University of Virginia Library

Allan H. Stevenson and the Bibliographical Uses of Paper
by
Paul Needham [*]

One of the great mathematicians of this century, j. e. Littlewood, wrote of his ancient Greek predecessors that they were to be thought of not as promising schoolboys or "scholarship candidates," but as "fellows of another college," who used a somewhat different vocabulary.[1] His remark suggests a potentially fruitful way of looking at the history of any intellectual discipline: when, within its course, do we find practitioners who are, essentially, our equals?—who, if they were resurrected, we feel we should be in more or less immediate communication with? Consider textual and historical criticism. Although an intelligent concern for the authenticity of written texts can be found even in the pre-Christian era, it seems to me that full "colleagues" in Littlewood's sense can scarcely be found earlier than in the seventeenth century, when they certainly did exist—witness Jean Mabillon and Richard Bentley; or perhaps, for textual criticism, in the sixteenth century in the shape of such scholars as Scaliger or Lambinus; or just possibly in the fifteenth century, with Angelo Poliziano and Lorenzo Valla. This is not, of course, to suggest that minds were less intelligent before these times, only that it becomes increasingly difficult to find true commonality of concerns. Indeed, textual critics today find commonality with the figures just mentioned in large measure because it is the latter who, primarily, shaped the methodology of the discipline.

Who are the corresponding figures, and what the consequent age, of the related discipline of (analytical) bibliography—by which word I

mean the reasoned study of printed books as artefacts?[2] To adhere to an argument already made, I should say that the time span back to our earliest bibliographical "colleagues"—predecessors whose thinking we treat as fully equal to the best of our own day—is considerably shorter. I see Henry Bradshaw (1831-1886) as the first and perhaps greatest of our colleagues.[3] By this very rigorous standard, analytical bibliography would be less than 150 years old. And yet of course, those 150 years of bibliography still have a complex history which is not a self-evident one: it must be studied with the tools of scholarship.

The tribe of bibliographers is not a particularly numerous one. It would be curious to attempt to list all those who have contributed significantly to the bibliography of books printed in European languages: how many names, one wonders, would result? Within that group, the number who have contributed in the most fundamental and long-lasting way, by introducing new concepts and categories of bibliographical investigation, is evidently very small. That honor can belong only to a relative handful. And yet, it is surely true to say that however short the list were made, Allan Stevenson's name would be on it. Stevenson's contribution can be summarized in a single sentence: he codified the scholarly study of paper as bibliographical evidence. If we place that contribution alongside the contributions of Henry Bradshaw, who similarly codified the study of codex structure, and of typography, it becomes evident that an extraordinary proportion of all our various strategies for studying books derives ultimately, in fact, from the thinking, and relatively few publications, of these two men.

The following pages are focused almost exclusively on Stevenson's bibliographical publications, rather than on his character, his career, or his life in general. My concern is to examine in some detail the development of Stevenson's thinking about paper evidence as it emerges from his published writings; and, to a degree, what reception his thoughts found among fellow book students. But it must be acknowledged, however little one explores the issue, that there is an unusually stark contrast

between Stevenson's importance to the thinking of the bibliographical world, and his relative lack of success within the academic community.

Allan Henry Stevenson was born 20 June 1903 in Merlin, Ontario. His family moved in his youth to Texas, where he graduated in 1924 from the Rice Institute, Houston, earning an M.A. there two years later. After several years of teaching at Rice, Stevenson moved to the University of Chicago to begin doctoral studies in English literature. This became a long-drawn-out process: it was twenty years before Stevenson was awarded his doctorate in 1949, at the age of 46. During these years he worked variously as instructor and assistant professor in English at De Paul University, the University of Chicago, and Illinois Institute of Technology. He never rose above assistant professor, and his teaching career ended ingloriously in 1952, tenure-less and untethered. The remainder of Stevenson's life was spent essentially as free-lance scholar, moving from one fellowship, study grant, or temporary appointment to another. His longest uninterrupted stint was of five years, devoted to cataloguing the botanical library of Rachel McMasters Miller Hunt, in Pittsburgh. I strongly suspect that the successful career in radio and television production of Stevenson's wife Rachel Waples (the dedicatee KIM of The Problem of the Missale Speciale) supplied his most reliable fellowship.

The topic of Stevenson's dissertation was the Anglo-Irish playwright James Shirley, and he published in the 1940s a few miscellaneous articles spun off from this work, relating to Shirley's years in Ireland and to his involvement with the first Dublin theater and with the London publishers William Crooke and Andrew Cooke. Stevenson's first purely bibliographical publication appeared in late 1948 in the inaugural volume of Fredson Bowers's annual, Studies in Bibliography (then titled Papers of the Bibliographical Society of the University of Virginia): "New Uses of Watermarks as Bibliographical Evidence." Stevenson's chief concern here was to chart the printing and forme-correction history of one quarto play, Shirley's The Opportunitie (1640). He merged his notes on the sheet-by-sheet changes of paper stock of a sample of a dozen copies with other bibliographical data such as identification of skeleton headlines.

There is nothing in "New Uses" to suggest any specific inspiration or mentor for this unusual concentration on paper stocks as evidence. Stevenson's involvement with paper must, however, have been gestating for some years, for a hint of it appears in his first scholarly article.[4] The

results of Stevenson's calculations as to such questions as edition size, order of printing, chronology, and whether the inner forme or the outer was printed first, are perhaps not entirely successful or convincing. For instance, he found reason to suppose, from the ratios of paper-stock distribution in his sample, that 1,500 copies of The Opportunitie were printed. Thus, where sheet B of his sample showed 6 examples of IHS-marked paper and six of Pot, he assumed that this represented three tokens (half-reams) of each stock, or 750 copies of each. If we grant the premiss, his conclusion follows that "Sheet E is easy: exactly two tokens of IHS and one of bird (8:4); and so is F, with five tokens of IHS and one of bird (10:2)" ("New Uses," p. 159).

But should we grant the premiss? Surely this is false exactitude. The neatness of Stevenson's numbers is primarily owing to the fact that his sample of copies (12) is a convenient factor of the hypothesized print run (1,500), the result being 125—a quarter-ream or half-token. If the sample had been, say, 13 copies, one could not have come up with such seemingly exact extrapolations; and yet the body of evidence would have been enlarged! With so small a sample—less than 1% of the copies (hypothetically) printed—it is entirely possible that the record of paper stocks appearing in a given edition sheet (for which term see below) will not correspond precisely, perhaps not even closely, to the ratio of paper stocks originally used. In fact, it is distinctly more likely that a series of small per-sheet samples will vary somewhat from the original proportions of paper stocks for their edition sheets, than that they will correspond exactly to the original proportions.[5]

A substantial portion of "New Uses," in which Stevenson attempted to use shifts of skeletons (including headlines) to determine order of printing of formes, has a certain whiff of the '40s to it. This was, so to speak, one of the standard ten-finger exercises for well-tempered bibliographers of the time, following the model of Charlton Hinman's and Fredson Bowers's earlier studies of headline evidence.[6] There was little awareness of the far preferable direct test for order of printing of formes:

of the two formes of a sheet, which pushes into the other, and hence comes second? Lest one suppose, by the way, that the test was not "discovered" until some years later by Kenneth Povey, it may be pointed out that German incunabulists had been using this diagnostic, since the turn of the century, to distinguish books printed page-by-page from those printed over multiple-page formes.[7]

In general, critically viewed, "New Uses" was not quite so pathbreaking a study as its title may immediately suggest. In its arguments for edition size of The Opportunitie, in its discussion of forme sequence, and elsewhere, it is a good standard example of the over-rationalization of assumed printing shop procedures that D. F. McKenzie warned against two decades later.[8] Sufficient evidence and warning of the tendency may be found in such of Stevenson's rhetorical formulations as: "We may suppose that [the printer] Cotes's well-ordered establishment . . ." ("New Uses," p. 166); or ". . . as will be seen presently, sheet C must have been printed around Easter; and on the morning of Easter Even or Easter Monday Tom Cotes or the corrector might come to the printing-house late. Now then . . ." (p. 168). The specific reasoning that follows hardly matters; one knows that it passes the bounds of the evidence.

On the other hand, an apparently solid if minor result of Stevenson's research seems to have been overlooked. Shirley's Opportunitie is one of a closely related group of seven play quartos (two by Fletcher, the remainder by Shirley) printed by Thomas Cotes, two of which bear title-page dates of 1639, the remainder of 1640. Stevenson gives good evidence that all seven, sharing a set of small job lots of paper, were printed in the winter to early spring of 1640. The two "1639" title-pages most probably

reflect Old Style convention, with year change on Lady Day 25 March, and are to be interpreted as 1640.[9] This argument has gone unnoticed both by Sir Walter Greg's Corrections to his Bibliography of the English Printed Drama, and by the Revised STC, where the two plays in question remain assigned to 1639.[10] Perhaps even more minor, but bibliographically very pretty, is Stevenson's correlation of the paper stocks with the corrections in sheet C of The Opportunitie, where uncorrected and corrected states may be identified of both the outer (o) and inner (i) formes. He found the following pattern:        

uncorrected o—i:	Pot (3x)
corrected o, uncorrected i:	Pot (3x)
(uncorrected o, corrected i):	none
corrected o—i:	Bird (3x), IHS (3x)

Here we seem almost to be handed a "retroscope," giving a momentary glimpse (if still a cloudy one) into the work patterns of a 1640 printing shop.

Essentially, this first appearance of Stevenson as a critical or analytical bibliographer reveals a strong general interest in and knowledge of paper, and an effective vocabulary for describing watermarks and paper stocks. Possibly, too, it is the first study to introduce the very useful concept of the "edition sheet" ("New Uses," p. 154)—all the copies printed of a particular sheet of a book. But judged by solidity of result, this still looks very much like the prentice effort of a somewhat aged (45-year-old) apprentice. It stands out only because the study of paper was an area where almost no one else in the bibliographical community (the most

notable exception would have been Allen Hazen in his study of the authentic and forged Strawberry Hill books) was even at that rudimentary stage.[11]

There is a significant and tantalizing "delayed opportunity" in the 1948 article. Stevenson wrote in passing ("New Uses," p. 156 note 15), "such [paper] trays were commonly made and used in matched pairs, so that one ordinarily cannot tell their watermarks apart." Here is an instance where a false statement, clearly and energetically phrased, can serve as a springboard to the truth. All that was needed was for that single word "cannot"—twin watermarks cannot be told apart—to change to a "can." For twins can be told apart, and that is the key to all adequate studies of handmade, watermarked paper. In 1948 Stevenson was on the verge of his great discovery; his mind was fully prepared to take interest in the question; but the time was not quite ripe.

There is, of course, no trace of discredit in Stevenson's initial failure to recognize that twin trays—moulds—with twin marks can be distinguished, and that there is a point to distinguishing them. The group of papers he was studying most intently are found in play quartos, hence with marks divided in the gutter, often in copies that would have been tightly bound; and he was dealing with very frequent changes of paper stocks, rather than long runs. It would have been difficult indeed to get pictures of these marks sufficiently accurate that the small details separating one twin from another could be grasped.

Stevenson's "Watermarks are Twins", appearing three years later, marks what is—or at least could have been—a new era in bibliographical study. For the first time, scholars were provided with a coherent conceptual structure for describing and analysing stocks of mould-made paper—a structure comparable with (and in fact closely related to) what Henry Bradshaw had provided, in association with William Blades and G. F. Tupper, a century earlier for handmade types. Paper stocks, Stevenson showed, were to be uniquely defined as products of the twin moulds that produced them—the result of the centuries-old paper mill tempo whereby a vat man and a coucher worked in tandem, interchanging in rotation two separate moulds sharing a single deckle. All complications of specific examples (and there are many such complications) flow in an orderly way from this underlying pattern. Equally new and important in this

study was Stevenson's careful description and illustrated documentation of the changing states that moulds and wiremarks characteristically undergo in the course of their use. Never before had watermarks been examined so closely.

Stevenson's article did not draw a single, striking conclusion from his new rationale for defining and studying paper. In essence, he provided bibliographers with a powerful new weapon for attacking any—that is, potentially any—problem involving documents printed (or drawn, or written) on handmade European paper. For those who have absorbed the lesson of twinship in handmade European paper—who have built on Stevenson's foundations—it is at first a little surprising to realize how unsettled Stevenson's vocabulary was in 1954 for distinguishing twins, and how unclear, at several points, were his underlying concepts. Until this time, virtually all studies of paper were based simply on traced or free-drawn—very occasionally photographed—images of the watermark itself. In essence, they were free-floating images. The best of the watermark albums, Briquet's, did indeed show also the positions of the chainlines adjacent to the marks, but this was still not sufficient to localize the mark, with unambiguous orientation, on its mould. It was assumed (unless otherwise stated, for other placements are possible), that the mark was centered in either the right or left half of its mould. I am not aware of any study which made a point of defining this situation further: which argued that there was any reason to distinguish which half of its mould a mark was centered in, or to take note of which side of the sheet one was examining.

When twin marks are to be identified, a more precise set of concepts and corresponding vocabulary become indispensable. In principle, with only the rarest of exceptions, twin marks are meant to look like each other. The idea, if we can call it that, must have been that to the casual but experienced eye—say, that of the paper maker, who had to keep track of his mould-pairs, or identify his own stocks—the twin marks sitting on a mould or in the sheets of a ream should look enough alike that it would be obvious that they belong together.[12] Thus, if twin marks are to be

distinguished for subsequent scholarly purposes, one must find consistent ways to describe their differences. To see how Stevenson, the first scholar to make this attempt, formed his concepts, it may be useful for comparison to outline my own descriptive system. It is not, of course, the only possible system, nor necessarily the best. But it has the virtue of being the direct descendant of Stevenson's system, and so is in communication with that system.

A sheet of handmade paper is a reflection of a paper mould in much the same way as a printed page is a reflection of the printing surface. The sheet has, of course, two sides: the mould side, which was actually in contact with the wire mesh of the paper mould; and the felt side, as it may conveniently be called, the side which was first laid onto the felt as the sheet was pressed out of the mould. If one examines a sheet with mould side facing, one sees in essence a mirror image of the upper surface of the mould; if one looks at the felt side, one sees in essence the upper surface of the mould itself. Similarly, if one looks at a printed page, one sees a reflection of the printing surface (type-page); if one looks through the page from the other side of the leaf, one sees in essence the raised surfaces of the type-page itself.

The mould, with wiremark sewn on, has a more or less indefinite number of individualizing features, and the question of which of these one records from its "mirror," the sheet, is to some extent arbitrary. One wants the most generally compendious, consistent, and useful details, but these will not, of course, be the necessary details for every conceivable investigation. Again, the analogy with founts of types and their descriptions is more or less obvious. Examples of data one might record are: sheet size, number of chainlines, placement of tranchefiles (supplementary chains placed adjacent to the two shorter sides of the mould), number or "density" of wirelines, apparent thickness of chainlines and wirelines, placement of the wiremark, various dimensions of the wiremark (greatest height and breadth, etc.), irregularities of the chainlines and wirelines, etc. Because a stock is built from twin moulds, one must make double records for each stock, one set for each mould. Besides these records, one needs of course pictures of the watermarks, which

might be freehand drawings, tracings, photographs or radiographs, or (only "discovered" after Stevenson's death) rubbings. Even these pictures, however accurate, are only a poor substitute for the ideal: an actual-size image of the entire sheet.

A record for a single paperstock (one of those supplying the Gutenberg Bible) could look like this:[13]

• Royal: Bull's Head
• a. mR4 | 19 | 14 chains (ave. 42.1 mm) | 18 | (6+)
• b. mL4 (3+) | 21 | 14 chains (ave. 42.3 mm) | 18 | (1+)

Such a record gives the sheet size (Royal), the watermark type (Bull's Head), and a rough description of the sheets turned out from each mould, a. and b. The a-sheets have watermarks located in the right (R) half, in the fourth chain-space in from the edge.[14] Looked at from the mould side, these have a left tranchefile space of 19 mm, then 14 chainlines with an average spacing of 42.1 mm, then a right tranchefile space of 18 mm, then a distance to the deckle edge of more than 6 mm.

In any number of ways the information could be expanded, but several rationales underlie the information given and the way it is given:

With these introductory comments to provide a standard of comparison, we can judge Stevenson's first remarks on identifying twin watermarks. He listed a variety of differentia by which twin marks can be told apart, such as variation in how they are sewn with respect to their neighboring chains, in their slant, in position of sewing dots, and in other significant detail. He outlined a system of measuring position of parts of the watermark with respect to adjacent chainlines: ". . . the crescent or fleuron in the superstructure of a pot may measure :3[13]5: or, more simply, 3[13]5—which means that the part is 13 mm. wide, centered between chains 21 mm. apart, 3 mm. from its outer chain and 5 from its inner chain . . ." ("Watermarks are Twins" p. 69). Such measurements formed a major part of Stevenson's private notes on paper, but the information they convey is so inferior to that supplied by actual-size representations of watermarks that I have difficulty in seeing them as a fundamental element of paperstock description.[17]

But Stevenson's first point, "Difference of Mould-end," is in fact unclear: he seems to have mixed up two independent forms of variation. He first stated, correctly, that twins are often centered in different halves of their moulds (i.e., one is mL and one mR). But as evidence for this, he stated (p. 66) "Actually, this is mainly an inference from the fact that some marks read in and some read out." That is, when marks bear letters, and are so oriented that their lettering reads forward, the lettering may either read "in" toward the center of the sheet, or "out" from the center. However, this distinction does not in fact reveal whether the mark is mR or mL. If the lettered mark reads "in," it is mL if one is looking at the mould side, and mR if one is looking at the felt side. In short, this is not direct evidence for which end of a paper mould contains the wiremark.

The primary evidence is rather what Stevenson here considered secondary: "Frequently the inference may be checked by the indentations made by the watermark or chain wires." The first step in "orienting" a watermark is to identify, from indentations, the mould side. Any mark, if it has an obvious or definable "up," can then be classed, when seen from the mould side, as mR or mL.[18] It makes no difference whether its lettering, seen from that side, reads forward or in reverse. Consider a stock marked with the Arms of England above a label with the maker's name JONES. The twins of such a stock can be any combination of mR

or mL, and forward-lettered or reverse-lettered: four possibilities in all for each. At this stage, Stevenson was still shaping his concepts and his vocabulary. He had no descriptive model to draw on, and it is not surprising if his reasoning was still in part inexact.

A like confusion seems to have led Stevenson astray in discussing the twins of the stock used in the 1619 Pavier quarto The Whole Contention between York and Lancaster. This stock is marked with a single-handed Pot with initials PA. Stevenson identified the twins, but his description (p. 80) is cryptic: ". . . in Pot a the letters consistently read up, and in b as consistently down." Very few readers, one suspects, could have visualized the situation here. What Stevenson apparently meant is that for one twin, the initials PA read forward from the mould-side ("up" meaning mould-side up), and for the other they read forward from the felt-side (mould-side "down"). Stevenson knew the distinction between mould-side and felt-side, but did not yet have the habit of referring to them clearly or the vocabulary for doing so. From Stevenson's description, one still has no way of knowing whether one twin was mR and one mL, both mR, or both mL.

Similar confusions arise in Stevenson's descriptions of several other Pot marks. Concerning a Pot stock with initials G / RO on its bowl he writes (p. 84), ". . . one mark reads up away from the handle, and the other also up towards the handle; so that both appear in the left halfsheet." The conclusion does not follow. If we specify a single-handle Pot with initials as being possibly (a) mR [in right half of sheet] or mL [in left half], (b) with hR [handle to right] or hL [handle to left], and (c) with lettering from mouldside reading forward ("up") or reversed ("down"), there are eight possible arrangements. There are two ways for a mark to read "up away from the handle," of which one way is mR (fig.6) and one mL (fig.2); and two ways to read "up towards the handle," of which again one way is mR (fig.5) and one mL (fig.1). An unambiguous way to distinguish these is by some such nomenclature as mRhL (G / RO), mRhR (G / RO reversed), etc.

Stevenson devoted almost no space in "Watermarks are Twins" to the question of why it is bibliographically useful "to know that watermarks like wrens go in pairs" (p. 88). He did note, however, that a recognition of the "binary nature" of watermarks would have strengthened Greg's famous demonstration that the eight or nine Shakespearean quartos published by Thomas Pavier with various dates 1600, 1608, and 1619 were all actually produced in 1619. Greg's studies on the Pavier quartos had considerable importance for Stevenson, being, although nearly a half-century old, still the most striking exemplification within the Anglo-American

bibliographical world of what Stevenson had yet to prove from his own work: that the analysis of watermarks could lead to important results.[19]

This group of quartos first became a "problem" in 1906, when A. W. Pollard took note of two separate volumes, each containing all nine editions bound together. When looked at together (most copies having been disbound in the past and treated as independent units), a family resemblance seemed to emerge. Eight of the nine contain the same titlepage device, with Welsh motto Heb Ddieu Heb Ddim, and none of the imprints give publishers' addresses.[20] How did these similarities in layout maintain themselves over so long a span of time, and how were at least several copies of all nine apparently sold together? Pollard tentatively supposed that a substantial number of sheets of the five "1600" and "1608" editions remained unsold until 1619, when Thomas Pavier marketed them with four newly-printed editions as a kind of Shakespeare set.

Greg exposed the weakness of this explanation—which Pollard happily relinquished—by studying the watermarks of the editions, and discovering that of some 27 separate marks, mostly Pots, a number seemed to be shared between editions variously dated 1600, 1608, and 1619. He concluded that this went beyond what the invocation of coincidence could support. All the editions must have been produced in 1619, by William Jaggard (possessor at that time of the Heb Ddieu Heb Ddim device). The earlier dates and several of their imprints ("by James Roberts 1600"; "for Nathaniel Butter 1608") were intentionally deceiving. Greg's arguments received independent confirmation two years later when an American scholar, W. J. Neidig, showed additionally that most of the title-pages of the Pavier quartos share standing type.[21]

Under an awareness of twinship, the situation obviously changes markedly, though Greg's argument from paper retains all its strength. Greg recorded 27 "different" marks, mostly Pots, in four sets of the quartos: thus four copies each of 81½ edition sheets, or about 310 watermarks (subtracting unwatermarked sheets). Simple probability suggests that within these parameters various consequences become highly likely. First: some of the recorded marks will surely pair off as twins; the number of stocks represented by the drawings must be less than 27. Second: there will very probably be twin marks either that did not turn up in this relatively small sample, or that Greg overlooked because their differentia are inconspicuous. Third: there could well be fugitive stocks that do not happen to show up in these four copies. After all, nine of the watermarks recorded by Greg appear in his sample in three or fewer examples each.

Stevenson was able to demonstrate at least the first and third of these predictions satisfactorily, and the second is fairly evident.[22] In "Watermarks are Twins" (p. 80) Stevenson "assembled" several stocks from Greg's drawings, e.g. Greg 2 + 5 are a single stock, Greg 15 + 16 another. In a supplementary note, "Shakespearian Dated Watermarks" (1952), Stevenson identified two "new" Pot stocks in the Pavier quartos, both rare and both in rather battered state, of which one Pot has a bowl bearing the damaged date 1608 (probably), and the other the date 1617 (probably). Since these appear in copies of Sir John Oldcastle "1600" and Henry V "1608" respectively, Greg's argument for the falsity of the title-page dates is corroborated.

The exemplary status of Greg's "False Dates" for establishing the bibliographical value of paper evidence was emphasized by Stevenson in various subsequent writings, and Greg seems increasingly to have advanced, in Stevenson's mind, as his bibliographical master. Greg's introductory words from "False Dates," "a happy inspiration led me to examine the paper," provided an epigraph for The Problem of the Missale Speciale.[23]

By an odd chance, in the same year that Stevenson announced paper-mould twinship to the English speaking book world, A. W. Kazmeier of the Gutenberg-Museum, Mainz, made a similar announcement to the German book world, with regard to the paper stocks of the Gutenberg

Bible.[24] Kazmeier made very accurate tracings of the watermarks in the copy of the Gutenberg Bible (volume II only), acquired in 1925 by the Gutenberg-Museum from Count Solms. He organized his tracings into twin marks representing four different Bull's Head stocks, two Grape stocks, and two Ox stocks. This was a major step forward in understanding the Bible's paper supply, building on the researches of Paul Schwenke, who had recorded the paper distribution of 27 copies, but whose cruder drawings had revealed only one Bull's Head stock.[25] The central virtue of Kazmeier's study was his clear recognition that each of the stocks is represented by twin marks. More consistently than Stevenson, he reproduced and discussed the marks according to whether, from the mouldside, they were centered in the right or left half of the sheet. Kazmeier did make an error in his treatment of the Ox stock, through a failure to realize that one of his tracings is a mirrored duplicate of another. He unwittingly took one tracing of an Ox from the felt side, and another of the same Ox as seen from the mould side, and on this basis created a "ghost" second Ox stock.[26]

What, finally, is the significance of twinship? Stevenson did not "discover" the twinship of moulds. Hand papermakers had always known, and continued to know, about twins. Twin moulds are listed in scattered business records of paper mills, and their use is described in various eighteenth-century accounts of papermaking. The great historian of paper Charles-Moïse Briquet wrote with precision of twin moulds, and reproduced several examples of twin marks. Even in Greg's classic study of "False Dates" there is clear reference to twinship, derived from Briquet, and yet Greg manifestly failed to see its application: ". . . technical evidence goes to show that a pair of frames could perhaps be made to last two years . . ." (p. 121). It was Stevenson's achievement to see that twinship had a methodological point: a paper stock is only defined when both its twins can be identified. Once stated, the point surely becomes self-evident. For in the absence of identified twins, one might easily (for example) record one watermark in one book, and a second similar but nonidentical

mark in another, with no way of judging whether one was dealing with the same or two different stocks.

During the following years, Stevenson's greatest energies were particularly applied to cataloguing Mrs. Hunt's important botanical library in Pittsburgh. His scholarly publications, late-begun and never numerous, seem to lack a clear developing shape during the eight or so years after "Watermarks are Twins." It should be remembered, however, that Stevenson's methodological statements on paper evidence were supported by a deep historical interest in all aspects of European papermaking and paper supply, based on widest reading in the published literature as well as a great quantity of original research. This profound groundwork makes itself felt in many peripheral remarks in "New Uses" and "Watermarks are Twins," but is especially evident in Stevenson's "Critical Study" of Heawood's watermark album (1951), in his review of The Briquet Album (1955), and in his lengthy introduction to Briquet's Opuscula (1955).[27] The broad-based but precise knowledge of handmade paper these studies reveal was then unique at least in the English-speaking world, and it may be doubted whether anyone of the present day can equal it. Throughout these years, and after, Stevenson projected an historical monograph which never eventuated, studying the paper supply of English printers in the fifteenth through seventeenth centuries. Its name and scope varied, from "Notes on Norman Papermakers" to Norman Paper and English Books to The Unicorns of Normandy.[28]

One further purely methodological study appeared in the early 1950s, "Chain-Indentations in Paper as Evidence" (1953-1954). Stevenson emphasized here, more forcefully than in "Watermarks are Twins," the importance of distinguishing the two sides of paper sheets. He did not yet possess, however, a consistent terminology. We find the expression "mould side" alongside the synonyms "right side" and "wire side." The term "felt side" does not appear, though its rather confusing trade synonym "wrong side" does. But Stevenson did not settle on any of these. One finds him using instead a variety of locutions: "sequence of grooves" and "groove pattern"; "indentations," "indented side," and "unindented side"; "smooth side" and "rough side."

Concepts need words, and the absence of a stable pair of expressions for the two sheet sides may be partly responsible for another "orientation

error" in Stevenson's discussion: "As always, paper was made on a pair of moulds, and these moulds varied in the position of the main mark—some pairs having their marks in opposite halves, some having them on the same side with one reversed or even inverted" (p. 183). This is illogical. As noted above, the position of a watermark in right or left (mould-side) half of a sheet is premissed on the mark being identifiably upright in one position. If one finds twin marks in two sheets of paper, of which one mark is upright in the right half of the sheet (mR), and the other inverted in the right half, that is equivalent to saying that the latter is mL. Therefore, "marks in opposite halves" represents the identical situation as "on the same side . . . with one . . . inverted."

As to the point of distinguishing sheet sides, Stevenson put most emphasis on the identification of cancels. Cancels may reveal themselves by bearing unorthodox or unexpected signatures, by typographic differences, etc. But the test of mould-side conjugacy (as I should call it) may signal the presence of cancel or supplied leaves even where other clues are lacking. In the general case, an established format for a printed book implies that all the leaves, or specified leaves, of a quire will form parts of the same sheet. This in turn entails that these leaves stand in orderly relation to each other with respect to conjugacies across the sewing fold and (in smaller than folio formats) over the top edge; to mould-side sequences; to watermark positions; and to location of tranchefiles and other indicators of sheet edges. Any disturbance to the required pattern must indicate a cancel or supplied leaf.[29]

The eighteenth-century volume of the Hunt Botanical Catalogue (1961) displays to advantage Stevenson's considerable general skills as a descriptive bibliographer, and his wide-ranging knowledge of European papers. Each copy description includes an expert note on the papers used, with distinction where necessary between the plate paper and the letter-press paper, in the form: "Demy, Genoese, marked FBcorner G and corner flower" (the long dividing marks from counter- and corner-marks). These are, indeed, intended as no more than preliminary notes, with identifications of size, source (papermaker or region of manufacture,) and watermarks. But the notes mark an advance, I believe, with respect to consistency and authority, over what may be found in any

earlier bibliography or catalogue, and continue to supply a useful descriptive model for catalogues which do not take the additional step of reproducing watermarks.

The two most significant methodological points of the Hunt Botanical Catalogue with respect to paper are the interrelated treatments of format and of paper sizes. The book-student interested in paper as a material object can hardly avoid being more sensitive to issues of format than the student for whom the leaves are merely a neutral whitish substance bearing the inked text. An accurate statement of format depends on identifying the fractional relation between the leaf and the sheet, which can hardly be done without a strong mental image of what a sheet is. Stevenson's bibliographical introduction to the Hunt Botanical Catalogue properly pointed out (p. cliv) the persistent error of many catalogues in misnaming as folios what are actually gatherings of broadsheets:

There are, however, additional complications with books constructed of assemblages of single leaves, in handling of which Stevenson's descriptions in the Hunt Botanical Catalogue are somewhat less sure-footed. The single leaves of such books may not necessarily be full sheets of paper, but rather half-sheets (or some smaller fraction); or, they may be a combination of, say, half-sheets of one paper size and broadsheets of paper of half that size. In these cases, Stevenson used such format descriptions as "F° in is" and "mainly F° in halfsheets".[30] A number of otherwise highly sophisticated catalogues and bibliographies continue to be misleading or inconsistent in their treatment of broadsheet/broadside books and of single-leaf broadsides.[31]

As to paper sizes: in his identifications of sizes in the Hunt catalogue, Stevenson was in essence contributing solidly to the post-war Anglo-American tradition already in progress—seemingly inaugurated by Philip Gaskell, The First Editions of William Mason (1951)—of designating paper sizes for eighteenth-century printing. The habit has still not established itself in bibliographical work on other centuries, although it is just starting to make its way in incunable descriptions. For the sixteenth and seventeenth centuries in particular, we continue to lack reliable guides to paper sizes and nomenclatures, such as Gaskell and Stevenson have supplied for the eighteenth century.[32]

An interest in paper sizes emerges at many scattered points in Stevenson's writings, but the process of synthesizing his data into a coherent picture moved slowly and apparently never coalesced into a full published statement. This was one of the chief topics planned for one chapter of his never-completed monograph, The Unicorns of Normandy. The central document for paper sizes in Stevenson's "original" century, the seventeenth, is the anonymous 1674-dated memorandum on prices of paper that was prepared for Bishop Fell.[33] It is instructive to note that Stevenson's first published use of the Oxford Price-List led him into venial error. In "New Uses" (p. 156) Stevenson referred to his study piece, James Shirley, The Opportunitie, 1640, as a "Demy" quarto, based

on its leaf size as compared with the various measures of named sheet sizes in the Oxford Price-List. But this is an excellent lesson in the pitfalls inherent in treating documents as purveyors of pure, unmediated truth. The Oxford Price-List designates, for each stock available, its size name (and sometimes quality), a description of its watermark (and countermark if any), a statement of how many sheets are in its quires, its dimensions in the form "How many inches Long" and "How many inches Broad," and the per-ream price.

However, "Long" and "Broad" must be interpreted. It is clear on examination that these figures represent not full sheets, but the height and breadth of the quire, each quire consisting of 24 or 25 folded sheets; the breadth must be doubled to get the full sheet size. With this correction, it becomes instantly apparent that the untrimmed leaf size of Shirley's Opportunitie conforms to Pott rather than Demy sheets. It would, indeed, have been a rare Elizabethan or Jacobean play quarto that was not printed on Pott-size sheets. By 1951, it is implicitly evident that Stevenson had caught his own error,[34] and he corrected it explicitly in "Watermarks are Twins" (p. 85), in his 1954 review of Labarre, and at some length in his 1955 essay, "Briquet and the Future of Paper Studies."[35]

One particular encounter with paper in 1954 resulted, eventually, in Stevenson's major scholarly achievement. In September 1953 (announced in early 1954) the Pierpont Morgan Library acquired from H. P. Kraus a copy of the famous Constance Missal or Missale speciale, the first (and only) copy to reach America. The Library's publicity suggested that this undated, unsigned book, printed in a variant version of the smaller of the two founts used in the 1457 Mainz Psalter, was to be seen as the first European printed book, pre-dating even the Gutenberg Bible. Its rather

shaky putative assignment was: [Mainz or Basel: Johann Gutenberg, not after 1450.][36]

This opinion was by no means a settled one even in 1954, although there were few alive at that time who were fully engaged in the question. In origin, this was the opinion of the discoverer and first serious student of the Missale speciale in the late nineteenth century, Otto Hupp. Hupp (1859-1949) was a talented graphic artist/engraver and proudly self-educated antiquary, with an exceptional talent for close reasoning and close observation. Besides his contributions to printing history, he became a major figure in the study of medieval German heraldry. As a young man, Hupp acquired an incomplete copy of the then-undescribed Missale speciale from a Munich bookdealer, but he did not publish on it until 1898.

The germ of Hupp's argument, propounded and defended in three trenchant monographs, is that the state of the fount in the Missale speciale is more primitive than in the 1457 Psalter.[37] The formal gothic script imitated or transmuted in the two Psalter founts has a rhythm based on equal-valued minim strokes. The letters c-e-f-g-r-t-x-y have rightward projections which, if not compensated for, disturb this even rhythm. Therefore, the letters written after these "projecting" letters regularly have slightly modified shapes, with reduced leftward projections compensating for their precursors and restoring the regular rhythm. This is virtually the fundamental rule of the gothic hand, and all the earliest Mainz types make attempts to obey the rule by means of filed or specially cast Anschlussbuchstaben—"abutting" letters of variant form—which were set after the projecting letters.

In the 1457 Psalter, both founts provide for these required variants by distinctively shaped sorts. The fount of the Gutenberg Bible, which is not after August 1456, similarly provided a full complement of such variant, abutting letter forms. But in the Missale speciale, using the smaller Psalter fount, abutting characters were generally not set. From these considerations one might easily, by leaning on a few suppositions, develop a kind of syllogistic argument for the primacy of the Missale speciale. The Gutenberg Bible is (let us say) essentially the first printed book, and its fount contains abutting sorts. The Mainz Psalter of (let us

say) the following year likewise contains abutting sorts.[38] But the Missale speciale, though printed with a Psalter type, contains virtually no abutting sorts. Therefore, we must be at the earliest stage of Gutenberg's typographic experiments.

Hupp's views were contested by such formidable scholars as Gottfried Zedler (1860-1945), Konrad Haebler (1857-1946), and, most significantly, Paul Schwenke (1853-1921). It would be impossible to summarize usefully their varying arguments, mixing typographical details with often-unconscious presumptions about what "The Master" Gutenberg would have done, the state of his finances and of his relations with Johann Fust, and so forth, in less than a dozen pages.[39] One seemingly minor and peripheral, but in the long run very prescient, point was raised by Schwenke: the unique copy of the Missale abbreviatum, a shorter issue of the Missale speciale, discovered in 1900, is in a binding surely of Basel provenance.[40] Basel "symptoms" came forward eventually in several other copy histories of the Missale speciale, in textual-liturgical studies of the Missale, and finally in the paper stocks. The idea that the book was printed in Basel was thrown into the mix of opinion.

By the 1950s all the chief disputants were dead, and their conflicting opinions had never reached a generally accepted resolution. There was no one left to argue the case against the antiquity of the Missale speciale from original evidence. The greatest work of Gutenberg scholarship in the immediate post-war years, Carl Wehmer's Mainzer Probedrucke (1948), although deeply concerned in questions of how to identify and date the earliest Mainz typography, circumspectly said nothing whatever about the Missale speciale. The influential historian of Gutenberg questions

Aloys Ruppel saw however in Hupp's thesis a "very considerable internal and external likelihood".[41] True, Ruppel was fundamentally an industrious compiler, with no special gift for bibliographical analysis; but his opinion carried weight as being that of the chief guardian of Gutenberg's flame.

When Stevenson read the news of the Morgan Library's great acquisition, his curiosity was immediately piqued. It must have seemed clear to him that paper evidence would be critical, and that he—perhaps he alone—possessed the mental tools to investigate that evidence. Otto Hupp had already published tracings or drawings of watermarks in the Missale speciale, and Stevenson, then resident on one of his several extended research visits to the Huntington Library, began comparing Hupp's pictures with what he could find in that rich store of incunables. He came across one Strassburg edition in particular, a text of Henricus Ariminensis (Goff H-19), whose long run of Cross on Mounts paper looked to be virtually identical to what Hupp illustrated. This edition, although undated, belonged surely to the early 1470s. At the very end of 1954, while in New York at the M.L.A. convention, Stevenson was finally able, in his words, to "keep his tryst with Constance" in the well-chaperoned Reading Room of the Morgan Library. He put alongside it the Morgan's copy of that same Strassburg incunable and, as he writes, "Within twenty minutes—or was it half an hour?—I knew beyond peradventure that some of the paper in the Missale speciale was the same as the main paper in the Henricus Ariminensis."[42]

And then a half-dozen years rolled by, years when as noted Stevenson's main energies were applied to eighteenth-century botanical plate books in Mrs. Hunt's library—a considerable, and demanding departure both from questions of fifteenth-century paper and printing, and from the editions of seventeenth-century English literature, printed mostly on mediocre Norman papers, with which he was most intimately familiar. I would guess that Allan Stevenson got the shock of his scholarly life when, in the spring of 1960, just as his work on the Hunt catalogue was drawing to a close, he learned that two German scholars, working independently of each other, had published separate arguments in support of a conclusion that he must have drawn for himself, but retained unpublished, a half-dozen years earlier: the paper stocks of the Missale speciale did not come into existence until the early 1470s, and consequently the book could not have been printed before that date. One

scholar, Gerhard Piccard, made his argument from finding the same papers in dated documents; the other, Theo Gerardy, from finding them in a group of five Strassburg and Basel incunables.

Stevenson soon embarked on the task of adding his quota of words to the question. By the beginning of the summer of 1960, he was planning and announcing his own monograph on the subject.[43] There was a definite point to this, first because the community of book scholars was by no means immediately convinced by Piccard's and Gerardy's arguments; and second because Stevenson had considerably stronger bibliographical knowledge in general than did the two German paper historians. Stevenson's evidence, rapidly assembled, was first laid out in a lecture read to the Bibliographical Society in London, February 1961; and was then published the next year, in differing forms, in The Library and in Gutenberg-Jahrbuch. These articles refer again to his projected monograph on the Missale speciale, to be published in 1962 by the Bibliographical Society. As it happens, that book was delayed for five more years, as Stevenson continually revised and expanded his researches.

Why, one must wonder, did Stevenson sit on his discovery for so long? He never explained why the startling find he made in 1954 incubated for so long, unseen to the world, apart from once remarking, "there were several years in which I could make but infrequent use of research libraries."[44] There are, however, several hints in the Kansas lecture of November 1959. Toward the end of this lecture Stevenson brought up the topic of the Missale speciale. He spoke about the Galliziani family, Basel paper makers whom he identified as the producers of the Missale speciale's stocks; he discussed the role of paper-stock distribution in assessing the question of whether the Canon of the Mass was the first section of the book to be printed. He went so far even as to state, "I have found the main pair of Missal Bull's heads in several early books printed at Strasbourg and Basel."[45] But he did not acknowledge directly

that there was a problem with the Missal's supposed "prototypographical" status. He turned away from the topic with a remark whose intent was perhaps nearly opposite to its strict sense: "I would not deceive you. I am not at this moment marshalling evidence whereby I might fix the date of the 'Constance' Missal." These words were spoken only a few weeks before Theo Gerardy's typescript on the Missale speciale arrived in the mailbox of the Mainz journal Papiergeschichte.

Stevenson's remark seems so clearly to call out for interpretation that I think it forgivable to attempt to supply one, although my remarks will be largely speculative. Obviously the Missale speciale had been much on Stevenson's mind since he first saw it in 1954. In view of his fateful day of discovery in the Morgan Library, how could he not have been "marshalling evidence" relevant to the dates of these books? Yet I doubt that Stevenson was simply being secretive or coy before his audience. I would guess that he was, at this moment, still doubtful that he knew enough to form a fully convincing argument against the "Hupp-Morgan Library" date. The generally separatist tradition of incunable studies has left many book-students, working on either side of the 1501 dateline, oddly uncertain of whether the principles of analytical bibliography can be applied to the earlier period. After all, the most eminent American incunabulist, Curt F. Bühler, had published what seemed to be a respectable argument that even the Greg-Bowers collational formula did not apply easily to incunables.[46] One would not have wanted to enter this seemingly mysterious realm of medieval printing unguardedly. Stevenson's hypothesized reluctance or uncertainty may be expressed in an earlier passage of the Kansas lecture, where, after discussing the Galliziani mills in Basel, Stevenson turns away with the remark, "As for me, I like to live in the seventeenth century" (Observations, p. 9).

The seeming authority of Bühler, a convinced "prototypograph" with regard to the Missale speciale, may well have been the most tangible obstacle standing between Stevenson and the gates of the fifteenth century. No reminiscences of Stevenson suggest that he was a shy or taciturn man. If at the end of 1954 he found in the Morgan reading room identical paper in the Missale speciale and in a Strassburg book of ca. 1473, we

may feel fairly certain that before he left the Library he shared the happy discovery with Bühler.[47] Several months later, in his prefatory essay to Briquet's Opuscula, Stevenson laid his first printed clue that something might be amiss in the dating of the Missale speciale. He pointed out the inadequacies of an early discussion of the Missal papers by Briquet (who did not personally examine the Missale speciale), but noted that Briquet "placed himself with those who have held that the Missal is later than the 42-line Bible." He concluded, "The matter is fascinating and might be explored further. . . ."[48]

The supposition that the eminent Bühler and the obscure Stevenson were already in conscious but unpublished conflict over the date of the Missale speciale suggests a point to what would otherwise be a somewhat inscrutable article by Bühler two years later, expressing deep reservations about the validity of paper evidence for dating books.[49] This article is, I think it fair to say, primarily a work of rhetoric rather than close reasoning. Bühler concatenated approvingly the published doubts of a variety of eminent scholars in several fields—Sir Henry Thomas, Dard Hunter, Arthur M. Hind, A. E. Popham, Karl Löffler, Fritz Milkau, Ronald B. McKerrow, Lawrence C. Wroth—regarding the usefulness of watermarks in dating manuscripts, printed books, maps and prints. He did not concern himself with whether any of these men had ever made a competent (by the standards of, say, Stevenson) study of paper stocks. It was sufficient that they were worthies and that they had expressed their doubts. Bühler cited, from other authorities, several apparent examples of decades-long use of the same papers in fifteenth-century printing shops; but these instances are, we now know, entirely inaccurate.[50] Perhaps

most curious of all is that, although the title of Bühler's article refers to "the dates of fifteenth-century books," no specific fifteenth-century book comes under discussion as being at issue. One has almost the impression that this was a sort of set-question from a medieval disputation.

Stevenson makes brief appearances in Bühler's arguments. Stevenson's inferred date of 1477 for Caxton's History of Jason (Goff L-112) was allowed as valid, on the grounds that eleven other writers had already proposed that same date. But since most of these authorities simply took their date from Blades, this does not encounter the question of how to assess paper evidence over against other evidence for the dates of books; it is not a matter of counting authorities. Bühler observed that Stevenson's own discovery of a 1608-dated watermark in one of the Pavier quartos would never in itself have pointed specifically to 1619 as the date of printing. Stevenson would undoubtedly have agreed with this, but would have seen immediately what Bühler had no inclination to consider: that there is a difference between the patterns and conditions of use of substantial stocks of paper, and those of exiguous remnants.[51] Bühler admitted, finally, that watermark evidence could be a tool for dating an incunable "possibly within a score or so of years"—which as it happens is just about the interval between his favored date (ca. 1450), and Stevenson's, for the still-unmentioned Missale speciale. He concluded, "It has not been demonstrated . . . that watermarks provide the incunabulists with that absolute criterion which some filigranologists believe to see in them." But which filigranologists are those, and where did they make their "absolute" remarks? The reader is left to wonder. I suspect that Bühler's goal was to present a construct of doubt about the use of watermarks in dating incunables sufficiently colorable that he should be relieved of the burden of ever having to think about them again.

Methodologically, the single greatest failing of Bühler's argument was a neglect to consider carefully the distinction between watermark evidence as providing a terminus a quo, and a terminus ante quem. Watermark evidence may reflect usefully on both termini, but it is much harder to oppose the indications for a terminus a quo than for a terminus

ante quem. Consider: one wants to date an undated book to, say, 1460, but evidence emerges that its paper stock cannot be found in dated use until 1470, when various examples occur. One may be tempted at first to argue that the search for dated uses has been an imperfect one, and that many examples earlier than 1470 may yet be found. But if the search continues and they do not emerge, the wished-for date of 1460 becomes greatly doubtful. The book cannot be printed until its paper is made, and if the evidence becomes strong that the paper was made in 1470, that becomes the starting point of a range of possible dates.

Suppose, on the other hand, that one wants to date the book to 1480, but again its paper is found in dated use only in 1470. It may be unreasonable or arbitrary to date a book ten years after the time its paper was made, but it will never seem impossible to make the argument. One simply postulates (whether correctly or not is a separate question) a period of storage for these particular reams. When Bühler allowed the possibility of using paper evidence to date an incunable "possibly within a score or so of years," he failed to define "within." Did this mean a score of years after the earliest evidence for the paper's existence, or a range of, say, ten years on either side of that date? The distinction is critical, and the latter interpretation could hardly be supported.

The question of chronological termini from paper evidence became critical when Piccard's and Gerardy's studies of the Missale speciale appeared in early 1960.[52] Before this pair of publications, over decades of heated controversy, the bibliographical world knew, and paid attention to, the Missale speciale papers only from three tracings supplied by Hupp, showing a Cross on Mounts and two Bull's Heads with Tau (S. Anthony Cross) surmounts, the one head notably broader than the other.[53] Piccard and Gerardy both succeeded first in defining the stocks adequately by identifying twins, and then in finding dated or datable uses of the same stocks in various paper documents and incunables. Hupp's Cross on Mounts tracing had, of course, a twin. His two Bull's Head tracings turned out to be twins of a single stock (hereafter BHT I), and Piccard and Gerardy found also a second, less common Bull's Head Tau stock (BHT II).[54]

Piccard, who had for years been taking tracings of watermarks from

documents in many archives, identified six documents written on BHT I, dated between 1472 and 1474; two on BHT II, dated 1473-1474; and seven on Cross, dated 1473-1476. Gerardy, whose archival researches were less far-ranging, found three more documentary uses of BHT I, dated 1474-1477. His main clue, however, was provided by an earlier album of watermarks in Strassburg incunables, edited by Paul Heitz.[55] Gerardy recognized (as had Stevenson) that all three watermarked stocks of the Missal were represented by Heitz's tracings. By tracking down Heitz's references, Gerardy found four Strassburg and Basel incunables using one or more of the Missal stocks (Goff E-166, G-399, H-19, V-2). Although the four editions are without imprint, the consensus of incunable literature dated all of them to approximately 1472-1474. Besides this group. Gerardy found one Basel incunable, dated 4 December 1478 (Goff P-452), whose stocks include a later state of BHT I, with both wiremarks showing deformations.

If one accepted that Gerardy and Piccard had accurately identified all these marks—and both of them were fully aware of the principle of defining a stock by its twin marks and moulds—then it must have been clear that there now existed massive evidence that the Missale speciale could not be earlier than 1472, and was very probably datable to about 1473-1474. Piccard's documentation and Gerardy's did not overlap at a single point, yet both arrived at the same dates. If these three paper stocks were hypothesized to have existed as early as 1450 or before, where could all the earlier datable uses be hiding?

The immediate response to Piccard's and Gerardy's findings was, however, rather dusty. At the moment their articles appeared, Schmidt-Künsemüller was just coming to the end of his laborious, highly detailed "Forschungsbericht" on the Missale speciale question. This study, distinguished by a clear grasp of the multitudinous opinions concerning the evidence of the Missal's type, seemed to have worked itself into an awkward corner. The Missal type could apparently be defined with some accuracy as an early state of the smaller fount used in the Mainz Psalter with colophon signed by Johann Fust and Peter Schöffer, 14 August 1457 (Goff P-1036). As Schmidt-Künsemüller correctly remarked, a distinction must be drawn, however, between the date of the fount, and its time of use.[56] The other chief factor to be considered was the wide

variety of Basel "symptoms" in the bindings and provenances of the known copies of the Missale speciale and Missale abbreviatum. This led Schmidt-Künsemüller to postulate two highly conflicting alternatives. The Missale speciale might date to ca. 1444-1448, the time of Gutenberg's "lost years" when he could be located neither in Strassburg nor in Mainz; or to ca. 1465-1468, when (as it then seemed) printing was first being introduced to Basel through the person of Berthold Ruppel, plausibly identifiable as the Bechtolf von Hanau who is recorded as one of Gutenberg's workmen in 1456.[57] Since a full two decades separate the alternatives, this was not an entirely happy conclusion.

It is fair to say that when the "Wasserzeichenforscher" laid their wares before Schmidt-Künsemüller, producing a date distinctly different from either of his painfully arrived-at alternatives, he was instinctively disposed to resist their conclusion. He was annoyed by Piccard's dogmatic declaration that watermark evidence is purely objective and conclusive; and he rightly rejected Gerardy's suggestion that the Missal was probably printed in Strassburg by Georg Reyser.[58] Yet his own dictum, that "the span of use [of watermarks] is certainly greater than the watermark specialists would have it," is just as "apodiktisch" as any of Piccard's remarks, and considerably less justified by an identifiable body of evidence.[59] It should have been clear from Piccard's and Gerardy's data, unless one doubted their fundamental accuracy, that 1472 stood forth strongly as a terminus a quo, however long after that date one might suppose that the paper stocks continued to be used. Schmidt-Künsemüller closed with a declaration that "the riddle of the Missale

speciale is unsolved," and that one can only hope for a "new archival find" to solve it.[60]

Stevenson's own researches on the Missale speciale were laid out briefly in his articles in The Library and in Gutenberg-Jahrbuch, 1962, and in full in the long-delayed Problem of the Missale speciale, 1967. In the Missale speciale he found the "problem" that most fully vindicated his earlier apostolic writings on the value of paper evidence, and that equaled or surpassed in significance the master Greg's study of the Pavier quartos. By normal standards of evidence, Stevenson's article in Gutenberg-Jahrbuch, coming in the trail of Piccard's and Gerardy's work and enlarging to nine the number of other incunables of the 1470s containing Missal paper, should have fully convinced the bibliographical world that the terminal year from which to begin considering possible dates for the Missale speciale was 1472.

If for no other reason, it should have been clear in 1962 that this "paper stock" date fits very well indeed with the scattered and independent copy-history indications. Long ago, as briefly noticed above (note 40), Paul Schwenke had shown that the binding of the unique Missale abbreviatum shares tools with several bindings surely localizable to Basel, these bindings being on Basel imprints of the 1470s. The Morgan copy of Missale speciale was in a binding which had once contained documentary waste datable to Basel, 1473.[61] Most recently, a fourth copy of the Missale speciale had been identified among the so-called duplicates of the City Library of Augsburg. This copy, formerly belonging to the Augustinian convent of Augsburg, is in its original binding from an anonymous shop of that city whose "floruit" was ca. 1470 to 1489. A binding from the shop in the City Library of virtually identical layout covers two incunables of 1475 and 1476.[62]

These elements of copy history had never been given their proper weight. As already suggested, the various "Basel symptoms" among them had been noticed, but the fact that they likewise suggested a relatively late date was generally overlooked. It surely is a valid criticism of Hupp

and other adherents of the "early" school that they were able so comfortably to see the Missale as both Gutenberg's first book, and as probably a Basel book, despite the entire absence of evidence that Gutenberg was ever in Basel. They should have seen a danger signal here, warning them not to place one assumption on top of another and call it a foundation. The warning was lost, however, on those who were determined to believe whatever it was necessary to believe in support of a preconceived conclusion.

However bright or flickering the light of pure reason may have been in the bibliographical minds of 1962, it is fortunate that Stevenson was impelled to proceed with a full monograph on the Missale speciale. The question of the Missal's date was one of the major cruxes in the history of the invention of printing, and well deserved, after more than six decades of controversy, a definitive statement. The resulting book, moreover, gave Stevenson an opportunity to display the full range of his bibliographical and historical talents, which were by no means confined to studying watermarks.

As to paper evidence, perhaps the chief methodological advance of The Problem of the Missale speciale was its thorough exposition of the evidential role of changing states of watermarks and consequently of paper stocks. Stevenson's "Watermarks are Twins" had illustrated and discussed the changing states of watermarks in Inigo Jones's Stone-Heng, 1655. He showed that the reams used to print that edition contained a variety of states of the handsome Pot watermarks, reflecting a process of shifting and deformation of the wiremarks, as the moulds suffered their quotidian knocks. Some of these changes of state must have involved resewing of elements of the wiremark. Although the Stone-Heng provided an instructive example of the pure phenomenon, Stevenson drew no bibliographical argument from it. But it is surely safe to assert that no paper stock had ever been so thoroughly investigated before, with regard to the small changes of state, over the passage of time, reflected in its reams.

A sensitivity to the possibility of finding variant states of a paper stock greatly expands the potential for using paper evidence effectively. When it is asserted, for instance, that "the same" paper is used over a long period of time, perhaps the first question to investigate (given that we are truly dealing with the product of a single pair of moulds) is whether this is all paper in the same state; or whether there are variations of state reflecting different times of manufacture. Changes of state lie on a wide spectrum of variation, from small shifts in the position of the wiremark, still preserving the locations of its sewing dots; to reshaping of the mark, often with changes of some of the sewing dots; to removal of the wiremark,

and complete resewing—perhaps with the mark put back on the mould front to back, hence mirrored; and finally, to the transfer of one or both of the wiremarks to a new mould or moulds. At this latter point, we are dealing with the same marks, but a new stock. The situation is further complicated by the fact that, as Stevenson showed clearly in Stone-Heng, each mould will exemplify a separate series of changes. Stevenson documented examples of all these types of variation at some place or other in his writings, and post-Stevensonians have come to realize that this is the common way with handmade paper.

In the Missale speciale, changing states play the significant role of vitiating one possible objection from the doubters. The Cross paper of the Missal apparently remains in the same state in all recorded uses; but the two Bull's Head stocks both witness changes, and in both instances the Missal states are not the earliest known. A clearly earlier state of BHT I is found in substantial runs in Eggestein's editio princeps of the Legenda aurea of Jacobus de Voragine (Goff J-81); and a slightly variant earlier state in the same printer's edition of S. Gregory's Dialogi (Goff G-399). This latter edition also shows an earlier state of BHT II than occurs in the Missal. The implications are useful. In the theoretical case, one might argue that a pair of moulds was put to use in, say, 1450, to produce a run of paper, then laid aside for a couple of decades before being used again. But when we find, as with the Missal stocks, that an inclusive range of states of the papers, including earlier states, came into use within a year or so of each other—and this is, indeed, the usual situation—the theoretical argument loses its force.

Since the appearance of The Problem of the Missale speciale, the Piccard-Gerardy-Stevenson interpretation of the chronology of the Missal stocks has been accepted as conclusive. Their combined investigations have discovered one or other of the Missal stocks in more than two dozen paper documents written in many parts of Germany and Switzerland, and (often in long runs) in eleven other incunables printed in Strassburg, Basel, Speyer, and Mainz. None of these "outside uses" can be dated before 1472. Although the Strassburg incunables in which much Missal paper is found are uniformly undated, no sceptical incunabulist has had the energy to try to put together an argument that these books, like the Missale speciale, in fact belong to the 1440s or 1450s, not 1470s! Chance subsequent discoveries have confirmed the chronology of use of these papers. Stevenson argued that Eggestein's edition of Voragine Legenda aurea probably dated to 1472, and was followed by the edition of Gregory, Dialogi. At the time he wrote, no dated purchaser's or rubricator's inscriptions were known for either book, to verify or cast doubt on these inferred dates from the state of the paper stocks. However, it may

now be noted that the Donaueschingen copy of Legenda aurea contains a purchase inscription of April 1472; and that a copy of Gregory's Dialogi acquired by the Morgan Library in 1986 contains a rubricator's inscription dated 1473.

Perhaps the sole informed doubter by the late 1960s was Curt Bühler, who had invested considerable emotional energy in his belief and argument that the Missale speciale was to be rightly seen as the first European printed book. He pushed hard a long-standing argument within the Missal Controversy that the Missal's omission of an office for the Feast of the Presentation of the Virgin (21 November) proved that it must have been printed before 30 August 1468, when the Archbishop of Mainz proclaimed that the Feast was of obligatory celebration.[63] And yet it is clear that the Missale speciale, a very flawed production in many ways, was not produced under diocesan authority or supervision of any kind. Few besides Bühler have experienced difficulty in believing that the Missale speciale was simply set from a manuscript which, rightly or wrongly, commonly or uncommonly, omitted this office. Bühler also came forward with more general doubts on the validity of paper evidence, but it may briefly be said that these do not meet the specific case.[64]

Although this study is not concerned specifically with the history of the Missale speciale controversy, it seems necessary to make some remark on whether the paper evidence is in conflict with the typographical evidence. An extensive line of scholars supposed, or at least felt the force of the argument, that the (general) absence of abutting forms in the Missal indicated its Psalter type to be in a very primitive state. The scholar most informed on the changing states of the Psalter founts, Sir Irvine Masson, essentially agreed with this interpretation, although he believed that this served to take the Missal-state of the fount back only to about 1454, not to the 1440s. Stevenson showed that even Masson overlooked a good many instances of abutting sorts in the Missale speciale, created by filing "primary" sorts rather than by special castings. It remains true, however, that broadly speaking the Missale speciale

lacks abutting forms, creating what to the sensitive eye is a strikingly loose, arrhythmic page that seems to violate the gothic modulus.

There is still work to be done with the Missal fount, now that we are released from the presumption that it must be primitive. It is possible that it does represent a primitive or experimental form of the Psalter type from the early 1450s, which migrated to Basel (very possibly via Gutenberg's workman, Bechtolf von Hanau = Berthold Ruppel) and was used about twenty years after its creation. It should be considered, however, whether we see rather in the Missal primarily the habits of a compositor who was in 1473 no longer used to obeying the "gothic rules." Although the history of the phenomenon remains to be fully explored, it is evident that the 1470s was a time when many type founts, and compositors, began to move away from the attempt to conform closely to scribal habits. Types tended to show a reduced number of sorts having special "positional" setting rules, and compositors became more careless of or oblivious to such rules.[65] It may be, therefore, that in the loose settings of the Missale speciale we see not signs of a primitive fount, but rather "1470s" compositorial symptoms.

In 1959, Stevenson told his audience, "I like to live in the seventeenth century." But in the last decade of his life, he lived mostly in the fifteenth, and in Europe. A series of grants and the support of many of the staff of the British Museum, and of the Hellingas in Amsterdam, permitted Stevenson to take up a variety of intensive researches in fifteenth-century topics besides his Missale speciale work. He was energized by the stimulus, after the Pittsburgh years, of access to the great European libraries. Perhaps equally stimulating to his imagination were visits to the traditional French papermaking centers of the fifteenth to seventeenth centuries, and the opportunities to work in a number of local archives where he often found particularly early examples of the output from the twinned moulds of neighboring papermills.

Much of the work undertaken in these years remained unfinished and unpublished. The most ambitious project was that of investigating the long-controverted dates of Netherlandish and German blockbooks. These picture-books, printed onto paper from woodblocks by a process of rubbing, are with very few exceptions unlocalized, unsigned, and undated. They seem characteristically to have been made not in single substantial editions, but in numerous smaller impressions—perhaps, even, to individual order in some cases—each impression being defined

by its own paper stock(s). When the record of the paper supply is correlated with that of progressive damage to one or more blocks in a blockbook set, many potential clues to both relative and absolute chronology and to localization will emerge. It was Stevenson's ambition to build up a corpus of paper-stock analysis of essentially all surviving copies of all blockbooks. His preliminary views, based on an examination of about one hundred blockbook copies, were made public in lectures at the University of Amsterdam at the end of 1965. Stevenson was able to verify from its paper stocks that the unique Rylands Library copy of Apocalypse I was indeed apparently printed in c. 1451-1452, but suggested convincingly that most impressions of the main series of blockbook texts—Apocalypse, Ars moriendi, and Biblia pauperum—belong rather to the 1460s. Stevenson's Amsterdam lectures represent work in progress; the full monograph was never completed.[66] Only recently have students of blockbooks begun to follow Stevenson's path of assembling and organizing the material evidence.[67]

Methodologically similar studies, combining examinations of the paper in books with archivally-based searches for the locations of the mills producing the papers, led Stevenson likewise into the way of Caxton studies, where his work, though suggestive and indicative, was so preliminary that we can see that its first results are almost all faulty.[68] Stevenson was increasingly ill in these years, and time was running out; his handwriting, always shaky, deteriorated strikingly. His most significant completed work was a study of the printed books containing paper from the mill of John Tate, the first English papermaker;[69] and his re-edition of Briquet's watermark album, where his lengthy introduction provides both a guide on how to use Briquet's album, and an attractive

sampling of case studies of how to put paper evidence to use. Stevenson's final publication was an argument, from the provenance of paper stocks, in support of the Hellingas' hint that an edition of Boccaccio, Genealogiae deorum (Goff B-750), conventionally localized to Cologne, might in fact be a Louvain book, representing the earliest stage of Jan Veldener's printing in Louvain.[70] Stevenson died in Chicago, 31 March 1970, shortly before this last piece appeared.

He left much unfinished work behind. The blockbook studies, if they had been brought to the state of maturation of The Problem of the Missale speciale, would very probably have been seen as Stevenson's crowning achievement, for here too is an area where the introduction of competent paper evidence drastically changes the status of the question. The uncompleted Unicorns of Normandy was an equally original attempt to write an historical study of how the paper trade affected the printing trade of a country over more than two centuries. No equivalent book seems to exist anywhere; and for the English printing trade, this is probably a study that only Stevenson could have carried through.[71] However, the work Stevenson did accomplish provided the bibliographical world with what is in essence a complete textbook, enlivened by numerous subtle and pretty examples, of how to use paper evidence effectively. Within the history of bibliography there is very distinctly a pre-Stevenson and a post-Stevenson era.

I have suggested—without, admittedly, being able to document this from Stevenson's own words—that the appearance of Piccard's and Gerardy's studies of the Missale speciale in 1960 was the critical event in Stevenson's intellectual life, impelling him, at the age of 57, to bring to fruition some major achievement. In the long run it matters little whether Stevenson was first, second, or third into print with an argument that the Missale speciale was printed in the early 1470s. His eventual monograph, The Problem of the Missale speciale, became the definitive statement of the question, and Stevenson lived to see the book acclaimed as a masterpiece.[72] It is, indeed, one of the classics of bibliographical

literature. To my mind, among those few classics, at least in the English language, The Problem of the Missale speciale stands alongside John Carter and Graham Pollard's Enquiry into the Nature of Certain Nineteenth-Century Pamphlets (1934) as a work that invites repeated reading not simply for the importance of its results, its decisive contributions to bibliographical method, its exemplification of the bibliographical mind at work, but also for the sheer pleasure and vitality of its prose.

Stevenson was not, quite obviously, a great success in the aspects of academia that have to do with finding jobs and establishing a respectable position for oneself in the company of what are sometimes loosely called peers. He was a slow developer, whose really significant work was not done until in his fifties; academia is not generally receptive to slow developers. His three applications to the Guggenheim Foundation were uniformly unsuccessful. But in the matter of knowing how to engage his readers, how to communicate to them that scholarship ought to be, whatever else it is, a source of mental pleasure—Stevenson was in a class by himself. He was a lord of language, projecting an authorial persona seemingly ever in high spirits at the sheer innocent joy of finding things out: "On a perfect morning in July, when I might have been bowling on the greens at Arroyo Seco, I made a minor Shakespearian discovery."[73] When he writes about the Bull's Heads of the Missale speciale, a fantasy world opens out and we are led into the Edenic fields of "Filigranistan":[74]

With regard to authorial persona, I am particularly struck by Stevenson's article in the 1962 Gutenberg-Jahrbuch, "Paper Evidence and the Missale speciale," where he first presented in some detail his own researches into this hitherto mysterious book. When one considers the disappointment Stevenson must have felt at having been anticipated, in

print, in his amazing discovery, and general disappointment at his very ambiguous career, one can think of a dozen ways that disgruntlement, envy, injured pride, defensiveness could have crept into his words. Add to this the undeniably more austere tradition of the German scholarly forum. Who but Stevenson, at that time and in that place, would have announced his arrival on the scene of Missale speciale battles with these words:

Allan H. Stevenson's Chief Writings on Paper Evidence

• "New Uses of Watermarks as Bibliographical Evidence," Papers of the Bibliographical Society University of Virginia [Studies in Bibliography] 1 (1948-49): 149-182.
• "A Critical Study of Heawood's Watermarks," PBSA 45 First Quarter (1951): 23-36 [offprint separately paginated].
• "Watermarks are Twins," Studies in Bibliography 4 (1951-52): 57-91; Addendum, p. 235.
• "Shakespearian Dated Watermarks," Studies in Bibliography 4 (1951-52): 159-164.
• "Chain-Indentations in Paper as Evidence," Studies in Bibliography 6 (1953-54): 181-195.
• [Review of E. J. Labarre, Dictionary and Encyclopaedia of Paper and Paper-Making (1952)], The Library 5th ser., 9 no. 1 (Mar. 1954): 59-63.
• "Briquet and the Future of Paper Studies," Introduction to Briquet's Opuscula (Hilversum, 1955): xv-l [and offprinted].
• [Review of The Briquet Album, ed. E. J. Labarre (1952)], Library Quarterly 25 no. 1 (Jan. 1955): 132-135.
• [Review of Thomas Balston, William Balston Paper Maker 1759-1849 (1954)], The Library 5th ser., 11 no. 3 (Sept. 1956): 213-215.
• [Review of Thomas Balston, James Whatman Father and Son (1957)], in The Library 5th ser., 14 no. 1 (Mar. 1959): 58-61.
• Observations on Paper as Evidence (University of Kansas, 1961). 28 pp.
• "A Bibliographical Method for the Description of Botanical Books," in Catalogue of Botanical Books in the Collection of Rachel McMasters Miller Hunt, vol. 2, parts 1-2, compiled by Allan H. Stevenson (Pittsburgh, 1961): part 1: cxli-ccxl [and offprinted with added half-title].
• "Paper Evidence and the Missale Speciale," Gutenberg-Jahrbuch 1962: 95-105.
• "Paper as Bibliographical Evidence," The Library 5th ser., 17 no. 3 (Sept. 1962): 197-212.

---

64




• "The Quincentennial of Netherlandish Blockbooks," British Museum Quarterly 31 nos. 3-4 (Spring 1967): 83-87 + plates 24-27 [and unpaginated "preprint", of same setting, but with different page layout and unnumbered plates (Autumn 1966)].
• The Problem of the Missale Speciale (London: Bibliographical Society, 1967). [The Bibliographical Society issue was printed on Warren Olde Style Laid; there was also an American issue of 150 copies on Flemish Laid Book, signed on p. 400 by Stevenson, and with imprint "PITTSBURGH | THOMAS C. PEARS III | 1967"].
• "Beta-Radiography and Paper Research," VII International Conference of Paper Historians, Trinity College Oxford, 24-29 September 1967, Communications (Oxford, 1967): 159-168 [and offprinted; reproduced from typescript].
• "Tudor Roses from John Tate," Studies in Bibliography 20 (1967): 15-34 [the unpaginated leaf of watermark reproductions was omitted from the offprint].
• "Caxton and the Unicorns," [London: British Museum, 1967; a mimeographed handout of 4 pages].
• "Introduction" to Charles-Moïse Briquet, Les filigranes (Jubilee Edition, 4 vols., Amsterdam, 1968), 1: 16 *-36 *, followed by further supplementary material.
• [Review of Peter Drach, Rechnungsbuch, ed. Ferdinand Geldner (AGB 5 [1962], 1-196)], The Library 5th ser., 23 no. 1 (Mar. 1968): 63-65.
• "The First Book Printed at Louvain," in D. E. Rhodes, ed., Essays in Honour of Victor Scholderer (Mainz, 1970): 402-406.
• "The Problem of the Blockbooks," in Blockbücher des Mittelalters (Mainz, 1991): 229-262.

Notes