s}e lookes faire, but fairer we it deeme
For that {s}weet odor,which doth in it liue: (54.1-4)
Truth needs no collour with his collour {fi}xt,
Beautie no pen{s}ell,beauties truth to lay: (101.6-7)
Or mine eyes {s}eeing this,{s}ay this is not
To put faire truth vpon {s}o foule a face, (137.11-12)
For I haue {s}worne thee faire,and thought thee bright, (147.13)

The word "gracious" varies "kind," the word "worth" rings a change on "truth", and the words "beauty" and "sweet" correspond to "fair." These five words, when added to the three theme words Shakespeare makes explicit, appear in 119 of the sonnets, four-fifths of the first three sequences. (See Appendix 4 for the data and Appendix 5 for the distribution of these words.) If we add other allied words to these, such as "constancy", "deare", "pity", and "prettie", the number of the first 154 sonnets centered on Shakespeare's theme, the three-in-one love, would grow. Any interpretation of these poems should first consider Shakespeare's own reading of them because computer analysis of this electronic text bears out what he says in sonnet 105.

What then of Thorpe's last "sonnet," A Louers Complaint? The poet overhears a "fickle maid" (5) complaining of her betrayal by a male youth, at first alone and then to a third party, a "reverend" man (57) who also listens in. Within these two embedded audiences, she tells her outraged, frantic, despairing story of the affair, her initial rejection of the young man, his heated wooing, and his ultimate success. She quotes the youth's wooing speeches, in which she finds hypocrisy and lies. In defending the genuine love he feels, the youth alludes to another young lady he wooed, one who denied him his wishes and, overcome by her own sexual feelings, became a nun, abandoning him and herself unsatisfied rather than remaining vulnerable to more of the same anguish in the future. After narrating how she was won over by his words, and how the sexual seduction was managed, the young lady ends by saying that her complaint is all the worse because she realizes that, even knowing the outcome of the affair, she would voluntarily succumb to the youth's wooing again.

In many ways this tale forms a diptych with, a complementary image to, the first 154 sonnets. The youth of Complaint, like Mr. W. H., receives sonnets from (ultimately unwelcome) wooers (209-10). The same three thematic adjectives describe him as do Shakespeare's "young boy": "fair" (83, 311), "kind" (186, 311), and "true" (105 "a pride of truth", 246; but cf. 169 "untrue," 186). The maid's deceptive young man covers a "naked and concealed fiend" with "grace" (316-17), and Shakespeare's youth possesses "Lascivious grace" (40.13). The "fickle maid," like the dark lady, seduces or is seduced by the youth. Female and male sexuality, love, is powerful in both sonnets and Complaint. Love unfolds despite the threats of Time's scythe (12.13-14; 11-14) and despite its own poisons (129.7-8; 301). In the sonnets, Shakespeare he describes himself as old and ready to die (e.g., 73), not unlike the aged man in Complaint (62, 70). Passed over by both the young man and the dark lady, Shakespeare increasingly describes love as an outsider, and sometimes in the past tense. As the nun loved the youth and rejected his demands by taking vows, so Shakespeare in the sonnets candidly explains that no sexual love is possible between the youth and himself (2) and accepts that the youth will favour other suitors. Perhaps Complaint emerges from Shakespeare's understanding that a "fickle maid" such as the dark lady has a perspective that deserves a hearing. The second group of sonnets, about the dark lady, makes bracing reading for its misogyny, but Shakespeare closes the quarto, in "Complaint," with eloquent testimony about the genderlessness of suffering in love.

Shakespeare had enough of a reputation as a matchmaker by 1604 to have been asked by his own landlord to forward a suit aimed at acquiring a husband for his daughter (Schoenbaum 211). Shakespeare succeeded all too well, for a lawsuit later erupted between the lovers and her parents, and the poet was called into court to testify--with none too good a memory for some details--about the whole affair. Shakespeare also extricates himself from two love affairs in the book entitled SHAKE-SPEARE SONNETS. Age makes him a witness, indifferent to party.

Compositor Characteristics

Another application of a RET electronic edition is to detect whether several spelling systems exist in a work. MacD. P. Jackson twenty years ago, in a study of the 1609 quarto spellings and punctuation, assigned most pages to two compositors in George Eld's printing shop, men known as Walker A and Walker B from their earlier identification in Alice Walker's study of the 1609 quarto of Shakespeare's Troilus and Cressida, also printed by George Eld. Walker based her analysis (129) on the following five variant spelling pairs, the last four of which also appear in the 1609 quarto:

        Eld A                   Eld B 

        Cressid, Cressida       Cresseid, Cresseida
        1. Hellen               Helen
        2. shalbe, wilbe        shall be, will be
        3. ritch                rich
        4. els                  else

Any master printer might mark off a manuscript for typesetting by two men because gatherings could be printed completely, without delay, in a "relay" system ensuring that the press itself would be more or less constantly in use. For example, one compositor might set the inner and outer formes of one gathering. Then, while the printing of both sides of this quarto sheet were done, a second compositor could set the formes of a second gathering. Those would in turn be printed as the first compositor dismantled the first gathering's two formes, distributed its type into his cases, and set the two formes of the third gathering. Jackson identifies four different skeleton-formes from "inspection and measurement of the headlines" or running titles of the Sonnets quarto (2-3). If each compositor had one pair of formes (inner and outer), these results would suggest that two compositors set this book.

Could inconsistencies in spellings through the quarto arise from Shakespeare himself, not the compositors? He may well have changed his mind about how he would spell some words over the years that the sonnets were evidently written, but he would not likely spell a group of words consistently in two different ways at any one time. Any variation in the author's spelling would be expected to reveal itself, not on separate pages, but in groups of poems or in discrete works written at different times in his life. Bibliographical variation should arise from a bibliographical factor.

Jackson correctly assigns pages in the Sonnets quarto to Eld A or Eld B according to whether or not they had the five spelling variants that Walker identified. However, Jackson's evidence needs a closer examination. Because his initial list of assigned pages (1975: 3, 17-22) neglects two of Walker's pairs ("Hellen"/"Helen", "els"/"else"), he does not assign D4r and L1v to Eld A (but see Jackson, p. 4), and G4r to Eld B. As a result, Jackson also does not see that B1r, B2v, and F3r have spellings characteristic of both Eld A and Eld B and so cannot be firmly attributed to either compositor. Further, Jackson attributes F1r to B on the basis of an occurrence of "rich" not actually found on that page. These minutiae affect seven assignments. Table 4 includes corrections in square brackets.

Eld A   Eld B           Cited Evidence                  Observed Evidence

B1r     [B1r]           shalbe, wilbe                   wil be (A) [else (B)]
        B1v             shall be, will be               will be
B2v     [B2v]           shalbe, wilbe                   wilbe (A) [else (B)]
        B4r             rich                            else, rich
        C1r             rich                            rich
        C3r             rich                            rich
C4r                     ritch                           ritch
D1r                     shalbe, wilbe                   shal be
        D3v             rich                            rich
[D4r]                                                   [Hellens] 
        E2r             shall be, will be, rich         rich, shall be
        [F1r]           rich                            [not found]
        F1v             shall be, will be               shall be, will be (B)
        F2r             shall be, will be, rich         rich, shall be
        F2v             rich                            richly
F3r     [F3r]           shalbe, wilbe, ritch            ritches, wil be [else(B)]
        F4r             rich                            richer
F4v                     ritch                           ritches
G1r                     shalbe, wilbe, ritch            ritch, wil be
G1v                     shalbe, wilbe                   shalbe
G2r                     ritch                           ritch
        [G4r]                                           [else] 
        H2v             shall be, will be               will be 
        I1r             rich                            rich
        I3v             (not noted on p. 22)            rich
L1r                     ritch                           ritchest
[L1v]                                                   [els (A)]

Table 4: Compositor Assignments: Version 1

Jackson's study continues when he argues that, because "A uses Oh 9 times (including three instances on L1), O not at all; [and] B uses Oh only once, but O 8 times," this variant pair "provides further evidence for distinguishing the shares of Compositors A and B" (4).

        A            B        

    5.  Oh           O

Jackson correctly identifies all the pages on which these two variants are found (pp. 17-23), but if we accept the above revised assignments of pages to Eld A and Eld B, A uses Oh 11 times and O once (L1v), and B uses O 9 times and Oh once (F4r). Thus Jackson cannot assign further pages to Eld A and Eld B--as he goes on to do--only on the basis of their use of O and Oh, because both compositors use both spellings. The best one can say is that A uses Oh, and B O, about nine times out of ten.

Jackson at this stage "very tentatively" identifies 18 pages as by Eld A, and 37 as by Eld B, but he does not give a breakdown by pages. However, my assignments--see the following table--work out to be less optimistic: only 15 pages by A or A-like, only 35 pages by B or B-like, and (inconveniently) 5 pages by both. There does appear to be some correlation between Eld A and Eld B, and Oh and O. The presence of Oh or O confirms 11 previous assignments, seven by A (D1r, D4r, F4v, G1v, G2r, L1r, and L1v) and--oddly, because nearly three times as many assignments were for B as for A--only four by B (C1r, D3v, G4r, and I3v). However, on two occasions the Oh/O test muddies the previous results: F4r appeared to be by B but had Oh (supposedly an A variant), and L1v appeared to be by A but had both Oh and O (the latter a B variant). Two instances of contradictory evidence out of 13 total instances (11 confirmations, plus these two problematic ones) means that, 15 percent of the time, the Oh/O test yields results that undermine the test itself. Note that the very same problem occurs previously when the five Walker variants were applied to the Sonnets. They assigned nine pages as by A, and 14 pages as by B, but also three pages as by both: this result undermines the test about 11 percent of the time (three of 26).

There is a lesson in common sense to be learned here. Assignments, even tentative ones, should only be made on the basis of at least two variants. This caution has to be taken because, between 11-15 percent of the time, the two tests applied so far (the five Walker variants, and Oh/O) not only assign a page inconclusively but also call into question whether the variant is a reliable way of distinguishing Eld A and Eld B. For this reason, instead of making 55 assignments, Jackson might have reduced the number he had previously made to 14, seven by A and seven by B. These pages have at least two spellings favouring either A or B and not both together. See the Table 5.

        Page            Cited Evidence      Jackson         This Edition

        B1r             wil be, else            A               A and/or B
        B1v             will be                 B               B-like
        B2v             wilbe, else             A               A and/or B
        B3r             O                       B               B-like
        B3v             O (2)                   B               B-like
        B4r             rich                    B               B-like
        B4v             O                       B               B-like
        C1r             O (2), rich             B               B
        C1v             O (3)                   B               B-like
        C3r             rich                    B               B-like
        C3v             Oh                      A               A-like
        C4r             ritch                   A               A-like
        C4v             Oh                      A               A-like
        D1r             Oh (2), shal be         A               A
        D3v             O, rich                 B               B
        D4r             Hellens, Oh             A               A
        E1r             Oh (3)                  A               A-like
        E1v             O                       B               B-like
        E2r             rich, shall be          B               B
        E2v             O (3)                   B               B-like
        E3r             O                       B               B-like
        E3v             O (2)                   B               B-like
        E4r             O                       B               B-like
        E4v             O                       B               B-like
        F1r             shall be, will be       B               B
        F1v             O                       B               B-like
        F2r             rich, shall be          B               B
        F2v             rich                    B               B-like
        F3r             ritches, wil be, else A                 A and/or B
        F4r             Oh, rich                A and B         A and/or B
        F4v             Oh (2), ritch           A               A
        G1r             ritch, wil be           A               A
        G1v             Oh, shalbe, wilbe       A               A
        G2r             Oh, ritch               A               A
        G3v             O                       B               B-like
        G4r             O, else                 B               B
        G4v             Oh                      A               A-like
        H1r             O                       B               B-like
        H1v             O                       B               B-like
        H2r             O                       B               B-like
        H2v             will be                 B               B-like
        H3r             O (2)                   B               B-like
        H4v             O                       B               B-like
        I1r             rich                    B               B-like
        I1v             O (2)                   B               B-like
        I2v             O                       B               B-like
        I3v             O (4), rich             B               B
        I4r             Oh (2)                  A               A-like
        K2r             O                       B               B-like
        K2v             O                       B               B-like
        K4r             O                       B               B-like
        L1r             Oh (3), ritch           A               A
        L1v             O, Oh, els              A               A and/or B
        L2r             Oh (2)                  A               A-like
        L2v             O (5)                   B               B-like

Table 5: Compositor Assignments Version 2

The rest of Jackson's analysis (pp. 4-5) depends on our accepting his 55 assignments at face value. He assigns most of the remaining pages on the basis of eight further variant pairs "preferred" by one of the compositors, and observed by looking at alternate spellings occurring on his 18 Eld A pages and 37 Eld B pages.

It is less risky to limit testing, at this stage, only to those pages confirmed by two or more spellings so far to be Eld A (D1r, D4r, F4v, G1r, G1v, G2r, L1v) or Eld B (C1r, D3v, E2r, F1r, F2r, G4r, I3v). In Table 6, asterisks surround "A" and "B" when Jackson assigns one of these as a compositor's preferred spelling, and plus-signs surround changes in these preferences arising from my more restricted testing. Note that Jackson conflates frequencies for different words in two of the five variants.

Variant                Jackson 55               Confirmed 14                 Compositor 
Pairs                  Word-Freq.               Word-Freq.                   Pref. (Revised)

6. doest               1 A; 3 B                 4 (1 A: F4v)                 1 +A+
   do'st                                        1
   doo'st              1 B                      1
   doost               9 *B*                    12 (2 B F1r)                 2 *B*
   dost                5 *A*; 3 B               12 (3 A: D1r, F4v, G2r)      3 *A*

7. eie                 9 A                      8 (1 A: L1v)                 1 +A+
   eies                                         15 (2 A: F4v, L1v)           2 +A+
   eielids                                      1
   eye                 36 *B*                   35 (1 A: L1v;                1 A and/or 4 *B*
                                                4 B: C1r, G4r, [2] I3v)
   eyes                                         48 (6 B: C1r, F1r, F2r,      6 +B+
                                                [3] I3v)
   eye-lids                                     1

8. flowre                 \                     3 (2 A: F4v)                 2 *A*
9. houre               6 B| = 13 *A*            12 (1 B: E2r)                1 *B*
10.powre                  /                     6 (5 A: D4r, F4v, G1v, 
                                                2 L1v)                       5 *A*
   hower               6 B                      6 (2 B: C1r, D3v)            2 *B*
   flower(s)           2 A | = 14 *B*           12 (2 A: G1r, G1v;           1 B
                                                (C1r) 2 +A+ / 1 *B*
   power                /                       8 (1 B: E2r)                 1 *B*

11.gold                 | = 4 *A*; 13 B         2                            no evidence
12.old                  /                       18 (1 A: G1r)                1 *A*
   gould                | = 10 B                5 (1 B: C1r)                 1 B
   ould                 /                       6 (1 B: C1r)                 1 B

13.:quatrain end       strongly *A*             17 A (2 D1r, 5 D4r,          17 *A* / 3 B
                                                4 F4v, 2 G1r, 1 G1v,
                                                2 G2r, 1 L1v); 3 B
                                                (C1r, F2r, I3v)
   ,quatrain end       slightly *B*             1 A (D1r); 14 B (3 C1r,      1 A / 14 +B+
                                                3 D3v, 1 E2r, 1 F1r, 2 F2r,
                                                2 G4r, 2 I3v)

14.tong-               7 A; 2 B                 9 (3 A: F4v, 2 G2r)          3 A
   toung-              13 B                     11 (1 B: G4r)                1 B

15.catchword: poem no. 7: 4 A; 3 B              7 (1 B: C1r)                 1 B
   catchword: 1st word 6: 6 B                   6

16.space/"," ends poem 13: 13 B                 14 (1 B: E2r). The others    1 B
                                                are: <space> 2 C2v, H1r;
                                                ",": B1v, B4r, B4v, C4r,
                                                D2r, D2v, 2 E4v, F1v, G3v.
                                                A colon ends one on D4v.
   punctuation ends poems 

17.noe                    \                     8 (3 B: C1r, 2 D3v)          3 B
18.loe                    | = 11: 10 B          4 
   no                                           74 (5 A: D1r, 4 G2r; 9 B:    5 A / 9 B
                                                2 D3v, F2r, G4r, 5 I3v)
   lo                                           3

Table 6: More Variant Spellings

At this point no account is taken of small differences in the centering of the sonnet numbers, a further variant cited by Jackson (p. 5).

Four of Jackson's new variants, this limited test shows, should join Alice Walker's original five variants as spellings distinguishing between Eld A and Eld B. Eld A prefers "doost," "powre," "old," and "tong-"; Eld B, in contrast, prefers "dost," "power," "ould," and "toung-". Neither Eld A nor Eld B uses both alternate spellings in each case. Two other distinguishing pairs proposed by Jackson are unclear: Eld A uses both "flowre" and "flower(s)," and Eld B both "noe" and "no." There is not enough evidence to associate the use of a sonnet number in a catchword, or of a space or comma to end the last line of a sonnet, with Eld B.

However, this limited test identifies two further possible distinct pairs: Eld A uses "eie(s)" three times as often as it does "eye(s)" (a result absent from Jackson's data) and also employs colons "strongly" (as Jackson says) at the end of quatrains, whereas Eld B uses "eye(s)" (as Jackson says) and commas at the end of quatrains (four times as often as it does colons, not one of Jackson's findings). However, because each of Eld A and Eld B uses both commas and colons at quatrain's end, and on four pages both punctuation marks appear, no firm distinction emerges.

Nonetheless, all occurrences of the first four variants (not just occurrences in the 14 test pages) can be used to update the compositor assignments. The "eie(s)" / "eyes" variant may also be added because they overlap on only one page, L1v, which also contained both "O" and "Oh" and may be a special case. In my opinion, colons and commas at the end of quatrains overlap too many times to be trustworthy discriminants.

Asterisks prefix changed lines in Table 7. Bars (|) around "A" and "B" indicate Jackson's assignments (p. 6).

Page    Cited Evidence                  Jackson                 This Edition 
                                        (pp. 6, 23-24)

A1r                                     unassigned              no evidence
A1v                                     unassigned              no evidence
A2r                                     unassigned              no evidence
A2v                                     unassigned              no evidence

["Sonnets" begin here]

B1r     wil be, else, old               2 |A|; 1 B              A and/or B
B1v     will be, dost (2), doost (2),
        ould                            2 A; 4 |B|              A and/or B [formerly B-like]
B2r                                     |B|                     no evidence
B2v     wilbe, else                     1 |A|; 1 B              A and/or B
B3r     O                               |B|                     B-like
B3v     O (2)                           |B|                     B-like
B4r     rich                            |B|                     B-like
B4v     O, old, tong-                   1 |B|; 2 A              A and/or B [formerly B-like]
*C1r    O (2), rich, ould, eye(s) (2),
        ould                            |B|                     B
C1v     O (3), ould, tong-              1 A; 4 |B|              A and/or B [formerly B-like]
C2r                                     |A| "slight"            no evidence
C2v     old                             1 A; |B|                A-like [formerly no evidence]
C3r     rich, old                       1 A; 1 |B|              A and/or B [formerly B-like]
C3v     Oh                              |A|                     A-like
C4r     ritch                           |A| "slight"            A-like
C4v     Oh, dost (2), old               |A|                     A [formerly A-like]
*D1r    Oh (2), shal be, dost           4 |A|                   A
D1v     doost                           |B|                     B-like [formerly no evidence]
D2r     doost                           |B|                     B-like [formerly no evidence]
D2v     doost                           |B| "slight"            B-like [formerly no evidence]
D3r                                     |B|                     no evidence
*D3v    O, rich                         |B|                     B
*D4r    Hellens, Oh, powre              |A|                     A
D4v                                     |A| "slight"            no evidence
E1r     Oh (3), old                     |A|                     A [formerly A-like]
E1v     O, dost (2)                     |B| "slight"            A and/or B [formerly B-like]
*E2r    rich, shall be, power           |B|                     B
E2v     O (3)                           |B|                     B-like
E3r     O, ould, toung- (2)             |B|                     B [formerly B-like]
E3v     O (2)                           |B|                     B-like
E4r     O                               |B|                     B-like
E4v     O, old (2)                      2 A; 1 |B|              A and/or B [formerly B-like]
*F1r    shall be, will be, doost (2),
        eyes, doost (2)                 |B|                     B
F1v     O, doost, toung- (2)            |B|                     B [formerly B-like]
*F2r    rich, shall be, eyes            |B|                     B
F2v     rich, toung-                    |B|                     B [formerly B-like]
F3r     ritches, wil be, else           |A|                     A and/or B
F3v     dost, old, tong-                |A|                     A [formerly no evidence]
F4r     Oh, rich                        1 A; 1 |B| "slight"     A and/or B
*F4v    Oh (2), ritch, dost, powre,
        eie, tong-                      |A|                     A
*G1r    ritch, wil be, old              |A|                     A
*G1v    Oh, shalbe, wilbe, powre        |A|                     A
*G2r    Oh, ritch, dost, tong- (2)      |A|                     A
G2v     old (A)                         A; |B| "slight"         A-like [formerly no evidence]
G3r     old, toung-                     1 A; 1 |B|              A and/or B [formerly no evidence]
G3v     O, old (4)                      4 A; |B|                A and/or B [formerly B-like]
*G4r    O, else, eye, toung-            |B|                     B
G4v     Oh                              |A|                     A-like
H1r     O                               |B|                     B-like
H1v     O                               |B|                     B-like
H2r     O                               |B|                     B-like
H2v     will be, dost, ould             1 A; 2 |B|              A and/or B [formerly B-like]
H3r     O (2), power (2), ould          |B|                     B [formerly B-like]
H3v     toung-                          |B|                     B-like [formerly no evidence]
H4r     power                           |B|                     B-like [formerly no evidence]
H4v     O                               |B|                     B-like
I1r     rich, doost                     |B|                     B [formerly B-like]
I1v     O (2), power (2), old, tong-,
        toung-                          2 A; 5 |B|              A and/or B [formerly B-like]
I2r     toung- (2)                      |B|                     B-like [formerly no evidence]
I2v     O, doost (2)                    |B|                     B [formerly B-like]
I3r     dost (2), powre, tong-          |A|                     A [formerly no evidence]
*I3v    O (4), rich, eye(s) (5)         |B|                     B
I4r     Oh (2), powre                   |A|                     A [formerly A-like]
I4v                                     |A|                     no evidence
K1r                                     |B|                     no evidence

["Complaint" begins here]

K1v                                     unassigned              no evidence
K2r     O, doost                        |B|                     B [formerly B-like]
K2v     O, power, old                   1 A; 2 |B|              A and/or B [formerly B-like]
K3r     tong-                           A                       A-like [formerly no evidence]
K3v     power, old                      1 A; 1 B                A and/or B [formerly no evidence]
K4r     O                               |B|                     B-like
K4v                                     unassigned              no evidence
L1r     Oh (3), ritch                   |A|                     A
*L1v    O, Oh, els, powre (2), 
        eie (2), eye                    6 |A|; 2 B              A and/or B
L2r     Oh (2)                          A [A and B]             A-like
L2v     O (5)                           [|B|]                   B-like

Table 7. Compositor Assignments: Version 3

The additional evidence gives these results:

        A               12: C4v, D1r, D4r, E1r, F3v, F4v, G1r, G1v, G2r, I3r, I4r, L1r
        A-like          7
        B               14: C1r, D3v, E2r, E3r, F1r, F1v, F2r, F2v, G4r, H3r, I1r, I2v, I3v, K2r
        B-like          18
        A and/or B      17
        no evidence     12

Mainly on this basis, Jackson assigns 22 pages to Eld A, and 49 to Eld B, out of a total of 80 pages. In my opinion, only 20 pages are unambiguously assignable to A or B. Their 12:14 ratio contrasts with Jackson's 22:49. Almost as many pages as can be assigned, 17, include spellings so far associated with both A and B. One lone spelling links 25 pages with one of the compositors, but, as we can see from the number of pages that have multiple spellings associated with both A and B, a single variant cannot be trusted. Jackson perhaps is overconfident in the "O" / "Oh" discriminant, which on six pages conflicts with the "ould" / "old" spelling (B4v, C1v, E4v, G3v, I1v, K2v).

Few firm assignments can be made, on this evidence, because many variant spellings turn up in pages already associated with both Eld A and Eld B. When both spellings in a variant pair occur together, and there is no additional orthographic evidence, assignments cannot be decided on the basis of that pair. We need more words that can be associated with Eld A and Eld B. Evidently the two compositors were not always consistent. Many pages will have to be assigned to A and B on a preponderance of discriminant spellings. Thus the more variant pairs we can find, the better. We can only find these words by creating a database of all variant spellings in the quarto and then by testing them for the first mark that they may be A- or B-specific, i.e., that they never occur on the same page. Then, if certain of these pairs fall into two networks, these networks can be tested for overlap with the existing Eld A and Eld B spellings. If overlap occurs, then we will probably have more pages that can be assigned firmly to A or B.

Jackson does not list dominant variant spellings from 1609 or test to see if they were distributed in such a way that the different forms of each variant pair never fell on the same page (and thus might be candidates for compositor-specific spellings).

My data collection begins by making a textual database of Sonnets with Text Analysis Computing Tools (TACT). The signature of the page on which each word occurs is part of its default reference citation (in Makebase). Then an alphabetical word-list for the text is produced with TACTFreq, and a word processor is used to reduce the list to possible variant spellings, and to print out that list of variant spelling pairs. With the printed list in hand, I produced KWIC displays of each spelling in question with Usebase. In this way I verified that each occurrence of a variant spelling arises only from orthography (rather than from the presence of a semantically different form), and finally added the signatures on which each variant spelling appears to the variant list.

The next step is to create a database of these variants. I imported the corrected list of genuine spelling variants into Excel, a spreadsheet program. Each spelling occupies the first column of each row, under the heading 1609 Word. The first row has a heading for each column in the spreadsheet. The other headings are: Explanation, Frequency, Signatures, No. Overlaps, Walker A or B, and then each signature in text-order, from A1r to L2v. Explanation speculates about the possible reason for the variant (e.g., a Shakespearean spelling described in Partridge), Frequency the number of occurrences of a variant spelling, Signatures the number of different pages on which this spelling appears, No. Overlaps the number of times the current variant spelling appears on the same page with its other spelling, and Walker A or B which Walker compositor the current variant spelling might be associated with (e.g., "1 Walker A" would mean "one occurrence of the current spelling occurs on a page where one of the five distinguishing spellings adduced by Walker also occurs). Then follow columns for each page in the quarto with a signature header. A plain-text file of the Excel database forms Appendix 15 of this edition.

This spreadsheet includes over 300 spelling variants. Many variants of the same word, however, occur on the same page and for that reason are not good candidates as spellings typical of only one of the two compositors. While many other spellings never overlap, they may be expected not to coincide because the forms often occur only on a single page each. The 35 variant spelling pairs that seldom or never overlap in the same page, and that both appear at least on two pages, are more plausible candidates as discriminants. See Table 10 for these pairs. Nine of them are already tested. Here are the results of further testing against the 26 Eld A and Eld B pages so far assigned, i.e.,

        A (12)          C4v, D1r, D4r, E1r, F3v, F4v, G1r, G1v, G2r, I3r, I4r, L1r
        B (14)          C1r, D3v, E2r, E3r, F1r, F1v, F2r, F2v, G4r, H3r, I1r, I2v, I3v, K2r

The numbers following each spelling are the frequency of the spelling.

Spelling No.            Difference                      Confirmed 26 
                                                        Test Group

19. audit (2)                                           A 1 (L1r)
    audite (2)          additional final "e"            B 1 (H3r)

20. back (4)                                            B 2 (2 I2v)
    backe (3)           additional final "e"            B 1 (H3r) 

21. beene (4)                                           3 A (E1r, 2 G1r)
    bene (3)            medial "ee"/"e" variation 

22. belou'd (2)                                         1 A (I4r)
    beloued (3)         elision of "e" in preterite form    1 A (F3v)

23. boast (4)                                           1 A (L1r); 1 B (F2v)
    bost (2)            "oa"/"o" variation 

24. cheekes (2)                                         1 B (F2r)
    cheeks (2)          additional final "e" 

25. child (4) 
    childe (4)          additional final "e"            1 B (C1r) 

26. eternitie (2)                                       1 B (F1r)
    eternity (2)        final "ie"/"y" variation        1 B (H3r)

27. euil (2) 
    euill (2)           doubling of final "l"           1 B (I2v)

28. far (6)                                             1 A (F4v)
    farre (11)          additional final "e"?           4 B (F1v, F2r, I1r, I2v)

8. flower (4) 
   flowre (2)           final "er"/"re" variation 

11. gold- (3) 
    gould- (5)          medial "o"/"ou" variation 

29. greef- (3)                                          1 B (D3v)
    grief- (5)          medial "e"/"ie" variation       2 A (D1r, F3v)

30. groane- (2) 
    grone- (4)          medial "oa"/"o" variation       1 B (D3v)

31. hold- (12)                                          1 A (G2r), 6 B (F1v, 2 F2v, 2 I1r, I2v)
    hould (2)           medial "o"/"ou" variation       1 B (H3r)

9. houre- (12) 
   hower- (6)           final "er"/"re" variation 

32. kind (5)                                            2 B (E3r, I2v)
    kinde (5)           additional final "e"            1 A (F4v) 

18. lo (2) 
    loe (4)             additional final "e" 

33. louers (5) 
    lovers (5)          medial "u"/"v" variation        2 B (E2r, H3r)

34. maid (2) 
    maide (2)           additional final "e" 

35. merit (2) 
    merrit(s) (4)       medial doubling of "r"          1 A (I4r); 1 B (I2v)

36. mind (6)                                            1 B (D3v) 
    minde (9)           additional final "e"            3 A (E1r, 2 I4r); 2 B (F1r, G4r)

5. o (28) 
   oh (14)              additional final "h" 

12. old (15) 
    ould (6)            medial "o"/"ou" variation 

10. power (6) 
    powre (7)           final "er"/"re" variation 

37. receau- (4) 
    receiu- (7)         i.e., forms of "receive";       2 A (D1r, L1r); 1 B (I1r)
                        medial "ea"/"ei" variation 

3. rich- (10) 
   ritch- (6)           final "ch"/"tch" variation 

38. song (2) 
    songe (2)           additional final "e"            2 A (G1v, G2r) 

39. sun (4) 
    sunne (9)           additional final "e"?           1 A (E1r); 2 B (C1r, I3v)

40. themselues (2)                                      2 A (D4r, F4v)
    them-selues (5)     additional medial hyphen 

14. tong- (9) 
    toung- (10)         medial "o"/"ou" variation 

41. vert- (7)                                           3 B (2 F1v, I2v)
    virt- (3)           i.e., forms of "virtue";        1 A (D4r)
                        medial "e"/"i" variation 

42. wealth (5)                                          1 A (F3v)
    welth (2)           medial "ea"/"e" variation       1 B (E3r)

43. whom (8)                                            3 A (D1r, F3v, I4r)
    whome (3)           additional final "e" 1 B (I2v) 

44. witnes (2) 
    witnesse (2)        final "s"/"sse" variation

Table 10. 35 Distinguishing Spelling Pairs

This variation seems to have a linguistic basis two out of three times: note the addition of final "e" (12 times), medial "o"/"ou" variation (4 times), final "er"/"re" variation (3 times), and medial "oa"/"o" variation (twice). Seven of these pairs discriminate between Eld A and Eld B: nos. 19 ("audit" / "audite"), 28 ("far" / "farre"), 29 ("grief-" / "greef-"), 32 ("kinde" / "kind"), 41 ("virt-" / "vert-"), 42 ("wealth" / "welth'), and 43 ("whom" / "whome").

Two variant pairs with a lower frequency also discriminate between Eld A and Eld B.

45. subdew- (2)                                         1 A (L1r)
    subdu- (2)          medial "ew"/"u" variation       1 B (G4r)

46. tis 1 A (G1r)
    'tis/t'is/ti's      use of apostrophe               3 B (E2r, 2 F2v) 

Table 11. Two Lower-frequency Spelling Pairs

Assignments are affected as follows in Table 12:

Page    Cited Evidence                  Jackson                         This Edition
                                        (pp. 6, 23-24)

A1r                                     |unassigned|            no evidence
A1v                                     |unassigned|            no evidence
A2r                                     |unassigned|            no evidence
A2v                                     |unassigned|            no evidence

["Sonnets" begin here]

B1r     wil be, else, old               2 |A|; 1 B              A and/or B
B1v     will be, dost (2), doost (2),
        ould, audit                     3 A; 4 |B|              A and/or B
B2r                                     |B|                     no evidence
B2v     wilbe, else                     1 |A|; 1 B              A and/or B
B3r     O, whom (2), kind 2A;           2 |B|                   A and/or B [formerly B-like]
B3v     O (2)                           |B|                     B-like
B4r     rich, vert-                     |B|                     B [formerly B-like]
B4v     O, old, tong-                   1 |B|; 2 A              A and/or B
C1r     O (2), rich, ould, eye(s) (2),
        ould                            |B|                     B
C1v     O (3), ould, tong-, ti's        1 A; 5 |B|              A and/or B
C2r     whome                           (2) 2 B; |A| "slight"   B-like [formerly no evidence]
C2v     old, greef-, far (2)            3 A; 1 |B|              A and/or B [formerly A-like]
C3r     rich, old                       1 A; 1 |B|              A and/or B
C3v     Oh                              |A|                     A-like
C4r     ritch, grief-, tis              |A| "slight"            A [formerly A-like]
C4v     Oh, dost (2), old, wealth       |A|                     A
D1r     Oh (2), shal be, dost, grief-,
        whom                            |A|                     A
D1v     doost, grief-                   1 A; 1 |B|              A and/or B [formerly B-like]
D2r     doost, farre                    |B|                     B [formerly B-like]
D2v     doost, whom                     1 A; 1 |B| "slight"     A and/or B [formerly B-like]
D3r     audite, grief-, farre           1A; 2 |B|               A and/or B [formerly no evidence]
D3v     O, rich, greef-                         |B|                     B
D4r     Hellens, Oh, powre, virt-       |A|                     A
D4v                                     |A| "slight"            no evidence
E1r     Oh (3), old                     |A|                     A
E1v     O, dost (2), farre (2)          1 A; 3 |B| "slight"     A and/or B
E2r     rich, shall be, power, t'is     |B|                     B
E2v     O (3), vert-                    |B|                     B [formerly B-like]
E3r     O, ould, toung- (2), kind       |B|                     B
E3v     O (2), vert-                    |B|                     B [formerly B-like]
E4r     O                               |B|                     B-like
E4v     O, old (2), wealth, far 4       A; 1 |B|                A and/or B
F1r     shall be, will be, doost (2),
        eyes, doost (2)                 |B|                     B
F1v     O, doost, toung- (2), 
        vert- (2), farre                |B|                     B
F2r     rich, shall be, eyes, farre     |B|                     B
F2v     rich, toung-, 'tis (2)          |B|                     B
F3r     ritches, wil be, else, virt-,
        whom                            4|A|; 1B                A and/or B
F3v     dost, old, tong-, wealth, 
        grief-, whom                    |A|                     A
F4r     Oh, rich, wealth, vert-         2 A; 2 |B| "slight"     A and/or B
F4v     Oh (2), ritch, dost, powre,
        eie, tong-, kinde, far          |A|                     A
G1r     ritch, wil be, old, tis         |A|                     A
G1v     Oh, shalbe, wilbe, powre        |A|                     A
G2r     Oh, ritch, dost, tong- (2)      |A|                     A
G2v     old, kinde (5)                  A; |B| "slight"         A [formerly A-like]
G3r     old, toung-                     1 A; 1 |B|              A and/or B
G3v     O, old (4), whom, 'tis          5 A; 2 |B|              A and/or B
G4r     O, else, eye, toung-,
        subdu-                          |B|                     B
G4v     Oh, tis (3)                     |A|                     A [formerly A-like]
H1r     O, virt-                        1 A; 1 |B|              A and/or B [formerly B-like]
H1v     O, kind, far                    1 A; 2 |B|              A and/or B [formerly B-like]
H2r     O, tis                          1 A; 1 |B|              A and/or B [formerly B-like]
H2v     will be, dost, ould, far        2 A; 2 |B|              A and/or B
H3r     O (2), power (2), ould,
        audite                          |B|                     B
H3v     toung-, whome                   |B|                     B [formerly B-like]
H4r     power, farre (2)                |B|                     B [formerly B-like]
H4v     O, kinde                        1 A; 1 |B|              A and/or B [formerly B-like]
I1r     rich, doost, farre              |B|                     B
I1v     O (2), power (2), old, tong-,
        toung-                          2 A; 5 |B|              A and/or B
I2r     toung- (2), farre, 'tis         |B|                     B [formerly B-like]
I2v     O, doost (2), vert-, whome,
        kind, farre                     |B|                     B
I3r     dost (2), powre, tong-          |A|                     A
I3v     O (4), rich, eye(s) (5)         |B|                     B
I4r     Oh (2), powre, whom             |A|                     A
I4v                                     |A|                     no evidence
K1r                                     |B|                     no evidence

["Complaint" begins here]

K1v                                     unassigned              no evidence
K2r     O, doost                        |B|                     B
K2v     O, power, old, tis              2 A; 2 |B|              A and/or B
K3r     tong-, subdu-                   1 A; 1 B                A and/or B
K3v     power, old, kinde               2 A; 1 B                A and/or B
K4r     O                               |B|                     B-like
K4v     greef-, kind (2)                3 B                     B [formerly no evidence]
L1r     Oh (3), ritch, audit, 
        subdew-                         |A|                     A
L1v     O, Oh, els, powre (2), 
        eie (2), eye, wealth
        subdew-                         8 |A|; 2 B              A and/or B
L2r     Oh (2)                          A [A and B]             A-like
L2v     O (5), whom, kinde              2 A; 5 [|B|]            A and/or B [formerly A-like]

Table 12. Compositor Assignments: Version 4
The additional evidence gives these results: A 15: C4r, C4v, D1r, D4r, E1r, F3v, F4v, G1r, G1v, G2r, G2v, G4v, I3r, I4r, L1r A-like 2 B 22: B4r, C1r, D2r, D3v, E2r, E2v, E3r, E3v, F1r, F1v, F2r, F2v, G4r, H3r, H3v, H4r, I1r, I2r, I2v, I3v, K2r, K4v B-like 4 A and/or B 28 no evidence 9

Nineteen discriminating variants thus identify 37 pages, just under one-half of all pages in the quarto containing poetic text, as being by Eld A and Eld B. These assignments are supported by 2-8 spellings each, or an average of 3.5 variants each. On each of these 37 pages, all spellings attest to one and the same compositor. On six pages we only have one variant, which is too little for assignment purposes. Nine pages lack evidence entirely. Conflicting evidence exists on 28 pages. Perhaps in these instances the copytext or copytexts on occasion offered A-spellings and B-spellings, and on these occasions neither Eld A nor Eld B consistently converted all those spellings to their own systems. Any printing shop that allowed its compositors to impose their own individual spelling during printing would hardly encourage them to be consistent in their changes. It is not impossible that some pages were shared by the two compositors.

The assignments in this edition thus differ from Jackson's. I attribute G2v to Eld A (not Eld B), and K4v to Eld B (not to Eld A), each based on two variants. Jackson assigns some pages, without explanation, on which I have found conflicting evidence. He leaves nine pages unassigned; I leave 43 unassigned for a reason.

The 37 assignments show something about Eld's printing shop. Clearly Eld A and Eld B did not split the setting of a gathering by inner and outer formes. For example, Eld B set F1r-F2v, and Eld A G1r-G2v. Neither did one of them take a forme or a gathering all to himself. B set I1r (outer forme), I2r (inner forme), I2v (outer forme), and I3v (inner forme), whereas A set I3r (outer forme) and I4r (inner forme). Although both compositors worked on the quarto more or less from start to finish, sometimes only one of them set type for an extended time, and at other times both laboured simultaneously. Eld does not seem to have divided the copytext up systematically between the two. He must have just left it to them to get on with the work as expeditiously as circumstances allowed.