 | |  |
Notes
The author wishes to thank Professor Ralph Hanna for an extensive critique of this paper and for many suggested improvements that have been incorporated into its argument. Thanks are also due to Thorlac Turville-Petre, Ian Doyle, and Charlotte Brewer, with whom exchanges on this topic have always been delightful as well as instructive. Responsibility for all judgments rendered here is, of course, exclusively my own. An earlier version of this paper was delivered as the inaugural R. E. Kaske Memorial Lecture at Cornell University, 9 November 1989. Funds for manuscript research were provided by the American Philosophical Society.
Some plan for systematically emending B from A and C had been intimated by Grattan and Chambers—and by Elsie Blackman—long before Kane ever came to the Athlone project (cf. Chambers 1910, 27; Blackman 1918, 518; and Chambers and Grattan 1931).
Patterson's inexperience with applied editorial reasoning is revealed in his inapposite critique of Anne Hudson's discussion of the notorious "bilyue/bely ioye" crux (215, n. 28). The logical force behind the principle of difficilior lectio is considerably dissipated when one is examining a set of variants where there is an a priori likelihood that more than one variant is authorial (cf. n. 18 below). Likewise naive is Patterson's notion of the role of hypothesis in the natural sciences, where—he tells us—"a theory can be disproven only by a better theory" (69). To the contrary, the history of physical science is strewn with theories wrecked by mere observation of embarrassing phenomena whose hypothetical position was not satisfactorily ascertained for decades or even centuries (Kuhn 66 ff.).
Those who doubt Skeat's "three-version" account of the Piers Plowman textual history often cite with approval Donaldson's own offhandedly skeptical remarks from thirty-five years ago: "I sometimes wonder whether the C-text, the B-text, and even the A-text are not merely historical accidents, haphazard milestones in the history of a poem that was begun but never finished, photographs that caught a static image of a living organism at a given but not necessarily significant moment of time" ("R and F" 211). Despite their engaging frankness, these remarks obscure some important distinctions between the extant textual states of Piers. Thus A's particular state of truncation seems to result from a cause that is neither accidental nor fleeting, viz., a major structural (and personal?) impasse that eventually forced the author to rethink and expand his whole plan. On the other hand, B has all the appearance of a finished (albeit not final) form of the poem. By contrast with both of the earlier versions, C manifests the incompleteness and indeterminacy typical of interrupted labor (Russell 1962, 44-45) and does indeed resemble the random snapshot of Donaldson's analogy.
In terms of the labels sometimes applied in New Testament text editing (Fee 174-175), my position roughly corresponds to "rational" eclecticism whereas Kane's coincides with "rigorous" eclecticism.
Cf. Donaldson's 1955 comments (presumably withdrawn later) regarding certain cases where the majority B reading parallels the reading of A and C but the minority B reading is perfectly acceptable as well: "These involve a substitution of similars so exact that there seems little to choose between them, even if the choice were put up to the poet himself" (196; italics mine). Later in the same discussion, Donaldson notes that "In no case is there anything to choose between the variant readings despite C's support of the majority: intertextual comparison on this level is futile" (196, n. 43; italics mine). Also cf. Tim Machan's recent evaluation of the same factor (the indeterminacy of the textual micro-organism) in his discussion of variants in Chaucer's Boece (192-193).
Kane and Donaldson seldom see several emendations as equally probable (cf. the discussion of their editorial monism on pp. 20-22), so the second kind of problem is rarely discussed. However, it is exemplified in their refreshingly candid treatment of 19.442 (a case discussed by Fowler [38-39]), which in Skeat (19.439) and all the B manuscripts read: "And god amende the pope · that pileþ holykirke." By contrast, Athlone reads: "And [Piers] amende þe pope, þat pileþ holy kirke." It is obvious (pace Fowler) that the alliteration of the line has suffered damage at the level of the archetype, and indeed beyond the level of the archetype since C also reflects this reading. However, there is almost no clue as to the proper way to repair the damage. In their discussion of this crux, Kane and Donaldson readily concede not only that their offering of "Piers" is purely speculative but that they themselves can suggest at least five other appropriate and likely emendations. Four of the five that they mention ("god haue pite on," "god pardon," "god make parfit," and "preie god amende") would have enjoyed (in my view) at least one advantage over their favored one, viz., preserving the line's essential meaning—which we have no cause to believe corrupt. The editors admit that, in the circumstances, "our decision is . . . arbitrary" (208).
I would define these as readings predicated on rejection of the copy-text or even of the archetype itself, depending on the available evidence in a particular textual situation.
Plausibility must always be assessed in terms of what we know about the genetic relationships among surviving witnesses. Thus in the case of Piers Plowman B, a plausible pattern of attestation might involve no more than one or two manuscripts, provided that key manuscripts supported our preferred reading.
Readers familiar with contemporary editorial theory will recognize a number of broad similarities between my proposed methodology for rationing emendations (with its correlative notion of the burden of proof) and that of Hans Zeller. Like my assessment of Kane-Donaldson, Zeller's critique of the Greg-Bowers approach to eclecticism focuses on the inability of that system to deal realistically with the multiple authorial variants so frequently encountered in works extant in several "authorised" versions (and its consequent tendency to produce contaminated scholarly editions). Nevertheless, my own thinking on these issues evolved independently of Zeller's influence, before I was familiar with his work—from the purely practical necessity of trying to re-edit Piers B. Moreover, his adherence to copy-text would, in application, be rather more conservative than my own—and for reasons that he well understands (240, n. 13): a nineteenth-century printer's fair copy makes for a much more consistent source of "authorised" readings than any late medieval vernacular manuscript, which, almost by definition, has suffered from contamination as well as all the usual scribal inadvertencies.
Whatever may have hindered him from adapting his A methodology to B's markedly different (and more restrictive) textual circumstances, it is certainly not my intention to imply that Kane was poorly informed, at the outset of his work, regarding the actual key differences in the textual traditions of A and B. His 1948 discussion, though brief, makes clear that he was very well aware of these (5).
In referring to the Crowley prints of 1550 as "authoritative," I am, of course, rendering no judgment on the relative authenticity of their text. My intention is solely to acknowledge them as a source not descended from any other extant copy and therefore capable (at least collectively) of independent testimony concerning the readings that once may have existed in the archetype. In that respect they are no different from any of the other surviving manuscripts of Piers B.
The obvious exceptions are the sophisticated MS Oxford Corpus Christi 201 (F), the thoroughly modernized MS Sion College Arc. L.40 2/E (S), and the heavily contaminated MS Hm 114.
This unevenness suggests that archetypal B may have been copied by two scribes or derived from two exemplars, one of which was much weaker than the other.
It is ironic that Brewer brings against Kane substantially the same objection (i.e., petitio principii) that Kane had earlier offered to classical recension. In neither case does the objection seem conclusive to me, but it underscores the problem of circularity implicit in all editorial reasoning and should remind us of the perils of any method when applied inflexibly. The pure recensionist believes that he can discover the controlling genetic relationship among his extant documents by concentrating on test cases where the relationship seems especially clear. He then merely surmises that in the more difficult or obscure cases the same fundamental relationship usually subsists. Such suppositions fly in the teeth of common sense and experience, which attest that many medieval documents are of contaminated parentage and that shared errors in them often occur through coincidence rather than affiliation. Despite its obvious differences, Kane's method involves a strikingly similar sort of rigid inductive generalization. By concentrating on sets of variants where originality seems indisputable, Kane believes that he can discern reliable, uni-directional scribal habits (a paradigm as idealized as any stemma) and thus resolve by appeals to usus scribendi the same textual anomalies that the recensionist would dissolve genetically. Yet such a programme offers its own offenses to common sense and experience, which aver that real scribal habits (as opposed to those described in manuals for editors) can go in several directions unpredictably, cancelling out or contradicting each other: "one may not assume scribes always to have gone in one direction" (Fee 185). Kane tries to place the recensionist in a logical box by asking why, if he is able to detect error without his stemma (and he must do so in order to develop the stemma in the first place), he doesn't just make all such decisions empirically, as the eclecticist would. But the question may be turned around on him, as Brewer does in essence, by asking why, if he must employ patterns of attestation to formulate his rules for scribal error, he doesn't just use attestation all the time, like a sensible recensionist or "best text" editor would. An illuminating recent discussion of ways to escape from the methodological circle faced by all editors may be found in Hoyt Duggan's "Langland's Dialect and Final -e," 160-170, esp. 163. Also cf. his "The Evidential Basis for Old English Metrics," 157-159.
Brewer has generously conceded this point to me—a matter of frequent past discussion between us—in her recent assessment of Kane's work (77, n. 10), and she is quite right in also noting that Kane and Donaldson never explicitly avail themselves of such a defense. I think they took it for granted. She, however, believes that they became completely skeptical about the value of majority attestation during their editing of B (thus undermining the principles behind Athlone A) and cites as evidence their statement that "unanimous attestation even in several traditions [is] by no means surely reliable evidence of originality" (129). Nevertheless, it seems clear to me that this statement applies, not to the unreliability of broad patterns of attestation throughout the manuscripts of the three traditions but rather to the tenuousness of even the most massive attestation as proof in any single case.
Lee Patterson seems momentarily oblivious to the existence of this fallacy, offering in defense of Kane-Donaldson a line of reasoning that involves both a tautology and an unintended parody of Kane: "this edition validates each individual reading in terms of every other reading, which means that if some of the readings are correct, then—unless the editorial principles have in an individual instance been misapplied—they must all be correct (69; italics mine).
It should be noted that this example is merely illustrative of a type of problem; in choosing to emend B's bely ioye with A's bilyue, Kane and Donaldson do not pose the question between these variants so simply as I have done here. For them the issue is indeed involved with the principle of authorial subtlety versus scribal overemphasis, but it is also bound up with a related question of possible mislineation in Bx at 7.129-130 (i.e., they believe that by foweles has been transposed to the end of 129 when its proper position is, as in A, at the opening of 130). Such an error, if it occurred, would have necessitated scribal adjustments to both the head and tail of line 130. They conclude, therefore, that bilyue and bely ioye are equally authorial in both versions but that Bx has dropped the subtler reading, bilyue, that properly attaches to Piers; has promoted the more emphatic one, bely ioye, from its original position (descriptive of the "foweles" of 129/30) to replace bilyue; and has smoothed the resultant truncated line with worldes blisse (of dubious metricality). Cf. Anne Hudson's observations on this crux in her essay on "Middle English."
See, for example, my discussion of baches versus bankes (B5.514) in "Editing and the Limitations of the Durior Lectio," Yearbook of Langland Studies 5 (1991), 7-15. In many instances where Kane and Donaldson invoke the "harder reading" to reject a B variant that I accept as authentic, the difference between us has less to do with the respective merits of the readings than with our assessments of the general textual situation. Often I would agree with their choice—if we had sufficient reason to believe no authorial revision had taken place. Among such cases, I would include 5.554 ("he"/"wy"), 6.190 ("dryue awey"/"ditte out"), 7.146 ("in þe"/"euene"), and 9.178 ("togideres"/"ysamme"). Unfortunately, we can never have that kind of security (unless we adopt the programmatic meliorism of Greg—cf. n. 24 below) since the surviving evidence suggests a pattern of sporadic and unpredictable revision, with some passages receiving only isolated, small changes while others were completely rethought.
The obvious case is that of Haukyn's wife. See Alford, "Quotations," 89-94, and Alford, "Review of Kane and Donaldson," 1003, for possible theological ramifications of this reference and for a discussion of how the Athlone editors arrived at their mistaken conclusion that "wife" is impossible.
Cf. Fee's discussion (194-197) of the many instances in New Testament textual criticism where one eclectic test (e.g., usus scribendi) may yield results that contradict those derived from another (e.g., which reading best accounts for the others?). In such examples, a "rigorist" will either follow an ad hoc rule of priorities or will resort to whim.
Consider, for example, Kane's treatment of such a clichéd and indifferent variant pair as ful/wel. These two adverbs occur in phrases like "And seide ful/wel softely," "wel/ful pensif in herte," "þriueþ wel/ful late," and "in a wel/ful perilous place." For all four of these examples, as well as two others, the unanimously attested B archetype reads ful. But Kane had chosen wel when editing each of these in the A version. Ignoring the impossibility of making any meaningful distinction between these two words on grounds of their intrinsic value, he managed to convince himself that ful was more emphatic than wel and thus always more likely to be scribal. Consequently, though a majority of A-version manuscripts also support ful in each of these instances, Kane chose wel as the authorial lection in that earlier edition. Therefore, when he edited B he had no choice but to veto all of these archetypal occurrences of ful. Not to have done so would have been to invite speculation about the accuracy of his judgment in editing A. There is, nonetheless, at least one instance in B where Kane and Donaldson allow ful, 6.109: "Oþere werkmen þere were þat wroʒten ful faste." Here, for some reason, Kane had chosen ful in the A version (though against 6 A mss attesting to wel) and thus had to choose ful in B. If he hadn't found it there among the B variants, he would have been forced to conclude that B was corrupt.
On the other hand, if one accepts (as I do in general) the results of Hoyt Duggan's and Thorlac Turville-Petre's recent application of metrical and syntactical analysis to the editing of Middle English alliterative verse (e.g., Turville-Petre, "Editing," 155ff.; Duggan, "Alliterative Patterning," "The Shape of the B-verse," and "Notes Toward a Theory," passim), it may be possible—in certain cases—to judge the competing claims of such tiny variants as those discussed above with some probability of success. Should Kane and Donaldson prove (by the yardstick of these newly devised criteria) to have guessed rightly more often than not, it may enhance their reputation for intuition but it can do nothing for their editorial method, whose rules are inapplicable to this level of analysis.
Cf. Brewer. One must, however, distinguish between variants in A manuscripts that are "authorial" because they originated (or were authorially endorsed for the first time) in Langland's B or C revisions and variants that are "authorial" because they reveal unresolved alternatives unique to the A tradition itself. The latter situation seems fairly inconsequential when compared to the problem involving variants shared with B or C, but it may, nevertheless, comprise several dozen cases.
An aversion to the notion of authorial variants is one of the most important threads running throughout Kane's illustrious career as an editor. First voiced in 1960 in the prolegomena to Piers Plowman A, his reluctance to concede the widespread existence of such variants—and to grapple with consequent complexities for editorial philosophy—is still apparent in two recent essays, one on the Manly-Rickert edition of the Canterbury Tales and the other on the two versions of the Prologue to the Legend of Good Women (which includes his unconvincing efforts to reclassify many variants in the Gg 4.27 text of the Legend as scribal). Yet he is hardly alone in adopting this viewpoint. In his essay on Manly and Rickert, Kane assures us that "when the author made a change he had an object in view and that it has been realized" (1989, 203); this merely restates an idealist editorial assumption traceable back to Sir Walter Greg, viz., that any authentic authorial variants will prove themselves genuine by their excellence and that the burden of proof is on their shoulders rather than the editor's: "a later variant that is either completely indifferent or manifestly inferior, or for the substitution of which no motive can be suggested, should be treated as fortuitous and refused admission to the text" (1966, 387; italics mine). Without apologizing for the inherent meliorism of this approach, one must, nonetheless, acknowledge that Greg's principle was formulated for dealing with a particular sort of problem (Elizabethan plays surviving in more than one edition) and that he explicitly excludes from its purview all cases where "complete re-writing" is involved. Unfortunately, Piers Plowman B overlaps Greg's categories since it exists as a complete rewriting of A but offers in addition many instances of isolated, minute revisions. John Bowers has proposed "hostility to authorial variants" as a defining characteristic of the entire last generation of English and American editors and sees it as a reaction against their own predecessors' credulity in such matters (445-447; cf. Hanna's evaluation of the work of George Pace on "Truth"). However that may be, authorial revision (with the resulting multiplication of "authorised" variants) constitutes a stumbling block to the theoretical framework of traditional eclecticism since it undercuts a major enabling hypothesis: the uniqueness of creative genius, for which there can never be more than one best way of phrasing a particular thought (cf. Pope, headnote). Zeller suggests that this misconception stems from a failure to appreciate the fundamentally different ways that authors and readers approach a given text: "The exegete presumes that the author's text arose not by chance but of necessity, that it is unique, unrepeatable, artistically complete, a text of maximum significance. Similarly the editor searches in the transmitted text for the one authentic text, in comparison with which all else will be textual corruption. . . . For the author on the other hand a text is something to be created by selection from the semantic inventory of the language, from the quantum of synonyms. . . . In the end the author has made a decision one way or the other, but he could have decided differently. For him creative writing does not mean necessity, but the possibility of variation" (258).
Ironically, Chambers and Grattan—by contrast—had decided long ago that it was essential (where possible) to establish the authorial text of B (whose author used a much better A manuscript than any now extant) before one could establish the text of A (1931, 1-2).
Cf. Pearsall, who derides the "belief in the poet as the practitioner of a divine mystery, attentive to the significance of very minute differentiation of wording" (1985, 99). While Hanna quite rightly contrasts the idiosyncratic nature of Langland's style to the purely stereotyped qualities of King Horn—in arguing that the difficilior lectio is more applicable to the former than the latter (1985, 937)—it also remains true that the compositional habits of any alliterative poet, however sophisticated, operate within a framework of stylized, formulaic, and even clichéd phrasal patterns. Many of these variations are probably attributable not to fussiness but to the author's attention having been focused at a higher semantic level than that of the individual word. Thus, even in the case of a Langland, our use of the lectio difficilior must be cautious, for this rule assumes a self-consciously bookish author (e.g., Vergil, not Homer) weighing every syllable (cf. Russell 1962, 43: "no medieval author in the alliterative tradition would have felt sufficiently strongly about the minutiae of his line structure to attempt a word-by-word check of his manuscript").
Despite solemnly avowing that his purpose "is not to cast doubt upon [Athlone B's] . . . editorial procedures," Patterson shrewdly puts his finger on this pattern of self-vindicating textual choices. He notes that when one compares A to B and C together, the latter versions appear to have corrupted A (if we accept Kane-Donaldson verbatim) at a rate of 1 mistake per 14 lines. However, when A is compared to B alone (in passages not found in C), the rate of B corruption would appear to jump to 1 mistake per 1.3 lines! How, wonders Patterson, can one credit the apparent implication of these and other relevant figures, which would suggest that B by itself is ten times more corrupt than when it shares readings with C? After all, when B and C alone are compared (where A has no reading), C shows a rate of superior originality that amounts to only 1 reading per 25 lines. Patterson concludes that A is thus being granted by Kane "a quite remarkable degree of originality . . ." (217, n. 52).
Unlike Charlotte Brewer, I incline to think the vast majority of Kane's A choices are valid and that one is not entitled to infer their wrongness there merely because they are likely to be wrong as emendations in B/C. It seems preferable to concede that revising authors do sometimes independently hit upon—and at other times simply endorse—changes, whether felicitous or merely fluid, already chosen (perhaps unconsciously) by their copyists (cf. Zeller 250-256, for examples from nineteenth-century German poetry). Pearsall implicitly concurs, noting that "An intelligent contemporary editor, with an intimate knowledge of his poet's language and idiom, may hit upon readings that seem preferable, not only to him and his modern counterpart, but that might even have been preferred by the poet himself if he had thought of them" (1985, 95).
In a number of cases, such as 10.142 (Kane A = [1] "dore" [2] "asid"; MS F = [1] "dore nail" [2] "aside"; Bx = [1] "deeþ and" [2] "arere"), only isolated agreements with A can be detected within an F text otherwise clearly B in origin. In discussing these cases Schmidt seems more flexible than Kane but similarly reluctant to acknowledge the likelihood of micro-contamination (lateral transmission at the level of the trivial isolated phrase or even the individual word) and thus cannot bring himself to reject altogether the hypothesis of F's having occasionally been proofed against a (now lost) superior B manuscript (280). And yet B MS C2's having been visibly "corrected" to the A readings at 10.142—which it shares with F (cf. my discussion of the "correction" of A MS H2)—probably attests to the presence of such micro-contamination. Chambers and Grattan allege B 19.236-237 in suggesting that F also occasionally contaminated his copy from a C manuscript (1931, 6).
Notable examples of this phenomenon include those cited by Kane and Donaldson at 8.49, 8.80, 8.103, 8.112, 9.33, and 10.142. In the last two instances, I also would support an emendation based on F; nevertheless, their treatment of each of these cases makes clear that F functions for Kane and Donaldson less as a witness to authorial B than as an excuse for printing Kane A. Cf. E. C. Colwell's scathing observation concerning the practice of "rigorous" eclecticism in editing the New Testament: "Such an editor relegates the manuscripts to the role of supplier of readings" (cited in Fee 179).
In the authorship controversy, it was Professor Manly who appealed to a standard very like Kane's "postulate of excellence" in arguing that the man who wrote the powerfully vivid and figurative Skeat A/B Prologue lines 11-16 could not possibly have spoiled it with the drearily explicit expansion found in the analogous lines of C. More circumspect, Chambers and Grattan contented themselves with observing that "we should ourselves hesitate to say how far it is inconceivable for an author, in revision, to spoil his own work" (1931, 11).
Hans Zeller's approach to the editorial problem posed by such extensive authorial revisions seems astute: "for the production of a new version the source of the elements which the author uses in the text is immaterial; it does not matter whether the variants are original or extraneous, misprints . . . or variants introduced by a publisher's editor. In other words, for the elements of a text to be authoritative, and thus for the text itself to be authorised, it is not necessary to assume that the author recognised extraneous elements included in the text (compositor's errors and editorial intervention) as such; he may have regarded them as original text. The necessary condition for our establishment of text is only that he should have registered [i.e., endorsed] the readings in question (256; italics mine). Zeller goes on to observe that our judgment that the author of a revised version did, in fact, "register" its characteristic readings represents "a generalisation made on the basis of the demonstrable cases" (256). He then concludes by drawing this distinction between his own method for treating such matters and that of Greg: "we assume that the demonstrable and the undemonstrable cases are fundamentally the same; that is to say, we assume that the author took note of individual readings in the text which formed the basis of the revision. . . . The contaminating edition, however, . . . assumes rather that the [revising] author not only ignored the misprint [= scribal error] as such, but also the whole of the passage in question including wording, etc., unless the opposite can be proved; thus it is assumed that the two cases belong to two fundamentally different categories, but that the demarcation line between them coincides with the borderline between the demonstrable and the undemonstrable. I cannot imagine how one could account for this agreeable coincidence" (257).
Examples of defective B passages convincingly emended by Kane and Donaldson from the other two versions of Piers may easily be found in all three possible sets of circumstances, viz., from A alone (e.g., 3.36, 3.71, 6.219, 6.249, 7.164, 8.43, 9.32, 9.123, 10.109, 10.166, 10.189, 10.207, and 10.353); from C alone (e.g., Prol. 201, 2.84, 3.316, 3.341, 5.152, 5.181, 5.329, 5.508, 11.233, 11.372-373, 12.204, 13.95, 15.98, and 15.102); and from A and C in tandem (e.g., Prol. 41, 1.165, 1.200, 3.222, 3.223-224, 5.47, 5.190, 5.617, 6.26, 7.96, 7.148, 7.180, 8.22, 9.7, 9.190, 10.54, 10.176, and 10.384).
Kane (and before him, Chambers) was aware of this possibility but discounted it (1948, 9, n. 2).
For example, at B2.10, Kane-Donaldson reads 'in' from AC, while B reads 'wiþ'. But some A and C manuscripts agree with the unanimous B lection. The same situation occurs in many other cases, viz., BProl.217 ('bondage' AC; 'bondemen' B), B1.103 ('ʒeftis' AC; 'siluer' B), B2.128 ('abigge' AC; 'abiggen it' B), B3.117 ('What' AC; 'To wite what' B), B3.210 ('men mede' AC; 'mede to men' B), B4.106 ('Lepe' AC; 'For lope' B), B5.214 ('softe' AC; 'oute' B), B5.220 ('beste' AC; 'beste ale' B), B5.514 ('baches' AC; 'bankes' B), B6.20 ('werche' AC; 'to werche' B), B6.92 ('my catel' AC; 'catel' B), B7.175 ('passiþ' AC; 'passed' B), and B10.55 ('take' AC; 'and taken' B).
In his most recent discussion of Piers textual questions, Kane offers an illuminating, though much too brief, survey of the problem of minute scribal correction—and thus contamination—between different limbs of the same version (1988, 187-188) but seems to overlook the possibility of the same phenomenon occurring interversionally. Hanna draws our attention to the potential complexity of such problems when he notes the difficulty in deciding whether the scribe of Hm 114 corrected trivial errors in his exemplar from free guesses or an independent source (127).
Examples of multi-versional lections where coincidental convergence seems, to varying degrees, less likely than micro-contamination include the following ones shared by A and B alone: B Prol.36, where Bx + MS H of A = 'Feynen' versus 'Fonden' of Ax; B1.103, where Bx + MSS R, U, and E of A = 'siluer' versus 'ʒeftis' of Ax; B2.160, where Bx + MS M of A = 'bad hem alle' versus 'alle' of Ax; B3.41, where Bx + MS E of A = 'bedeman' versus 'baudekyn' of Ax; B5.126, where Bx + MSS E, A, M, and H of A = 'þe beste' versus 'good' of Ax; B5.214, where Bx + MSS E, A, M, and H3 of A = 'oute' versus 'softe' of Ax; B6.16, where Bx + MSS M and H3 of A = 'comaundeþ' versus 'wile' of Ax; B7.17; where Bx + MSS H, M, and K of A = 'þe heiʒe' versus 'here' of Ax; and B9.179, where Bx + MSS A, M, H3, W, and H2 of A = 'þe same' versus 'riʒt also' of Ax. Those found in passages shared by all three versions include B1.165, where Bx + MSS E and H of A = 'bygynneþ' versus 'comsiþ' of Ax/Cx; B2.200, where Bx + MSS A, M, and W of A = 'þyng' versus 'tresour' of Ax/Cx; B3.222, where Bx + MS M of A = 'teche children' versus 'kenne clerkis' of Ax/Cx; B3.223-224, where Bx + MSS D, E, and M of A misdivide 323 by adding the beginning of 324 to it and MSS E and M then pad the ensuing short line with the same filler phrase used in Bx; B5.190, where Bx + MSS A, M, and H3 of A (presumably contaminated in their shared ancestor since their sibling, E reads a distorted version of the same phrase) = 'eiʒen as a blynde hagge'/'eyne blynd as an hagge versus 'eiʒen' of Ax/Cx; B5.303, where Bx + MSS H, N, and V of A = 'Hastow (ouʒt) in þi purs' versus 'Hast þou' of Ax/Cx; B6.5, where Bx/Cx + MSS E, A, M, and H3 of A = 'and (also) sowen (it) after' versus 'scribal omission of entire b-verse' in Ax; B10.54, where Bx + MS K of A = 'a tale ouþer tweye' versus 'how two slowe þe þridde' of Ax/Cx. Because no complete critical edition of C is yet available, it seems imprudent to attempt compiling a similar list of variants shared exclusively between B and C. The evidence available from materials ready to hand (i.e., Pearsall, Skeat, and Kane-Donaldson) suggests that there is comparatively little interversional contamination of C manuscripts from B.
Hanna believes the scribe of Hm 114 was aware of the existence of different versions of Piers: "The conflated Piers Plowman presupposes a scribe who knew that there were three versions of the poem, who knew how to recognize them with some facility, and who knew where to find copies of them" (123). The Hm 114 scribe appears to have operated within the London book trade (Doyle 41), and either he or his sponsor may, indeed, have known that the author of Piers Plowman had completely reworked the poem several times. The mere existence of Hm 114 certainly proves that, whether by chance or choice, he had his hands on copies of all three versions and carefully noted many of the differences between them, but it still seems to me unlikely that he understood them to be (as we now take for granted) distinctive authorial drafts of the poem. His conflating the three suggests instead his perception that none of his copies preserved the poem in its entirety, or, as Hanna describes it, his "awareness of deviant textual detail which needs to be concorded to produce a full and . . . acceptable text" (128). An identical motive probably accounts for the much higher incidence of contamination (both gross and micro) among the A manuscripts generally than among the B and C ones, viz., the perception that this version of the text, on account of its abbreviated and truncated appearance, lacked authoritative status and required careful corroboration (and sometimes expansion) from the longer forms.
I have no firm opinion on the question of whether the unique material in Bodley 851 is authorial or evinces a proto-A version of Piers. However, the statistical pattern formed by the unique B variants in this manuscript (i.e., those not witnessed by any other A manuscript while being clearly archetypal in the B tradition) is very similar to the pattern of this class of variants found in M (Society of Antiquaries MS 687), an A copy whose exemplar (as well as a more distant ancestor) appears to have been sporadically "corrected" from the B tradition.
Elsie Blackman raised this question concerning G more than half a century ago (513-515) but ultimately rejected contamination as a significant factor in accounting for G's peculiarities. Instead she preferred to regard all of G's trivial agreements with A as coincidental (many doubtless must be) and most of the striking ones as evincing G's partial derivation from a B exemplar superior to the common archetype of the other extant copies (she was unaware of MS H, a BA splice that sometimes shares G's eccentric readings—also, presumably, in part, through lateral transmission from the A tradition). Blackman's reluctance to conceptualize the problem as one involving intermittent contamination of G from the A version is, of course, understandable. She was accustomed to thinking of contamination as both more purposeful and larger in scale than what she detected in MS G (cf. Chambers and Grattan: "It is absurd to imagine a collator comparing his MS systematially with another one, and altering . . . trivial details quite frequently, whilst leaving untouched . . . large common blunders" [1931, 14]). Her training would have led her to expect isolated transfers of words or phrases only through memorial contamination, and that process would automatically have restricted itself to memorable linguistic units (cf. Donaldson 1955, 184, n. 17, where Blackman's view is rejected and G's actual character is attributed to correction plus memorial contamination). However, the still visible work of the H2 corrector (and that of the Hm 114 scribe) shows that interversional collation of Piers is practicable and can be sporadic as well as systematic. Similarly, the numerous trivial "corrections" observable in B MS M (no less than those of H2) demonstrate that a reader / corrector need not have been governed by any coherent purpose in order to alter considerably the apparent parentage of his "corrected" manuscript. Once such a copy produces progeny of its own and, in the course of time, disappears, the way is clear for someone to argue that its descendants represent a "uniquely valuable witness" to the authorial text.
Such behavior may seem less improbable when set beside that of the corrector(s) of MS Harley 6579 (a copy of Hilton's Scale of Perfection). Here one version of a prose work has been cross-checked and corrected (both minutely and sporadically—most heavily in Book I) against a significantly different version of the same work. The most prolific "editor" of the Scale manuscripts, James Grenehalgh (a Carthusian monk of the late fifteenth century), is said to have made nearly 1,000 critical annotations and corrections to the text of four different copies of the work. In all, notes Michael Sargent, "The scribes and annotators of nearly one-fifth of the surviving MSS [43] seem, then, to have been aware of more than one form of the text, and to have gone out of their way to draw attention to the fact. This in itself demonstrates a degree of lateral contamination among the MSS. . . ."
If so, this would certainly be ironic since he has managed to replace a line whose metrical form (though rare) is attested by many other examples in the Langland corpus with one whose form (ax/ay) is unquestionably deficient. Cf. Duggan, "Notes Toward a Theory," 53-67.
Cf. another microscopic example long ago noted by Chambers and Grattan (1931,, 13-14), where the Hm 114 scribe "corrects" an obviously meaningless phrase in B, "by þe souþe" (5.329), by substituting A's parallel half-line, "was red for to ryse." Unfortunately, he should have used his copy of C here, for it would have revealed to him that Bx was corrupted from "þei bi souhte."
That is, ancestors of R/F on the one hand and N on the other might both have suffered loss of text at this point and been repaired, quite by accident, through consulting a manuscript of an alien version. In the case of N, damage to the exemplar appears less likely here than a deliberate intention to enhance this A copy with material from one of the longer versions (cf. Kane 1960, 30ff., esp. 35-36). Similar phenomena involving versional exchanges of the same brief phrases or whole sets of lines are too frequent for coincidence alone to seem a comprehensively plausible hypothesis.
Donaldson's observation (1955, 193) that archetypal corrections or insertions tend to remain insertions "in subsequent transcripts" seems relevant in this regard since it implies that such confusions could be perpetuated and amplified by descendant copies for several generations after the authorial foul copy had ceased to be available.
For purposes of this argument, it is immaterial whether one, two, or none of the variants in a particular shared set is authorial in origin.
Russell is more concerned with the publication history of B and C than with that of A (39-41, 45), so that one is simply left to infer that A was the earliest version publicly available: "the short and essentially incomplete A-text finds readers and copyists in numbers from the beginning" (39; cf. the more forthright, but utterly unsupported, formulation of Chambers and Grattan: "During the fifteen years intervening between the two texts [i.e., A and B], friends and admirers no doubt copied the author's unfinished MS" [1931, 10]). Russell acknowledges (36) that A was particularly prone to contamination from the later, longer versions, but he overlooks possible implications for A's place in the versions' sequence of appearance. As for B, the uniformity of its text and the narrow scope of its stemma lead him to postulate that, unlike A and C, B was never widely distributed and was, presumably, withdrawn from circulation almost immediately (by its author or by the authorities) on account of its radicalism and was quickly superseded by C. C, we are told, "circulated among a large number of readers without any of the disadvantages of A- or B-" (41). Nevertheless, Russell candidly concedes that this theory of C's broader circulation and B's early withdrawal (based on a higher number of surviving C copies?) seems contradicted by dialect evidence (cf. Samuels, 240). Seen from this perspective, the audience for C was mostly limited to "the valley of the Severn and its vicinity" (41) while A and B manifest a much wider popularity!
Positing extensive interversional contamination might seem to rule out any critical edition for Piers Plowman, but in actuality we are far from such an impasse—for two reasons. First, many of the effects presumably produced by this process are so trivial that one could almost never hope to distinguish them by rational criticism anyway (i.e., matters on which the only sane policy is following one's copy-text). Secondly, where it is most palpable, micro-contamination appears to move predictably in the same direction as that seen in the large-scale borrowings (i.e., from B toward the other versions, especially toward A). What this means for the editor of A is that, when faced with variants of approximately equal merit, he should prefer those which are not present (or are less well attested) in B manuscripts. Fortunately, the broad scope of A witnesses makes it unlikely that any particular B micro-contaminant will appear in more than two genetically distinct groups of A manuscripts. In the case of B, the problem is much less acute since most of the lateral transmission is outbound rather than incoming. Even so, one must be on guard against the isolated evidence of several key manuscripts (F, G, and H) presumed to have been collated against A copies. Here one's hardest choices occur when F, G, or H offers the only satisfactory reading (a reading identical or similar to that of A), but one seemingly incapable of accounting plausibly for the scribal production of the common B reading.
In his recent study of "Langland's Dialect," M. L. Samuels points to a wide dialect dispersion for the A copies, observing that "They demonstrate a situation that is found elsewhere in the history of textual criticism, in which the oldest MSS of a work (or their descendants) are found on the periphery of the culture" (238). It should, however, be noted that Samuels is not arguing for A's precedence as the "first edition" of Piers). He is simply invoking the "commonsense view" as a means of accounting for the striking fact that many extant copies of A appear to have originated in the geographic and cultural hinterlands. In reality, this distribution may tell us nothing about the relative antiquity of A's circulation and evince no more than its locus of popularity. This provincial popularity, in turn, may owe much—as Ian Doyle has suggested to me—to its having filled a role as a sort of Reader's Digest condensed version of Piers Plowman, with much of the dreary exposition and hairsplitting theology left out. Such a humble function and audience may also be implied in the fact that nearly 40% of extant A copies are on paper, "while it is almost unknown for the B- or C-Texts to be written on paper" (Lyall 14).
That all traces of the earliest copies of Piers A should have disappeared is not without parallel. Doyle alleges the example of Richard Rolle and suggests (36-37) that early A copies may have been quickly displaced by the reading public's preference for the longer versions. Nevertheless, one is not obliged to offer any explanation for the disappearance of such materials, whose very existence is very much open to question.
Cf. Doyle's observation about Caxton's "business sense endorsing contemporary taste" in omitting Piers from his printing programme (36); also his conclusion that Piers "was not a leading article of commerce in the metropolis after the earlier years" of the fifteenth century (47-48).
I incline to agree with Chambers and Grattan that the "A-MSS. have very much the appearance of having been derived from originals which go back, if not to the author's actual autograph, at any rate to something very near it. There are passages where all the extant A-MSS. are [variously] confused or imperfect. This may sometimes be due to confusion in the author's autograph" (1931, 10; my italics and interpolation—cf. n. 48 above and related text). Chambers and Grattan proceed to contrast this state of affairs with that of the B manuscripts, all apparently descended from a carefully controlled copy full of invisible scribal errors. They fail to see, however, that this set of circumstances suggests a late release of the A autograph (Vernon, the earliest extant copy, is the only A witness to reflect a Worcestershire dialect, but it can be dated only to the 1390s) and a very informal process of publication, as compared to B.
The exclusive use of C to complete various A manuscripts (occurring on at least four separate occasions) is, perhaps, indicative of an early cohabitation between these two texts but may be an unrelated accident, especially since the inferior text states of W, N, and K (A sigils), as opposed to that of the inferred T-Ch-H2 ancestor, suggest an advanced degree of decay. Unfortunately we have no way to ascertain whether the corruptions in W, N, and K (some of which are shared with unmixed A copies) gradually accreted through the reproduction of a long series of AC parents or whether the splicings took place after most of the blatant textual error had been introduced. The mere fact that these manuscripts sometimes share errors with A manuscripts of unmixed parentage need not, as Professor Kane has often reminded us, be taken to prove common descent (and thus a late date for their splicing)—it may only demonstrate the insidious effects of convergent variation.
Though Kane appears not to have taken much account of this possibility in his editing of A, Chambers and Grattan were very well aware of it (i.e., "backwash of B- and C-readings upon MSS. of the A-text") and of its cause ("all the extant MSS. of the A-text are later than the B-text, and the majority are much later"). Moreover, their refutation of Mabel Day's argument for multiple authorship depends, in part, on their contention that individual readings, as well as occasional lines, from the B archetype imposed themselves randomly on isolated A manuscripts as well as on the ancestors of certain extant A families (1931, 28-40, esp. 36-37, 39).
Although "coincidental convergence" is incapable of accounting for a number of unexpected and striking agreements in error and, in the light of extant patterns of sporadic contamination, is unlikely to have caused as many of the trivial ones as we might otherwise have supposed, there still can be no doubt that, in thousands of cases, genetically unrelated Piers Plowman manuscripts agree by coincidence in trivial errors generated merely by their scribes having shared a common grammar and lexicon (cf. Chambers and Grattan 1931, 15-16).
| | |