In a recent monograph and series of essays, Lawrence Warner argues that early C-text drafts on loose leaves contaminated an original ‘ur-B’ Piers Plowman that lacked, among other key passages, the two final passus. Warner’s argument depends upon the peculiar state of the text in National Library of Wales, MS 733B (N), and he claims that the alternative explanation, that this copy was contaminated by a B-text manuscript, relies on ‘unprecedented’ scribal behaviours. Ample parallels for the hypothesized behaviour of N may be found, however, in the most dramatically conflated of all the Piers Plowman manuscripts, Huntington Library, MS HM 114 (Ht), with which N shares a number of readings. Like John Manly’s theory of multiple authorship, which also depended on a hypothesized ‘lost leaf’, Warner’s arguments against the integrity of Piers Plowman B might ultimately prompt a more intense formal, rather than textual, analysis.
Mr jusserand warns us that if my methods are adopted, the whole history of literature will have to be rewritten’.1 So wrote John Manly at the height of the controversy over his theory that Piers Plowman was the work of five different poets. Manly's theory, according to which only A Prologue to passus 8 (‘A1’) were composed by the original author, depended heavily on his hypothesis of a lost leaf that had originally contained the confession of Wrath. Since Manly's claims for multiple authorship rested on his perception of the superior poetic powers of the author of ‘A1’, only such a scribal accident could explain the omission of Wrath from the sequence of confessions by the seven deadly sins in the A version.2 As Charlotte Brewer points out, this ‘lost leaf’ was a purely hypothetical construct for which there was no textual evidence, yet ‘the stress on the physical aspects of his argument diverted attention from the purely subjective impression from which the argument sprang’.3
Yet in spite of its implausibility, Manly's ‘lost leaf’ has continued, at some level, to fascinate scholars. Since Manly, a number of textual conundrums have been ‘resolved’, with varying degrees of plausibility, by similar hypothetical constructions.4 Because of their seeming tangibility, loose and lost leaves possess a peculiar allure. They lend to the arguments they are conjured to support an apparent connection to the real material world of historical textual production. Yet as hypothetical solutions to otherwise intractable problems, their existence cannot be proven.
I here address a new effort at rewriting the history of ‘the 1000-odd lines of Middle English poetry with the greatest impact upon our understanding of English culture c. 1380’—that is, the two final passus of Piers Plowman B, but by extension the whole history of the development of the text—that also depends heavily upon hypothesized loose and possibly lost Piers Plowman leaves.5 According to Lawrence Warner's theory, set out in his monograph The Lost History of Piers Plowman and recently defended in an essay in the Review of English Studies, early C-text drafts on loose leaves contaminated an original, ‘ur-B’ text at the earliest stages of the transmission of the B version.6 The ‘B text’ of Piers Plowman as we know it in all the surviving manuscripts and in all editions results, in this argument, from a series of scribal conflations, by which the earliest form of B absorbed materials—including the two final passus, B.19–20=C.21–22—that had originated in Langland's C-text revisions. However, this new ‘loose leaf’ hypothesis presents, as I will demonstrate, problems of a similar order to those of Manly's ‘lost leaf’ and other arguments that have theorized loose and misplaced materials.
Central to the case for the simultaneous production of the B and C versions is the peculiar state of the text of Piers Plowman in Aberystwyth, National Library of Wales, MS 733B (N).7 This manuscript is one of seven copies that combines an A text with a C conclusion, together with a total of eighteen passages from C within its A-text portion. It also contains individual readings as well as whole lines and passages from the B version.8 To most scholars, N's B-text readings reflect scribal contamination of an AC text by consultation of a B-text manuscript. In Warner's account, however, these B-text readings attest N's status as a witness to an early draft of C that contaminated B. The contamination, in other words, was in Warner's view in the opposite direction from the one that most critics have supposed: from C to B, rather than from a manuscript of B to N.
Loose leaves of authorial-revision material, Warner hypothesizes, lie behind N's agreement with one of the two genetic branches of Piers Plowman B for a number of lines and passages where the other is absent. Students of Langland's text have agreed that the transmission of the B version divides into two distinct branches. One branch, alpha, is represented by two manuscripts, Oxford, Bodleian Library, MS Rawlinson poetry 38 (R) and Oxford, Corpus Christi College, MS 201 (F). The remainder of the B- text manuscripts descend from the common archetype, Bx, by a second branch, beta. Each branch is characterized, as well as in individual readings, by around 170 lines in substantial passages not witnessed by the other branch.9 Some (although not all) of the discrete passages unique to the beta branch play a critical role in Warner's theory. He claims that the inclusion in N of ten passages witnessed by the beta family where alpha is absent or else contains spurious lines points to the status of these passages as an early draft of the C version now witnessed in part only by N. These ‘ur-C’ or ‘N version’ passages, he claims, were incorporated into the archetype of all B-text manuscripts, Bx, as loose sheets, where they were copied by the beta scribe. But the loose leaves had been ‘removed or lost’ by the time the alpha scribe came to copy the manuscript, explaining the absence of these passages from R and F.10
Lost or loose leaves, then, play a central role in Warner's revolutionary claims about the development of the B and C versions, as they did in Manly's controversial arguments about authorship and as they have done in various other textual theories since. Yet the hypothesis of the loose Langlandian leaf lost to history proves ultimately as distracting to Warner as it did to Manly's contemporaries a century ago. Warner's preoccupation with the idea of the discrete passage of Langlandian draft leads him to overstate the case for what he presents as an ‘elegant program of textual affiliations, over the course of entire passages of up to forty lines long, between a C-character manuscript and the W-M group [i.e., beta] where the RF group has nothing or is spurious’.11 The textual evidence appears considerably more ambiguous than such a description implies. At the same time, Warner recently claims that the alternative hypothesis, that N was contaminated by a B-text manuscript, relies upon a ‘series of unprecedented’ scribal behaviours.12 But such an argument overlooks the ample parallels for the hypothesized behaviour of N that may be found in the most dramatically conflated of all the Piers Plowman manuscripts, Huntington Library, MS HM 114 (sigil Ht).
This manuscript, a B text into which have been inserted passages from A and C as well as a number of unique lines, attempts to create a copy of all the ‘best bits’ of Piers Plowman.13 It shares with N, in its C-text excerpts from passus 10 onwards, a number of readings otherwise unique in the C tradition, but it also offers a more general parallel to N as the product of a situation in which ‘consultation of other copies was easy’.14 This important copy has been the focus of increasing critical attention but remains underexamined in detail. It reveals much about the behaviours of scribes who enjoyed access to multiple exemplars and made use of them in ways that often defy logical analysis. The scribal behaviour on display in this manuscript, while appearing baffling and illogical to a modern textual critic, nevertheless proves that the hypothesis for N's contamination by the B text is not without precedent elsewhere in the transmission of the poem. Warner's alternative proposal about the relationship between N and Ht, that both in part reflect early authorial draft materials on loose leaves, simply ignores the textual evidence that points, as I will demonstrate, over-whelmingly against such a conclusion. The seductive idea of the Langlandian loose leaf distracts Warner from undertaking any systematic analysis of the actual variants of the passages that he would propose as authorial draft.
Only the individual variants of the passages in question, of course, form a legitimate basis for determining their status as authorial draft or otherwise. As George Kane wrote axiomatically in his edition of the A text, ‘The sole source of authority is the variants themselves’, whose originality must be determined according to whether they are satisfactory in themselves and in their ability to explain how the competing erroneous readings arose.15 Kane here rejects attestation, that is, the agreement of manuscript copies, as being of any value in establishing the original text.16 Yet Warner consistently overlooks such a basic axiom, arguing for the existence of ‘ur-C’ drafts on loose leaves solely on the basis of a pattern of attestation, N's inclusion of passages that appear in beta but not in alpha. He reports that the N+beta passages contain eighty-five individual shared readings, but he analyses none of these.17 As I will show below, a glance at Warner's key ‘loose leaf’ passage, B. 15.533–69, reveals that the readings that distinguish this passage as copied by N and beta from the C-text version are overwhelmingly scribal.
Ultimately, such is the allure of the hypothesized loose leaf that Warner seems determined to find, as Manly had done, discontinuities in the production of Piers Plowman even in the absence of any textual evidence. Warner thus eventually reinvokes at the end of his monograph The Lost History of Piers Plowman what he had earlier identified and rejected, in his argument that the two final passus of B are not integral to that version, as purely literary criteria. Like Manly's argument for multiple authorship, Warner's for the C version origins of the two final passus of Piers Plowman (among other important passages) might, then, best prompt a literary rather than textual response: a more careful and patient reading of the rhetorical and structural properties of Piers Plowman.
Loose leaves and the ‘ur-C’ Piers Plowman
The recent controversy over the status of N as either a witness to an early version of C or (the traditional view) an A-C splice contaminated by a B manuscript has focused on its inclusion of ten passages comprising eighty-one lines witnessed by the beta family of the B version in places where the alpha family (RF) has either nothing or else spurious lines.18 N also agrees with the beta family in ninety-three individual readings in its C-text portion where alpha agrees with C—the opposite of the scenario we would expect, given that N is supposed to be a representative of C and both alpha and beta witnesses to a common ancestor, Bx. N additionally presents sixty-four agreements with Bx against C.19 The most obvious explanation for these agreements with the B version, and particularly with its beta branch, at places where we would expect N to attest C is that they witness contamination by a manuscript of the beta family of B. In their edition of C, George Russell and George Kane had supposed that forty-five instances in which N preserves the correct B-text reading against the erroneous reading of all the other C-text manuscripts offered evidence that N had been ‘corrected’ from a pre-archetypal copy of C. Carl Schmidt, however, explains these readings as contaminations from a B manuscript, and Robert Adams and Thorlac Turville-Petre argue, following Ralph Hanna, that Russell and Kane's notion that N contained ‘authoritative’ corrections was a critical convenience designed to avoid the suggestion that they had emended C by importing B- text readings.20
Warner, however, offers an alternative explanation of N's agreements with B. He proposes that the passages attested by N and beta but absent from the alpha manuscripts were not an integral part of the original, ‘ur-B’, text. Rather, he claims, they represent materials from an early draft of the C version, now witnessed in part only by N. These draft materials were incorporated into the common ancestor of all B manuscripts, Bx, on loose leaves and were copied by the scribe of beta, the ancestor of all the manuscripts of that genetic branch. The passages were subsequently revised into their final, C-text form. The loose sheets, having been recalled by the poet to make his final revisions or else lost, had been removed from Bx by the time the alpha scribe came to copy it, explaining the absence of these passages from R and F, the sole witnesses to this branch of transmission. The ninety- three individual readings where N agrees with beta against the shared reading of RF and C witness a further stage of development: the scribe of alpha was provided with a set of new readings, perhaps, Warner speculates, to compensate for his lack of access to the passages provided to the beta scribe, with the readings of N+beta representing the unrevised, ‘ur-C’ forms. The ‘ur-B’ version of Piers Plowman was thus ‘contaminated’ by C-text materials at two distinct stages in the development of Langland's final version: ‘in the first stage loose sheets went to Bx as copied by the scribe of the W-M ancestral copy [i.e., beta]; in the second, a final set of revisions of individual lections were recorded in Bx as copied by the RF scribe’.21
Warner finds further evidence for the contamination of an ‘ur-B’ Piers Plowman by ‘C-text’ material in the two final passus of the poem, which he claims also originated in Langland's C-text revision. Warner notes one hundred or so RF or F agreements with C against the readings of beta in B19–20 (for a substantial portion of which R is absent, having lost a quire). In his view, these (R)F+C versus beta agreements indicate, again, two stages of C revision, with beta attesting the unrevised, ‘ur-C’ version and (R)F the final, ‘C’ version.22
Another part of the poem central to Warner's argument is the sequence of lines on the Donation of Constantine in B.15.533–69, one of the ten passages witnessed by N where alpha is lacking. In their edition of B, George Kane and Talbot Donaldson had argued that this passage had been erroneously placed by a pre-archetypal scribe as the result of a misfolded bifolium (it appears in all beta manuscripts after 15.503α).23 This apparent misplacement, together with the absence of the passage in the two alpha manuscripts, had already led Robert Adams to propose that these lines originated in a loose sheet of last-minute B revision. The scribe of Bx, Adams hypothesized, left the passage as a loose insertion when he came to copy his exemplar, which was probably a scribal copy but with late authorial insertions and additions, some on loose leaves and scraps. The physical condition of the passage meant that it was subsequently overlooked by the scribe of alpha.24 In Warner's alternative theory, the lines were also a loose leaf, but from an early C-text draft that was ‘available to the ancestral scribe Nx (the first post-holograph copy, which possibly turned into Cx upon further revision), confused the W-M [beta] scribe ... and had been removed or lost by the time RF copied Bx’.25 His hypothesized loose leaf of C-text draft relies upon Russell and Kane's characterisation of Langland's C revision copy, which they argue contained over a dozen ‘single leaves or bifolia of new material, interleaved or loose’.26 Warner draws further support for his claims from Wendy Scase's earlier proposal (discussed below) that draft passages from the C text may have circulated independently of the rest of the poem.27
Warner himself acknowledges that he offers a complicated narrative, and it seems on the face of it inherently unlikely, as Adams and Turville-Petre have pointed out, that draft C-text lines could have been preserved in a copy of C that represents a particularly advanced stage of the transmission of that version, removed by many generations of copying from the shared archetype of all C-text manuscripts. The transmission of the C text divides into two ‘great families’, x and p, and N's C text, which begins in passus 10, shares errors with both the two families.28 N's inclusion of readings from both x and p suggests consultation of more than one C-text manuscript at some point in its history. Moreover, as I will discuss further below, some of N's shared errors reflect readings that are confined to genetic sub-groups representing the third or fourth generation of copying after the common ancestor of all the manuscripts, Cx.29 N is thus separated, Hanna argues, by at least five generations from the shared archetype of all C-text manuscripts; it is evidently, as Adams and Turville-Petre observe, ‘the product of a series of conflations’, and it is therefore extremely unlikely to have preserved draft materials prior to the archetypal text Cx.30
Adams and Turville-Petre also point to evidence of beta readings in the A-version portion of N's text. As they observe, these cannot be ‘ur-C’ readings, which in Warner's theory were transmitted to ‘Nx’, the first copy of the draft C version now witnessed only in N's C-text portion, and to the scribe of beta. The obvious conclusion is that the N scribe kept a copy of B in front of him throughout the copying of his A-C text.31 Unlike his ‘degraded’ copy of C, this exemplar of B was evidently, on the quality of its readings, an ‘excellent beta copy’.32
Most importantly in the context of the present essay, the evidence suggests that the pattern of textual affiliations upon which Warner bases his argument might not be so exact or ‘elegant’ as the opening pages of his book would imply.33 Warner hypothesizes that whole runs of lines witnessed by N and beta but not by alpha reflect loose leaves of Langlandian draft that were not made available to the alpha scribe. But his investment in this idea of loose drafts obscures the sometimes inexact match between N's beta readings and the passages absent in the alpha manuscripts. Adams and Turville-Petre point out that in the case of the passage most prominent in Warner's argument, B. 15.533–69 on the Donation of Constantine, N continues to witness beta readings after RF resume, casting doubt on its status as a witness only to the discrete passage absent from alpha.34 Similarly, the first two lines of another key passage quoted by Warner in his book (C.10.283–89/B.9.182–88) are only ambiguously witnesses to beta, since with the exception of the unique reading ‘now’ for ‘nat’ in C.10.283, N's variants in C.10.283 and 284 are also shared by MS D of C, alone or in combination with another C manuscript.35 N's readings in these two lines might well, then, reflect what we would expect, its status as a witness to the C version (Warner's statement that ‘N2's now for received nat at C.10.283 brings it into line with Kane and Donaldson's B.9.182’ is, so far as I can see, simply erroneous.)
Moreover, another manuscript of C also in part attests one of Warner's N+beta-only passages at a point where RF have spurious lines based upon the equivalent A-text section (B.3.51–63, part of Meed's expanded conversation with her friar-confessor; cf. C.3.56–67/A.3.50–52). British Library, MS Royal 18 B. XVII, sigil R of C, also inserts, after C.3.52, the first line of the relevant passage and the line immediately preceding it, B.3.50–51. Warner himself points out and discusses this phenomenon in his original essay, where he suggests that ‘it is clear that W + [i.e., beta] or N itself has influenced’ this and three other manuscripts that he notes also contain beta readings in this section (where by N he means, apparently, what he calls the ‘N version’, the early draft of C contained in the ancestor of NLW 733B).36 Warner does not mention, however, that R also contains three other lines from B passus 3 in what is otherwise a C text, B.3.37, 107, and 109. These cannot represent ‘ur-C’ lines, since RF also attest them. The existence of an ‘ur-C’ text whose new passages were loaned to the beta but denied the alpha scribe does not explain how R of C, too, incorporates part of one of the passages involved in the ‘pattern’ Warner observes in relation to N and alpha. Warner's commitment to the idea that N+beta witness loose-leaf draft passages unavailable to the scribe of alpha leads him to over-simplify a textual situation that appears considerably murkier than his presentation might at times suggest. The phenomenon of parallel versional switches in several manuscripts awaits full explanation.37
Perplexing conflation in Huntington Library, MS HM 114
Warner's presentation tends, then, to overstate the extent to which the agreements between N+beta fall into discrete passages that might reflect loose-leaf authorial drafts to which the scribe of alpha lacked access. At the same time, his recent claim that the alternative explanation— that N was simply contaminated by a manuscript of B—relies on a ‘series of unprecedented’ scribal behaviours simply ignores, as I will show in this section, the parallel examples of the behaviour of a conflating scribe that can be found in Huntington Library, MS HM 114 (Ht), an ABC splice with which N shares a number of unique C-text readings.
In his recent response to Adams's and Turville-Petre's defence of the traditional view of N, Warner rightly points out that contamination of N by a manuscript of B implies a very complex scenario. The major complicating factor is the preponderance in N of specifically beta-family readings as opposed to B-text readings shared with both genetic branches. Warner finds sixty-four readings in which N agrees with the shared ancestor of all B-text manuscripts Bx, compared with ninety-three readings shared with the beta family where alpha instead agrees with C, and eighty-five individual readings among the eighty-one lines shared by N and beta where alpha is deficient.38 ‘[D]istinctive, “beta” readings’, he notes, ‘make up only 5 percent of the B text available to the NLW 733B scribe, and yet over 70 per cent of the readings this scribe supposedly selects’.39 This apparent selectiveness, favouring beta-only lines and readings rather than readings common to both B-text families, leads Adams and Turville-Petre to propose that the scribe might have consulted an alpha manuscript as well, noting points of difference.40 Warner is surely right to suggest that the argument for contamination of N involving both a beta and an alpha manuscript points to a very complex and perhaps irrecoverable sequence of events, for the scribe of N must have used his alpha manuscript highly selectively indeed, or, as Warner puts it, ‘only negatively’. He apparently noted, that is, some—but not all— of the passages in beta that were missing in alpha, marking these for insertion into his copy of A-C, but failed to note, for some reason, the roughly equal number of alpha-only lines.41 The case for contamination of N by a B manuscript certainly points to a very obscure situation.
But Warner surely errs when he suggests that since contamination of N involving both a beta and an alpha manuscript must necessarily involve a sequence of events and motivations that may prove impossible to recover, it simply could not have occurred. His recent refutation of Adams's and Turville-Petre's defence of the traditional view of N ultimately comes to rest on what he takes to be the improbability, first, of any text witnessing comparison with both an alpha and a beta manuscript and, second, of various kinds of scribal behaviour that the account offered by Adams and Turville-Petre would have to assume. Warner points, for instance, to the inconsistency that the scribe of N apparently substituted, in a way that Warner finds ‘perplexing’, small variants from his beta copy while simultaneously ignoring the larger passages that he would have found in his B manuscript not present in his exemplar of AC, such as B's expanded portraits of the seven deadly sins. Warner also questions why the scribe of N, if he were, as Adams and Turville-Petre argue, in possession of ‘an excellent beta copy’ of B, persevered in copying his corrupted C exemplar at all. To accept the proposal of Adams and Turville-Petre, we would also have to accept, Warner observes, that the scribe used one copy effectively to cancel lines in another exemplar, since in switching to his B-text exemplar for B. 15.533–69 on the Donation of Constantine, N replaces the equivalent passage in C with text four lines shorter. Adams and Turville-Petre thus propose an alternative narrative that depends, in Warner's view, upon a ‘series of unprecedented’ scribal behaviours.42
Yet ‘impossible to reconstruct’ is not the same as ‘impossible’, as Kane pointed out in his discussion of the conflations from B and C in N's A-text portion.43 While the motives and the exact sequence of events that ultimately produced N's text perhaps defy explanation, none of the individual scribal behaviours that Warner calls ‘unprecedented’ is beyond the realms of possibility. Indeed, the incorporation of apparently inconsequential minor variants while overlooking larger passages; decisions about the relative authority of multiple exemplars that look manifestly absurd to the modern textual critic; the omission of lines from one exemplar on the authority of another; and the repeated comparison of texts witnessing different versions and branches of transmission all find ample parallels in the very manuscript with which N itself shares a number of C readings, Huntington Library, MS HM 114.
Huntington Library, MS HM 114 (hereafter Ht), the work of a London professional scribe also responsible for two other literary manuscripts, London, British Library, MS Harley 3943 and London, Lambeth Palace Library, MS 491, is one of the most notorious Piers Plowman manuscripts. Interest in the manuscript looks only likely to increase following the possible identification of the scribe as Richard Osbarn, clerk of the London Guildhall. It is evidently the product of a redactor who had access to at least one exemplar of each of the three versions of Piers Plowman and who was strongly motivated to preserve as many lines as possible from each, sometimes at the expense of even the most elementary literary cohesion.44 Among its insertions are passages from the C Prologue and from C passus 9 that appear in a textual form unique to this copy and to another manuscript by a prolific London scribe, the ‘Ilchester’ manuscript (London, University Library, MS S. L. V. 88, sigil J). Some critics have proposed that these interpolations in Ht and J might reflect early C-text drafts circulating independently. As we will see, Ht's textual relationship to Ilchester and the earlier proposals about the status of these two manuscripts as possible witnesses to loose-leaf draft material prove central to Warner's own arguments about the relationship of Ht and N, but these arguments are not supported by the textual evidence.
As well as sharing about eighty readings in their C-text portions, Ht and N both apparently involved close line-by-line comparison of multiple exemplars with an eye towards textual completeness. Both scribes, as Adams and Turville-Petre have argued, ‘had access to various versions of the poem’ and were ‘keen to include them where appropriate’. Adams and Turville- Petre point out, for example, that Ht's exemplar of B, a member of the YOC2CB subgroup of the beta family, lacked the passages missing from all the beta manuscripts and that the scribe managed to make good some of these omissions by inserting the parallel lines from his exemplar of C. Similarly, they argue, the scribe of N introduced readings from his B-text exemplar into his text of C, although in this case they were ‘not insertions but replacements’.45 Warner, however, rejects such a comparison:
The authors’ presentation of this as somehow in line with Osbarn's work is inapposite: no other instance of Middle English manuscript contamination takes the form of substitution rather than supplementation.46
This is a rather peculiar statement, not least because Adams had used Ht itself as one example in his earlier demonstration of the prevalence of contamination at the level of small substitutions throughout the manuscript tradition of Piers Plowman. Adams cites as an instance of this phenomenon Ht's substitution of the b-verse of A.3.89 for that of the parallel line in B (B.3.100).47 Another example occurs in B.8.98, where the a-verse of that line is followed in Ht by the b-verse of C. 10.95. Here the substitution was perhaps occasioned by the redactor's preference for the macaronic ‘putte adoun preuaricatores legis’ of C over the English B, ‘punge adown þe wikked’48 Such small substitutions in fact occur repeatedly; another telling example, because it is one where another scribe has made the same substitution through independent comparison of the same two versions of the poem, occurs at B.2.11. Here, for ‘gold wyr’ in B, both Ht and G (Cambridge, University Library, MS Gg.4.31) substitute the reading of the A text, ‘ringes’ (A.2.11). G's substitution appears to be an example of the ‘microcontamination’ of this copy by A at some point in its history.49 Although Warner casts doubt on the idea of trivial variants in N's A portion representing the work of a contaminating scribe, copyists of Piers Plowman evidently could and did contaminate at a microscopic level; moreover, the example of B.2.11 suggests the possibility of scribes independently contaminating their texts at the same locations.
While the Ht redactor's primary mode of engagement with the text of Piers Plowman was simply to accumulate as many lines as possible from A and C and insert them into his exemplar of B, his behaviour is neither uniform nor, often, very comprehensible. Many of the other kinds of scribal behaviour that Warner dismisses as improbable when hypothesized in relation to N are prevalent in Ht as well. For example, Warner finds it doubtful that the scribe of N could have incorporated into his A text very minor variant readings while apparently passing over such attractive B-text passages as the confession of Wrath (not present in A).50 Yet a parallel for such erratic scribal behaviour lies close at hand in the Prologue of Ht. The Ht redactor evidently had a C-text exemplar that included the Prologue, yet he passed over the explosive new C-text lines in which Conscience attacks the negligence of the contemporary clergy (C.Prol.95–124). Apart from the C-text reading ‘And charge’ at B.Prol.89 for B ‘And signe’, the first evidence of the redactor's use of his C-text exemplar occurs almost at the very end of the Prologue, where for B.Prol. 219 he substitutes ‘Bakers & brewers bochers and other’ (fol. 4r).51 Ht's readings here correspond, as George Russell and Venetia Nathan observe in their description of the manuscript, with those of C.Prol.225 (‘Bothe Bakeres and Breweres, Bochers and other’) against those of B (‘Baksteres and Brewesteres and Bochiers manye’).52 And yet the variants involved appear utterly trivial. Why did the scribe ignore all the additional matter available in his C-text exemplar up to this point, only to quibble over such tiny variants in a line about bakers and brewers? No motivation readily reveals itself, and yet that is precisely what the redactor apparently did.
If the Prologue of Ht reveals one instance of inconsistent scribal behaviour, passus 6 and 7 of the manuscript together offer another clear example of the kind of unfathomable decisions a scribe with access to multiple copies might make. Warner questions why, if he had a good exemplar of B to hand as Adams and Turville-Petre propose, the scribe of N would continue with his execrable C-text exemplar at all.53 The Ht redactor's activities in his copy of the second vision raise similar questions. Here, the redactor inserted, after B.6.158, a version of C.9.66–87, 96–163, 189–280 plus C Prol.91, 95–127 that appears in a textual form unique to Ht and to one other copy, sigil J, the ‘Ilchester’ manuscript. This was, as I discuss further below, a poor version of the passages in question, many generations removed from the original text. And yet the scribe saw fit to use this peculiar collection of excerpts, despite having access to at least one complete exemplar of C. From this complete C-text exemplar he selected, in passus 9, only three lines (C.9.164–66) that were not present in his exemplar of the interpolations shared with J. These he inserted in the same location that Langland himself had added them, following B.7.89, whereas the interpolations shared with J he intruded into an entirely other location, the middle of the ploughing scene in passus 6. The logic behind such a series of decisions involving competing exemplars again largely defies analysis.54
Of course, a scribe comparing exemplars of different versions of the poem is unlikely to have had any inkling that they represented distinct sequentially composed versions as opposed to alternative representations of the same original text. Nor, lacking a full conspectus of variants like a modern textual critic, would he have had any way of determining the relative authority of lines or readings unique to any copy. In such circumstances, it is easier than Warner allows to understand how, for instance, a scribe might effectively use one exemplar to cancel lines present in another, as we must suppose the N scribe to have done in substituting B.15.533–69, the lines on the donation of Constantine, for the C-text equivalent of the same passage, which includes four additional lines (C.17.202, 212, 228, and 230). Counterintuitive as it might seem to us, particularly when a scribe appears otherwise to be in possession of ‘various versions of the poem and . . . keen to include them where appropriate’,55 this is precisely what the Ht redactor did. Russell and Nathan drew attention to one such example when they pointed out that in passus 8 of B, Ht omits line 69 ‘as the C reviser had done’,56 although line 25 is also omitted in this passus, despite finding a parallel in C. Clearer examples occur, however, in passus 13, where Ht omits B.13.178, also omitted in C, and in passus 16, where Ht omits B.16.187, also deleted in the C text. The most intriguing example of the phenomenon occurs in passus 11. Here is Ht's version of B.11.207–211α:
And aftir his resureccioun Redemptor was his name (B.11.207)
And we by hym were bought boþe riche and pore (B.11.208)
Forþi loue we as bretheren shold & eche man laughe on oþere (B.11.209)
Alter alterius onera portate (B.11.211 α)
And euery man help oþer for hennys shull we alle (B.11.211)
And þei we lyve longe at last shul we wende (unique)
The inversion here of B.11.211 and 211α may have been inherited from Ht's B-text exemplar, for OC2 from the same genetic subgroup of B also show signs of disruption here, with the Latin added in the margin. The omission of B.11.210 and the unique line added after B.11.211 are a different matter, however, and would both appear to be clearly the work of the same redactor. One might perhaps suppose the omission of one line and insertion of another to be random and unconnected, until one considers the passage in Ht alongside the parallel lines in C:
And aftur his resureccoun redemptor was his name
And we his blody bretherne, as wel beggares as lordes.
Forthy loue we as leue childerne, lene hem þat nedeth
And euery man helpe other for hennes shal we alle
To haue as we haen serued, as holy chirche witnesseth:
Et qui bona egerunt, ibunt &c
Here, unlike in the example from the Prologue discussed above, the redactor did not incorporate any of the minor rewritings of the lines in C. Neither did he substitute C's alternative Latin line for the one he found in his exemplar of B. He did, however, apparently omit B.11.210 because it was also omitted in C, and his unique line appears to respond to Langland's insertion of the additional line C.12.118. Thus although the redactor has, at this point in the text, been copying B steadily since B.11.96, and although he incorporates no new C matter here, nevertheless he appears to have kept his C-text exemplar continuously to hand, removing one line and adding his own unique line where C's version ran closely in parallel. Why he did not simply copy C.12.118 rather than supplying his own spurious line is another mystery.
Finally, whereas Warner casts doubt on the idea that N's text could have been produced by contamination involving both a beta and an alpha copy, there is nothing inherently unlikely about this possibility, and Ht offers one piece of evidence that suggests that comparison between a beta and an alpha exemplar might have occurred at an earlier stage of this manuscript's transmission.57 The Ht scribe copied the line B.13.391 twice, once in its proper location in the fourth vision of B, and once as part of an insertion into the confession of Coveitise in the second vision (after B.5.198, Ht inserts B.13.356–98, imitating Langland's own incorporation of much of the material from the Haukin episode into the second vision in C). The second time the scribe copied the line, it appears correctly. The first time, however, the line is erroneous and even nonsensical, reading ‘conscience’ for ‘And if I sente’ at the start of the line. The reading ‘conscience’ properly belongs to the line above, B.13.390, ‘Vpon a cruwel coueitise my conscience gan hange’. But ‘my conscience’ is the reading only of alpha, and in B.13.390 Ht reads ‘my herte’ with the rest of the beta manuscripts. We can be more or less certain that the Ht redactor did not have access to an alpha manuscript, since as noted above he made good some of the omissions that characterize the beta family by inserting the corresponding passages from a C-text copy.
The obvious explanation for the apparently accidental intrusion of an alpha reading here is that Ht's beta exemplar had already been ‘corrected’ in places from an alpha copy by the time it passed into the redactor's hands. The reading ‘conscience’ could have been marked interlinearly or in the margin, leading the Ht scribe to overlook it the second time he copied the line, but accidently to incorporate it into the subsequent line on his first attempt.
This example illustrates that the chances of comparison at some point in N's history with both an alpha and a beta copy may not be so remote as Warner suggests. It provides a clear instance, too, of the complex forms that contamination could take: the Ht redactor worked with three different versions of the poem, but his beta exemplar shows some sign of having been itself checked, at a previous stage in transmission, against a manuscript of the other branch of B. Finally, the example illustrates that if the motives of a conflating scribe are frequently impossible to discern, at times contamination may be quite unmotivated. Even a redactor with an apparently clear agenda to produce a ‘complete’ Piers Plowman has made thousands of not necessarily compatible decisions when confronted with three texts that themselves differed at multiple levels, both in large-scale revisions and removals and in meticulous word-by-word rewriting. Confronted with a mass of variants, the redactor behaved variously and unpredictably, sometimes ignoring whole chunks while fussing over seemingly inconsequential details, sometimes working line by line inserting his own compositions and omitting lines that Langland had cancelled, sometimes apparently treating the two verses of an alliterative line as mix-and-match units to be switched out as preferred. The motivation for any one of these decisions is obscure, but our inability to comprehend inconsistent scribal behaviour provides no evidence that would undermine the thesis that comparison of multiple exemplars might also explain the disposition of the text in N.
N, Ht, and J: Langlandian loose-leaf drafts?
Warner's own discussions of the relationship between N and Ht illustrate again how his preoccupation with the idea of the loose leaf of authorial draft leads him to overlook or discount the textual evidence that would render the authorial status of such passages highly unlikely. Because he apparently did not realize that Ht included substitutions as well as additions, Warner rejects the comparison that Adams and Turville-Petre proposed between this manuscript and N as products of scribes making use of multiple exemplars of Piers Plowman. Instead, he proposes a different significance to the textual relationship between N and Ht in their C-text portions from passus 10 onwards. His argument depends on the textual relationship between Ht and J (the ‘Ilchester’ manuscript) in the interpolations from C.Prol and C passus 9 that appear in Ht in passus 6 and in the Prologue of J (Ht includes C.9.66–87, 96–163, 189–280 plus C.Prol.91,95–127; J includes C.9.75–87, 96–163, 189–280, plus C.Prol. 91–159).58 As noted above, these passages appear in this pair of manuscripts in a textual form otherwise unique in the C manuscript tradition. Noting the relationship between Ht and J in these interpolations, Warner argues for a larger textual relationship between all three manuscripts, Ht, J and N:
Since N2 is extant only from Passus 10, no positive evidence indicates that it agreed with Ht and J for this passage, though so far as I know no one has suggested the existence of two separate C traditions behind Ht. The simplest postulate is that the HtN2 relationship and the HtJ relationship both point to a larger HtJN2 relationship.59
Warner apparently draws here upon Wendy Scase's well-known argument that the interpolations shared by Ht and J reflect authorial draft materials circulating on loose sheets. In his book, Warner argues that the textual affiliations of the interpolated material from C.Prol and C.9 in Ht and J ‘suggest strongly that it might descend from Nx itself’.60 All three manuscripts Ht, J and N, Warner claims, then, might witness to the same early stage of C, material that Langland composed on loose leaves (Nx being Warner's term for the hypothesized first scribal copy of this draft ‘N version’ or ‘ur-C’ text).
Yet such a proposal draws on Scase's argument for the circulation of early drafts of C on loose leaves while overlooking the textual evidence that makes, first, the suggestion of a larger relationship between HtJN unlikely and, second, the argument that this relationship points to authorial draft material impossible. Warner seems to misunderstand Russell and Kane's comments on Ht's textual affiliations, for the existence of more than one— indeed, three—separate C traditions lying behind Ht is precisely what their remarks imply:
In the insertions from the earlier part of C most variation from the adopted C text is with the X family or across families. But in the insertions from C X onward agreement is much more often with the P family, but also, notably, with N2.61
For example, in C.10.256–69 Ht, alone or in combination with N, agrees in error, by my count, four times with p-family manuscripts, once with a reading confined to the x family, once in a reading found in Cx, the ancestor of all the manuscripts, and four times in errors that are split across the two great families p and x. There are also several errors unique to Ht and N. By contrast, in the first fifty lines of C passus 5 (which Ht copies to line 103), I find six readings in which Ht agrees in error with p-family manuscripts, seven with the x family, and fifteen split across both families, together with one erroneous reading found in Cx. The obvious conclusion is that Ht either used two C manuscripts, each of a different character, switching in passus 10 where the first agreement with N occurs (C.10.237–41), or that he employed a single manuscript that had similarly changed affiliations at some earlier point in its history.
Contrary to Warner's sense that Ht drew on only one textual tradition of C, the interpolations shared with J in fact imply a third C-text source and one which was, like the source of the other interpolations prior to passus 10, also unrelated to the branch of transmission shared with N from that point onwards. As Russell and Kane point out, the character of the HtJ interpolations is obscure: none of the agreements in error with particular manuscripts is persistent and the variants shared with other copies are of a minor order. Examples include:
9.202 they] þes HtJPERVAQFKGN
9.225 frithes] frith HtJRM forestes] forest HtJM
C.Prol.99 hangeth there] þer hangiþ HtJMF
C.Prol.101 30w] om. HtJOLBSF
C.Prol.106 thorwe] for HtJQSF
Although conclusive proof is impossible since N does not witness the lines shared by HtJ, all indications would imply that Ht used at least two different C-text exemplars (one containing the interpolations shared with J, together with at least one complete exemplar of C). These exemplars between them represented three different branches of transmission, only one of which (from passus 10 onwards) is shared with N. It is, in short, impossible to identify a common source of all three manuscripts, Ht, J, and N, beyond Cx itself, the shared ancestor of all C-text manuscripts.
Cx is evidently not the common source that Warner has in mind, of course, in proposing that all three might descend from ‘Nx’, the first copy of an early C-text draft. But his view that N, Ht, and J might all share access to early authorial drafts circulating on loose leaves simply repeats Scase's original error of failing to take proper account of the readings of the passages in question. In his response to Scase, Ralph Hanna demonstrated with numerous examples how the HtJ text, including its omission and mislineation of lines in the C Prologue portion, descends by the usual processes of scribal error from the shared archetype of all other C-text manuscripts. The now-published C-text collations confirm his analysis of the scribal omission of lines 110b–13a in the C Prologue by showing that four C-text manuscripts, P2OLB, share in part the same mislineation (in lines 112–16) that Hanna argued had caused the ancestor of HtJ to drop several lines.62 The HtJ interpolations are thus removed by many generations of copying from the C archetype and cannot witness pre-archetypal drafts. Those who would maintain the status of the HtJ interpolations as authorial draft must explain those manuscripts’ rendition of lines such as C.9.206, which appears in the adopted text of C as ‘And carteres knaues and Clerkes withouten grace’. Ht and J represent the line as follows:
Ht Clerkis without grace & carters knaves
J Clerkes connyngles of scole and carters knaues
The HtJ version differs from the adopted text of C solely through a clear scribal error, whereby the reversal of the noun phrases on either side of the conjunction renders the line unmetrical, with J, as often, ‘editing’ to restore the alliteration lost in the common source of the two manuscripts.63
What prompts Scase's and Warner's willingness to overlook such infelicities in the individual readings of the HtJ interpolations is the allure of their apparent ‘shape’ as passages new to C that could be accommodated on loose leaves.64 Scase had claimed that the HtJ passages took the form of lines new to C, with a couple of contextual lines carried over from B as cues for insertion. She argued that it was unlikely that the HtJ materials were excerpts from a complete text of C, because the lines that Langland had carried over from B into C between the blocks of new C-text lines were not included in the interpolations. A scribe making excerpts from a complete copy of C would have no way of knowing, Scase pointed out, which the new C-text lines were.65 But Vance Smith has since comprehensively demolished this argument about the ‘shape’ of the passages as reflecting loose leaves of C-text draft. As he points out, the large block of omitted lines C.9.164–88 contains not only B-text lines carried over into C, but also several entirely new to C, lines 164–66, 179–83, and 188. The extended passage of new C- text lines at C.9.88–95 is also omitted from the HtJ interpolations. The ‘shape’ of the interpolations therefore in fact tells against their status as draft C-text lines. The omission of some of Langland's most humane new writing about the sufferings of the poor in C.9.88–95 suggests rather, as Smith convincingly argues, a scribal editorial agenda.66 There simply remains no compelling evidence to suggest that HtJ might reflect loose leaves of authorial draft associated with the same early draft materials that Warner would find in N.
Loose leaves and lost histories
The allure of the idea of the loose-leaf draft passage, as Hanna argued, led to Scase's failure to undertake any systematic analysis of the variants in the HtJ passages which would have pointed to their ultimate source as the archetype of all other C-text manuscripts rather than pre-archetypal materials. Similarly, Warner's attraction to the imagined shape of the N+beta-only passages as loose leaves of authorial draft is associated with a silence about the status of individual readings within these supposed drafts. It is central, for instance, to Warner's argument that one of his key passages attested by N and beta only, the lines on the Donation of Constantine (B.15.533–69), was revised between its ‘ur-C’ state and its final appearance in C. It is impossible to imagine, he claims, that such an ‘incendiary’ passage, in which Anima proposes that the clergy might be stripped of their temporal wealth, went unrevised by Langland in C. He points to the four additional lines that the passage contains in C (C.17.202, 212, 228, and 230), as well as a block of eighteen new lines spliced in towards the conclusion of the passage (C.17.233–49, inserted between B.15.567 and 568). He also notes ‘about twenty-seven other readings not in the N-W + version’.67 Warner offers, however, no explicit comment on the nature of the revision, if any, that he supposes this loose-leaf passage to have undergone in the final C version, beyond the addition of the new lines.
Examining the readings that differentiate the version found in N and beta from the C-text form of the passage, one can find a few readings that apparently evidence authorial tinkering: ‘ri3twise’ in N+beta (B.15.533) for ‘riht holy’ in C.17.194, for instance, or ‘plentee and pees’ in N+beta (B.15.538) for ‘pees and plente’ in archetypal C.17.199 (though the latter is a reading that Russell and Kane emend in their edition of C to agree with B).68 Yet the vast majority of the readings that distinguish N+beta's text from C are surely scribal, the result of errors in copying by the beta or the Bx scribe, as Kane and Donaldson's emendations in their edition of B indicate. For example, several of the unique lections in the N+beta form of the passage represent readings that render the line defective in alliteration (N+beta readings are given here in bold, with Kane-Donaldson's emendations and the corresponding reading of C in square brackets at the end of each line):
B. 15.546 For coueitise of þat cros men of holy kirke [KD + C: clerkes]
B.15.548 Wite ye no3t, wise men, how þo men honoured [KD + C Mynne]
B.15.552 Shal þei demen dos ecclesie, and youre pride depose [KD + C depose yow for youre pride]
B.15.555 The lordshipe of londes for euere shul ye lese [KD + C lese ye shul for euere]
B.15.566 Good were to deschargen hem for holy chirches sake [KD + C Charite]
This last example is particularly telling against any view that these variants might represent minor authorial revisions between an ‘ur-C’ version witnessed by N and beta and the final, C version. Are we to suppose that, having already composed around 600 lines on the theme of charity in B15— a passus followed by the vision of the Tree of Charity, and one that uses the word ‘charity’ some thirteen further times (out of a total of thirty-five in Piers Plowman B)—Langland faltered in his early ‘ur-C’ draft, somehow failing to find the word that would give him three full staves and settling instead for ‘Good’ until he finally hit on the alliterating term in his final, C revisions? It is true that a passage like C.Prol. 108–24 might offer evidence that Langland's draft materials were not regular in alliteration. But that passage offers no support whatever for the view that a line like B. 15.551 as copied by N+beta, ‘Right so, ye clerkes, for youre coueitise er longe’, lacking a stave in the b-verse, might be authorial (Kane and Donaldson here emend to the C reading ‘er come au3t longe’). To suppose that such a line were authorial draft, we would have to imagine that alliterative poets composed in prose that they subsequently translated into verse. Such would of course be a fundamental misapprehension of the metrical ‘rules’ of alliterative poetry, which represent not an ornament a poet might add later but rather a compositional vocabulary.
Warner does implicitly acknowledge the scribal character of N+beta's reading ‘Good’ when he later cites Kane and Donaldson's emendation approvingly,69 but in doing so he does not spell out the agent he believes responsible for any of these twenty-plus readings, most of them seemingly scribal in character, distinguishing N+beta from C. Since Warner's hypothesized ancestor of N, Nx, is the ‘the first post-holograph copy’,70 perhaps some of the conspicuous scribalisms in N+beta's readings in the passage are, in Warner's theory, to be attributed to errors made by the Nx scribe in copying Langland's ‘ur-C’ papers. Yet this situation is not what is implied by Warner's listing of all the instances that I have just discussed as ‘readings where C revises from N-W+’ (my emphasis).71 Warner's failure to address in any coherent way the status of the variants in this key N+beta passage is suggestive: throughout, Warner argues from patterns of attestation rather than scrutinising readings, the sole basis for determining originality. The alluring idea that the shape of the passage might point to its status as a loose leaf of Langlandian draft means that Warner neglects to consider systematically whether or not it looks, in the details of its individual readings, like draft authorial matter.
A similar failure to scrutinize readings attends Warner's proposal that some of the apparently heavily corrupted C-text lines in N's continuation might in fact witness, where they are lines new to C, ‘the poet's initial, “ur-C” efforts’.72 He cites the example of N's version of C.17.241, ‘which scans well and makes good sense, and in my opinion is a better line than received C’.73 Taken in isolation, there is perhaps nothing to suggest that N's version of the line could not have been written by Langland, although Warner's preference can hardly prove that it was. But viewed in its context in the manuscript, the likelihood of the line witnessing the poet's initial draft looks exceedingly slender. In the following line, for instance, also new to C, N's omission of the word ‘priue’ means that its version lacks a second stave in the a-verse: ‘Bote thorw pacience and priue gyle he was prince ouer hem all’ (C.17.242). The scribal character of the omission is confirmed by the fact that this error is shared by MS G of C. In the same line, N shares with DZWF the reading ‘of’ for ‘ouer’ in the b-verse. (N agrees in error with W ninety-two times, forming one of the twenty-six ‘relatively persistent’ variational groups identified by Russell and Kane).74 These four manuscripts represent both the two branches of the C-text transmission (D is a representative of x, the others of p) and reflect the further divisions into subfamilies that occurred in subsequent generations of copying, after the split into the two ‘great families’. The pair ZW, for instance, witnesses the third generation of copying after Cx, the shared ancestor of all C copies.75 N's readings in this line confirm its status as a copy separated by many generations from the archetypal text and extremely unlikely to have recorded pre-archetypal material.
The line that precedes Warner's example, 240, is also clearly scribal. In the received text of C, this line reads, ‘Hadde al surie as hymsulue wolde and sarrasines in equitee’. N's unique reading, ‘at his wille’ for ‘as hymsulue wolde’, again drops the second stave in the a-verse. Another line in the same eighteen-line addition in C offers clear evidence of the sub-archetypal character of N's readings: the placement of 235α in the margin in N reflects similar disruption to this line in PERMZW of the p family and XYJP2D2 of the x family.
A final example, this time taken from lines shared with Ht, illustrates again the scribal character of N's C-text lines. C.19.245, another line new to that version, reads in the received text: ‘Lordliche for to lyue and lykyngliche be clothed’. Ht and N share a very different form of this line:
N Wan his lordeship here & 3it is demed to helle
Ht Wanne his lordship here & 3et is dampnyd in helle
As it appears in N and Ht, the line lacks alliteration, and its scribal character is indicated tellingly by its relationship to the previous line, given here in the adopted version of the C text with the variants beneath:
Sethe he withoute wyles wan and wel myhte atymye (C.19.244)
Sethe] And sethe XYJP2D2DN2Ht he] om. DHt wyles] wyle N2Ht wan] om. ZWN2Ht and … atyme] wiþ trwþe & wiþ resoun N2; in treuþe and yn resoun Ht
The omission of the verb ‘wan’ in N and Ht, perhaps occasioned by confusion over the correct scansion of this line, has clearly inspired the rewriting of the following line 245 in the common ancestor of those two manuscripts. Moreover, the original omission in line 244 that has occasioned the editorial activity visible in N and Ht is shared, again, with the subgroup ZW. As already noted, this pair belongs to the third generation of copying after the C archetype, and at least a further two generations are required to account for, first, the subsequent smoothing unique to N and Ht, and second, the divergent readings of these two copies (N's demed versus Ht dampnyd).
N's unique readings in its C-text lines confirm, then, its status as a copy removed by many generations from archetypal materials, let alone prearchetypal drafts. Although Warner is again attracted to the idea of the loose leaf draft passage, suggesting that the state of N could be explained by draft C-text lines having been supplied on loose sheets ‘to a copy of ur-C whose scribe had already ravaged its text’, it is difficult to see how such a scenario could account for Warner's own example of C.17.241. Here the material that Warner proposes as authorial draft appears in the form of a single line, which we would thus have to imagine was alone spliced between two clearly ruined lines.76
Ultimately, the idea of the loose leaf of authorial draft apparently has such a powerful hold on Warner's imagination that he seems determined to dismember the B-text of Piers Plowman into a series of loose-leaf-sized passages, each of which we must now consider suspect as potential infiltrations from early C-text drafts, even in the absence of any textual evidence. In his final chapter he argues, partly on the basis of thematic connections with other ‘ur-C’ passages including B.15.533–69 on the Donation of Constantine and B.19–20, that another group of lines, B.15.417–28α, might also have been intruded into B from Langland's early C-text drafts:
The two characteristics that lead me to suspect that this was not integral to B are that, like the Donation of Constantine passage, in received B this interrupts a developing discourse about evangelism, not alms … and that half of these lines end up revised in C passus 17, suggesting its character as draft material … If my hunch is right, then we need to look beyond the final two passus and their concluding rubric when considering the extent of C's impact on Bx, even where both B families attest the passage.77
Warner has here abandoned, of course, the very category of textual evidence, the agreements of N+beta against RF+C, that had led him to hypothesize the existence of an ‘ur-C’ text in the first place. Here, besides the poor ‘fit’ of the passage into the ongoing argument, its subsequent revision in the C text alone points to the status of the passage as ‘draft’ material not integral to B. Such a criterion must be problematic, unless of course we take the whole of B to be a draft of C.
A similar sense of literary discontinuity had, of course, prompted Kane and Donaldson's original proposal that the Donation of Constantine passage had been misplaced by a pre-archetypal scribe as the result of a misfolded bifolium. The Athlone editors cited its ‘inconsequence, taking the form of discourse interrupted for no apparent homiletic or dramatic purpose or effect’. In that instance, however, the state of the same material in the C text lent some support to the argument for B's disarray.78 With much less support, a similar perception of literary discontinuity had also led Manly to propose his own radical theory that five different authors had composed the three versions of Piers Plowman. Where Manly had constructed a hypothetical ‘lost leaf’ to explain his perceptions of the poem's literary disruptions, Warner works in the concluding chapter of his book not from the ‘objective’ textual evidence of his earlier chapters but from what he acknowledges to be a mere ‘hunch’ about the disunity of B.15. He had earlier dispensed with literary criteria in arguing against the view that B19–20 are integral to the themes and structure of the B version, as almost all readers of the poem have perceived them to be. A feeling for the integrity of B.19–20 to the meaning and design of the poem, Warner says, ‘does not constitute an argument against all the textual indicators that these passus were not integral to B’.79 But in his final pages, Warner reinvokes the category of the literary in order to discover the purely literary discontinuities in passus 15 that in his eyes point to the discontinuity of its textual production—in the absence, however, of any textual evidence of the kind that, in his view, overrides any sense that readers might have that passus 19–20 are essential to the structure of B.
Such a move is ill taken, and not only because of the warning that the example of Manly provides against constructing a textual argument out of a subjective literary impression. B passus 15 is throughout, not only in the two passages Warner believes intruded from C-text drafts, marked by an extraordinary digressiveness, with Anima pursuing several distinct strands only obliquely linked by his theme of charity. The digressive quality of the sequence is established from the beginning, with Anima leaping from an account of his own various functions to the perils of the prideful quest for knowledge (B.15.50–69), prompted by Will's jest about his many names. This theme segues into an attack on friars’ intellectual presumption (70ff), followed by a more general denunciation of corrupt clergy (82ff) and hypocrisy (110ff). At line 149 Will, perhaps growing impatient like the reader, interrupts this already meandering discourse with a question that picks up the key word at the end of line 148, charity, prompting another change of course with a long definition of that virtue that eventually encompasses the lives of the desert fathers. Their exemplary mildness (258) and ‘patient poverty’ implicitly contrasts with the prideful acquisition of both knowledge and material goods by the contemporary clergy, who should rather follow the example of God's miraculous sustenance of Mary Magdalene on roots and dew by providing for the deserving (307). Anima here resumes his earlier attacks on the wealth of the clergy and their collusion with the rich and on human knowledge—with the discussion of the latter now dominated by its contemporary decline rather than its pride- fulness (354ff).
The introduction of the theme of the conversion of Muslims seems to spring organically, but again digressively, from the complaint about contemporary clergy who ignorantly skip over parts of the divine office, meaning that the salvation of the ‘lewed peple’, like that of Saracens and Jews, must depend on their faith alone (389–90). Muhammed himself as a deceitful learned man offers a parallel with contemporary clergy, and by an ingenious conceit the dove with which he tricks his followers is likened to the ‘dove of covetousness’ fed by English clerks (415), in contrast to Antony and the founders of the friars who lived humbly by alms (417–28α the passage Warner believes to be an intrusion).
In short, Anima continually, not only in the lines Warner highlights, interrupts himself as he pursues obliquely several overlapping but not explicitly linked subjects. The logic of inserting a passage about alms into a discussion of evangelism is apparent to a reader who has been following the strands of the discussion thus far and who has understood the implicit contrast that develops throughout the sequence between a non-grasping charity, which ‘ne chaffareþ no3t, ne chalangeþ, ne craueþ’ (165) and which is embodied in figures such as Antony and Francis, and the intellectual ambition, intellectual failings, and covetousness of the contemporary clergy. The discussion of Antony living on alms, though it might at first glance appear another digression, does not interrupt the discussion of the conversion of Muslims but rather, in the complex harmonies of Anima's sermon-like discussion, underscores the relevance of the theme of conversion to the earlier discussion of the desert fathers, whose example the contemporary clergy betray with their greed, thus leading Christians from right belief.
In the history of Piers Plowman studies, textual controversies have often overshadowed efforts at literary interpretation. Yet textual controversy can also stimulate a more attentive formal analysis of the poem. As Anne Middleton and Brewer have observed, Manly's challenge to the assumption that a single author was responsible for the three versions of the poem had the unintended effect of forcing his opponents to define more carefully the literary properties of Piers Plowman, and Manly had himself perceived a greater structural coherence to his ‘A1’ than it had hitherto been thought to possess.80 Warner, in offering a challenge to the established paradigm as provocative as Manly's a hundred years ago, has powerfully indicated how many textual problems remain—like Conscience at the end of the poem— crying out for answers. But the closing pages of The Lost History of Piers Plowman might stimulate another kind of response as well. My proposal is thus more modest than Warner's. I suggest that before we abandon the notion of an integral B version of Piers Plowman, we ask how many of the discontinuities like the one Warner finds in B. 15.417–28α and like those other textual scholars have found elsewhere might appear purposeful when subjected to a more patient reading. Langland's logic—or, in Kane and Donaldson's terms, his ‘homiletic or dramatic purpose or effect’—need not be always immediately evident. Likewise, I propose that we do not dismiss out of hand the possibility that repeated scribal contamination (rather than repeated releases of authorial drafts and revisions) might have produced the text of National Library of Wales MS 733B, even though the motives and logic behind such activity are necessarily lost to history.