University of Virginia Library

"THIS INSTANCE WILL NOT DO": GEORGE STEEVENS, SHAKESPEARE, AND THE REVISION(S) OF JOHNSON'S DICTIONARY
by
R. Carter Hailey [*]

In 1773 Samuel Johnson published revised editions of his monumental Dictionary of the English Language (4th ed., folio),[1] and, with George Steevens, of The Plays of William Shakespeare. The working relationship of the two men dated from at least 1765 when Johnson had published his first Shakespeare edition, a project to which Steevens contributed forty-nine textual and usage notes.[2] Johnson in turn encouraged and praised Steevens's 1766 edition of Twenty of the Plays of Shakespeare, Being the whole Number printed in Quarto During his Lifetime, or before the Restoration. While the 1773 Shakespeare was an ostensible collaboration, both the impetus for the edition and the bulk of the work were provided by Steevens. Johnson contributed some eighty new notes to the commentary, while Steevens established the text, wrote General Observations for The Comedy of Errors and Much Ado About Nothing (which had been lacking in the 1765 edition), and added hundreds of explanatory notes, many citing parallel passages from other Renaissance dramatists.[3] Conversely, although Steevens is nowhere

identified as a contributor to the revised Dictionary, Arthur Sherbo has discerned an element of cross-fertilization between the two projects by identifying a number of instances in which Steevens's notes to the Shakespeare edition resulted in Johnsonian revisions in the Dictionary. [4]

Steevens's scholarly abilities were eminently suited to both textual criticism and lexicography, and contemporary praises for his intelligence, industry, and wide acquaintance with antiquarian literature—especially that of English Renaissance drama—are plentiful.[5] Steevens's scholarly interests coincided with the aims of both the Dictionary and the Shakespeare editions, and his rationale for reprinting the Shakespeare quartos recognizes the affinity between the two projects:

The last sentence might serve as a concise statement of Johnson's methodology in compiling the Dictionary, and appears to have been designed both to flatter Johnson and to demonstrate Steevens's own grasp of the project's nature and scope. If the reference was not sufficiently transparent, Steevens continued:

This better understanding is implicitly the product of Johnson's efforts in producing the laborious catalogue and judicious glossary that is the Dictionary of the English Language. By reprinting the Shakespearean quartos Steevens contributes to the larger cultural project: he not only offers the public "the poet's first thoughts" (14) but also helps to preserve the literary heritage by providing the raw materials from which dictionaries can be made. Steevens has thus in a few lines insinuated himself into a productive scholarly

association with Dr. Johnson, William Shakespeare, and the art of lexicography.

John Middendorf has speculated that "from time to time evidence may still emerge to add to our knowledge of the possible extent of Steevens's anonymous assistance" to the 1773 Shakespeare edition (129). It is now possible to expand this statement to include Steeven's activity in assisting Johnson with the Dictionary during "The Year of Revision" (Sherbo, "1773," 18). I will argue that George Steevens, dubbed the "Puck of Commentators" by a contemporary for his literary quarrels and hoaxes, played a much more active role in the revision of the Dictionary than has hitherto been suspected, a role that, without acknowledgment, continued long past Johnson's death.

I

Two sets of materials offer a fascinating glimpse into Johnson's working method in preparing the revisions for the fourth edition of the Dictionary. [6] The first is the Sneyd-Gimbel copy—so called for its previous owners—now held by the Beinecke Library at Yale University (shelfmark MS Vault Johnson). This copy contains first edition (1755) sheets from the letter A through the entry for Pumpion (with several gaps), corrected and annotated in the hands of Johnson and an amanuensis and interleaved with more than 1,800 slips containing additional illustrative quotations. In The Making of Johnson's Dictionary, Allen Reddick mounts a persuasive argument that the Sneyd-Gimbel materials represent a preliminary stage in Johnson's preparations for the revised fourth edition.[7] A heretofore perplexing set of materials is a second partial and annotated copy of the dictionary, British Library C.45.k3,[8] which consists of mixed first and third edition sheets from "A"Jailer," interleaved, with manuscript notes for revisions to the text. The sheets for letter "B" are from the first edition (1755) and are annotated by Johnson and an amanuensis; the remaining letters, "A" and "C"-"I/J," are third edition sheets (1765) annotated by George Steevens,[9] whose work is

markedly different from Johnson's. Johnson reorders entries, adds new lemmas and new senses of words, and deletes some quotations while adding others. (The fourth edition would eventually include some 3000 new illustrative quotations.) Steevens on the other hand, although he offers numerous suggestions for additional illustrative quotations, concentrates on usage notes and textual corrections to Shakespeare quotations.[10] How these apparently disparate materials might have come together has been an enduring mystery. Prior to their identification of Steevens's hand, Sledd and Kolb described the problem:

While some of the mystery must remain, a reconsideration of the evidence will show that Steevens almost certainly annotated the BL copy in order to assist Johnson in revising for the fourth edition of the Dictionary, that at least some of these annotations resulted in printed revisions in the fourth edition, and that the missing third edition sheets of the letter B—whose place is taken by the first edition sheets annotated by Johnson—served as the printer's copy for that letter in F4.

Reddick argues convincingly that the first edition sheets in the BL copy— those annotated by Johnson—represent a second stage in the revision process and were intended as printer's copy for the revised edition, but were for some reason never used. How they might have been mislaid or made otherwise unavailable is unknown and, absent new documentary evidence, probably unknowable. But, in the event, Johnson was forced hastily to reedit the letter "B,"[12] with the result that the printed revisions in "B" are of a very different order and magnitude from those found in the rest of the Dictionary. There are for example fewer than half the number of new quotations that on average appear in other letters of the revised Dictionary, and authors like Bacon, Spenser, and Browne whose works frequently supply new illustrations in adjacent portions of the revision do not appear at all in the new illustrations in "B." Additionally, "B" is the only letter in the revision which is shorter (by four pages) than in previous editions, "because very little new material was added, while long quotations and references (as in other sections of the text) were deleted or abridged."[13]

Reddick also asserts that "the corrections and additions" from the portion of the BL copy annotated by Steevens "were never incorporated into the Dictionary," indicating that "Johnson probably did not have access to them" (193). And, after offering several possible scenarios to explain the odd juxtaposition of Johnson's and Steevens's work in the BL copy, including one that implies some chicanery on Steevens's part, Reddick concludes that Steevens annotated the third edition sheets only after Johnson's death, at the request of Charles Marsh, a fellow member of the Society of Antiquaries who had purchased the materials when Johnson's library was dispersed in 1785 (193194). But my own collation of Steevens's annotations in the BL copy against the printed text of the revised fourth edition of the Dictionary (1773) indicates that a number of Steevens's suggestions were in fact adopted by Johnson. While he made no systematic use of the annotations—only a small proportion of the more than 760 changes Steevens proposed appear to have influenced the text of F4—in some forty-four instances Johnson either accepts Steevens's suggestion outright, or alters an entry in a way which suggests that Steevens's note was the probable source of the revision.[14] Why Johnson admitted only a small number of Steevens's suggestions and what procedure he used to incorporate those that were accepted remain obscure. But a substantial number of Steevens's proposed revisions—around 310 of the 760— are Shakespeare quotations which offer supplementary examples for words and senses which were already illustrated in F1. While one of Johnson's purposes for the revision was to provide additional illustrations, Shakespeare was already heavily represented in the word list and only one of the printed Steevensian revisions adds a Shakespeare quotation to an existing sense of a word.[15] (For additional authorities in F4 Johnson relied heavily on the Bible and Milton, at least in part because concordances were available.) Additionally, Steevens for some reason annotated a copy of the third edition (1765), while Johnson had prepared his revisions (rightly) from a copy of the first edition (1755). Dozens of corrections suggested by Steevens in the BL copy are to corruptions that had crept into the text in the transmission from first to third editions, and were thus unnecessary to an edition set from annotated first edition sheets. (They amply attest, however, to Steevens's perspicacity

as a proofreader.) Thus if the unnecessary corrections and Shakespeare quotations are excepted from the total, Johnson adopted in some form roughly 10% of Steevens's suggestions.[16] I say "in some form" because, as will be seen below, some of the printed revisions vary to a greater or lesser degree from Steevens manuscript notes. This is not altogether surprising considering that, whatever Johnson's method for selecting instances to incorporate in F4, Steevens's notes, corrections, and quotations would need first to be transcribed on the interleaves of the printer's copy, probably by an amanuensis, and then set in type by compositors. Variation could have crept in at each stage. The Steevensian revisions that Johnson did choose to print take a variety of forms: some correct textual errors, others rectify problems with usage and etymology, provide additional illustrative quotations, or introduce new senses of words. A few cases add new lemmata to the word list.

Several of the revisions correct small textual errors in Shakespeare quotations and thereby attest to Steevens's meticulous attention to detail and prodigious memory, for no Shakespeare concordance was available until 1787. Under To Abut, for example, F1's reading "Perilous the narrow" from Henry V is corrected in F4 to "The narrow perilous," and under To Croak 2, F4 corrects F1's "The raven himself not hoarse" to "is hoarse" in the first line of the Macbeth quotation. Steevens's expertise is not limited to the works of Shakespeare; in the F1 entry for Almond Tree, the first line of a Fairy Queen quotation reads: "Like to an almond tree, you're mounted high." Steevens, thoroughly familiar with Spenser's characteristic archaizing, noted that "you're mounted" should properly read "ymounted." In F4 the reading is simply "mounted high," with "you're" deleted. The original error had probably arisen from a compositor's failure to recognize Spencer's use of the archaic past tense "ymounted" and thus mistaking the "y" for the abbreviation "yr."

A more complex example occurs in the entry for Cannibal, where Steevens alters the F3 reading of an Othello quotation from "It was my hent to speak" to "hint" (my emphasis). In F1 the word had read "bent," an apparent transcriber's or compositor's error that had been corrected to "hent" in F3; since F4 was set from a copy of F1, without Steevens's suggestion the reading might well have remained "bent" in the later editions. This series of readings effectively illustrates the interconnection between the Dictionary and the Shakespeare edition. The Shakespeare First Folio (1623) and Second Quarto (1630) both read "hint," while the First Quarto (1622) reads "hent." William Warburton, whose 1747 edition was Johnson's source for Shakespeare quotations,[17] had accepted "hent," appending a note defining it as "use, custom." In collecting "authorities" for the first edition of the Dictionary

(1755) Johnson had uncritically accepted Warburton's reading, but when he came to edit Shakespeare ten years later, he rejected the Second Quarto reading, adopting instead the First Folio's "hint." Johnson prints Warburton's note on the reading, followed by one of his own: "Hent is not use [sic] in Shakespeare, nor, I believe in any other author. . . ."[18] The 1773 Johnson/ Steevens edition also prints both, followed by an additional note from Steevens which indicates that he was fully aware of the conflicting readings which had given rise to his suggestion in the BL copy. (The same quotation had been used to illustrate Antre, and again Steevens questioned the reading "hent," but the correction was never made.)

Among the corrections suggested by Steevens and accepted by Johnson, roughly a quarter involve the addition of new illustrative quotations.[19] In several instances Steevens provides a quotation for a word not otherwise illustrated. Finding that Coigne "1. A corner." lacked an example in F1, Steevens suggests a quotation from Macbeth: "—no jutting frieze, / Buttress or coigne of vantage, but this bird / Hath made his pendant bed or procreant cradle." Reading further, Steevens found that Johnson had used this very quotation (and another from Shakespeare) to illustrate the same sense, but under the spelling Coin. There Steevens notes: "Coigne is not spelt coin in the ancient Editions or in Dr. Johnson's." In F4 both quotations are moved from Coin to Coigne, and the anglicized lemma, though retained, is not illustrated. For Dexterous, not illustrated in F1, Steevens offers "For both his dextrous hands the lance could wield. Pope's Homer." Johnson duly adds the quotation in F4 as "For both their dext'rous hands their lance could wield. Pope."; the source of the error "their" for "his" is unclear. Similarly, for Downhil adj., Steevens proposes "and the first stage a downhill green-sword yields. Congreve." In F4, Johnson prints "And the first steps a downhil greensward yields. Congreve."[20] This instance meets only the minimum requirement for illustrations: as Johnson acknowledged in the "Preface" to the Dictionary, "Many quotations serve no other purpose than that of proving the bare existence of words." But ideally quotations would be "useful to some other end than the illustration of the word" by the additional performance of an aesthetic or didactic function. "It is not sufficient that a word is found unless it be so combined as that its meaning is apparently determined by the tract and tenor of the sentence . . ." (F1, B2v-C1r).[21] That is, quotations should preferably not only illustrate the existence of a particular word or sense, but additionally, through contextual clues, assist in defining

it. Steevens was throughly familiar with the aims and methods of compiling "authorities" as Johnson repeatedly calls them. Under To Defeat, for example, Steevens argues that: "To defeat. Perhaps in the following instance means to alter, to change. Defeat thy favour with an usurped beard. Shakespeare." In F4 Johnson, satisfied that the conjunction of "defeat" and "usurp" fulfilled the "tenor and tract" test, adopts Steevens's suggestion by adding "to undo" to the definition and illustrating it: "Defeat thy favour with usurped beard. Shakespeare."

The revision of the lemma Eisel again attests to the complex interactions between Steevens's editorial activities and his work on the Dictionary. In F1 Johnson had defied eisel as "Vinegar; verjuice; any acid" and illustrated it with a quotation from Hamlet: "Woo't drink up eisel, eat a crocodile? / I'll do it." In the BL copy, Steevens crosses through the quotation, adds a note which insists that "The passage from Hamlet is controverted," and offers several alternatives: "Whilst like a willing patient, I will drink / Potions of Eisel 'gainst my strong infection. The Complaint a Poem attributed to Shakespeare";[22] "—Remember therewithal / how Christ for the fasted with Eisel & gall. Sir Thomas More"; "Kned in with Eisel strong and egre; / And there to she was lene and megre. Chaucer's Romaunt of the Rose." In F4 the Hamlet passage is dropped, replaced by the More quotation which comes from his translation of Pico della Mirandola's life of Johan Picus. Atypically, Johnson expanded the passage Steevens had suggested: "Cast in thy mind / How thou resemblest Christ, as with sower poison, / If thou paine thy taste; remember therewithall / How Christ for thee tasted eisel and gall. Sir. Thomas More." Johnson corrects Steevens "fasted with" to "tasted," while introducing the reading "poison" for "potion," an easy transcription or compositorial error through metathesis and a mistaking of long s- for t- (if indeed "potion" had been the spelling of his source[23]).

It had been Steevens himself who had controverted the reading eisel in the 1773 Johnson/Steevens edition (10:321-322 n. 5), where the line is given as "Woo't drink up Esil? Eat a crocodile?" Steevens appended a long note on the reading "Esil," quoting Theobald's edition of 1733 in which "Eisel"—a reading accepted by nearly all modern editors—was first conjectured as the original of Q2's Esill and F1's Esile (Q1 had read "vessels"). Theobald rejects earlier commentators' notion that Esill/Esile is the name of a river in Denmark, contending that "there is none there so called."[24] Steevens, however, argues otherwise, citing an unnamed "old Latin account of Denmark" which includes "the names of several rivers little differing from Esil, or Elsill,

in spelling or pronunciation." Steevens goes on to allege that "no" authours later than Chaucer or Skelton make use of eysel for vinegar: nor has Shakespeare employed it in any other of his plays." His comments are curious in light of the passages he had suggested in the BL copy which, as I have shown, must have been annotated prior to the publication of the revised 1773 editions of both the Dictionary and the Plays of William Shakespeare. The More quotation is roughly contemporary with Skelton, but, although "Eisel" does not in fact appear elsewhere in Shakespeare's plays, it is found in Sonnet 111: "Whilst like a willing pacient I will drinke, / Potions of Eysell gainst my strong infection."[25] This is of course the very passage Steevens had cited in the BL copy, where its source is wrongly identified as "The Complaint a Poem attributed to Shakespeare." How Steevens could have known the line when annotating the BL copy, but not when writing the textual note for the Shakespeare edition, is curious. But Steevens clearly uses the Dictionary to substantiate his own reading of the passage, suppressing counter-evidence by advising Johnson to delete the Hamlet quotation. And Steevens's note seems almost certainly to have prompted Johnson's revision of the entry, although the means by which a longer version of the More quotation was used is unclear.

In several instances Steevens suggests that an additional sense of a word be defined and illustrated. For Affection, Steevens's note on the interleaf is keyed by an "x" to follow the eighth sense of affection and offers: "affection. Affectatation [sic] / x No sallets in the lines &c this [sic] / might indite the Author of Affection. Shak. Hamlet" and "—witty without affection Shak. Love's L.L." And in the 4th edition Johnson adds a 9th sense: "9. It is used by Shakespeare sometimes for affectation. / There was nothing in it that could indict the authour of affection. [26] Shakespeare." At To Clamour, Steevens observes of the second Shakespeare quotation—"Clamour your tongues, and not a word more"—that "This instance proves just the contrary to what was intended by it. To clamour bells is to silence them by raising them upright." In F4 Johnson adds a second sense of the word: "2. In Shakespeare it seems to mean actively, to stop from noise" and illustrates it with the suggested quotation. (It is worth noting in passing that "clamour" is thus one of the rare words like "sanction," "cleave," and "let" that can mean the opposite of itself.)

When Johnson accepts a suggestion, the revised text is generally straightforward

enough. For example under Caisson Steevens notes: "Likewise the floating frame in which the Piers of Bridges are built," and in F4 Johnson adds a second sense of the word: "2. A wooden case in which the piers of bridges are built within the water." In several instances, however, a suggestion is accepted, but with a bit of editorial commentary which indicates that Johnson remains unconvinced by Steevens's argument. In F1 Johnson, citing Pope, derives Argosy from the name of Jason's ship Argo. Steevens's note reads: "Argosy is as probably derived from Ragusa formerly famous for Pirates (hence Shakespeare's Argosine) but now tributary to the Porte." In F4 the etymology is altered to read: "Derived by Pope from Argo, the name of Jason's shop; supposed by others to be a vessel of Ragusa or Ragosa, a Ragozine, corrupted." Here one may detect a hint of disapprobation in Johnson's "supposed by others." In F1 the word Comart n.s. is not defined, although it is illustrated by a quotation from Hamlet. Steevens crosses through the quotation and on the interleaf writes: "The word was inserted by Warburton, but no Dict. that I have seen gives any example of it." Johnson nevertheless retains both lemma and quotation in F4, but modifies the definition: "This word, which I have only met with in one place, seems to signify; treaty; article from con and mart, or market."[27] That one place had been in the Second Quarto of Hamlet (1604-5).[28] Warburton's Shakespeare (1747)—the edition Johnson used for collecting quotations for the Dictionary—had accepted the reading "comart" instead of the First Folio's "cou'nant." Johnson's edition of 1765 follows the First Folio in reading "cov'nant," although Warburton's note defending Q2's "comart" is included; the 1773 Johnson/ Steevens edition reads "covenant." Warburton's note is appended, but is now followed by Steevens's rebuttal which echoes his note in the BL copy: "I can find no such word as comart in any dictionary" (10:151). Again the intersection of the two projects is manifest.

In several instances Johnson adds a lemma to F4 that had been suggested by Steevens in the BL copy. On an interleaf Steevens proposes "Chargeful a word employed by Shakespeare," although, uncharacteristically, he does not cite a specific passage. (It is in fact used only once, in Comedy of Errors 4.1.29.) Johnson adds the lemma in F4, defining it as "Expensive, costly. Not in use." Here Johnson accepts Steevens's suggestion almost grudgingly, as if to say, "Well yes, it may well be found in Shakespeare, but I didn't include it because it's not current." Johnson's tone may be in part a reaction to the brusqueness of some of Steevens's notes. Under To Chance, for example, Steevens crossed

through the first Shakespeare quotation, adding a note that "This instance will not do," and under Commiseration, where again a quotation is marked through, the note reads simply, "This is a bad instance." Steevens tone should not, however, be taken as evidence that his annotations were not intended for Johnson's eyes. Reddick, for example, argues that the annotations were probably made after Johnson's death since in several of the annotations "Steevens seems to speak of Johnson as a third party (i.e. `Compliments in this passage is the same as compliments, and is so explained by Dr. Johnson in his Shakespeare'), rather than to address the lexicographer directly, [which] seems to suggest that they were not executed for Johnson's use" (193-194). But Steevens's use of the third person results from the habitual manner in which both Johnson and Steevens referred to each other in their notes to the Shakespeare edition. A search of the Chadwyck-Healey electronic text of the second Johnson/Steevens edition (1778) reveals that, in the textual notes, Johnson refers to "Mr. Steevens" twenty-five times and that Steevens refers to "Dr. Johnson" on one hundred and forty occasions.[29] While the great majority of Johnson's notes cite "Mr. Steevens" with approbation, on at least one occasion (cited above), Johnson observes that "Mr. Steevens appears to have forgot our author's 111th sonnet."

Steevens is much more apt to contradict Johnson and his tone is reminiscent of that in the BL annotations:

In exchange for Steevens's invaluable assistance, Johnson was apparently willing to tolerate a bit of brusqueness.

In addition to the previous examples of corrections and additions inspired by Steevens's notes in the BL copy, there are also a number of instances where Johnson apparently responds to Steevens's suggestions not by revising but by deleting the passage in question.[30] One major stage of the Dictionary revision had, in fact, been the deletion of superfluities and the truncation or compression of entries to allow space for new material to be included. The extent of Johnson's editorial excisions is apparent when considering that although some 3,000 new quotations were added, the fourth

edition is only eighteen pages longer than the first.[31] In one or two instances it could be argued that the deletion would have been made independent of Steevens's suggestion. Under Aroynt, for example, Steevens proposes an alteration to the quotation which was not accepted, but also notes that it had been wrongly attributed to Shak. King Lear, rather than to Macbeth. In F4 it is attributed simply to Shakespeare. Since Johnson frequently truncates attributions in F4, this particular instance may not have been a response to Steevens's note, but no space is saved by the shortening. In other cases, however, the deletion was apparently prompted by Steevens's notes. In the first Shakespeare quotation illustrating Addition 4, the third line in F1 reads "The sway, revenue, execution of th'last." Steevens indicates that the last words should read "the rest." In the fourth edition "of th'last" is simply excised, although no space is saved. (This is a rare instance where the attribution is expanded in F4, to Shakes. King Lear from F1's Shakespeare.) The primary sense of Chamber 1 is defined as "An apartment in a house; generally used for those appropriated to lodging." Steevens crosses through the first Shakespeare quotation, from Richard III: "Welcome, sweet prince, to London, to your chamber." His note states: "Chamber, in this instance, signifies London anciently called Camera regea." In F4 the quotation is deleted. In F1 Johnson defines Commere as "A common mother" and illustrates it with a quotation from Hamlet: "As peace should still her wheaten garland wear, / And stand a commere 'tween their amities." Steevens crosses through the entry, and notes on the interleaf: "There is no such word in Shakespeare. Commere was instead of comma by Warburton." Johnson drops the entry entirely in F4.

One final example again indicates the cross-fertilization between the Dictionary and the Shakespeare edition. For the word Hail n.s. Steevens marked for deletion the illustration "As thick as hail / Came post on post. Shakespeare's Macbeth." This had been the reading, first proposed by Rowe, of Warburton's edition (1747) from which Johnson had chosen his Shakespeare illustrations. And Johnson had printed the line thus in his 1765 Shakespeare edition, but appended a note reading:

In the 1773 edition Johnson, perhaps prompted by Steevens's corroborative note in the BL copy, adopts his own suggestion and prints "As thick as tale / Came post with post."

As the preceding series of examples demonstrate, Steevens's sedulous labors in annotating the BL copy were utilized by Johnson, although not systematically so. The strong suggestion is that Steevens worked closely with Johnson over a period of many years on both the Shakespeare editions and the Dictionary. [32] And while Allen Reddick has brilliantly reconstructed the several stages in Johnson's lexicographic development, this new awareness of Steeven's deep involvement provides fresh clues about the nature and provenance of the British Library copy. The first edition sheets were certainly, as Reddick suggests, prepared by Johnson as printer's copy for the letter "B." But why were they never used for that purpose, and how did they come to be inserted between the third edition sheets annotated by Steevens? How Johnson or the printer might have mislaid the material must remain mysterious, but it is possible to say with some assurance what was used in its stead. It is evident that where the Johnsonian material for the letter "B" is inserted in the BL copy, Steevens's annotated third edition sheets for the same letter are absent; it would be difficult to imagine a situation in which Steevens would have annotated only the letters "A" and "C-I/J." I consequently suspected that when the printer's copy for the letter "B" was mislaid, Johnson, rather than beginning from scratch, took what was nearest to hand—Steevens's annotated third edition sheets—hastily augmented them with a few additional revisions, and sent the copy off to the printers. This scenario would account for the different nature and magnitude of revisions in the printed text of the letter "B" as compared to the rest of the fourth edition of the Dictionary.

To confirm my suspicions I collated the first thirty pages of the letter "B" in the first, third, and fourth editions and through recension determined that, unlike the rest of the fourth edition, "B" was set from third rather than first edition sheets. The following is a sample of the more than sixty instances of shared F3-F4 readings in the collated pages:

                     

Location	F1 reading	Shared F3/F4 reading
To Baa (Sidney)	treble baas⁁ for	treble baas, for
Babble (Milton)	mere	meer
Baby 1 (1st Locke)	plumbs	plums
Baccivorous (definition)	A devourer of berries	Devouring berries
Balance (Sir John Davies)	burden'd	burden
Ballad-singer (definition)	employment it is	employment is
Bank n.s. 4 (South)	Their	There
Barbarian n.s. (etymology)	only foreign	only a foreign
Barm (Shakespeare)	drink to bear	drink bear
Base n 7 (Dryden)	trebles squeak	trebles queak

The first three are among the many examples where the readings of (relatively) indifferent accidentals in F3/F4 agree against F1. In serving to determine the copy from which F4 was set, they are valuable only in their sheer

frequency, for in individual instances it is always possible that the compositors of separate editions coincidentally produced the same divergence from copy. More suggestive is the entry for Baccivorous, because it represents a rare instance in which a substantive change had been made in the third edition. Since the word is an adjective, Johnson had properly altered the first edition reading "A devourer of berries" to "Devouring berries" for the third edition. Had the letter "B" in the fourth edition been set from first rather than third-edition sheets, this change would in all likelihood have been lost. Most conclusive, however, are instances where F3/F4 agree in manifest error, where the F1 reading had been correct; it is highly improbable that the compositors and proofreaders of separate editions would repeatedly and coincidently commit and allow to stand identical variance from copy.[33] In order to assure myself that the letter "B" had been set throughout from third edition sheets, I additionally collated six pages near the end, from 3R1r through 3S2r, and found a dozen instances where again the text of F3/F4 agreed against F1. Since shared error is more conclusive than shared correction, the most telling example occurs under Bunch 4, where a line from Fairy Queen appears correctly in F1 as "A bunch of hairs discoloured"; in both F3 and F4, "discoloured" becomes the nonsensical (in context) "discover'd." Additional support is found in the shared misspellings of several headwords: F1's "Bivalvular" appears in both F3 and F4 as "Bivalvula," and F1 "Bobbin" as F3/F4 "Bobin." The letter "B" in F4 was thus demonstrably set not from F1, as is the rest of the Dictionary, but from F3.

As noted above, the new illustrative quotations found in the letter "B" are both fewer in number and different in source from those found in adjacent letters. It is thus useful to compare the new illustrations which appear in the printed text of the letter "B" with the suggestion for new quotations Steevens had made in the BL copy. In "B," as with most of Vol. I, the largest number of new illustrations in F4 come from the Bible (there are 25), due largely to the availability of Cruden's concordance. But second to the Bible are Shakespeare and Pope with 16 new quotations each: this compares with 21 new Shakespeare quotations and 80 new Pope quotations in the whole of Volume II.[34] The over-representation of Shakespeare is particularly striking, and provides

an additional indication that Steevens's annotated F3 sheets for "B" must have served as printer's copy for the revised edition.

Additionally, I have collated approximately the first quarter of the letter "B" in F1 and F4 and found a comparatively large proportion of the sorts of changes Steevens had proposed in the BL copy (and that, on the basis of the BL copy for "B," Johnson himself rarely made).[35] There are, for example, small corrections to Shakespeare quotations under To Bait v.a. l, To Bait v.n., Bank 1, and To Beguile; a longish usage note following the Shakespeare quotations which illustrates Bankrupt adj.; new "authorities" (Shakespeare of course) for Baa and Beggarly; and the new lemmas To Bass and Beef-witted and a third sense of Bastard, all illustrated with Shakespeare quotations. Remarkably, there are two instances in the letter "B" of F4 where Johnson cites Steevens by name. The first, discovered during collation, occurs in the etymology for Beef-Eater n.s.: "from beef and eat, because the commons is beef when on waiting. Mr. Steevens derives it thus: Beefeater may come from beaufetier, one who attends the sideboard, which was anciently placed in a beaufet. The business of the beef-eaters was, and perhaps is still, to attend the king at meals." A search of both F1 and F4 of A Dictionary of the English Language on CD-ROM[36] revealed a second example in F4 under Bumbast: "falsely written for bombast; bombast and bombasine being mentioned, with great probability, by Junius, as coming from boom, a tree, and sein, silk; the silk or cotton of a tree. Mr. Steevens, with much more probability, deduces them all from bombycinus."[37] Nowhere else in either F1 or F4 does Johnson mention Steevens. The confluence of circumstantial evidence supports my original conjecture that Steevens's annotated third edition sheets for the letter "B" served as the basis, and ultimately as printer's copy, for Johnson's revision of this letter in F4. At some point after the letter "B" was printed off, the missing sheets Johnson originally had prepared must have been located and were inserted in their proper alphabetical order in Steevens's annotated 3rd edition copy, replacing what had now become copy for that letter. That explains why the BL copy is made up of mixed third and first edition sheets, annotated by Steevens and Johnson respectively.

II

Following Johnson's death in 1784, Steevens continued his work on Shakespeare. In 1785 a new edition of the Johnson/Steevens Shakespeare

appeared, and although much of the editorial work had been entrusted to Isaac Reed, Steevens contributed new notes to all of the plays.[38] Steevens would later resume direct editorial control and in 1793 published another edition, in large part as a response to Malone's edition of 1790. In 1800 Steevens died while working on still another edition, which Reed published in 1803 as containing "the last improvements and corrections of Mr. Steevens."[39] What may be surprising is that Steevens's involvement with the Dictionary also continued long after Johnson's death.

The copyright held by the original proprietors of the Dictionary (Longman, et al.) had expired[40] and, while their fifth edition folio—essentially a reprint of the fourth—was still selling at £4.10.0, two substantially cheaper competing editions began to appear in the fall of 1785. James Harrison offered a folio reprint of the first edition supplemented by a life of the author in one hundred weekly numbers at six pence each, while John Fielding published the first English quarto edition in forty-eight weekly numbers at a shilling apiece.[41] Longman quickly responded:

Before the end of the year the proprietors published new and competitively priced quarto (Q6) and folio (F7) editions of the Dictionary, both in weekly numbers, which incorporated Johnson's "last improvements." That the changes were mostly minor and numbered only about two hundred and fifty went understandably unremarked. But such was the cachet of these "final revisions" that Harrison began to advertise, falsely, that his edition was also based on the Reynolds copy. (Since Harrison's was actually a reprint of the

first edition of 1755, it failed in fact to include even those revisions Johnson had made for the fourth edition of 1773.) If the number of surviving copies is any indication, Longman and the other proprietors, with their concurrently published sixth edition quarto and seventh edition folio, easily prevailed over the competition.

Equally unremarked was that the proprietors had procured "a corrector for the press [who took] some independent and fairly intelligent care that the sixth and seventh editions should be accurately printed."[43] It is highly probable that this corrector was George Steevens whose diligence in this regard was renowned, Malone having called him "one of the most accurate correctors of the press . . . in the world."[44] The evidence consists of changes which had long before been suggested by Steevens in the BL copy but which appear for the first time in the 6th edition.[45] Several are relatively minor corrections in punctuation and spelling. At Aim 5 in the last line of the quotation from Henry IV Part II "intreasur'd" is emended to "intreasured," restoring a metrically necessary syllable. At Air 5 (which had been sense 4 in F1), in the first line of the Lear quotation "vengencies" is altered to "vengences." At Anatomy 5, in the quotation from Comedy of Errors "a needy hollow-ey'd" gains a needed comma. And at To Deck 3, in the second line of a quotation from Spenser, "Fit to adorn the head, and deck the dreary tomb," "head" is corrected to "dead." Such seemingly insignificant variants might easily be laid to the vagaries of compositorial practice if we did not have Steevens annotations in the BL copy to indicate that they were in fact deliberate corrections.

A more substantive example is found at To Case 2 "To cover as a case" which Johnson illustrates with a quotation from Steevens's favorite play: "Then comes my fit again, I had else been perfect; / As broad, and gen'ral, as the casing air, / To saucy doubts and fears. Shakesp. Macbeth." Johnson had, as he so often did, abridged the quotation. In the BL copy Steevens marked the third line for deletion—not to save space, but to preserve sense—but the line is retained in F4. In the 1773 Johnson/Steevens the passage reads in full: "Macb. Then comes my fit again: I had else been perfect; / Whole as the marble, founded as the rock: / As broad, and general, as the casing air: / But now, I am cabin'd, cribb'd, confin'd, bound in / To saucy doubts and fears. But Banquo's safe." Steevens certainly understood Johnson's method of compressing quotations; but in the "Preface" to the Dictionary Johnson himself acknowledged that "The examples are too often injudiciously truncated," and this is just such an example. The excision of the second line creates no difficulty, but without the fourth line the final phrase is rendered nonsensical. In Q6 the quotation, although further truncated, for the first time makes passable sense. The range of Steevens's literary expertise is further illustrated by his handling of Direful. In F1-F4 an illustration from Pope read: "The

wrath of Peleus' son, the direful spring / Of all the Grecian woes, O goddess, sing." In the BL copy Steevens crosses through the quotation, and offers a different version: "Achilles' wrath, to Greece the direful spring / Of woes unnumber'd, heav'nly Goddess sing." As it happens, both versions are authorial. The former had been the original opening of Pope's translation of the Iliad, and the lines were printed thus in the quartos and folio editions of 1715 and the first and second octavo editions of 1720. The latter version represents a revision which first appeared in the third octavo edition of 1732.[46] While Steevens may have been unaware of the earlier version and believed Johnson's rendering to be erroneous, I find it more likely that he preferred the revised lines as representing Pop's final thoughts. Whatever the case, in Q6 the lines read: "Achilles' wrath, to Greeks the direful spring / Of woes unnumber'd, heavenly Goddess! sing."

The evidence of Steevens's hand is apparent in Q6, but by 1785 the BL materials were in the possession of Charles Marsh, who procured them when Johnson's library was sold. Since Steevens and Marsh were both Fellows of the Society of Antiquaries, however, they must have been well acquainted, and it seems plausible that March would have allowed Steevens direct access to the materials he had annotated.[47] It is also possible that, with his prodigious memory, Steevens simply recalled some of the infelicities that had drawn his editorial attention in the BL copy but had remained uncorrected in the fourth edition. Whatever the case, it appears certain that Steevens himself saw the new editions through the press. And the Puck does not stop here.

Over fifty years ago William Todd characterized "the proper discrimination and ordering of multiple editions" as "perhaps the most vexing" problem facing the bibliographer of eighteenth-century literature, asserting that "undenominated reprints are to be suspected everywhere," and that one may suspect "the presence of two or three concealed editions in practically every major production in the eighteenth century. . . ."[48] Such is the case with the sixth edition of A Dictionary of the English Language. In his magisterial A Bibliography of the Works of Samuel Johnson, J.D. Fleeman notes that this edition

Fleeman's explanation that variant press figures signal reimpression or reimposition depends on the unspoken assumption that the printing of the sixth edition took place exclusively from 1785-87, with the further implication that large quantities of type were kept standing. But in fact both physical and archival evidence shows that while the title-page date of 1785 is maintained, the sixth "edition" was reprinted from new settings of type several times over the next ten years. A full account of the complex printing history of this edition is beyond the scope of this essay, but several preliminary observations may be made.

The evidence that type was reimposed from one format to another comes directly from the Strahan ledger, where there are notations indicating savings from what Strahan calls "overrunning," both from quarto to folio and vice versa (fol. 97v). But, within the quarto itself, I have found no evidence that suggests copies issued in parts were from impressions separate from those issued complete.[50] Publication in weekly numbers began 19 November 1785 and ran through 30 June 1787. But on 17 February 1787 the Morning Chronicle and London Advertiser ran the following advertisement:

With printing now completed, the sixth edition was available in several forms. While distribution of the printing-in-parts proceeded, the proprietors issued the edition as completed volumes as well, following the general practice with serial publication.[51] Later gatherings—those of the latter parts of Vol. I, most of Vol. II, and perhaps the title page and preliminaries as well— would then have been printed in sufficient quantities to satisfy both forms of publication, while Vol. I gatherings which had previously been sold out in parts would have been reset and printed off as necessary to make up complete sets. (Type for the earlier gatherings is unlikely to have been left standing

for sixteen months.) Six of the nine Q6 copies I have examined exhibit a pattern that supports this hypothesis: they are essentially invariant in the latter portions of Volume I and the whole of Volume II, while in the first two-thirds of Volume I most gatherings are found in two different type settings and a few in three.[52]

Q6 provides a tremendous quantity of press-figure data since there are 141 gatherings in Vol. I, 137 in Vol. II, with the majority figured in both formes. One feature that may have contributed to Fleeman's puzzlement is that some sheets from a single typesetting are differently figured, most often with identical figures in one forme, and variant figures in the other. In comparing the BL and DVM copies I noted ten such occurrences in Vol. I; perhaps significantly, all but one occur between 5Q and 6T and the other is found in gathering `a' of the preliminaries, which would have been printed shortly after the completion of 6X, the final sheet of Vol. I. (In Vol. II there are only four examples, more randomly assorted.) I have also found examples where, coincidentally, formes from different settings of type have the same figure in roughly the same location on the page. And on at least one occasion a press figure actually undergoes a stop-press correction.[53] But in most cases in the sixth edition, different press figures imply different settings of type, and individual copies may be composed of a dizzying variety of reset and identical gatherings, depending on the number of copies of particular earlier gatherings that were still on hand when the decision was made to issue the edition in complete volumes. This possibility is further complicated by an advertisement that appeared on 21 December 1787: "GENUINE EDITIONS, . . . On Saturday, January 5, 1788, will be published,/Number I. of the Folio, and Number I. of the Quarto/Edition of/A Dictionary of the English/Language . . ." (Morning Chronicle and London Advertiser). The advertised "Conditions" are that the parts, consisting alternately of three and four sheets, will be issued weekly until completion, the folio to be composed of eighty sixpenny numbers, the quarto of seventy-eight. The process of serial publication appears to have begun again, while complete editions are still being offered at two guineas. Whether this new issue represents an actual reprinting of the parts or a new offering of parts left from the earlier printing has yet to be established, but entries in the Strahan ledgers which are not cited by Fleeman support the former interpretation. Under the heading "Johnson's Dictionary 4to" is an entry dated January 1789 which records charges for printing a run of 2000 copies of signatures c-k—the preliminary gatherings of Vol. I which cover "The History of the English Language"—

and signatures "B-3Z" of Vol. II.[54] These are the same signatures Strahan had printed in 1786 when the run had been between 3000-3500 copies. In a second ledger, again under the heading "Johnson's Dictionary 4to," are additional entries which record the printing of these same signatures yet again, with the prelims printed in January 1791 (2000 copies), sigs. "B" and "C" in July of 1793 (2000 and 1500 copies, respectively), and "D-Z" in December of 1795 (1275 to 1325 copies).[55] The copy of the sixth edition held by the University of Virginia must come from this last printing, since much of it is printed on paper whose watermark is the date "1794".[56] This edition (for it comes from a separate setting of type) is apparently intended to be indistinguishable from the earlier printings. The title-page date remains 1785 and the list of proprietors is identical to that in Vol. II of the earlier printing, even though at least four—L. Davis, W. Owen, W. Richardson, and J. Murray—were dead by 1794.[57] The collation also duplicates that of the earlier issues:

Even the anomalous parenthetical signings which intrude in the alphabetical sequence of the preliminaries and the curious signature "4F-M" which appears on 4F1 are repeated. This puzzling state of affairs may be explained in part by a comment made by Andrew Millar—bookseller, publisher, and one of the original proprietors of the Dictionary—concerning Hume's essays: that he regarded them as classics and never numbered editions when there was little need to "puff" them.[58] Todd concludes that "A close facsimile was, then, what [the booksellers] oftentimes attempted to produce.[59] The Dictionary would certainly have been considered a classic—and, particularly in the unabridged quarto format, it continued to be quite a profitable venture for its proprietors.[60] With the cachet of containing the "last improvements"

from Dr. Johnson's hand, the sixth edition would have required no additional puffing. But the proprietors apparently saw the Dictionary as something more than a cash cow, for as they reprinted it they continued to have it improved— almost certainly by George Steevens. In the 1795 printing of the sixth edition several corrections which had been proposed by Steevens over twenty years previously appear in a printed edition of the Dictionary for the first time. All are textual corrections to Shakespeare quotations: Under Censer 2, in a quotation from Taming of the Shrew, "Here's snip, and nip, and cut, and slish, and slush," "slush" is corrected to "slash." Under Chamberlain 3, in the quotation, from Macbeth, "We will with wine and wassail convince," the word "so" is inserted after "wassail." In the Macbeth quotation under Farrow, "Pour in sow's blood that hath litter'd / Her nine farrow," "litter'd" is altered to "eaten." Under To Fast, in the Cymbeline quotation "Last night the very god shew'd me a vision," "god" is corrected to "gods." And under Green adj. 4, in a quotation from Romeo and Juliet, "Lies festering in his blood," "blood" is corrected to "shroud."

Since the annotated British Library materials end with the letter "I/J" they do not include even all of Volume I of the folio Dictionary, which ends with the letter "K", and Vol. II is not represented at all. It is possible that Steevens annotated a complete copy, and that the end of the first and all of the second volume have been lost. I have as yet made no attempt to collate Vol. II, either between F4 and Q6, or between the early and late printings of Q6. But I have done so for the letter "K," which runs only to a dozen pages in the quarto edition. I was not surprised to find that between the early and late printings of Q6, there are two substantive corrections, both to quotations from Macbeth. [61]

Most if not all of the changes Steevens introduced in seeing the successive editions through the press would have gone unremarked by readers. As Johnson had reassured the public in the "Advertisement" to the revised fourth edition of the Dictionary, the owner of a previous edition "needs not repent; he will not, without nice collation, perceive how they differ. . . ." Yet the regard in which the proprietors held Johnson's Dictionary as a cultural icon apparently dictated that if they were to reprint it, they should also improve it. And George Steevens, through his anonymous ministrations carried out over a period of nearly twenty-five years, sought "not merely to obtain justice to Shakespeare," but justice to Johnson as well.

Notes