copyright 2014 by SEENET
Web edition published by
The Society for Early English and Norse Electronic Texts
www.seenet.org
SEENET A.10
The edition aims to establish the archetypal readings of the witnesses to the B-version of
The practice of recension as the basis of editing is sometimes called into question on two principal grounds. The first is purely practical: that the relationship between witnesses may be so thoroughly obscured by contamination and coincidental variation that it is impossible to establish the stemma. The second is a more fundamental objection: that since the relationships can only be established on the basis of identifying erroneous readings, the stemma is not a tool for the editor but a product of the edition. We find the first objection surmountable and the second invalid. A number of scribes of
In its essential features, the stemma of the B-version was in fact established by Kane and Donaldson (1975), and their collations were refined by Schmidt (1995, 2008),B-text alone, and in any case are more widely accessible. We comment occasionally on substantive revisions to the notes.2)C][Cr(WHm)]. L is independent of beta1, while M, though corrected from a CrWHm-family exemplar, is essentially an independent witness for the first part of the poem (see III.2). Beta2 consists of CrWHm, beta3 of WHm alone, beta4 of GYOC2CBmBoCot, beta5 of BmBoCot. Although some of the details are uncertain, Adams supplies us with the following stemma (modifying Schmidt's and slightly revising his own published version):
Although Kane and Donaldson's collations actually establish a stemma, their view was that "the extreme frequency of convergent variation in the transmission of
As demonstrated by Adams (2000), the two key witnesses are L and R. Skeat was so impressed by the quality of L that he at first thought it might be Langland's holograph, and Kane and Donaldson, though choosing W as their copy-text, recognised the "superior originality" of
L (p. 214). We believe the L scribe to be a remarkably accurate copyist of the beta hyparchetype, uncontaminated by either of the other versions of the poem. Of the only two alpha manuscripts, F is a very eccentric text that reflects the activity of a scribe reshaping his text in both broad features and individual details, and one who had access at certain points to a text of A and probably also of C. We see no evidence to support the view that F had access to a manuscript of B
anterior to Bx as argued by Kane and Donaldson (pp. 165-72); see e.g. note to Bx.14.25AC versions. It follows that agreement of L and R is very strong evidence for Bx.Bx.2.121
Where the text of R conflicts with that of F, the only other alpha witness, its reading can often be seen to be the origin of F's invention or misunderstanding (see notes to Bx.10.3881WHm form a close group derived from a reasonably good text, beta2, with Crowley's print Cr1 as its most reliable witness, although account has to be taken of its modernisations and Crowley's access to C-text manuscripts. The much less faithful beta4 is represented by GYOC as well as C2, probably a direct copy of O. Apart from C2, which has no independent authority, G is the least useful of this group, a sixteenth-century manuscript showing conflation with AC and introducing quite a number of independent revisions and modernisations. We have not found that the beta5 derivatives BmBoCot offer useful evidence in constructing Bx, and while we have considered their readings at every point, we have not displayed them. Although we have examined the variant readings of all the Bx witnesses, we display and in general cite the readings of only the ten manuscripts LMCr1WHmGOCRF, in that order.
In addition to evidence drawn from stemmatic relationships, readings that are scribal can generally be identified by the usual characteristics of reversion to a commonplace or easier reading, adoption of a prose word-order, omission and misunderstanding. By the use of this "direct method" as well as genetic evidence, the establishment of the alpha and beta hyparchetypes is generally
pretty straightforward except where R's text is lost and the determination of alpha depends upon the very unreliable F. We have taken agreement of alpha and beta as conclusive evidence for the readings of Bx.
The difficulty is greater where alpha and beta lections are divided, since they are of equal standing for the establishment of Bx. In these circumstances, corroboration of one or other may be provided by the support of one or both of the other versions of the poem. This needs to be distinguished from the process for which the Athlone edition has been criticised, that is, importing readings of A and/or C into their critical text of the B-version. Only in those cases where alpha and beta, the two prime witnesses of the B tradition, are in disagreement are we justified in looking in the other traditions for support for one over the other so far as Bx is concerned.
A significant difference between alpha and beta is the absence of lines and longer passages from one or the other: alpha lacks 37 such units that are recorded in beta, and beta lacks 45 in alpha, a total of nearly 200 lines absent from each. All but one are listed in Kane and Donaldson, pp. 66-9, who regard them as accidental omissions usually caused by eyeskip (p. 65). Kane and Donaldson point to the likelihood of scribes skipping from paraph to paraph, which we regard as a major factor in very many cases, but, since they do not record paraphs in their edition, critics have focussed on their additional explanations of other kinds of eyeskip, some of which are unconvincing (Hanna [1996], 219-22, Schmidt [2008], 148-9). Partly for this reason, the view that at least some of these alpha- and beta-only passages represent "rolling revision" has gained currency.
Advocates of "rolling revision," that is, sporadic alteration by the author over a period of time, might argue that a B-text containing both these alpha and beta passages conflates (as traditional editions of
This seems a strong argument in favour of the Athlone interpretation of the evidence, viz. that both alpha-only and beta-only passages were present in Bx, with some of them lost in the alpha hyparchetype and some in that of beta. But there are difficulties. It has been objected that such extensive omissions in both alpha and beta posit two copyists more careless than any of the
On the other hand, there is nothing inherently improbable about "rolling revision," as Hanna (1996), 222-3, describes it: "even after B version 'publication,' but before
embarking on the major overhaul that produced the C version, Langland continued to tinker." If he thus altered his house-copy bit by bit over a period of time, this manuscript might have served the scribes of alpha and beta and the poet of C as copy-text at different stages
of its revision. Hanna (1996), 215-29, in a discussion that makes a powerful argument for "rolling revision," offers a detailed hypothesis as to how this might have happened (esp. 223-5). It is not a new idea. Skeat (1869), xii, first suggested that beta was the earlier copy of the text, while Donaldson (1955) initially argued that the alpha copy came first, in an article that makes some shrewd points. More recent interest in the mouvance of texts has given added impetus to such hypotheses, though Schmidt (2008), 149-51, opposes it with considerable vehemence. The curious fact of the late entry of beta's omitted passages might be explained as a consequence of "rolling revision": the passages were not part of the text
when the beta scribe had access to the house-copy, and he copied faithfully enough; but the more careless alpha scribe, making his copy at a stage when revisions had been entered, lost (by eyeskip, loose sheets that had been mislaid, or whatever) passages distributed throughout the text.
In whatever way the alpha- and beta-only passages are explained, the fact remains that both types of passages figure in the B-text upon which the C-poet worked, and to that extent the editor of Bx is justified in including them all.
This issue of the additional passages in either alpha or beta needs to be kept distinct from a matter more troubling for the editor: the variant alpha and beta readings of a word or phrase. In this case the editor is faced with alternatives rather than additions. Some, no doubt the majority, represent misreadings in one hyparchetype or the other, and comparison with AC versions will provide evidence of support for one or the other. However, in a small number of cases the readings of one hyparchetype will be supported by one of the other versions, and that of the other text by the other version. Alpha may thus be supported by Ax and beta by Cx, or more commonly vice versa. Since we see no evidence for contamination of either hyparchetype with the other versions, such shared readings must represent either coincidental error or authorial revision.
To take a clear example: in Bx.1.46Ac the moneye of þis molde . þat men so faste holdeth
Alpha (R) has the same line but with kepeth in place of holdeth. Such variation of synonyms
looks typically scribal, except that AC have exactly the same line, with A agreeing with beta and C with alpha. Of course, alpha and C could share coincidental error, especially as it is not easy to see Langland as the sort of poet who would fuss over such an apparently inconsequential change. Yet the whole passage is full of such small alterations to C, especially in b-verses; indeed the very next line presents another example:
Telle me to whom Madame . þat tresore appendeth (
The last word corresponds to Bx.1.47bylongeth in C. In this line, however, alpha shares the beta reading so that Bx is not in doubt. Bx.1.84A and beta, but is dropped in alpha and C. Most apparent instances of A/beta agreements against alpha/C are far more trivial, with such variation as bouȝte / he bouȝte (Bx.5.225He is / is (for / of (
Perhaps Bx.10.198A/alpha agreement against
beta/C. A has a b-verse deficient in alliteration:
Leue lelly þeron ȝif þou þenke do wel (K.11.144).
Alpha's b-verse is almost identical:
Loue þou lelly if þou thenke do wel (as R; F reads
Beta, however, has a correctly alliterating line:
þynke to). Loke þow loue lelly ȝif þe lyketh dowel.
This seems to lie behind the revised C-version:
Lerne for to louie yf þe like dowel (RK.11.133).
Yet there are so few instances of A/alpha agreements against beta/C, and the A/alpha reading is so obviously
inferior, that we suppose it more likely that alpha presents a coincidental error here.
In theory the editor is presented with an insoluble problem in such cases: once "rolling revision" of individual lections is admitted, it must follow that any alpha reading could be a revision and, especially after passus 10 when the A-text ends, "all unique RF readings present in C could be argued to reflect the intermediate version" (Hanna [1996], 229).beta, we would not be able to draw the line at the few large patches of apparently added or deleted materials but would have to accept every tiny pair of alpha/beta variants as potential evidence of fluctuating authorial purposes. Nevertheless, whatever may be true in theory, in actuality the vast majority of small variations between R/F and beta are readily distinguishable (i.e. not textually neutral at all)."B continuation, who must inevitably take agreement of either alpha or beta variant with Cx as strong evidence in its favour. Also inevitably, there may be a few occasions when the editor of B will be rejecting an unrevised reading and adopting a lection from the later version. Yet Hanna, who is an advocate for "rolling revision," concludes: "The poet made some extensive additions to his fair-copy version, but in doing so, he seems almost never to have queried its standing readings" (Hanna [1996], 229). In fact to the end of Passus 10 we find no convincing instances of A/alpha agreement against beta/C, and fewer than thirty instances of A/beta agreement against alpha/C, and many of these can easily be
explained as coincidental variation.Ax / alpha = Cx, see notes to Bx.1.46Cx / alpha = Ax, see notes to: Bx.1.72
The manuscripts are described in order of their citation, LMCr1WHmCGORF, followed more briefly by those not regularly cited, YC2BmBoCotHS. Dates and dialects are from Hanna (1993), 39-40. See also Doyle (1986), 35-48.
Edited Hoyt N. Duggan and Ralph Hanna,
s. xiv ex.; SW Worcs relict forms. Duggan-Hanna (Introduction, section III) suggest that the manuscript was actually written in London.
The text is divided by blue paraphs with blank lines between them. (There are no paraphs on f. 1r because there is a red and blue spray along the left margin.) Latin lines and words are boxed in red. The half-line is generally marked with a raised point, though with a punctus elevatus in the opening 25 lines and occasionally thereafter. The punctuation is quite often overlooked by the scribe.
L evidently offers a very careful and accurate copy of beta; in passus 6, for instance, we identify only three certain (Bx.6.9Bx.6.139Bx.11.357a for as and an for and. Later hands have for some reason marked lines or passages with large crosses. Still mainly legible are the scribe's marginal instructions for the rubricator as a guide to passus headings, which the rubricator often abbreviated (see IV.2 below).
Schmidt (1995), lxvii-iii, argues that L provides independent witness of the B-text citing the following lines (on which see our notes): Bx.14.194Bx: Bx.5.267Bx error), Bx), Bx), Bx). We suppose the following to be
easy coincidental error: or
for of), I in LR, but possibly a Bx error), I in LR). In Bx.13.166deme as sen, but the L scribe spotted the misreading and corrected it. For fuller discussion see our notes to these lines.
Edited Eric Eliason, Thorlac Turville-Petre, and Hoyt N. Duggan,
s. xv in.; SW Midl dialect.
The text is divided into verse paragraphs with blank spaces between (as in LWRY), but no coloured paraphs were added. Latin lines are generally boxed in red or in text ink. The original scribe usually punctuated with a mid-line raised point, but the corrector commonly altered this to a punctus elevatus for the first half of the text (up to the middle of passus 13).
M's textual relationships are more complex than the stemma can indicate. For much of the poem its text is relatively independent of other beta witnesses. A good example is Bx.5.209ware LMR, supported by AC,
against chaffare in beta1. Yet there are sporadic indications as early as passus 13 and 14 that M inherited errors shared with beta1; see e.g. notes to Bx.13.781WHm. Its text has
subsequently been very extensively revised, with corrections from a beta2 text and perhaps other copies as well, although the majority of its revisions are spelling changes to bring the forms into line with those used by sophisticated London scribes such as the W scribe (Turville-Petre [2002], 41-65). For the
argument that the corrector is the W scribe, see Horobin (2009). Crucially for our purposes, M is independent of L, so that they
cannot inherit wrong readings against correct readings in other beta
manuscripts. They can, of course, and frequently do, uniquely share right readings with alpha against all other beta manuscripts.
References are to the first edition (Cr1). Hailey (2007), 143-70, discusses the alterations in Crowley's second and third editions (listing them on pp. 165-70).
Crowley makes it clear in his Preface that he has consulted manuscripts of the C-version, and the lines he quotes (RK.8.348-9) share the readings of the early 16th-century manuscript, London, BL Royal 18 B.xvii (a text of the P family). For instance, he adds a C-text line after Bx.3.30
Crowley uses syntactic punctuation, with a comma to mark pauses within the line, and sometimes full-stops at the end of the line. The line is indented at the start of a paragraph. The text is the best representative of beta2, but with some
contamination from C.
Edited Thorlac Turville-Petre and Hoyt N. Duggan,
s. xiv ex. London.
Paraphs (more than in any other text) alternate in red and blue, with a blank line between paragraphs, as in LMRY. Latin lines and words are in enlarged script and usually enclosed in a red box. The scribe uses a very regular raised point in mid line.
W is a fully professional metropolitan production. The presentation of the text, including its spellings and grammatical forms, conforms very systematically to the best London practice, as in the Ellesmere-Hengwrt scribe of the
Edited Michael Calabrese, Hoyt N. Duggan, and Thorlac Turville-Petre,
s. xv in. SW Warwicks. (
There are two main text scribes, hand1 copies up to Bx.2.210Bx.P.177
Hm belongs to the beta2 group, with Cr and W. The scribes are often careless, and have corrected their own work quite heavily. In addition, the rubricator of the manuscript has
made numerous corrections, and a corrector or correctors has made hundreds of alterations to spelling (see Turville-Petre [2002], 41-65]. A major disruption has caused the reversal of Bx.11.445
s. xiv/xv.
Latin lines are in enlarged script and often underlined in red. Paraphs are in alternating red and blue. The mid-line punctuation is generally a raised point. With GOC2YBmBoCot,
C is a beta4 text.
Edited Judith Jefferson,
s. xvi1.
In later passus, particularly 19-20, the scribe uses a mid-line solidus, and in earlier passus some have been added in brown ink. There are no paraphs, but enlarged capitals, used at the beginning of each passus, also occur fourteen times at significant junctures.
The text is contaminated from A and perhaps also from C, with such contaminated readings sometimes in clusters: e.g. Bx.1.8wilne, G and A kepe; transgressores, G and A (and the P family of C) trespacers; wolden al, G and A asketh þe; þe, G and AC syght off thes; fresshe, G and the X family of C flessche &. Table of
contents on 101v-103r (Jefferson, 2010).
Edited Katherine Heinrichs,
s. xv1. N Herts (LALME LP 6550).
Latin lines are underlined in red and are sometimes placed in the margin, especially after Bx.10.119Bx.5.80
Edited Robert Adams,
s. xiv ex.; SW Worcs relict forms.
Samuels (1985), 232-47, notes its "south-west Worcestershire dialect" relict forms. Hanna (1996), 40, supposes "several stages of copying; in succession: SW Worcestershire, London, Suffolk." It might be observed that this mix could have been achieved by a Suffolk scribe copying an authorial blend of London and SWMidl. dialects, and need not imply a long chain of copies. The punctuation is generally a mid-line punctus elevatus and often a punctus at the end of the line. There are some paraphs in alternating red and blue, but many are indicated only by "cc". A blank space is left between paragraphs (as in LMWY).
R is a professionally produced manuscript by an extremely literal-minded scribe providing a very accurate copy of alpha and rarely tempted to improve on his exemplar, to the extent that he will copy nonsense (e.g. Bx.10.291Bx.5.594Bx.P.1-124
Edited Robert Adams, et al.,
s. xiv/xv; Essex (
F is written by an Essex scribe, using vellum of very variable quality with many singletons, perhaps indicating a limited supply in the area. Yet it was not a cheap production, since it uniquely has an illustration within the first capital. Paraphs are first in red and green, later in red and blue, others indicated but not executed. The scribe uses a mid-line solidus. He also copied the Prick of Conscience, fragments of which are preserved as Ushaw College, Durham, MS 50 (see
The text of
We have relied on Kane and Donaldson's collations of the following texts:
Y: Cambridge, Newnham College, MS 4 (the Yates-Thompson manuscript), ff. 1r-104r.
s. xv2/4. London, with underlying N Oxon forms.
C 2: Cambridge, Cambridge University Library, MS Ll.4.14.
s. xv2-4. Ely (LALME, LP 673).
Bm: London, BL MS Additional 10574, ff. 1r-99v.
s. xv1-4.
Bo: Oxford, Bodleian Library, MS Bodley 814, ff. 1r-92v.
s. xv1-4.
Cot: London,
BL MS Cotton Caligula A.xi, part 2, ff. 170r-286r.
s. xv1-4.
Bm, Bo, and Cot are so closely related that they are referred to by the common sigil B. Their B-text begins at passus 3; before that their source is a C-text, with a passage also from an A-text.
H: London, BL MS Harley 3954, ff. 92r-123v.
s. xv med. Norfolk (LALME LP 638).
After Bx.5.128A-text, and its B-text
is heavily contaminated from A.
Schmidt (2008), 135-6, rather oddly calls it "a significant witness" to B, while admitting that its "superior"
readings could be derived from A. At
any rate it is of no use for establishing Bx.
S: Tokyo, Toshiyuki Takamiya MS 23 (olim London, Sion College MS Arc. L. 40).
s.xvi med.
We can determine very little of the language of the archetype and have made no attempt to reconstruct it here. Scribes can never be trusted to have followed the forms and spellings of their exemplars. Indeed the scribe of W has been shown to have imposed with considerable consistency and intelligence a system reflecting the best London usage (
However, forms that differ from a scribe's usual repertoire may go back to his exemplar, and, where such forms are shared by beta and alpha (in particular L and R), they probably derive from Bx. Thus Samuels (1985), 241, observed patterns of distribution in the forms of any / eny, ȝif / if, come / cam, byȝunde / byȝende in L and R, concluding that these "must be presumed to be archetypal", and on the basis of spot-checks, Duggan and Hanna (2004), Introduction III.1, confirm these findings. Again, it is telling that -ende as the present participle ending only occurs on the same three occasions in L and in R, at Bx.17.50Bx.17.50
Samuels (1985), 232-47; (1986), 40; (1988), 70-85, argued that Langland's dialect can be placed in SW Worcestershire on the basis of a number of relict spellings, and if this is the case (for reservations see Hanna [1993], 5-8), such forms are necessarily among those transmitted by Bx. Some of these can be established as authorial where they satisfy the requirements of an alliterative pattern. We give three examples of words in Bx where this consideration applies: words for "she", "they", and "church".
The form he/heo occasionally contributes to the alliteration of the line, and in such cases it must be a relict spelling. L and some other scribes write this in alliterative position in place of their regular she: at Bx.1.75she). At Bx.18.175he, as in alpha; but this has been altered in L by the addition of the initial s-.
In Bx.18.156he over beta's she. In Bx.20.198h-. Evidently Bx had h- forms in all these lines. Where such forms appear in non-alliterating positions, however, they may or may not be archetypal. L has non-alliterating he at Bx.1.144he(o) for "she" more often than beta. The form is common in R up to the end of passus 5 – it appears eight times between 5.219 and 256, for example – but heo is not used after 5.648, though he(e) appears sporadically after this point.
This probably means that R was at first copying out the form of his exemplar, later shifting to that in his active repertoire (Benskin and Laing [1981], 66), which suggests that the form was common at least in alpha.
L uses hij for "they" on ten occasions:
Bx must have used the <h-> form in the three lines it alliterates, and it seems reasonable to suppose that L, whose own form is þei, reproduces beta on all ten
occasions. L is usually supported by M (twice corrected to thei), but M uses it nowhere else, implying that beta only used hij on those ten occasions. The pattern
in W is characteristic of the carefully regularising practice of that scribe: he has hij only where it alliterates,
otherwise altering to þei. The form
is rarely used in alpha. R has hij
just three times (a. In he (perhaps by misunderstanding). Otherwise F never has the h- form, and the evidence suggests that alpha generally avoided it. In he (F is absent), though the scribes probably misunderstood it as a sg. form.
Significant patterns emerge in the distributions of chirche (OE), in its various spellings, and the synonymous kirke (ON). These occur the following
number of times in the six most important manuscripts:
Kirke is a word of the Danelaw, restricted to northern and eastern dialects. Chaucer, Gower and William of Palerne never use it; The Wars of Alexander and the Gawain-poet never use cherche. Langland uses kirke as an alliterative convenience. L has it 21 times,kirke occurs six times does not take account of 18 examples of holikirke and 3 of kyrke.holy-kirke agreeing with MCr1W (kyrke in final position in Bx.19.2cherche. R is strongly resistant, and almost always substitutes cherche, though in kerkes in final position. Of the beta witnesses, Hm avoids kirke (once it has the odd spelling kurke). It is also evident that scribes are particular about the spelling of "church": M has the SW form churche, W and F the Midland/London form chirche, and LHmR cherche, a Midland form found especially in eastern counties (
No other scribes, not even the careful L and R, are systematic in these respects. The editors of L point out that the scribe often writes <-e> as a marker of length in the preceding vowel, so that it no longer has a grammatical function (
The evidence suggests that L and R reproduced many of the spelling features of their exemplars, beta and alpha, with some fidelity, and that they may be closer to the spellings of Bx than other copies. It may also be (as suggested by R's treatment of heo and beta's avoidance) that alpha's language was more distinctively SW Midlands than beta's.
The B witnesses apart from F are unanimous
in their division of the poem into a prologue and 20 passus. Each passus has a heading, and though the precise wording of these headings varies from one manuscript to another, we find good evidence to believe that the marginal guidewords left by the scribe of L for the rubricator originate with Bx. From Passus 3 onwards the rubricator has abbreviated these instructions. The editors of L say: "Presumably, the guides record the form of the exemplar, edited later by the scribe when he went through the manuscript with red ink to add the actual headings" (
The headings are set out in Adams (1985), 216-31, and Benson and Blanchfield (1997), passim. Below we revise their
descriptions of the ten copies LMCr1WHmCGORF, in particular to include the guidewords still legible in M.
The heading to the Prologue in L is no longer legible, but Skeat read it as Incipit liber de Petro Plowman. F has Incipit pers þe plowman in a later hand.
R is lost, and most manuscripts including MW have no heading. Hence Bx is irrecoverable, and we follow
copy-text as read by Skeat for want of anything better.
1C. O omits petri plowman.
F has the rubric
1 lack
1C lack
quartus for iiijus), and probably beta. (HmCr1C lack
v for quintus) and W, and probably beta. L's rubric and MHmCr1C lack petri plowman.
Passus vjus), MW and R; evidently therefore Bx. (HmCr1C lack
Bx. (MHmCr1 lack
The variants in the petri plowman, except in Passus 6-7 where it is identical to beta.
Passus 8 and 9 cause difficulties. It seems clear that the heading Bx passus 8 is
both the end of the Visio and "the beginning of the search (
inquisicio see Hanna (2005),
243-304, esp. 246-7.1 lack
1 (which however does not include
1W. Hm has the same, but advancing by one as described above. L's rubric is
Bx. MCr1 add 1 continues in this style until Passus 14, but M drops this addition henceforth. As before, R adds
1R have their standard additions, and Hm continues as before.
1 has its standard additions, but R miscounts from this point, losing one. Hm continues as before.
1R have their standard additions, and C has the same as R (which continues to miscount). Hm continues as before.
&).1 and M's guide have the same with minor variations; M's rubric and O have merely
1, and M's guidewords. All these count this as the first Dobet passus, regarding Passus 15 as transitional between the Dowel and Dobet sections. However, HmG both refer to this as the second passus of Dobet. CO and M's rubric abbreviate to
&c after the passus number). L's rubric and MOC abbreviate to 1 adds de
visione; HmG continue as before, but from here R miscounts by two
(corrected by a later hand adding j):
&c after the passus number). What is legible of M's guide suggests it had the same, though the LMCO rubrics abbreviate to 1 adds de visione. HmG and R continue as before.
1 adds
1C drop
At the end of the text LMWGOCC2Y have: dialogus, see Hanna (2005), 243-304, esp. 246-7.Bx for
all that, since L also has in the left margin below the last line the
guidewords Bx, perhaps receiving his copy passus by passus, wrote the guide in expectation of a further section.C-text manuscript Laud Misc 656 (late 14th c.) which has as its "explicit": Explicit passus secundus de dobest incipit passus tercius. Seven other manuscripts of C also end with explicit passus secundus. See Adams, Bx, and also that L had some other source for his rubricated explicit which became that standard in beta copies. On our principle of following the guidewords in L, we adopt this as the first explicit, and follow that with the rubricated explicit.
All B witnesses have some indication of paragraphing. The archetype evidently had paraph signs reinforced by blank lines between paragraphs, as in LWYR, though in R the paraphs have often been left unrubricated. The same arrangement was planned for M, which has the blank lines, though the paraphs are not usually executed. Hm has paraphs but does not leave blank lines. O has "cc" marks at the start, but progressively fewer as the text proceeds. G has sporadic enlarged capitals. Cr has paragraph indents, frequent in Prol. (21 times, corresponding most often to W), but much fewer thereafter. Paragraphs vary in frequency from manuscript to manuscript; figures for Passus 5-7, assembled from Benson and Blanchfield (1997), 316, are: R 257, W 246, L 223, Hm 214, M 209, C 189, Y 184, F 171. From this it can be seen that R and its alpha partner F vary considerably, not so surprisingly in view of F's recasting of the passus structure. R and W have more paragraphs than other manuscripts, with W using them to emphasise the other aspects of his spacious display. L and M are close, both in total number and in position, suggesting, as in other respects, that their paragraphing is a reliable indication of beta's. What does not emerge from the figures, but is very striking when the positioning is examined, is that L's paragraphs are almost invariably confirmed by R's, though, since R has more of them, the reverse is not true. As a result, LR agreement is, as in other aspects of the text, a secure witness to the paragraphing of Bx.Bx. R less often has a paraph following a Latin line where beta has it. This might be explained by a scribe leaving a gap after the Latin line to allow for red boxing, as in W, so that a rubricator might assume a paraph was needed.
The only thing that can be said with certainty about the punctuation of Bx is that it marked the caesura, for all manuscripts of B have mid-line punctuation (G only sporadically). However, the marks used vary between a solidus (as F), a punctus elevatus (as O), and a raised point. R consistently has a mid-line punctus elevatus, perhaps representing alpha, and alpha perhaps also marked line-ends with a point, as R does; but most beta manuscripts have no regular punctuation at line-end. Crowley uses syntactic punctuation, usually a comma, but the manuscripts have it only rarely, and we have not often found it to be archetypal. For details of L's punctuation see V.2.3.
In all manuscripts except Cot the Latin lines, and sometimes individual Latin words, are distinguished in some way, with underlining in black or red, or enclosing in a box, and sometimes written red or in enlarged textura. In LMWR the lines are boxed in red ink and often written in display script, so that it is likely
these features were inherited from Bx.
Our survey of alliterative patterns in all 7217 lines of Bx (excluding those entirely or mainly in Latin) shows the expected great preponderance of the "normative" pattern aa/ax. We found this in 5796 lines, 80% of the whole. For the rest, the commonest patterns were as follows:
The frequencies for rare types are: aa/bb 43 lines, xa/aa 26, ax/aa 19, ax/xa 3, aaa/xa 2. There are also 26 lines in which we could see no alliterative pattern at all.
No two such surveys can be expected to arrive at quite the same results. Although the majority of Bx lines show clear and indisputable patterns of alliteration, a sizeable minority are open to more than one possible scansion. Bx.19.331crosse and croune. Schmidt's text follows Bx, but, because he allows what he calls "cognative" alliteration between /g/ and /k/, he scans the line aaa/ax: Schmidt (2008), 460. Like Schmidt, Kane-Donaldson reject xa/ax, as a pattern that Langland did not allow; but, because they do not accept Schmidt's cognative alliterations, they conjecture garland for croune, giving aa/ax (KD.19.321).Bx, issues concerning the poet's own metrical usage arise only occasionally, as one factor in choosing between manuscript alternatives; but, in order to determine the archetypal patterns, they have to decide what alliterations to recognise. The present editors have allowed those between vowels and /h/, /f/ and /v/, /s/ and /ʃ/, and /w/ and /hw/, but not that between /f/ and /θ/, accepted by both Kane and Donaldson, pp. 132-3, and Schmidt,
Issues of a different kind arise with those lines that have, or appear to have, three alliterating staves in their first half. We have recorded 153 lines with aaa/xx, having no alliteration after the mid-line pointing. This pattern (generally accepted as authorial) raises no problems: examples are Bx.15.383Bx.15.440Bx.16.7beth in this context, so we have registered the line as aa/ax.
Duggan (1987), 41-70, recapitulates the Metrical Rules that he had identified in b-verses of classical alliterative verse, and lists lines from Bx that do not accord to these patterns (and see Duggan [2009]). Rule 5 states that the b-verse requires between four and eight syllables and that one of the dips preceding either lift must be strong and one must be weak. A weak dip is of one syllable, a strong dip is of two or more. In contravention
to this rule are archetypal lines with two strong dips: e.g. Bx.P.56in mysbileue, as archetypal, and another example is discussed below, V.3.5.
Scribes using the forms of their own dialect can often alter the syllable-count of their exemplar, by dropping or supplying an -e(n) ending, adopting a syncopated form of a word for the full form or vice versa, and so on. It is often impossible to be sure of the forms of Bx, but agreement of the most reliable manuscripts, LMR, is a guide. From this it appears that Bx happily transmitted b-verses that contravene Metrical Rule 5. Many of the forms of W are unsupported by LMR and, whether or not they "improve" the metre, they are unlikely to be those of Bx; it seems clear that the highly professional London scribe of W has adopted the forms used by metropolitan scribes of Chaucer and Gower and by Hoccleve. It is these forms that we can observe the corrector inserting into M's original copying.
Our choice of L as copy-text is really inevitable. It was the basis for Skeat's edition, who was so impressed by the quality of its text and the nature of its corrections that he thought it might even be written by Langland himself, a view which he later retracted (Brewer [1996], 141 and note). Kane-Donaldson, followed by Schmidt, based themselves on W, while admitting that it had more errors than L (Kane and Donaldson, p. 214). They chose W because of its consistent spelling and systematic grammar (Kane and Donaldson, pp. 214-15), although they admitted that "the ideal basic manuscript or copy-text is the one which first provides the closest dialectal and chronological approximation to the poet's language, and then second, most accurately reflects his original in substantive readings" (Kane and Donaldson, p. 214). Later study suggests that the sophistication of W's hand is matched by the sophisticated accidentals of his text, which has been processed for a London audience by imposing standard London spellings and Chaucerian grammar (see
We believe that both L and W (and perhaps R as well) are London productions, but L is on a much more modest scale with none of W's consistently imposed spelling system and grammar. While it cannot be relied on to represent the forms of Bx, still less of Langland himself, it is clear that many forms significant for alliteration and metre are well preserved. Thus Langland uses the <g-> forms of "give, forgive" for alliteration and the SW Midlands <ȝ-> forms elsewhere; of the fifty occasions on which L has the <g> spelling, all but one alliterate, and that one (Bx.19.341gaf in all manuscripts and therefore must be archetypal. In contrast, Hm very frequently, and W quite often, use the <ȝ> forms in lines alliterating on /g/. Parallels with R, who also preserves many of the spellings of his exemplar, reinforce the impression that L accurately represents Bx in this respect. Though the spelling system of L is not an entirely consistent representative of a single dialect, it is probable that Bx was also inconsistent, and likely enough that Langland himself, as a London immigrant, wrote forms reflecting his new environs as well as his native dialect.
Copy-text is therefore the text of L as presented in our diplomatic edition (
We follow the guide headings of L, our copy-text, which we believe to offer the closest representation of the headings of beta. For reasons explained above IV.2, there is poorer evidence for the headings in alpha. On extra-textual grounds put forward in Burrow (2008), we believe that beta's headings are also those of the archetype, and that these were reduced by alpha.
Our documentary editions have reproduced as far as possible in print the layout and pointing of the manuscripts, and we have followed the practice here in so far as we can infer the format of Bx. We have found no information about its page-breaks, so they are not recorded. There is considerable agreement among the manuscripts as to the division into paragraphs and to the placing of paraphs (see Benson and Blanchfield (1997), 240-313), and the four important witnesses LMWR emphasise these divisions by a blank space between line-groups. We believe this arrangement to be archetypal, and indeed eyeskip from paraph to paraph to be a frequent explanation of major omissions in alpha and beta (see Burrow [2010]; Burrow and Turville-Petre [2012]), and so we have marked the line-groups with blank spaces and with paraphs. L's paragraphs are consistently matched in M, and so give good evidence of beta's scheme. More surprisingly, L's paragraphs are almost invariably matched by R's, and give us confidence that in these cases we are reproducing the arrangement in Bx (see earlier, IV.3.). Where L is not supported by R, there is often some identifiable reason for this, suggesting why one of the two scribes might have dropped or supplied a paragraph break. Where we have not followed the scheme in L, we have added a note to explain our reason for departing from it, and also noted where L is not supported by R.
All the early manuscripts of Langland's poem have mid-line pointing, whether it is a solidus, a raised point or a punctus elevatus. Modern editors, unlike Skeat, do not reproduce the mid-line pointing, and Kane and Donaldson, 138 n. 37, speak of "a subjective element in the reading of verse which may affect identification of the caesural pause." But the manuscripts largely agree in their placement of punctuation, offering evidence of Bx's usage which editors are wrong to neglect. At Bx.15.120prestes:
Riȝt so many prestes . prechoures and prelates
but Schmidt (2008), 425, says that prelates is to be "pronounced with two full stresses," evidently placing the caesura after prechoures to give the normative scansion aa/ax. Yet the pointing in the manuscripts provides evidence that the line should be scanned xa/aa, a pattern which both Kane and Donaldson and Schmidt regard as non-authorial.
Our copy text, L, begins with a punctus elevatus, perhaps because it was more formal, but switches to a raised point after P.25, perhaps because it was quicker. The L scribe occasionally uses a punctus elevatus thereafter in particular circumstances. Inserted punctuation is usually by punctus elevatus, presumably because it is clearer (e.g. Bx.12.287Bx, but since we suppose that this pointing was a formal device and part of the Bx; see notes to these lines (and Smith [2008]). L also has raised points at the ends of 24 English lines, but we have not recorded it. There is sporadic end-line pointing in some manuscripts (notably R), but the distribution does not
suggest derivation from Bx, so we have not recorded it. Latin lines in prose and verse are commonly punctuated, and we have retained the punctuation of copy text.
Beta manuscripts identify the seven sins in Passus 5 in the margin in the scribal hand, formal and boxed in LWHm, and we have presented them in our text. These marginal identifications are not in alpha, though in F the names of the sins are highlighted and underlined in red in the text. We also keep L's interlinear glosses to Bx.15.605
We have almost always retained the forms and spellings of L. So we do not generally alter L's forms of the personal pronouns. Bx uses hij for "they" in at least three lines where it alliterates (see above, IV.1). The L scribe, whose own form is þei, uses it on these occasions and in seven further lines, and we suppose L reproduces beta on all ten occasions. In
other lines the attestation suggests that beta uses the th- form. There is no evidence to suggest that alpha differs. In all cases, therefore, we take L's form of the third person nominative plural pronoun. For reasons explained above, IV.1, we also follow L's form of the nominative feminine pronoun, except on two occasions where
alliteration demands an h- form (Bx.18.156Bx.1.75heo is supported by MR and participates in the alliteration; in he which is not necessary for alliteration even though the line alliterates on /h/ + vowel, so we do not emend L's she; in quod he against quod she in all other manuscripts in a line alliterating on /t/, and again we take L's pronoun. It seems that alpha had the h- form for
"she" much more often than beta, and we can only determine when Bx used the h- form where it is necessary for the alliteration. Similarly in Bx.3.47a for "he," we have no reason for thinking that this reduced form was that in Bx. L several times has an
for "and," sometimes in lines that are marked for correction, as Bx.5.361Bx and perhaps also
representing Langland. In Bx.5.550soules (gen.) where other reliable manuscripts have the form soule (so MHmG + alpha, as in AC), but we have not emended L's form. If we were to make alterations in cases of this sort, it would be difficult to know where to stop, and we have recognised that we cannot recreate the spellings of Bx with any assurance. Where L is lost, we have adapted the spellings to L's usual forms, so that where beta has lost a passage extant in alpha, we have modified alpha's (usually R's) forms without notice.
Wherever the reading of Bx is in doubt, we
compare the readings of Ax and/or Cx if there is an equivalent line. Our
references to Ax and Cx are inevitably provisional, since neither has yet been established. We therefore rely on our own assessment of the evidence presented by Kane and Russell-Kane, and report in more detail where the evidence is divided. This applies most often to variation between the two great families of C, X and P. It follows that references to lines in Kane and Russell-Kane are to our reconstructions of archetypal readings, not to the edited texts.
Readings from Ax and Cx cannot, of course, be imported when evidence for Bx is supported by a consensus of the
better B manuscripts, in particular L, M and R. We recognise that other texts, especially Cr1GF, are at times contaminated from A and/or C, so that their witness can be discounted or at least regarded with suspicion if not confirmed by others. AC attestation becomes an important factor, however, where alpha and beta are divided. As a straightforward example: in the a-verse of Bx.5.569Ich haue myn huire wel, against alpha's Ich haue my huyre of hym wel, but AC have the line as in alpha, and therefore we conclude that beta has dropped of hym. The beta reading in Bx.5.185bi contenaunce ne bi riȝt makes poor sense: Repentance instructs Wrath not to reveal privities by contenance ne by speche as alpha has it, in a reading supported by C (the passage has no parallel in A). More uncertain is Bx.5.579a longe tyme þere-after, attestation that would usually secure Bx. However, there is conflicting support from AC for the reading of other manuscripts, a long tyme after. It is perhaps relevant that this b-verse, though presumably authorial, is unmetrical (x / x / x), so we suppose that either Bx or LR and beta2 supplied the necessary syllable. If the former, we should retain þere; if the latter we should emend L. In the face of such equal attestation on either side, we follow copy-text. At Bx.14.116it against he in other beta manuscripts and omission in F. Cx has he in a revised line, but LR agreement is usually decisive, and we follow it here.
On a number of occasions where the alpha/beta variation is paralleled by A/C variation, and where we think authorial revision within the B tradition is a strong possibility, we have presented both variants. For an example see Bx.1.46A or C, variation within alpha or beta, the inherent probability of coincidental agreement, and so on — is conducted in a note. After passus 10, where there is no A-text to compare, we have used the parallel text in C to help determine the reading of Bx. In a number of cases this may mean that we have designated an unrevised but authorial variant preserved in beta as an error, and we are thus presenting the revised text as in alpha and C, but there is no reason to suppose that revision was much more thorough in the second half of the poem than the very sporadic revision in the first half. We have, of course, discussed the possibilities in individual notes.
The passage from Bx.18.428C-text
is only lightly revised in the last two passus of the poem, we initially followed the principle that readings of F supported by Cx generally represented Bx,
and we adopted most of those readings against beta. Further considerations, however, caused us to rethink. Kane and Donaldson and Russell-Kane state categorically that C is unrevised in its passus 21-2:
These passus were untouched by the revising poet, and the manuscripts of the two traditions were treated as constituting two great families with an exclusive common ancestor, a single scribal B copy, and as differentiated only scribally. (RK, 118-19)
But it became apparent to us that the C-text has
indeed undergone a certain amount of revision (see notes to Bx.19.150Bx.20.371Cx is at times corrupt. Having found instances of F and Cx agreeing in error, we were forced to conclude that we could not unquestioningly regard F/Cx agreement as representing Bx. Coincidence might account for a number of such agreements, but some were of a character that suggested contamination. There are sporadic indications of F's contamination from a C-text elsewhere in the poem: a notable
instance is Bx.13.361-8Bx.19Pieres beta, cristis F/ Cx; conquest beta, his conquest F/ Cx; holy beta, onely F/ Cx; seke beta, tooken F and Cx?; þi bylyf beta, lyve by F / Cx.
There is no wholly satisfactory way of restoring Bx in these circumstances. Where beta readings and those of F/Cx are equally strong, or the differences are trivial, we have preferred beta on our usual grounds that L is copy text. For instances see the notes to Bx.19.39Bx.19.96Bx, though this is perhaps not a high price to pay. For a particularly difficult case see the note to Bx.19.493
It might be supposed that conjectural emendation had no place in an edition of the archetype, and was appropriate only in the reconstructed B-version. In theory, however, reconstruction of the archetype from its misrepresentation
in all extant witnesses must be a possibility. In practice there are very few occasions where we had the confidence to do this. One case is Bx.13.92wynked and alpha bad. There can be little doubt that the authorial reading was the rather rare and dialectal verb preynte, "winked in admonition" as conjectured by Kane and Donaldson. But it was
probably also the reading of Bx; how can the alpha and beta variation be accounted for otherwise? Beta substituted a common synonym for the movement of the eye, while alpha replaced with a word conveying the intention of the winking. There is a similar situation at Bx.11.190owre hele, while R has the nonsensical oure euel. We suppose that alpha and Bx had iuel, meaning "jewel," misread by R and replaced by beta. This is a little more problematical because F replaces it with helthe, but we see a Bx reading "jewel" as the most probable source of the variants. Again the reading was conjectured by Kane and Donaldson. In þe borre in þe throte, "the burr in the throat," i.e. "hoarse," as adopted by Kane and Donaldson from C. We suppose that, in order to avoid a rare word, alpha altered alliterating þe borre to cowȝe, "coughing," and beta to hors. In conjecturing þe borre we are aware of the danger of emending to B rather than Bx, but it provides an explanation for the variant readings of alpha and beta, neither of which is likely to have given rise to the other. In a quite exceptional case in Passus 19 we introduce a line from Cx that is not in any B manuscript: see the note to
Irregular alliteration is not sufficient evidence on its own to cast doubt on Bx (see IV.6 above), but it may be one factor in the choice of variants. See the variants in Bx.11.6Ax and Cx has been used to guide a choice between variants.
We have of course not corrected an unmetrical b-verse where the distribution of variants suggests that it was the reading of Bx. L's unmetrical b-verse in Bx.5.610or þow beest noȝte ysaued (x x / x x / x) could easily be "improved" by dropping the y- prefix, but it has good support for Bx from MR. Although other B
manuscripts drop the prefix, this presumably reflects the scribal dialects rather than metrical sensitivities. Though A manuscripts and the P family of C are without the prefix here,
the X family has it, as would be standard in their SW Midland dialect (LALME dot map 1195). Exceptionally, we
have emended L in Bx.10.115þei went to helle (x / x / x), since all other manuscripts have disyllabic wente(n).
Our aim is to restore the readings of Bx, not of the poem that Langland wrote. This means that we are bound at times to preserve a text that we know or suspect to be corrupt if it is what Bx wrote, and even, though it goes
against the grain for editors to do so, to preserve nonsense. In Bx.5.341arise þe southe
to be another example of that scribe copying exactly what he had before him. F typically rewrites, while beta makes a small adjustment to give superficial sense with arose bi þe southe. The evidently correct reading is that offered by C, aryse they bisouhte. We follow R's meaningless reading. In Bx.10.291For goddis worde wolde nouȝt be boste, then
corrected the last word to loste,
thus agreeing with all other manuscripts except one. In the usual
circumstances, we would suppose boste
to be a mere miswriting. However, R also has boste, without correction. We know that R is both accurate and
often mindless, copying whatever is in front of him,
however nonsensical; we therefore suppose that Bx had the reading boste,
easily corrected by all scribes except R to loste.
How else could we otherwise account for the agreement of the two most careful scribes? Since L is self-corrected, we record the reading as [b]oste. Fortunately, there are not many occasions on which we have to preserve self-evident nonsense.
For each line of the edited text, transcripts of the ten most important manuscripts are presented in parallel. Clicking on the line number invokes them below the line. The annotations discuss all variants we take to be significant for the establishment of the text.