University of Virginia Library


Page 48

Fists and Filiations in Early Chaucer Folios, 1532-1602
Joseph A. Dane

Thynne's 1532, double-column folio edition of Chaucer is the first of the series of black-letter folios constituting the early "vulgate" Chaucer.[1] The interest of Chaucerians in these editions has been two-fold: for some, they provide a record of canon-formation, with each subsequent edition adding to its predecessor.[2] For others, the interest is almost exclusively textual-critical, with the value of each edition based on the off-chance that it may be an independent witness for manuscript readings now lost. The following article looks at what might be called the extra-textual tradition of these editions--their page layout, line composition, and the mysterious marginalia typeset in the early editions' texts of the House of Fame.[3] Examination of these elements first provides evidence for printer's copy used in the various editions (a question often obscured by editorial concentration on textual matters) and secondly illustrates a process of rationalization, whereby printers reinterpreted details of their tradition they understood no better than we do.

The history of this series of editions was sketched as early as the eighteenth century, and finally presented with uncommon clarity by Thomas R. Lounsbury in 1892.[4] What is generally accepted today is the following: the


Page 49
1542 and 1550 editions (TH2 and TH3) are set from the 1532 edition; the Stow edition of 1561 (ST) is set from the 1550 edition; the Speght edition of 1598 (SP1) is set from the 1561 edition; the Speght edition of 1602 (SP2), although a reprint of 1598, is to some extent based on the 1561 edition as well. This simplified description shows that the relation between the various editions is not strictly linear, but sometimes "leapfrog," although even the most careful scholars of these editions occasionally imply otherwise.[5]

The complex relations among the later folios need clarification, particularly in view of the recent collations provided by the Variorum Chaucer. And in the initial sections below, I focus on what can be conjectured as the printer's copy for the two Speght editions--an entity to be distinguished from the various textual-critical entities known as "copy-text."[6] I will suggest that the printer's copy for a great part of the 1602 edition is not simply a copy of the 1561 edition, but the same copy previously marked up to serve as printer's copy for the 1598 edition.

Printer's Copy for the two Speght editions: Prose Sections

Since the 1602 edition contains revised versions of much of the preliminary matter from the 1598 edition and two texts not in Stow's 1561 edition ("Chaucer's Dream" and "Flower and the Leaf," fols. 355-368), some form of the 1598 text served as printer's copy for these sections in 1602. If we consider the 1602 edition alone, it might seem improbable that Speght would use the 1598 edition to set these sections, then use a 1561 edition as printer's copy for the remainder of the edition. Yet the bibliographical evidence from prose


Page 50
sections proves that he did precisely that. And even though there is no a priori reason to assume Speght acted rationally, if Speght had a copy of the 1561 edition readily available--the one used to set the 1598 edition--this procedure makes perfect sense.

Because the collations so far done by Lounsbury and by Variorum editors have been confined to the verse sections, the only kind of printing influence that has been detected is a "page-by-page" set-up or what Andrew calls "articulation" (meaning, I think, headings, section breaks, etc.). It is easy enough to determine, say, that the 1598 edition is set from the 1561 edition (since the page length is the same), but since the 1602 edition employs a longer column, it seems to have no relation in terms of page layout to the earlier editions.[7]

A comparison of the uncollated prose sections is, however, decisive. Throughout most of the prose, the 1598 edition is a perfect, line-by-line reprint of the 1561 edition. And since the 1561 and 1598 editions are so similar, any apparent relation of the 1602 edition to either seems to prove nothing. Consequently, some Chaucerians have concluded that the printer's copy cannot be determined for individual sections (so Plummer, Summoner's Tale, 85). Yet examination of the compositing details in these prose sections proves absolutely that the 1602 edition used the edition of 1561 as a printer's copy, just as did the 1598 edition.

Although the 1598 edition is a line-by-line reprint of the 1561 edition, at the end of sections of prose, that line-by-line correspondence occasionally grows slack. The reason is fairly obvious. The purpose of a line-for-line reprint is not primarily to aid a compositor (that it actually interfered with the 1598 compositor's work is shown by the erratic spacing between words occasionally necessary to keep that line-by-line correspondence).[8] It is rather an aid for the printer in casting off copy--something that can be done very precisely if compositors are instructed to reproduce the copy line for line. A compositor working this way can relax when approaching a section break, since there will be some leeway in the setting of the final line preceding a new paragraph. Consequently, in prose sections, the line-for-line correspondence between the 1598 and 1561 editions every few pages tends to drift apart toward the end of paragraphs and section divisions. I have checked those particular sections where the two editions differ, and here, the 1602 edition


Page 51
often comes into perfect line-for-line correspondence with the edition 1561.[9] (Had the 1602 edition been set from a copy of the 1598 edition, we would expect the two to agree in line composition against the 1561 edition at the end of paragraphs; but they do not.)

In determining printer's copy, the evidence from line composition is more decisive than evidence from accidentals or substantives. In the prose sections, Speght indisputably printed from a copy of the 1561 edition in 1602 just as he did in 1598. Any agreement in accidentals grouping the 1598 and 1602 editions against the 1561 edition is thus the result of the same compositorial or house style being imposed on the printer's copy of both; any agreement in substantives producing these same groupings (SP1 and SP2 against ST) must be coincidental or the result of contamination--either the result of specific corrections introduced from the 1598 edition into a copy of the 1561 edition, which in turn served as printer's copy for the 1602 edition, or, more probably, the result of a correction introduced into the copy of the 1561 edition that was then used as printer's copy for both the 1598 and 1602 editions.

The preliminary matter of the two editions presents a different situation, since much of this appears only in the 1598 edition and not in that of 1561. The Life of Chaucer and all sections in roman type of the preliminaries of the 1602 edition are set directly from 1598. Although the 1602 edition is not a line-for-line reprint, specific correspondences in layout can be seen in the italicized marginal notes (see for both editions sig. b2r and the marginal note on Canterbury College at the top of sig. b3r). These show that the printer's copy for the 1602 edition was a copy of the 1598 edition, not, say, the manuscript that was itself the basis of that earlier edition.

One might assume, then, that all the preliminary matter in the 1602 edition would be set from the 1598 edition, but that, surprisingly, is not the case. Much of the preliminary matter differs in content (for example, the dedicatory letter to Cecil and the section "To the Readers"). Clearly, behind these is a manuscript, not the 1598 printed text. More surprising is the case of the introductory letter from Thynne to Henry VIII (reprinted in all sixteenth-century folio editions since the 1532 edition), where Speght follows what seems an unnecessarily complicated procedure. The number of lines per column in 1602 is different from the number of lines in 1598 and 1561. Comparing the 1550, 1561, 1598 and 1602 editions shows very clearly that the 1602 edition used the 1598 edition for its heading--a margin-to-margin heading in three sizes of roman type: both the typeface and the layout are identical in the two. In addition, the ornament running across the top of the page is the same (the 1602 version having an additional section). It seems preposterous that Speght in 1602, after using the 1598 edition for the Life of Chaucer and for the very heading to the letter of Thynne, then shifted to a 1561 edition for the printer's copy of the text of that letter; yet he did precisely that. The


Page 52
evidence from the layout and line composition is unambiguous. Unlike prose in the rest of the volume, the preliminaries of the 1598 edition are not reprinted line for line from the 1561 edition. And for the most part, the three editions of 1561, 1598, and 1602 vary in layout here. But at the end of paragraph sections in the letter to Thynne on sig. C5v, the line lengths and breaks for the 1602 edition are identical to those of 1561 and bear no relation to those of 1598. This cannot be fortuitous; the prose of the 1602 edition, both in text and in its preliminaries, was set from the 1561 edition whenever a 1561 text was available. The only reasonable explanation for this is a readily available copy of the 1561 edition already in the possession of the press.

As noted earlier, variation in substantives and accidentals[10] is less important in determining printer's copy than matters of page layout and line composition. And as far as I know, none of these prose sections--either those of the text or the preliminaries--has been thoroughly collated for such variation. In those prose sections I have collated, I have found no substantive disagreements in the three editions (although I presume they exist, and that the Variorum editors will in the future find them). If and when such disagreement is found, it will indicate something other than printer's copy: because the 1561 and 1598 editions generally correspond line for line, any corrector set to the task of introducing the readings from a copy of one into a copy of the other would have a relatively easy time of it.[11] As for accidentals, the only pattern of variation that could challenge the above hypothesis would be the overwhelming agreement of SP1 and SP2 against ST. The collations I have done show only what is expected: general agreement of SP1 and SP2 in accidentals; general agreement of both with ST; numerous cases of ST and SP1 against SP2; numerous cases of ST and SP2 against SP1.

Textual-Critical Evidence of Printer's Copy for Sp2: Verse Sections

The collations of the Variorum editors for verse sections of the Canterbury Tales support the conclusions above, although some of their claims are unnecessarily understated. All are based to some extent upon Pearsall: "The conclusion seems clear that SP2 was reset from ST ... but SP2 has been so extensively edited that the evidence cannot be so decisive as it usually is with early printed editions" (Nun's Priest's Tale, 114). Only Baker in his edition of the Squire's Tale is categorical about the 1561 edition as printer's copy; yet the evidence is far stronger than other editors seem to believe.[12] In only


Page 53
one or two instances is the substantive variation such as even to suggest that either printer's copy or textual source for the 1602 edition was anything other than a copy of the 1561 edition, and even these examples are capable of alternative explanation. Original manuscript readings supposedly "restored" in 1602 are, as Pearsall states, "commonplace" and "widely attested."[13] The manner of presenting statistics in these volumes occasionally implies some uncertainty about the printer's copy or copy-text for the 1602 edition. But at least for the Tales already collated by the Variorum editors (the General Prologue is a special case), there should be no hesitation whatsoever, as examination of the exceptions cited by these editors quickly reveals.[14]


Page 54

Where some question remains is in the General Prologue. The evidence presented by Lounsbury in 1892 and the more detailed collations recently published by Andrew show that the situation is not the same as in the texts of the Tales. To begin with, printer's copy, which editors of other Canterbury Tales sections describe simply as ST, must be defined more carefully, since Stow's edition contains two versions of the General Prologue, one with woodcuts (STw) and one without (ST).[15] Andrew quite correctly concludes that the only version "behind" the 1598 edition is ST, not STw (Andrew calls this a "copytext," 94) and the same is true for the 1602 edition; that is, particular readings of STw can be disregarded.[16] I extend this here by suggesting that we are dealing not with a textual copy-text (ST rather than STw) but more likely a specific copy of ST--a specific copy that may have served as printer's copy for both Speght editions.

More than in the Tales, the General Prologue shows apparently mixed filiation--a conflation of ST and SP1. Andrew follows Lounsbury and concludes: "It seems then that ST and SP1 had nearly equal influence on the GP text of SP2" (100). The influence Andrew notes here is textual, and the word "equal" misleading. The statistics Andrew provides supporting this statement actually show that these two texts (ST and SP1) influenced SP2 in demonstrably different ways.[17] Printer's copy for SP2 is ST. Any substantive agreement of SP1 and SP2 against ST is coincidental, or possibly the result of an


Page 55
editorial change made in the particular copy of ST that served as printer's copy for both.[18]

Lounsbury noted that Speght states in his introductory note "To the Readers" (1598) that he did not involve himself in the edition until late in the printing process: "three parts thereof alreadie printed." The various sections of the book should thus differ textually, reflecting Speght's care (or interference) (Lounsbury, 270-271). What Speght's work amounted to is unknown, but it could well account for the apparent mixed textual affiliation of the General Prologue. The copies Speght refers to when he claims the text is "by old written copies corrected" ("To the Readers" 1598) or "by old copies reformed" (1602), are probably no older than 1561 (so Pearsall, "Speght," 86-87).

The Problem of the Fists

The editions under discussion here have a long-noted oddity that bears on the tradition of printer's copy for these editions. The earlier folios contain some typographically peculiar marginalia in the House of Fame; in the 1561 edition, these typographical marks seem to be replaced for the most part with marginal fists, and they are again confined to the House of Fame.[19] The 1598 edition has no marginalia of any kind, but the 1602 edition marks proverbs and sententiae with marginal fists throughout (the two texts printed from the 1598 edition, texts not found in the 1561 edition, are unmarked). Hetherington is one of few bibliographers to address these marginalia:

The earlier Chaucer folios themselves contain some markings which could be regarded as anticipating the formal and practical use of the fist. Curiously, in each edition these only appear with The House of Fame. In 1542 and 1561 they only occur in the 'middle' margin. In 1550 they seem to be 'end of line' rather than marginal. I have not found any single line to be marked in more than one edition. (8-9)[20]
No explanation of these marginal marks has been offered beyond this. And the only sixteenth-century books where I am aware of having encountered


Page 56
similar marks in any significant numbers are the Great Bibles from the early 1540s printed by Grafton and Whitchurch (Grafton is the printer responsible for the 1542 Chaucer, although not named in the colophon). Here, they appear sporadically in the Kalendar, varying in location from edition to edition, and disappearing from later editions.[21]

In the first Thynne edition of 1532, three marks occur, all easily reproducible on a keyboard: on fol. 316, House of Fame, lines 837, 848 and 858:

(:):) (:):) (:(:):)

In 1542, this same section is marked at lines 848 and 853:

(:)(:) (:)(:)

In 1550, the most heavily marked version, no marks appear in this section.[22]

In the 1542 edition, in the middle sections of House of Fame, there are two more marks of this kind--a punctus elevatus (I represent this mark below with a modern question mark) within parentheses at lines 1108 and 1127. Toward the beginning and end of the work, several other differently-constructed marks occur: three small joined o's, a leaf, and a fist in lines 77, 83, 98 respectively. In the final section, the mark consisting of three o's recurs at line 1947, and right-pointing fists appear in the interior margin, set with the left column of type, at lines 1923, 1955, 2130, and 2140.

I will state now that I am not about to offer an explication of the specific function of these marks. I am certain that if some dozen or so lines in any text were marked entirely at random, most bibliographers, textual critics, or literary critics worthy of their profession could come up with reasons why those particular lines (and only those lines) were of significance. My argument here is rather that the very tradition of these marks was in itself stronger than their perceived function and that only in 1602 were they finally rationalized. In short: sixteenth-century printers had no clearer idea of the original function of these marks than we do.

With that caveat, several aspects of these marks should be noted. I begin with the 1542 edition. Here the right-pointing fists appear in the interior margin, but they are set within the borders of the left column of type. They thus point to lines in the right column, not to the left-column lines in which they are clearly set. Are they pointing the wrong way? The last two marginal


Page 57
fists seem to point to line 2154 in the right column--a line about stamping for eels (!, see note in Riverside 990)--and a line in the spurious continuation reading "Of that god of Thondre." Eel-stamping may be noteworthy, but these fists almost certainly originate as responses to the lines in the left margin: "With boxes crommed ful of lyes / As euer vessel was with lyes" (lines 2129-30) and "For al mote out late or rathe / Al the sheues in the rathe" (lines 2139-40). Each involves an apparent example of rime riche, and such a rime type may well have attracted editorial concern. In the second couplet, the rime is clearly an error (for rathe read lathe). In the first, the rime depends in part on the meaning of boxes--if, that is, boxes belongs here. MS. F reads boystes, MS. B bowgys. I haven't a clue what lines 2129-30 mean or what Thynne in 1532 thought his variant meant: the Riverside gloss at p. 373 suggests a rime on lies and lees: "With containers full of lies as ever vessel was with dregs."

The other odd mark, the joined o's at line 1947, is also not a mark for a reader; the reading "In sommer whan they ben grene" reads in other copies "In sommer whan they grene bene." This mark is quite possibly a compositor's rendition of what was intended as a mark of transposition.[23] Like the fists at the end of the text, this is something produced during the printing process, and represents (or originates in) notes to a printer or editor. Taken together, these marks are not in any way coherent guides for a reader; for any reader taking them seriously would be alternately baffled by the text in the vicinity or left wondering what the fuss was all about.

The 1550 edition contains many complex and bizarre-looking marginal marks and ornaments. Again, they are confined to the House of Fame. In some cases, they appear like those in the 1542 edition (there are numerous instances where they are constructed with type sorts, rather than ornamental sorts). For the first two books, these consist almost exclusively of marks formed from parentheses, the punctus elevatus, and colons: many are as easily constructed on a keyboard as from a typecase and include the following, at lines 171, 520, 704 and 930:

):( (?) )?( (?)?(?)

At the end of the second book, with the beginning of a new quire (3G), these become much more elaborate, formed by multiple commas, inverted commas, colons, etc. Among these are the marks at lines 1185 and 1128 and 1467:

(,:,:,:,:,) (:()?) (:)(:)(:

These are consistently well formed and quite carefully done; they are not a haphazard collection of sorts, but formed, for example, with alternating commas and right-facing or left-facing inverted commas, all within parentheses. The following occurs at line 1429 (the third inverted comma should face right here):



Page 58

I have conjectured a number of functions for these: annotations for inexistent notes, illegible corrections for text, casting-off marks for an projected edition. Most of my conjectures, however, contain the word inexistent or a near variant: there are no annotations; the lines marked are in no obvious need of correction; the intervals would not be appropriate for casting off copy. The precise function of the (lost) original marks that inspired the compositor for the 1550 edition to create these variant marks is quite simply lost.

The distribution of these marks does, however, reflect the physical structure of the book, and thus not what must have been the distribution of whatever marks were in the printer's copy (an annotated copy of the 1532 edition). The pattern of their distribution is a product of the creation of this edition. The more elaborate marks appear only in a single quire, 3G, and the independence of that quire is shown by the initial line, which erroneously repeats the last line of quire 3F and thus makes the catchword inaccurate; some sort of interruption occurs here. Another indication that these are products of processes at the printing house is the presence of marks in quire 3F. They occur only in particular formes: 3F2r/3F5v and 3F3r/3F4v. They do not occur in the reverse formes: 3F2v/3F5r and 3F3r/3F4r. All this indicates that whatever these marks may have represented or have been thought to represent, their presence in the 1550 edition is a function of something in the printer's copy that one compositor could have interpreted as "something to be printed" and another could have interpreted as "something to be ignored" (the hypothetical compositors could of course be one compositor working under a different directive).

When we look at the edition of 1561, we can see another step in the apparent rationalization of an oddity in a printer's copy. For this edition, the printer's copy was a copy of the 1550 edition, containing the bizarre printed marginalia in the House of Fame. The 1561 edition puts fists in its text of House of Fame (and in no other text). Pace Hetherington, there appears to be some relation of these fists to the marks in 1550, and it might be argued that in some cases, they represent an interpretation of the marginalia of 1550.[24]

In the first two books of House of Fame, there is no apparent correspondence. There are a number of fists at lines 250ff., but no mark corresponds to them in the 1550 or 1542 editions (at the end of the first book). There are none in the 1561 edition on sig. 3e2r-3e3v (an entire inner and outer forme), whereas numerous marks appear in this section in the 1550 edition. From the beginning of book 3, there are numerous fists in the 1561 edition, and many marks in the edition of 1550. There is some correspondence here: a mark oc-


Page 59
curs at line 1169 in the 1550 edition and an upward-pointing fist in the 1561 edition; another mark at line 1206 might correspond to an upward-pointing fist in the same line in 1561, and another mark at line 1217 might correspond to an upward-pointing fist at line 1219 in 1561. So also at line 1465, an upward-pointing fist in the 1561 edition is at least in the vicinity of a strange mark in the 1550 edition at line 1467; a downward-pointing fist at line 1479 might correspond to another mark in the 1550 edition. But of the eleven fists that follow, there is only a rough correspondence in one or two cases.

A few additional peculiarities about the fists in the 1561 edition are noteworthy. To begin with, there is only one such fist in the typefont, a right-pointing fist. This is set to point right on only one occasion. In all other uses, the fist appears between the two columns, set with the left column of print, and points either up or down. There may be a correspondence with the text, but trying to find one appears to have the intellectual validity of the Sortes Virgilianae. What can be said is the same that can be said of the 1550 marginalia: they respond to something noticeable in the printer's copy, but not something the compositor fully understood or interpreted in any intelligible way.

What happens in the two Speght editions is a response: in the 1598 edition, no marginal fists appear, although Speght claims in his introductory note "To the Readers" that he has marked all sentences and proverbs. In 1602, marginal fists appear throughout the book (with the exception of the two texts that the 1598 edition added to the contents of earlier editions: "Chaucer's Dream" and "Flower and the Leaf"). And they are perfectly rationalized--marking sentences and proverbs, precisely what Speght claims on his title page and in the preliminaries.[25] Because they are rational, they have no relation to any of the marginalia in earlier editions.

The history of these marks confirms other evidence for printer's copy, supporting the hypothesis of the retention of particular books or manuscripts as printer's copy, first for the editions of 1542 and 1550, and second for the edition of 1602. There is no doubt concerning the general relations of the 1542 and 1550 editions: the collations show that both were set from the first Thynne edition (1532). How, then, does the 1550 edition happen to have so many marks, some of which correspond exactly to the 1542 edition? And why do both editions have marks that are formed typographically like the three in the 1532 edition?

The only explanation for this is that the 1550 and 1542 editions were set not simply from what a textual critic might call TH (the text of the 1532 edition) but rather from a specific copy of that edition, a marked-up copy.[26]


Page 60
Such a specifically marked copy would explain all apparent correspondence with these two editions as far as particular extraneous marginalia are concerned. The relations of the various printers involved in the two editions are close. According to Isaac, the printing of the 1550 edition, in all its issues, is done by Nicholas Hill, who shared type with Whitchurch and Grafton (Grafton is responsible for the 1542 edition); the printers named in the colophons (Bonham and Reynes for 1542; Bonham, Kele, Petit and Toye for 1550) are part of a consortium.[27] These close relations provide additional support for the notion of a single marked-up copy of the 1532 edition used as printer's copy for both editions. Had another copy been chosen as printer's copy, there would be no reason to suspect any correspondence between the 1542 and 1550 editions as far as the marginal markings are concerned, unless there was a printed mark in the 1532 edition at the point of these correspondences.

The marginalia also help describe the nature of the printer's copy for the 1602 edition. The 1561 edition served as printer's copy for this edition, and its marginal fists may well have provided the inspiration for the fists in 1602, just as the incoherence of their placement could have inspired the rejection of the fists in 1598. The marginal fists in the 1602 edition, thus, do not represent the marginal fists actually printed in its printer's copy (otherwise they would correspond to those in the 1561 House of Fame) but rather indicate lines marked by hand in that copy. This would explain why no such fists appear in those sections of the book set from the edition of 1598--only the printed source text (ST) was marked, not the source text for these two poems. It also can explain why Speght claims in 1598 to have marked the text's proverbs ("Seuenthly, sentences noted"), when only the 1602 edition is so marked (see n. 25 above). The printer's copy used in 1598 may well have been so marked, just as Speght claims, but those marks were not introduced into the printed text (as marginal fists) until that copy was used again as printer's copy for the 1602 edition.


In 1892, Lounsbury examined lines from the General Prologue, looking primarily for evidence of Speght's claim that he had consulted earlier manu-


Page 61
scripts. Although his conclusion was negative, in the process Lounsbury discovered many of the variants that provided an outline for the printing history of these editions, and in particular, the relation of the 1602 edition to the earlier editions of 1561 and 1598. Lounsbury concluded: "A full examination, which has never been made and hardly seems worth making, would be necessary to settle the matter beyond all dispute" (277). The examination "hardly worth making" has now, with the work of the Variorum editors, begun in earnest. Most of the collations confirm the evidence provided by details of layout and line composition, but the interest of editors in possible original manuscript readings contained in these late editions has obscured what the evidence for many of them shows and what Lounsbury suspected: clear and exclusive dependence of later editions on earlier editions whenever those editions were available.[28]

The examination of layout and composition suggests a few areas in which editorial language regarding sources for certain texts could be clarified. Manuscript sigla refer to readings contained in specific material objects. But the sigla for printed editions (TH, ST, SP)--sigla I have of necessity adopted here--refer to entire editions, individual copies of which have different readings, either due to ordinary press-variation, or more significantly to the intervention of an annotator. In the present case, one could imagine, say, a copy of the 1561 edition, in which all the readings of the 1598 edition had been entered, or in which numerous changes had been made that were eventually to become the 1598 version (in terms of textual substantives, the two might be identical). Any text copied closely from this might resemble the 1561 edition in details of layout and perhaps in accidentals, but in terms of substantives, it would duplicate the 1598 edition. A textual critic might reasonably claim that the 1598 edition served as 'copy-text' or 'base-text'. And under certain understandings of these terms, that could be the case. But it would not serve as 'printer's copy' in any sense--whether we mean by that phrase a physical object in the press room, or more abstractly an edition, one of whose representative copies is in the press room.

Evidence for actual printer's copy is more likely to come from extra-textual matters than from the level of textual substantives of interest to most editors--a level to which modern Chaucer editors have found themselves increasingly committed, especially since the publication of Manly and Rickert's Table of Variants in 1940. Yet the precise definition of a substantive, and the textual-critical value of such substantives varies considerably (the transposition he said/said he is common in Chaucer texts and by definition classi-


Page 62
fied as a substantive; but the textual-critical value of such variation is no more than that of accidentals).[29]

The books examined here show that details of layout persist quite apart from their textual or intellectual functions, just as the use of black-letter itself for medieval texts persisted as the implications of that typeface changed dramatically. Later printers were faced not only with making sense of a text growing increasingly archaic and inaccessible, but with making sense of at least one mystery of their own making.



The principal editions (several with variant issues) are as follows: the 1532 Thynne edition (STC 5068); an edition of 1542 adding the Plowman's Tale (STC 5069-5070); an undated edition where the Plowman's Tale is moved to precede the Parson's Tale (1550? STC 5071-5074); the Stow edition of 1561 (STC 5075-5076.3); two editions credited to Thomas Speght of 1598 and 1602 (STC 5077-79; STC 5080-81). I will use the now standard abbreviations adopted by the Variorum Chaucer, although I have reservations (expressed in my conclusion below) about what these abbreviations sometimes mean: TH1 TH2 TH3 (= the Thynne editions of 1532, 1542, and 1550), ST (= Stow's edition of 1561), and SP1 and SP2 (= the two Speght editions of 1598 and 1602).


See, e.g., Walter W. Skeat, "Early Editions of Chaucer's Works," The Complete Works of Geoffrey Chaucer, vol. 1 (1894): 27-46 . On the manuscript sources for the many additions to the canon made by Stow, see esp. Bradford Y. Fletcher, "Printer's Copy for Stow's Chaucer," Studies in Bibliography 31 (1978): 184-201.


I am not concerned here with the typography; see my Who is Buried in Chaucer's Tomb? Studies in the Reception of Chaucer's Book (1998), chap. 3: "Toward a Typographical History of Chaucer."


Thomas R. Lounsbury, Studies in Chaucer, 3 vols. (1892), 1: 265-80, based in large part on Francis Thynne, Animadversions uppon the Annotacions and Corrections of some imperfections of impressiones of Chaucers workes (1598), ed. G. H. Kingsley, rev. J. F. Furnivall (1875). See more recently the excellent bibliographies by Eleanor Prescott Hammond, Chaucer: A Bibliographical Manual (1908), 116-127, John R. Hetherington, Chaucer, 1532-1602: Notes and Facsimile Texts (1964), and the discussion by Charles Muscatine, The Book of Geoffrey Chaucer (1963), and Derek Pearsall, "Thomas Speght (ca. 1550--?)," in Paul G. Ruggiers, ed., Editing Chaucer: The Great Tradition (1984), 71-92. See also: W. W. Greg, "The Early Printed Editions of the Canterbury Tales," PMLA 39 (1924): 737-761; Alice S. Miskimin, The Renaissance Chaucer (1975), 242-261; and Tim William Machan, "Speght's Works and the Invention of Chaucer," Text 8 (1995): 145-170. The importance of Lounsbury is often underestimated, but the influence of his narrative and the infectiousness of his style are easily seen, as in the following statement on the relation of Speght to William and Francis Thynne: "In return Speght spoke with the profoundest deference of the Thynnes, father and son" (273); the sentence finds its way into Pearsall virtually unchanged: "He speaks with the profoundest deference of the Thynnes, father and son ... " (Pearsall, 85). The word then finds its way into a recent volume of the Variorum Chaucer: "Perhaps in deference to Thynne..."; Malcolm Andrew et al., The General Prologue, Variorum Chaucer II, 1 (1993), 100.


"They are set up, line by line, from their predecessor, diverging from it only insofar as the text undergoes the usual mechanical degeneration at the hands of the compositor" (Pearsall, "Speght," 71).


See the brief notes in G. Thomas Tanselle, "The Meaning of Copy-text: A Further Note," SB 23 (1970): 191-192. No specific document exists in the sense of those studied long ago by Gavin Bone, "Extant MSS. printed from by Wynkyn de Worde," The Library ser. 4, 12 (1931): 284-306, or more recently by James E. Blodgett, "Some Printer's Copy for William Thynne's 1532 Edition of Chaucer," The Library ser 6, 1 (1979): 97-113, and by N. F. Blake, "Aftermath: Manuscript to Print," in Jeremy Griffiths and Derek Pearsall, ed., Book Production and Publishing in Britain, 1375-1475 (1989), Appendix A: "Caxton Prints for which a copy-text survives or which were used as copy," 419-425 (Blake's word 'copy' has the meaning 'printer's copy').


The volumes of the Variorum consulted here are the following: Derek Pearsall, The Nun's Priest's Tale (1984); Thomas W. Ross, The Miller's Tale (1983); Donald C. Baker, The Manciple's Tale (1984); Helen Storm Corsa, The Physician's Tale (1987); Beverly Boyd, The Prioress's Tale (1987); Donald C. Baker, The Squire's Tale (1990); Andrew et al., General Prologue (1993); John F. Plummer III, The Summoner's Tale (1995). In his notes (267, n. 26), Pearsall cites collations by Moorman on the General Prologue; I assume these are what is included in the published version, although Ransom seems to be claiming to have thoroughly revised them (Andrew et al., General Prologue, xv). Ransom's statement is none too clear here, and I am uncertain as to whose "Collations," Ransom's or Moorman's, were checked by Dr. Levy. I will refer to this edition simply as Andrew's. On the setting of the 1598 edition, see Pearsall, "Speght," 79, 84-85, with reference to collations contained in his excellent edition of The Floure and the Leafe and The Assembly of Ladies (1962).


Aesthetically, the prose sections of the 1598 edition seem badly set as a result. The spacing in the 1602 edition is far more uniform.


See, e.g., the first paragraphs of "Pars secunda penitencie" of the Parson's Tale in ST and SP1 (sig. S5r). The second paragraph of SP2 (1602, sig. R6r) is a line-for-line reprint of ST, not SP1. See also, ST and SP1, sig. T2va. Again, SP2 (sig. S2va) agrees with ST only. The same correspondences exist in the Testament of Love.


The distinction is that found in W. W. Greg, "The Rationale of Copy-Text," SB 3 (1950-51): 19-36. Although much criticized, it is of crucial importance in twentieth-century Chaucer editing; see below, n. 29.


This accounts in part for the general agreement of the prose sections in these editions: errors in a line-for-line reprint of prose will be much more apparent to compositors and proof-readers than in ordinary reprints of prose, or even in line-for-line reprints of verse.


Baker, Squire's Tale, 106; Baker cites as reference the somewhat variable statements in Pearsall, Nun's Priest'sTale, 114, Corsa, Physician'sTale, 78, and Boyd, Prioress'sTale, 102. Cf. Plummer, the most recent Variorum editor: "Because ST and SP1 are nearly identical, one cannot demonstrate beyond doubt which of the two served as copytext for SP2" (Summoner's Tale, 85); and Corsa, Physician'sTale, "The making of SP2's text ... remains a mystery" (79).


Nun's Priest'sTale, 114. Pearsall claims that such restorations occasionally involve complete lines. If that were true, it would be decisive evidence of a textual source beyond that of the 1561 edition. But the sole example Pearsall cites involves line 4117:

When humors ben too habundant in a wight:
certes this dreme, which ye haue met to night
Commeth of the great superfluitie
Of red color that is in you parde ... (1602 edition)

The third line here seems to replace what in all editions from Caxton to SP1 is some version of the following: "I tel you trouthe ye may trust me." The full context of this supposed restoration shows that the authentic line does not come from an independent source. See, e.g., the 1532 reading:
Whan humours ben to habudant in a wight
Certes this dreme/ which ye haue met to night
I tel you trouthe/ ye may trust me
Cometh of superfluyte / & reed colour parde
Whiche cause folke to drede in her dremes.

The 1602 edition produces a reading (recorded in earlier sources) by cancelling a superfluous line, not by adding a line.


Baker, Manciples Tale, 68: "Whatever the exemplar from which SP2 was set up, either ST or SP1, SP2 contributes 25 variants from SP1." The figures are misleading, and Baker provides only two instances where SP2 agrees with SP1 in substantive variation against ST: at line 105, Stow has "in pearth," a misreading of a thorn, easily corrected to "in this world" (SP1 and SP2) (many manuscripts read "in this earth"). The clear error "criyng" at 301 requires no source to correct to "crye." A similar error occurs at 262, where SP2 corrects a clear error "prien" to "wryen" (a reading that also appears in MS); the error involves only the reading of a scribal w as p. See also, the correction at line 105, the first line of the tale--a correction likely to come not from manuscript but from SP1. Corsa, Physician'sTale, 78-79, claims that SP1 and ST diverge substantively ten times, and that in six of those instances, SP2 reads with ST (lines 59, 82, 190, 216, 271, 276). This to me certainly implies that SP2 does not read with ST at 125, 138, 165, or 190. But that is not the case. SP2 agrees with ST in all ten instances. I agree with the conclusions of Boyd, Prioress'sTale, 102, which are those of Pearsall; but the reference to the Variorum Table of Correspondences ("SP2 agrees with ST in 95 variants and with SP1 in 93") is misleading in that such agreement is only with Hg (the readings to which the Variorum editors are more or less committed as original). As far as I can determine, every example Boyd cites is evidence that the printer's copy for 1602 was ST, and there is no evidence of mixed agreement (i.e., agreement with SP1 against ST). See finally Baker, Squire's Tale, 106: "In 7 of the 8 divergences of SP1 from ST, SP2 reads with ST; the sole exception is at line 440, where SP2 has a unique reading." The evidence is unanimous: since the SP2 reading at 440 is unique, this is not an exception to the notion of ST as printer's copy.


The revised STC lists three issues here; the title-pages and the two versions of the prologue distinguish STC 5075 from 5076. STC 5076.3 has a different colophon; one of its listed copies contains woodcuts, the other (apparently) does not. The abbreviations STw and ST in this context should be construed as referring simply to the printed pages containing the General Prologue.


Andrew's statistics supporting this are much more decisive than Andrew claims, and the apparent evidence for mixed agreement is illusory. Andrew enumerates 23 substantive differences between STw and ST (95), and concludes "ST is clearly the copytext for SP1" (96). On four of these, SP1 seems to agree with ST. But only one of these involves substantive variation (and/and a at 558): plaien/plain (236) is a spelling variant; meserable/miserable (435) is another accidental variant elevated to a substantive; out of/out oft (487) is another accidental. Andrew is the first Variorum editor to discuss in detail the problems of the definition of a variant (122-24). Nonetheless, the statistics themselves are by nature uncritical, and many of the supposed substantives tabulated are products of easily-correctable accidental variation. See, e.g., the following instances of variation between STw and ST: ensired/espired; porte/sporte; pleasaunt/pleasaunce; no/not; he/she.


The evidence for "equality" Andrew presents is in the following statistical statement: SP2 and ST agree against SP1 58 times; SP1 and SP2 agree against ST 60 times (Andrew, 100). But again, Andrew's analysis (which is careful and detailed) actually yields a somewhat different statement, with none of the illusory balance of the above: where SP2 and ST agree against SP1, 2 cases involve SP1 omission, and 23 involve SP1 additions; where SP1 and SP2 agree against ST, 26 cases involve ST omission, and 14 involve ST additions. Obviously, the influence of each text on SP2 is different. I believe the conclusion to be drawn here is the following: the agreement of SP1 and SP2 against ST is random; the agreement of SP2 and ST against SP1 is systematic (most involve SP1 additions); the implication is that ST is printer's copy for SP2.


See, e.g., the apparent agreement of SP1 and SP2 against ST at line 147: only when variants are considered singly do SP1 and SP2 agree; there is no agreement if the entire line is taken as lemma. The agreement at line 176 (pace for ST's space) is quite possibly the consequence of SP1 following a marked copy of 1561, not a matter of a 1598 reading being introduced into that copy. Many of the more radical differences are cited by Lounsbury, 275-276; but of these fifteen, only two show agreement of SP1 and SP2 against ST. And most of the more substantial variants in Andrew show the same thing (see, e.g., line 73). The agreement of SP1 and SP2 against ST often involves individual words, and such agreement is quite possibly the result of a gloss (called/clepyd, line 121; wenden/goon, line 21). Again, this is perfectly consistent with a single marked-up copy of ST serving as printer's copy for both Speght editions.


There is at least one exception to this in Stow, and there may be more. On fol. 243r of ST (sig. Y3r), there are five small ornaments in the inner margin. I have not found these elsewhere in the edition.


The statement that these marks occur in the 1542 and 1561 editions only in the middle margin is not accurate.


See the 1539 Bible (STC 5068), and the series of Great Bibles printed by Grafton and/or Whitchurch from 1540-1541 (STC 2070-2076). Ornaments constructed from type-sorts much like those in the Chaucer editions appear frequently in the prefatory matter, with no discernable relation from edition to edition. For a convenient overview, see Francis Fry, A Description of the Great Bible, 1539, and the Six Editions of Cranmer's Bible, 1540 and 1541, printed by Grafton and Whitchurch (1865), plates 2-4. Later Bibles printed by Whitchurch in 1549 (STC 2079) and 1550 (STC 2081) do not contain a calendar; these contain sporadic ornaments and marginal fists, but nothing similar to those discussed here. Stephen Tabor of the Clark Library has pointed out to me as well the signature marks indicating half-sheet collation noted by David J. Shaw, "Quire and Sheet Numbers in Sixteenth-Century France," The Library, ser. 6, 17 (1995): 311-320.


For ease of reference, I cite these by the line reference in Larry D. Benson, ed., The Riverside Chaucer (1987).


For early printers' marks of transposition, see the examples in Peter W. M. Blayney, The Texts of King Lear and Their Origins, vol. 1: Nicholas Okes and the First Quarto (1982), 224, fig. 17d, 225 n.1, and 237, figs. 31f.


See, however, the downward pointing fist at "So gyue hem ioye that it here" in the 1561 edition; there is a leaf there in 1542, but nothing in 1550. Yet 1542 is not the printer's copy, and it may be that relations between the 1550 and 1561 placement of marginalia are products of pure chance. Anne Hudson, "John Stow (1525? - 1605)," in Ruggiers, 60, rightly concludes that the 1550 edition is the source for that of 1561, but collates, unnecessarily, against various earlier 'issues' of 1542 and 1550 for press-variants. There is no likelihood that divergences among various issues in terms of press variation will be any greater than divergences in individual copies of the same issue, since issue is not defined by variant sheets.


In "To the Readers" (1598) Speght claims as one of his eight "undertakings": "Sentences noted." But this is not included among the seven "Additions" listed on the title page and the sentences are in fact not noted in the 1598 edition. The title page of 1602 states "Sentences and Prouerbes noted" and Speght adds to the earlier dedicatory letter to Cecil of 1598 the statement that he has "noted withall most of his Sentences and Prouerbs."


I have done some checking of press variants in various copies of this edition, and the conjecture of a single material printer's copy for both editions would be refuted if the 1550 and 1542 editions were shown to differ on the variants that exist in different copies of 1532. They do not so differ in any I have checked so far. See my "On 'Correctness': A Note on Some Press Variants in Thynne's 1532 Edition of Chaucer," The Library ser. 6, 17 (1995): 156-167.


Frank Isaac, English and Scottish Printing Types, 1535-58, 1552-58 (1932), see Appendix: William Bonham, Robert Toy, and entry under Richard Kele, Richard Grafton, 29-39, Edward Whitchurch, 40-50 and Nicholas Hill, 88-91 (for 94T, see figs. 50 and 88a). The type identification and size is the same (94T; see figs. 50 and 88a), and the two are very similar (the upper case is different). See, however, the notes by Hetherington, 3-4, and discussion in Muscatine, Book of Geoffrey Chaucer, 23-24. The Variorum editors often refer to unspecified evidence in favor of the later date for the third Thynne edition (1550, instead of the 1545 date in the first edition of STC), but none has detailed what that evidence is. See, e.g., Andrew, 94. (I believe the reference is to Hetherington, 3-4, although Andrew, unlike other Variorum editors, does not include Hetherington in his bibliography). Isaac's work is cited by Muscatine, but not by any of the Variorum editors.


The reluctance of editors to give up the notion of such manuscript sources can be seen as manuscript readings hypothesized by one scholar become actual historical manuscripts for another. Where Pearsall spoke of possible manuscript readings in Speght (readings he characterizes as "commonplace"), Corsa, citing Pearsall, speaks of specific manuscripts: "The difficulty in such research, however, is increased by Speght's inconsistent use of more than one manuscript" (Corsa, 79, with reference to Pearsall, "Speght," 87) There is no evidence that Speght used any manuscript, unless we include as manuscript hand-written notes (perhaps his own) in a copy of the 1561 edition.


John M. Manly and Edith Rickert, The Text of the Canterbury Tales, 8 vols. (1940); Andrew, 122-124, and notes on 122. Andrew is the first Variorum editor to provide a clear presentation along with examples of what Manly-Rickert and later Variorum editors define as a variant. Some points of value can also be extracted (with difficulty) from an earlier study by Kurt Rydland, "The Meaning of 'Variant Reading' in the Manly-Rickert Canterbury Tales," Neuphilologische Mitteilungen 73 (1972), 805-14.