University of Virginia Library

Search this document 


  

expand section 
collapse section 
 1. 
 2. 
 3. 
expand section4. 
 5. 
 6. 
 7. 
collapse section8. 
 01. 
 02. 
 9. 
 10. 
 11. 
 12. 
  
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 

expand section 

Notes

 
[*]

The research for this paper was supported by a generous grant from the National Endowment for the Humanities, Division of College Teachers, for which I am also indebted to Fredson Bowers, Peter Blayney, and Donald H. Reiman for their supporting recommendations. This paper would not have been possible without this support. I also thank Fredson Bowers for suggesting the approach of this paper and encouraging and guiding its progress, and Peter Blayney for sharing his vast knowledge of the subject. The staff of Reader Services and the Special Reading Room of The Huntington Library also deserve a note of appreciation for their cheerful and efficient assistance.

[1]

G. Thomas Tanselle's wide-ranging discussion in "The Use of Type Damage as Evidence in Bibliographical Description," The Library, 5th ser., 23 (1968), 329-351, includes an especially relevant comment: "By definition a descriptive bibliographer must describe; and description entails more than the notation of the minimum number of apparently 'significant' features for the bibliographical analysis of one work. The bibliographer is further obliged to contribute to a larger body of information, and any descriptive bibliography should be, in effect, a partial history of printing. Details which may turn out to be unimportant in analysing the printing of a particular book or determining the number of impressions it went through may nevertheless furnish important corroborative evidence to another bibliographer dealing with a different book of the same period," p. 336. In this context, the kinds of evidence which suggest or demonstrate shared printing of a book should be given special attention. Although Tanselle's observations on the importance of evidence from type damage for reconstructing the printing of an issue or edition are certainly sound, I have difficulty understanding how damage to specific types can be defined in a bibliographical description so that they would be recognizable in another book. This seems a bit too idealistic. The precise image of a damaged type as seen at high magnification is necessary for identifying its recurrence. No degree of precise verbal description or recording of measurements can serve the purpose. This paper assumes a familiarity with concepts, principles, methods and examples presented in two earlier papers: "Reproductions of Early Dramatic Texts as a Source of Bibliographical Evidence" (hereafter "Reproductions"), TEXT, 4 (1989), 237-268; and "Font Analysis as a Bibliographical Method: The Elizabethan Play-Quarto Printers and Compositors" (hereafter "Font Analysis"), Studies in Bibliography, 43 (1990), 95-164 (see note 12, p. 152, for an explanation of my method of labeling specific fonts). Important back-ground information about shared printing and typographical study is found in Peter W. M. Blayney, "The Prevalence of Shared Printing in the Early Seventeenth Century," Papers of the Bibliographical Society of America, 67 (1973), 437-442, and The Texts of 'King Lear' and Their Origins (hereafter Texts), vol. 1 (1981); Antony Hammond, "The White Devil in Nicholas Okes's Shop," Studies in Bibliography, 39 (1986), 135-176, and Hammond's review of Blayney's Texts, The Library, 6th ser., 6 (1984), 89-93; and Charlton Hinman, The Printing and Proof-Reading of the First Folio of Shakespeare (1963).

[2]

See discussion, Principles of Bibliographical Description (1949), pp. 300-306, 344-347. G. Thomas Tanselle, "The Identification of Typefaces in Bibliographical Description," Papers of the Bibliographical Society of America, 60 (1966), 185-202, is rather non-specific about early typography but establishes useful principles for including information about typefaces in a bibliographical description. Evolving knowledge should eventually produce a more specific basis for such descriptions. A potential problem exists in as much as W. Craig Ferguson's Pica Roman Type in Elizabethan England (1989), described as a "vital reference source," may be used unwittingly by bibliographers in the manner described by Bowers, e.g., "The font of type . . . should be identified by reference to books on printing types," p. 305. Ferguson's misconception of the actual complexity of typefaces and fonts underlies his overly simplistic method of typographical analysis which, as a consequence, lacks the precision needed for sorting out extremely similar letters in different typefaces and, moreover, eliminates from consideration most of the lower-case, all capitals except the 'M', and all ligatures and punctuation. In general, the magnitude and frequency of demonstrable errors and oversights that are found in Pica Roman Type disqualify it as a dependable reference for the identification of typefaces and specific fonts. Of particular concern are problems encountered in Tables 2, 3, 8, and 9 (all duplicated in Table 1 entries). For details, see my review (forthcoming in Papers of the Bibliographical Society of America, December, 1989) as well as comments included below. See "Font Analysis," note 10, pp. 151-152, for references to contemporary typeface specimens; and Frank S. Isaac, English and Scottish Types 1535-58, 1552-1558 (1932). My suggestion that bibliographers check the fonts in every book calls for a word of encouragement since a well-known bibliographer (who shall go unnamed) responded with what may be a typical reaction: "You can't expect someone to examine the type in over a thousand books." Indeed, this seems an enormous burden to impose upon an already devastatingly tedious task. However, once one learns to recognize typefaces by stylistic traits and differentiate fonts by easily observed "gross features," it becomes a simple matter to note typographical facts while paging through a book in search of ornamental stock, anomalies in signatures and/or pagination, and other kinds of bibliographical data that should be recorded as a matter of course. A method of learning typeface characteristics is described in "Reproductions," pp. 252-254. Special stylistic traits and variant letter-forms are discussed in detail in "Font Analysis," esp. pp. 97-119 and note 28.

[3]

The unfortunate fact, however, is that during the past quarter century or so the "publish or perish" pressure in academia has left little leisure for this degree of editorial responsibility. The discipline must eventually recognize the futility of the continuing production of new editions of shared Elizabethan play-quartos which are treated as if they are the work of a single printer. The fact that the search for a sharing printer can require a considerable amount of time and end in failure is not a valid reason for avoiding the minimal task of detecting shared printing and identifying the primary printer's section(s) of a book.

[4]

See W. W. Greg, "'The Honest Whore' or 'The Converted Courtesan,'" The Library, 4th ser., 15 (1935), 54-60; Fredson Bowers, "Notes on Running Titles as Bibliographical Evidence," The Library, 4th ser., 19 (1938), 315-338, and "The Headline in Early Books," Essays in Bibliography, Text, and Editing (1975), pp. 199-211.

[5]

See discussion in "Reproductions," pp. 242-244, and "Font Analysis," note 40. In addition, I should note that the font used in running-titles may suggest the identity of a suspected printer or lead to a rejection of him as a candidate. The font may contain unique discriminants sufficient to demonstrate ownership. On the other hand, a survey may reveal that the candidate printer never used the typeface seen in the titles. For example, STC14377 and STC14381 are assigned respectively to Simmes and Windet, but the running-titles are set in a Y-font which neither used elsewhere. Both conclude with the Read/Eld McKerrow 320b "Gilley flower" (I have seen only the final page of STC14381 in British Library copy 720.a.32(3)). STC14377 also uses a Read/Eld 'V' (A2) and 'A' (A6), and Eld-Y1 in the letter (A2-2v) and as emphasis in the black letter text.

[6]

The ingenious technique devised by Randall McLeod which employs transparencies of running-titles is a valuable improvement over the unaided eye. It seems that the added precision offsets the expenditure of time required to produce the transparencies. See "A Technique of Headline Analysis, with Application to Shakespeares Sonnets, 1609," Studies in Bibliography, 32 (1979), 197-210.

[7]

As a matter of course, correct procedure entails moving from the general to the more specific in stages. Having ascertained that the Y-face appears throughout, the next step involves a quick survey aimed at detecting obvious "gross features" differences, beginning with the capitals, punctuation, and foul-case cluster. The font of A-E contains both Y- and S-face capitals throughout, while F-H uses Y-capitals exclusively. This in itself indicates the strong probability of two same-face fonts. Further weight is added by a survey of foul-case italic capitals that yields different clusters which overlap in just a few sorts. By this point, there can be little doubt that a boundary marked by a shift in fonts occurs at signatures E/F and implies the possibility of shared printing.

[8]

See recent studies noted in "Reproductions" and "Font Analysis" as well as studies of the shared play-quartos noted below.

[9]

See, for example, Devil's Charter STC1466, Byron STC4968, T. Travailes STC6417, Northward Hoe. STC6539, Westward Hoe! STC6540, Volpone STC14783, What You Will STC17487, A Tricke STC17896, Hamlet Q3 STC22277, Troilus STC22331.

[10]

References that reproduce and identify ornamental stock are given in Texts, pp. 458-459, largely duplicated in "Font Analysis," note 1; see also notes 11 and 15 below. I follow Blayney's method of citing these references, e.g., the author of the reference separated by a slash "/" either from the owner of an ornament or the title of the reference; McKerrow's identifications in Printers' and Publishers' Devices are cited simply by number.

[11]

The back-and-forth process of comparing ornaments and searching for printers is reflected in Katherine V. Pantzer's apt selection of the title "The Serpentine Progress of the STC Revision" for her report in Papers of the Bibliographical Society of America, 62 (1968), 297-311, which includes reproductions of six initials and one factotum (p. 303) and valuable discussion of the search process, pp. 299-309, ending with a confession: "I like catching the printers with their ornaments and initials showing; a bibliographical Peeping Tom, so to speak." For the average bibliographer who lacks access to the STC office's resources for identifying ornamental stock, the peeping-tom analogy leaves something of a bitter taste since the initial glimpse is sometimes the last, as if the shade were pulled or the lights switched off.

[12]

Ferguson's oversight of identifiable ornamental stock at divisions between shared sections here and in other works such as Antichrist STC7120 and Essays STC18041 is puzzling, but the end result is that the shared printing is overlooked. More puzzling is his comment on STC19733: ". . . sheet C is neatly divided between C4v and C5r, the centre of the sheet. The division in sheet B is not so neat: the Lyon(b) was used on B1r, the Lyon(a) was used to the middle of page B5r, and then the use of the Lyon(b) fount recommenced at a new paragraph. There is no evidence of mixing later in the book [my emphasis] so the seven and a half pages of Lyon(a) must have been carefully distributed back into its own cases. One can answer the question of whether compositors sometimes stuck to their own cases and distributed back to them in the affirmative," Pica Roman Type, p. 15. In fact, Roberts's final section is sheet C, and the rest of the book is printed in Creede-4, so quite naturally the type from sheet C would not be mixed in with the type seen "later in the book." Creede's 1605 reprint of the book (19733a), moreover, also uses the same Yamada/Creede I1 initial and prints in Creede-4. This should have been enough to occasion a closer look at the fonts in BC and 2B-K, provided Ferguson checked 19733a. There is the outside possibility that we looked at two different issues, one printed entirely by Roberts, but the presence of Yamada/Creede I1 in a hypothetical second issue hardly could have escaped detection by STC researchers since it appears so frequently in Creede's books. Since Ferguson consistently fails to provide essential information as to which copy of a book he means, the hypothesis of two issues is left hanging as an improbability. In fact, one fatal weakness of Pica Roman Type as a reference work is the absence of information about the books that were actually examined in the surveys of the various printers. The reader is left with the task of surveying a specific printer's books from a particular year to determine just where "in 1602 another pica was used." Beyond that, it is clear that oversights of three kinds undercut the reliability and usefulness of Pica Roman Type: (1) pica roman books listed in Morrison's Index were overlooked; (2) although some pica roman books in Morrison were checked for possible reassignment in the new STC, many others were not; and (3) the new STC was not searched as a whole for newly assigned or reassigned pica roman books. As a result, errors in font identifications are compounded by the omission from the Tables of both identified fonts and at least a dozen unidentified fonts from 1595-1610. Aside from such deficiencies, the proposed printer-search method (see pp. 18-19) using the Tables and facsimiles cannot substitute for reliable empirical methodology. On the one hand, no system of typographical analysis, however sophisticated, can identify a font on the basis of the evidence seen in a single page reproduction. Conversely, recurrent-types which are required for conclusive identification of a font can be absolutely identified only in originals.

[13]

These initials include: the M5 at 2N7, from 2G2v, 2K2, 2R8v; the Fish-S at 2O8, from 1D5v, I4v, 2D2, 2G1, 2R3; the Winged-Horse-E at 2S1, from 1O6v, S6v, 2B1v, 2C6v, 2E2v; the 'V' at 2S4, from 1D4, P1v, Q4v, 2I6; the N5 at 2S5v, from 1A5, 2A5, 2C4; and a 'V' from 1D4 recurs at 2S4.

[14]

For example, the Vine-I at T6, Lute-G at X5, the Vine-T at Z6, 3H3 and 4K1, the Stag-C at Q2v and 4X5, the Rabbit-W at 4Y2v, the Standing Cherub-M at S6, 3H5v,6, and 4I6v, and the Satyr-D at 3A1 and 3F2v.

[15]

See William A. Jackson, "Counterfeit Printing in Jacobean Times," The Library, 4th ser., 15 (1934), 364-376, with reproductions of the forged items at p. 367 and pp. 368-369.

[16]

I am indebted to Peter Blayney for pointing out appearances of this 'W' in STC 18230 (1598), A2, and STC7188 (1608), B1.

[17]

An effective method of comparing similar ornaments or initials, especially large complex ones, employs transparencies produced on a Xerox machine, provided that the machine permits a continuous adjustment of the reproduction ratio so that the two images can be reproduced at an identical size. When the transparencies are placed one on the other on a sheet of glossy white paper (or a light-table if available), differences attributable to design, flawed copying or damage usually are easily detected.

[18]

The differences are as follows: A5: upper-right corner, detail in a rectangular area 10mm down, 4mm in: damage in Islip's A5 produced a 1mm gap between the berry and base of the leaf; the nearly horizontal portion of the vine at the bottom of the area is blunted and prints quite bold. Okes's A5 is undamaged. See Islip's A5: STC17291 (1598), D8, L4v, L8; STC13200 (1601), T5; STC15170 (1602), Y4v. I5: the vine splits into a flaring double-line 2mm below the junction with the top leaf (extreme upper right); the outer line is sheared at the bifurcation point and is pushed away from the inner line in Islip's I5. Okes's I5 is normal. See Islip's I5: STC11196 (1599), A2; STC23456 (1604), P8; STC5504 (1605), A2. T5: the vine above the left edge of the cross of Islip's T5 (see enclosed area of T4 State 2, p. 450) is straightened to form a wide-angle inverted 'V'; the vine above the right portion of the cross exhibits a 1mm gap at 7.5-8.5mm in from the right edge. See Islip's T5: STC5493 (1600), Q1v; STC13200 (1601), O2; STC17291 (1598), B3v.

[19]

The internal frame of a factotum into which an ordinary initial was inserted was especially vulnerable to damage incurred during insertion. Framing rules are frequently missing; in factotums lacking internal framing rules, the portions of the design abutting the frame are frequently damaged or obliterated. This area of a factotum can sometimes be quite useful for distinguishing castings but seems rather unreliable on the whole for identification purposes.

[20]

I must note that a comprehensive survey is needed to determine whether both versions were used in books produced by the Eliot's Court group. I have not performed the survey at this time but serendipitously stumbled onto the appearances in STC6164, STC 10548, STC24719 and STC18191.

[21]

It can be noted in Plomer's reproduction that the outer half and tip of the leaf leaving the stock at 11mm down from the top is intact. This leaf and the first leaf above the bulb (larger in the Eld/Jaggard version than in Plomer 52, another difference from imperfect copying) have been sheared along a straight-line drawn from the outer edge of the top-most leaf through the outer edge of the bulb (13mm down). This state obtains in STC 12988 and an early Eld book (Darius STC 350, A1, late 1604). The left edge of the top leaf is blunted inward and the terminating curl of the tendril immediately below is lost in a slightly later state (Obedience STC25633, A2 and B3, early 1605). In addition, the tip of the second leaf at the top-right corner has been blunted with a slight cut above the tip.

[22]

At bottom center, the tip of the leaf is severed from the long center petal with a precise cut; the entire left edge has been blunted or sheared downward and the tip of the middle leaf has been compressed to a vertical line about 1mm in length; a gap in the line from the center occurs prior to the first leaf at the top-right; and an extraneous 1mm line extends beyond the left top-serif. Additional progressive damage is seen in Spider STC5693.5: the leaf tips at the upper right are sheared or blunted. (See also De Vnione STC13951, 1604; Times Anatomie STC20342, 1606).

[23]

The STC12988 state lacks the gap in the outer line forming the mustache at the junction with the left frame. STC12988 was entered 7 April 1603, a few months before the gap appears in a Read/Eld book (Isahaacs STC25643, B4v) which was entered 16 June 1603, suggesting progressive damage; it remains in later appearances (see Darius STC350, 1604, A4; Ostend STC18895, 1604, 2G3v; and in Fools STC4963, 1605, A3v. Damage to two other areas is identical. The bold line forming the underside of the leaf at the lower-right seems to have been depressed so that a portion prints thinly. A wide gap appears in the outer line of the frame just to the right of the right mustache.

[24]

A smaller version of the Goat-head flanked by feeding-squirrels is reproduced in Woodfield/Field #10; although it exhibits major design differences, the general concept is the same. Version #1 appears in An Answer STC23451 (1595) at B3, with the imprint "Deputies of R. Barker"; and in Les Reports STC5493 (1600), A1, with the imprint "in aed. T. Wight" but reassigned in the new STC to A. Islip on the basis of ornamental stock, an assignment which can be confirmed by the appearance of many ornaments in the context of Islip's fonts in other books. Version #2 appears in An Answer STC12988 (1603), 1A4; and frequently in Eld's books, including A Panegyric STC12061 (7 June 1603), B1; Darius STC 350 (1604), B1; Ostend STC18895 (1604), A1, 2G3; and Fools STC4963 (1605), A3. Absolute design differences differentiate the two versions as follows:

  • 1. Leaves of the top right flower: in #1, the first small petal to the outside is a short nib; it is longer in #2 with a curved tip. This difference cannot be due to progressive damage, since elements can be shortened but not lengthened.
  • 2. The third vine 9mm in from the left bottom corner is formed by three lines (double lines form the top) in version #1; it is formed by two lines in #2.
  • 3. Area at top left, 27mm in, 9mm down: in #1, a 1mm gap separates the short, blunt leaf from the large scroll structure; the leaf is larger and pointed in #2 and touches the scroll.
  • 4. At left center: in #2, the beak of the grotesque animal figure touches the leaf to the left; the two are separated by a 1mm gap in #1.
  • 5. Upper left corner, 15mm in, 7mm down; the small tendril ends in simple curl in #2, but in #1 terminates in a four-loop bud.
  • 6. Upper right corner: in #1 the coiled tendril is formed with distinct first and second loops, a compressed third loop, and terminates like a monkey-tail hanging down; in #2, the termination is a simple, thin vine which could conceivably be a remnant of the monkey-tail structure following damage, except that the overall structure of the tendril suggests otherwise.

    Some differences that could result from damage should be noted; these details also suggest that Eld's #2 appears in STC12988.

  • 7. The outer line of the first loop of the coiled tendril (noted immediately above) exhibits a 1mm gap in #2 not in #1.
  • 8. Upper right corner: #1 has a short curled leaf to the right of the stem at the base of the flower bulb; its absence in #2 could be due to blunting or shearing damage.
  • 9. Bottom of outermost right vine: #1 shows a downward thorn or petal that is absent in #2; damage is highly unlikely as a perfectly clean cut exactly parallel to the line would be required.
  • 10. Bottom center: three hairs extend from the tip of the beard downward: in #1 they flare and terminate at the same level; in #2, the left hair is longer than the right.
  • 11. Bottom center: in #1, a two-leaf cluster extends below the arches which support the squirrel on both sides of the goat head; the left leaves are lacking in #2 (they could have been cut away).

[25]

The survey process will benefit immensely when P. G. Morrison's outdated Index to Printers is superseded by the appearance of Volume 3 of the revised STC in late 1990 or early 1991. Scholars attracted to printer research by its inherent sense of mystery (not to mention the flush of sublime fulfillment that accompanies each identification of an unknown printer) will no longer have to "read the STC" in search of newly assigned books by a suspected sharing printer. (Incidentally, I must note that my research was guided primarily by Morrison's Index and supplemented by searches of portions of the new STC that promised new leads; it can be assumed that I have inevitably overlooked assignments that will emerge in Volume 3.) Katharine V. Pantzer has graciously supplied information about the format and contents of Volume 3, which is an accumulation of an incredible amount of information relevant to early printing. Two indexes are most pertinent to printer research. "Index 1: Printers and Publishers" lists printers and publishers identified in new STC in alphabetical order, each headed by supplemental information garnered from numerous references followed by a chronological list of books by STC number. "A Chronological Index" rearranges STC books by author's surname and STC number for each year to 1640. Items with erroneous, conflicting or multiple dates are re-entered under each date. The compiler of "A Chronological Index," Philip R. Rider (University of Northern Illinois), richly deserves the gratitude of the scholarly community for accomplishing this monumental task. The survey process begins by compiling from Index 1 a list of proximate books that are assigned to a suspected printer. The books are then checked for samples of a font in the correct size and style that matches the target font. In general, it is wise to continue until two or three font samples are located in order to confirm tentatively that the printer used the candidate font more than once. Comparative analysis leads either to the rejection of the printer or to the final stage of locating recurrent-types and confirming his work in the target book. The search process can sometimes be more narrowly focused by the consideration of circumstantial factors. The kind of text can lead to the elimination of potential candidates who would otherwise be surveyed because of their use of a correct style and size of font. For example, Robert Barker used a Y-font but never printed a play-quarto. Or again, Islip frequently printed privileged legal texts courtesy of Wight's patent and then for the Stationers' Company after September 1605. A glance at STC entries for theological authors such as Matthew Sutcliffe or Henry Smith can suggest potential printers for this type of book. In contrast, trade printers such as Allde, Braddock, Creede, Eld, Jaggard and Purfoot seem to have printed anything that came through the door and are likely candidates in any case, provided that the fonts match. Format sometimes suggests candidates. Islip seems to have favored folios after about 1605. Similarly, Eld undertook some heavily illuminated and illustrated folio texts that were remarkable challenges (see especially Birth of Mankinde STC21161 and Admirable & Memorable Histories STC12135). Many of the trade printers produced play-quartos: see the repeated involvement of the group in STC sections on playwrights such as George Chapman, Ben Jonson, John Marston, Thomas Middleton, Shakespeare and others. Translations form another class of books that might focus a search; Eld, for example, printed several translations by Edward Grimeston. In this area, Virginia Renner, Reader Services Librarian at The Huntington Library, is currently working on a bibliography of STC translations extending to over a thousand items; once completed, the relationships among translators, publishers and printers that emerge should be a valuable resource in printer research. (See also Julia G. Ebel, "A Numerical Survey of Elizabethan Translations," The Library, 5th ser., 22 (1967), 104-127, which is limited to items in the unrevised STC for 1560-1603.) Similarly, the publisher-printer relationship has not received much attention to date, but quite valuable information and analysis is found in Gerald D. Johnson, "Nicholas Ling, Publisher 1580-1607," Studies in Bibliography, 38 (1985), 203-214, and "John Busby and the Stationers' Trade," The Library, 6th ser., 1 (1985), 1-15; and Blayney's Texts. Known sharing relationships provide another network of cooperation. For example, William Aspley and Edmond Blount published books with either Eld or Simmes as primary printer, several of which were shared between the two. Then again, Eld as primary printer shared with Braddock, Creede, Jaggard, Purfoot, Simmes, Stafford and White, Purfoot with Jaggard and Windet, Creede with Windet, and so on. Other printers such as Felix Kingston seem not to have taken much to sharing. Overall, a printer search is pure detective work and unpredictable. The sometimes radical variations in shop output imply one disturbing factor whose significance cannot even be estimated, namely, the probability that some portion of the books produced in the period are no longer extant.

[26]

For discussion of the impact of several classes of foul-case letters upon the purging process, see "Font Analysis," pp. 131-134.

[27]

See "Font Analysis," pp. 133-134, for discussion of the process of contamination affecting one case of Windet's S-font (Windet-S2). Incidentally, at no point was Windet-F mixed wholesale into Windet-S1,2 as claimed by Ferguson, i.e., "In 1607 a fifth fount [5] was created by mixing together the recast Garamond [F-hybrid] and the Lyon(c)," Pica Roman Type, p. 32. The process of low-density contamination of Windet-S2 dates from at least as early as 1605, was nearly up to peak levels by March 1606, and therefore was not a new phenomenon in 1607. The level of contamination seen in Plate 164 which purports to reproduce the "new" "Windet-5" is essentially the same seen, for example, in Fawne Q2 STC17484 (12 March 1606) G3, G4v, or Sophonisba STC17488 (17 March 1606) B3-4, D1,2, especially D2v-3, F2,3,4,4v. Furthermore, the mis-identified "Garamond" typeface and fonts (Tables 2, 9) is actually a hybrid which combines Haultin's Y-face capitals and '?' with a second lower-case long-since attributed also to Haultin (see Dreyfus, Type Spec. Facs. Vol. I, notes for facsimile 14, specimen 393, and facsimile 15, specimen "Mediaen Romeyn, No. 1"). It was used as early as the 1550's in Venice by Paulo Manuzio and Angelo Gardano, and in Holland throughout the 17th century. Similarly, the mis-labeled "Le Be" typeface and fonts (Tables 8, 9) do not match Le Be's Garamond punches/matrices, but are hybrids which combine sorts from dissimilar sets of non-Garamond punches. Actually, one mis-labeled "Tavernier" font (Sutton-1, Plate 142) is the only example (as far as I know) of a Garamond pica roman (see Type Spec. Facs. Vol. II, "Index 1567" and "Le Be" specimens 12, 13) to appear in England during the period (excepting in imported books by R. Schillers, Middelburg); nor is Bynneman-4 (Plate 29) a Tavernier, but simply the Haultin Y-face whose first (?) appearance in A Dictionarie STC6832 (1570) predates Jerome Haultin's arrival by four years. Shipment directly from the Haultin establishment at La Rochelle seems the inevitable inference.

[28]

See "Reproductions," pp. 250-251, and "Font Analysis," note 57, for a description of the patterns of alternation. Incidentally, Ferguson cites Edward 3 STC7502 (1599) as an example of "irregular," or alternating, stints (p. 16), and later offers an explanation based upon STC7502 for the composition of Stafford-EFb which he considers "very much a jumble": "behind the mixture probably lay some Haultin [Y-face] sorts and some Lyon(a) [C2-hybrid] which were used more or less unmixed in printing parts of the 1599 edition of [Edward 3] STC7502, into which was mixed some Lyon(c) [S-face] as time went on," p. 31. Assuming (again) that STC7502 exists in only one state represented by The Huntington's copy, it can be noted categorically that the text prints entirely in Stafford-EFb and nothing else. See "Font Analysis," pp. 122-125, and Appendixes III and IV, pp. 145-148, for an exhaustive analysis of Stafford-EFb's actual composition.

[29]

See "Font Analysis," pp. 135 and 139, for discussion of this phenomenon which implies a printing sequence of Whore Q1-2 STC6501-6501a, Malcontent Q3 STC17481, Fools STC4963, Eastward Q1 STC4971.

[30]

For example, Sejanus STC14782 was entered twice (first entry 2 November 1604, reentered 10 August 1605) and bears an imprint of 1605. Fools STC4963 bears an imprint of 1605 but was not entered; the performance at Court in January 1605 suggests the earliest printing date possible. This leaves the problem of dating the printing of Malcontent Q3 STC17481 (entered 5 July 1604; imprint: 1604) and of Eastward STC4970-73 (entered 10 September 1605; imprint: 1605).

[31]

The distribution pattern of Eld-S1 and -Y1 in Eastward Q1 STC4971 is: Eld-Y1: A2, A3, A4v-B2, B4v-C2v, D4v-E2, E4v, F2v-3v, H2v-4, I1-2; Eld-S1: the rest including I2v-4v. Note the typographical error in the last line of the table in "Reproductions," p. 248, reading "F2v-F3r" instead of "F3v".

[32]

The migration of standing type from Eastward Q1 STC4971 to Q2 STC4972 is as follows: E4v:1-32 to E2:12-36 [stick Q1 = 92mm, Q2 = 96mm]; E4v:33-37 to E2v:1-5; H2v: 3-30 to G3:17-37; H2v:30-37 to G3v:1-5 [G3v:6-37 reset in Eld-S1]; H3v:2-H4:7 to G4:1-37 [speech prefixes mostly reset]; H4:8-37 to G4v:1-27 [28-37 reset in Eld-S1; stick = 100 mm]; I1:1-23 to H1:22-37; I1:24-39 to H2:1-14; I1v:1-29 to H2:15-39; I2:1-38 to H2v:9-39 [H2v: 1-8 reset in Eld-S1; 38 lines of Q1 are crammed into about 30 lines in Q2 with turn-overs etc.].

[33]

The pattern of alternation in T. Miseries STC25635 (1607): Jaggard-S1 at A2-4v, B(i), C, D1v-3, E, F1-2v, G, H1-2, I1-2v, I3:25-36, I3v-4v, K1,2v,3-4; Jaggard-C2 at B(o), D1,3v-4v, F3-4v, H2v-3v [4-4v lacking], I3:1-24, K1v,2,4v.

[34]

See Ferguson's explanation that a single Y-font, obtained in 1582, "was used through 1610. . . . From 1594 there was a mixture of 'M's in at least one case of the Haultin, but Purfoot kept at least one case unmixed," Pica Roman Type, p. 29. Twenty-eight years of usage would indeed be remarkable. Incidentally, Plate 117 purporting to reproduce Purfoot-2 ("unmixed") is actually Jaggard-Y1b. See also the explanation of the Middleton-3 to Robinson-2 to Braddock-1 lineage, 1579-1610, pp. 29 and 30.

[35]

Ferguson concluded that Read-S1 passed to Eld-S1 along with the shop and explained: "When Eld obtained the materials [Read-S1] he discarded those letter forms which did not belong, and produced an 'i' with the dot a trifle higher," Pica Roman Type, p. 30. This proposition ignores the virtual impossibility of a compositor sorting through the approximately 1000-3000 types in a sort (see Blayney's sort-counts, Texts, pp. 145-148) and attempting visually to distinguish wrong-face pica types and discard them. See also discussion of purging in "Font Analysis" (see note 25 above).

[36]

Widow Simson, AElohim-triune STC5329 (2 January 1601). Read, Ten Learned Personages STC1074 (16 January 1601); A Sermon STC1454 (imprint: 1601; entered 18 March 1600); Contemplations STC4662 (18 October 1601); Cynthia's Revels STC14773 (23 May 1601).

[37]

The distinction between Robinson-Y1a (1590-1594) and -Y1b (1595-1597) as possible states of one font ignores the probability that two fonts as a whole are involved. However, the relationship between the two is not entirely clear at this point since some sorts of Y1a could have been transferred to Y1b. Adequate samples of both are available and reveal that a major decontamination was necessary to transform Y1a into Y1b, including the purging of oversized capitals (A C D2 E4 H I M O P R S T V), Guyot 'L S', a few low-density lowercase S-face sorts (d1 g1 f1 n1 x2), and over-sized 'e h o r', and discarding the entire, new S-face 'ſt' ligature. The presence of the latter can be attributed to almost total replenishment since a very few original Y-face 'ſt' ligatures still remain in Y1a. In addition, the proportion of damaged sorts drops radically when Y1b appears in 1595 in A Comparison STC4098 and the oversized capitals have been reduced to the 'A P S'; the presence of these capitals is not significant in itself as evidence of identity with Y1a since several Y-fonts in the 1590's seem to have been purchased or replenished almost immediately with an assortment of these capitals which range from 2.85-3.15mm in height. (This phenomenon requires further study, but it is worth noting that the cluster of oversized capitals was expanded in Brad-dock-Y1 after 1598 along with the addition of moderate-density new S-face 'ſt' ligature, although the lower-case remained uncontaminated). However, the appearance of eleven S-face 'ſt' ligatures as well as the 'R' on L1v-2 suggests the possibility that Y1a had not yet been discarded and was used to set this final Y-face passage. Moreover, a search of 45 prose pages of Tetrastylon STC25701 (1593) revealed only three battered types with any probability of recurrence in Y1b. Of these, only the 'w' at p. 37:25 of STC25701 is almost certainly that seen in Salomon STC18194 (D4:16, O1:26, P1v:36) and Two Treatises STC12322 (1598) (B2:16, D2v:29), but another recurrence or two in STC25701 would be helpful. The same holds for the 'a' at p. 32:3 of STC25701 and O4:18 of STC12322. Overall, the issue of the identity of Robinson-Y1a,b is left hanging, but the general principle implied by the unsuccessful recurrent-types survey is clear: a sample text in Robinson-Y1a cannot be used to establish his ownership of a target font if that font happens to be Robinson-Y1b.

[38]

See below, note 48.

[39]

In this instance, circumstances of the Read/Eld shop's history invite speculation that makes some sense of the situation. The transfer of the shop to Read occurred in January 1601, bounded by Widow Simson's AElohim STC5329 with the shop's McKerrow 320a "Gilley flower" ornament (the initials "G[abriel]. S[imson]." still in place) on the title page and the text in Simson/Read-S1, used also in 1601-1602 Read books Ten Learned Personages STC1074, Revels STC14773, Contemplations STC4662, as emphasis in C-E of Sermon STC 1454 (18 March 1601), and in Read's section (H-M2v) of EMIH STC14766 (1601). McKerrow 320b with the initials "G. S." effaced appears as well in STC1454 (E1v), Essayes Pt. 2 STC 5775 (O3), Tragedies STC26076 (title), A Dialogue STC18892 (title), and Cromwell STC 21532 (title). If Essayes STC5775 Pt. 2 (October 1601) was actually printed in Read's shop, it marks the virtual cessation of production until Cromwell (11 August 1602) since Epitaph STC3415 and A Dialogue STC18892 are such negligible, short emphemera. In general, it is difficult to reject the probability that Eld took over the shop in early 1603. Eld served his apprenticeship (1592-1600) to Robert Bolton in Richard Jugge's old shop at the sign of The Printers Press in Fleet Lane (as did Read), but his activities during 1600-1603 are unknown. Read apparently trainferred Simson's materials from The White Horse to The Printer's Press address when he took over. According to the title pages of Revenger's Tragedy STC24149 (1607) and A Trick STC17896 (first issue, 1608), and the Stationers' census of 1615, the shop remained at this location. The appearance of Eld-Y1 in Nero STC12551 (entered 23 February 1603) commences a dramatic shift from the relative dormancy of Read's shop during 1602 to the consistently high output of Eld's mainstream printing activities during the subsequent two decades. Overall, 1603 production included at least fourteen titles (STC1117, 6260, 7120, 7594, 10800, 11086, 12061, 12551, 14377, 14381, 18292, 24041, 24343, 25643; and as sharing printer: 18041, Book 3). Volume 3 of revised STC may reveal other books assigned to Read. It is interesting to note the coincidental first appearance (as far as I know) of Eld's favorite title page xylographic ornament (the opposed scrolls described earlier) in T. Opinion STC24343 along with other stock; unfortunately, T. Opinion was not entered, so the date of printing is unclear, although a survey of papers in 1603 would probably establish this book's position in the shop's output. A second remarkable change occurs in shop activity with respect to the initiation of business relationships with a new group of publishers. As far as we know, Read printed 1601-1602 for J. Brome, W. Burre, M. Law, W. Jones, and E. Mattes, a small group that reflects the sparse output from the shop. It seems significant that none of these publishers is included in the group commencing with Nero STC12551 (23 February 1603) and extending through 1609. The obvious inference is that a kind of radically new marketing environment emerged in 1603 which, in business circles, is usually attributed to new management. It seems safe to assume that publishers "shopped around" for the best deal then just as they do now, and furthermore, that printers had a choice either of passively waiting for business to walk through the door or of aggressively marketing their services. The latter is suggested by the fact that, in 1603 alone, nine publishers without prior involvement contracted printing jobs (W. Aspley, E. Blount, C. Burby, F. Burton, G. Chorlton, J. Harrison IV, J. Norton, T. Pavier, G. Seton). Of these, Aspley and Blount remained frequent customers, while Burton brought only two other books (1604, 1606) to Eld. During 1604-1608, Eld printed four or more editions for T. Adams (5), W. Aspley (8, and four others as sharing printer), E. Blount (7), T. Thorpe (7), S. Waterson (4), and J. Wright (8). Furthermore, a new business strategem emerges in the form of Eld's activity as a sharing printer, first with Simmes in Essayes STC18041, a relationship that would endure for at least two years. It is inconceivable that the veteran Simmes, backed by the equally astute publisher Edmund Blount, would approach a printer with Read's track record to undertake the printing of Book 3 of Essayes STC18041, an investment risk of the first magnitude for which no expense seems to have been spared. This job alone represents a large multiple of the total known output of Read's tenure. Blunt and Cuthburt Burby also contracted other large jobs in 1603. Parts of Burby's Antichrist STC7120 (entered 22 April 1603), moreover, were then farmed out to Braddock and Creede, an unprecedented move that probably indicates that the shop's resources were pushed to the limit for the first time. One wonders, finally, what in Read's track record led Blount to feel confident about bringing the latin text of Nero STC12551 to the shop. In short, everything points to an Eld takeover of the shop in early 1603. It should be borne in mind that ownership of the shop passed to Eld via marriage to Widow Read, probably after a respectable mourning period. Perhaps the plague of 1603 removed Read from active involvement in the shop's operation, but this is pure ungrounded speculation of the kind so frequently encountered in hypothetical dating efforts and should be treated as such until a church record of Read's interment adds the essential fact that can settle the issue.

[40]

In the absence of an original copy which could yield identified recurrent-types in the target font, the composition of the complementary italic font used for emphasis, speech prefixes and stage directions in dramatic quartos such as Cromwell STC21532 sometimes provides corroborative evidence that the target font is being used by the suspected printer (see discussion, "Font Analysis," pp. 126-127, and notes 42-46). The cogency of such evidence, which usually consists of wrong-face letters in the font, depends upon the uniqueness and longevity of the mixture. This evidence must be approached with caution since mixing of italic capitals is such a common phenomenon. In this instance, the mixed composition of the Read/Eld Granjon pica italic font is fairly distinct and long-term and consistently recurs through Nero STC12551 but not later in Isahaacs STC25643 (entered 16 June 1603). Prose texts such as Ten Learned STC1074 and Essayes Pt. 2 STC5775 place minimal pressure on the italic font, but most capitals appear at least a few times in each so that opportunity exists for the random recurrence of the set of Guyot italic letters (see Vervliet, Fig. 231, specimen IT10, p. 301). Mixed into the Granjon italic are Guyot capitals (swash-A swash-B C E F G H N swash-R S T) and Guyot lower-case 'v w'. The 'C E T' are present in moderate-density proportions and appear in STC1074, Cynthia's Revels STC14773, STC5775, STC26076, STC21532 and STC12551. The 'N' appears in all but STC1074 and STC26076; the swash-'B' skips STC5775 and STC12551; the swash-'A' appears only in STC14773, STC 26076, and STC5775; the 'F' only in STC14773 and STC21532; the 'G' in STC1074 and STC14773; the 'H' in STC14773, STC21532, STC26076, and STC12551; and the 'S' in STC 1074, STC5775, STC26076, and STC12551. Oddly enough, the Guyot swash-'R' appears only in STC5775 (T7v:18). The lower-case 'v' appears in all except STC26076 while the 'w' skips the latin emphasis of STC5775 and latin text of STC12551. An additional discriminant is provided by the normal tall-'C' and swash-'I' which appear frequently in all. Overall, these appearances conform to the pattern of random recurrence that is to be expected in the context of sort-pressure generated by the kinds of texts involved. There can be little doubt that the Read/Eld italic font appears in Cromwell STC21532.

[41]

The minimal font sample found in emphasis appearances presents a special problem that merits comment. The emphasis use in long quotations such as Short-Y1 in English Secretary STC6404 and Robinson-Y1b in A comparison STC18246 or in catechetical dialogues such as Read-S1 in Contemplations STC4662 and Purfoot-Y1 in Caesar's STC18432 usually permits a complete analysis of composition and possibly the discovery of recurrenttypes that lead to the identification of a font. However, an emphasis use limited to names and the like often results in such a limited typographical sample that it is impossible to define a font's composition completely. It is possible to sort out shared sections on the basis of this limited evidence in some cases if different typefaces are used by two or more sharing printers. For example, Stafford-EFb can be distinguished from Allde-S2 in Seven Deadly Sinnes STC6522 in emphasis appearances of the former in sheets BC and O, with Allde-S2 in D-N. Similarly, Jaggard-C2 can be distinguished in T. Deade Tearme STC6496 in D3,4v, E3, F2, G1,2v from Jaggard-S1 in A3-4v, BC, D1-2v,3v, E1-2v,3v-4v, F, G1v,2,3,3v and in many pages of Sermons STC15282 so that the alternating pattern serves as evidence of ownership. If these were shared books, the evidence would be adequate to point a search in the direction of a printer who simultaneously used the two typefaces. Otherwise, the usefulness of emphasis appearances is defined by the quantity of "gross features" evidence that can be located. Read-S1's differentiae, for example, can be noted (along with ornamental stock) in C-E of Sermon STC1454, but an unfamiliar head-piece and factotum appear at B1 along with a shift to 96mm roman emphasis that indicates a second printer. The quite limited Y-font emphasis appearance in A Dialogue STC18892, on the other hand, reveals a few S-face sorts that are not seen in Eld-Y1 until much later. In some instances, the uniqueness of composition is suggestive. Given the mix of Windet-F sorts in Windet-S2, especially the short second-stem 'w' and condensed 'ſh' ligature, an appearance in short preliminaries or as the emphasis font should produce enough evidence to suggest a look at Windet-S2 and thence to a comparison of the target book's font to Windet's equivalent font. For example, several books previously assigned to William Stansby contain short samples of Windet-S2 (see T. Jesuites Play STC21514, A3).

[42]

Scourge STC17486.5 illustrates the potential impact of undetected font samples hidden in ghosted books or shared sections upon font analysis and bibliographical analysis. Despite the imprint of 1599, the actual date of the printing of Scourge STC17486.5 is unclear. Given the fact that the first edition (STC17485) was entered 8 September 1598 and Roberts printed another edition dated 1599 (STC17486), it seems safe to assume that STC 17486.5 was printed sometime later than 1599. The foul-case cluster of Braddock-Y1 in STC 17486.5 suggests a transitional state between earlier works and MND Q1 STC22302 (see "Font Analysis", p. 140 and Table 5, p. 149). First, the resident S-face cluster is consistent throughout 1598-1600 with most sorts appearing in STC17486.5 (A E H I M Q R S V; lacking: P). A similar consistency is seen in the proportion of the damaged 'ſſ' ligature fouling the 'ſl' ligature sort. The italic foul-case cluster is substantially the same seen in earlier works (A B E H I L P S T) and in approximately the same proportions, previously unseen letters (swash-B D F O) appear at low-density levels, and several low-density sorts (C G N Q) do not occur, primarily due to low sort-pressure, but purging also is possible. However, the state of Braddock-Y1 in Downefall of Robert STC18269 and Death of Robert STC18271 (both entered 1 December 1600 after MND Q1), two other texts from 1601 not listed in Morrison/STC, indicates that purging did not occur. Although Braddock-Y1 is used here only as the emphasis font in speech prefixes and stage directions, the complete S-face and italic (excepting 'C G O') clusters occur. In addition, prefixes pull italic 'K R' into the font. It is clear from the increase in the density of italic cluster members that Braddock-Y1 was severely depleted in the upper-case. The transitional state between 1599-1601 is clear from the presence of two portions of the cluster. My ignorance of the hidden sample of Braddock-Y1 in STC17486.5, 18271, and 18269 led to the erroneous conclusion that the 96mm 'C' were purged at the same time as the 96mm 'T' ("Font Analysis," p. 140). Although the 96mm 'T' is absent and presumably was purged, the 96mm 'C' seen in A Short Forme STC12312 (imprint: 1599; see 2D3v, 2E4v, 2G1, 2I2) is still resident in STC17486.5 (see A4, C6v, C7, D7v) and Death (see G1v). In fact, the 96mm 'C' were not purged after A Short Forme but simply skipped MND Q1. Furthermore, the presence of a new portion of the cluster in STC17486.5, along with the residence of the 96mm 'C', provides important evidence about the printing of MND Q1. As I noted in "Font Analysis," viewing the occurrences of small capitals 'a h i t' as "repeated responses to newly developed shortages in these sorts (transient fouling) totally distorts the picture of presswork that emerges, if, in fact, these foul-case letters are resident in the font, i.e., already in the respective sort compartments before setting began" (p. 129; see also note 46). The small capitals 'a h i t' appear in STC17486.5 at roughly the same proportions as in MND Q1, although comparing densities in quarto vs. octavo settings is difficult. In addition, a small capital 'M' appears twice in STC17486.5 but not in MND Q1, although the non-appearance is to be expected if but a single type is involved (as seems to be the case). Death and Downefall, however, exerted heavy pressure on the depleted 'M' sort and required as many as seven small capital 'M' per page (see B4v of Downefall). In addition, small capitals 'f p' appear. The black letter colon likewise appears with the low frequency seen in MND Q1 A and C so that the deluge of 20 such colons in B clearly represents transient fouling that was deliberately purged during the distribution of B (see discussion of a similar fouling/purging process by Short's compositors, "Font Analysis," p. 131). Death and Downefall are ambiguous in this context since no colons were required. (I should note that STC17438 is incorrectly listed as "17483" in Table 4 and Table 5.)

[43]

Noted Islip-Y1a,b appearances: A Treatise STC17291 (1598), T. Dialogues STC21576 (1598), La Nowel STC10964 (1598), Prenobilis STC11196 (1599), A Booke STC3345 (1600), Four Bookes STC13200 (1601), and Le Necessarie STC4719 (1601). Noted Islip-S1 appearances: Abrahams STC18538 (1602), Eirenarcha STC15170 (1602), A Booke STC3346 (1604), Sanctuary STC7113 (1604), Lytylton STC15753.4 (1604), T. Duties STC15155 (1604).

[44]

See comment in "Font Analysis," pp. 122-123, regarding sharing patterns (note as well the alternating pattern of Islip-EFb and the unidentified Y-font in De Missa STC23456 discussed earlier). For example, it makes sense that, in Fools, Eld would print AB (Eld-Y1) and farm out the two-gathering sections CD (White-M), EF (Stafford-EFb) and GH (Eld-Y1) to printers 2, 3, and 4, the final full gathering I (White-M) to printer 5, and do half-sheet K himself. What actually happened makes very little sense: he resumed printing with GH, but then gave I to White, and finished with half-sheet K. The many instances of sharing in anomalous divisions and separated sections suggests that flexibility was more important than tidy logic in Elizabethan sharing strategies. The seriatim order of analysis should be followed as a matter of course.

[45]

In Book 2 of STC19295, for example, Jaggard's fonts alternate within gatherings as follows. In sig. 2B, Jaggard-C2 appears in 1, 2v, 3v-6, 7, 7v; Jaggard-Y1b in 1v; Jaggard-S1 in 2, 3, 6v, 8, 8v. In sig. 2I, Jaggard-S1 appears in 1-2v, 5v, 6, 7-8v; Jaggard-C2 in 3-5, 6v.

[46]

New STC tentatively assigns the book to Lownes, Kingston, and East (in that order). However, Kingston did not use an S-font (section 2) nor did East use a Y-font (section 3). The alternation of Roberts-S1 and -C2 in Hamlet Q2 STC22276 in units of a gathering or more (Roberts-C2: B-D, F, I, N-O2v) creates the appearance of sharing. The only other noted use of Roberts-C2 is in Of the Calling STC19733 where it alternates within gatherings (see note 12 above), a pattern which usually indicates ownership. In general, several appearances of a font in signed books are desirable and should be expected. In this case, supplemental evidence from running-titles could be decisive. In "The Printing of Hamlet Q2," Studies in Bibliography, 7 (1955), 41-50, Fredson Bowers describes an anomalous movement of running-titles across the boundaries established by the alternating fonts: title V moves from I1 (C2) to O1 (C2) to M1 (S1), VII moves from I3 to M4, VIII moves from I2v to M4v, XII moves from L2 (S1) to N4 (C2), XIII moves from L3 to N3, and XIV moves from L4v to N4v. If Bowers's identifications are correct, they indicate setting in one shop and hence confirm Roberts's ownership of the C2 font.

[47]

Incidentally, new STC lists Q3 and Q4 in reverse order on the basis of the corrected signature B3 in 14430.5 (missigned C3 in 14430) which is taken as the later state and thus the later edition. 14430.5 is actually Q3. 14430.5 Q3 is printed on the same job lot of papers found in 14429.5 Q2 except for sheet H while Q4 is printed on entirely unrelated papers. Q4 presents a rare instance in which a short section (bottom half of M3v) had to have been reset in Braddock's shop and transported (or vice versa, M4 transported from Simmes's shop).

[48]

See: k1: F2:15, F4v:21; b1: F1v:32, F4v:1; y1: F1v:11, F3:7; C1: F1:20, F2:1; M1: F1:31, F3v:9; W1: F1v:6, F4v:31. A similar recurrence situation is seen in Fools STC4963 where embossing evidence indicates that the inner forme was first through the press: b1: B1:9, B1v:21; d3: B1v:30, B2v:30; f1: B1v:29, B3:36; f6: B2v:25, B4:34; l3: B1:3, B1v:9; r4: B2v:9, B4:7; w1: B1v:12, B2v:5; y4: B3:6, B4:35; (a7: B2:11, B3v:18?).

[49]

See "A Proposal: The English Short Title Catalogue", Papers of the Bibliographical Society of America, 82 (1988), 333-336.