Though recent work by Lister Matheson offers fascinating glimpses of William Langland's family background,Lister M. Matheson, paper delivered 3 May 1997 during the Medieval Congress at Kalamazoo. For what is known of the life of William Langland, also see Ralph Hanna III, , Authors of the Middle Ages, no. 3 (Aldershot and Brookfield, VT, 1993): pp. 1-24. we know almost nothing directly concerning the man who wrote these poems. Certain general resemblances to the fictional circumstances of Long Will, the dreamer-narrator of the poem, seem plausible (e.g., that Langland was married, that his formal education was interrupted, that he lived for a time in London by performing non-sacramental spiritual ministries), but the author's actual life cannot be reliably correlated to that of his character. The linguistic evidence in the surviving witnesses tends to corroborate Langland's identity as a Southwesterner, and many descriptive details in all three versions demonstrate that he must have lived for at least a while in London. We cannot prove more than that. Though The Piers Plowman Electronic Archive ultimately aims to determine and restore the authorial texts, we do not attempt at this level of the Archive systematically to distinguish the work of scribes from his.

At the second level R reflects the work of a scribe who produced the already defective manuscript copy from which all extant B manuscripts descend. Detailed reconstruction of the Bx-scribe's work lies outside the primary concerns of this edition of R. It will then appear paradoxical that we nevertheless cite in our textual notes the readings of Bx. When we refer, as we will numerous times in this edition, to the readings of Bx, those lections represent a preliminary working hypothesis about that text, constructed on the basis of our own careful collation of ten primary B manuscripts as well as from analysis of the variants provided in the Kane-Donaldson edition.Our first draft of Bx is based upon collation of manuscripts CCr1FGHmLMORW. Having already transcribed all of the B manuscripts, we have had many occasions to check the Athlone collations, and we have found them remarkably accurate. Moreover, preliminary collations by Adams and Hanna in the years before the Archive began clearly supported the essential correctness of the now conventional view that there are two major manuscript families in the B tradition, alpha and beta. For consideration of the two families, see George Kane and E. Talbot Donaldson, eds., , 2d ed. (London, 1988), pp. 57-59, 70-97; A. V. C. Schmidt, ed., , 2d ed. (London, 1995), pp. lx-lxv. Though some cruxes remain in those lines where R's relation to Bx is at issue, in general it is a simple matter to determine the Bx reading, at least in the light of present evidence. We expect that in a number of details our working hypotheses about Bx will require correction, and it is one advantage of the electronic text that it so readily permits that kind of adjustment.

Though our reconstructions of the readings of Bx must appear logically circular, a moment's thought should suggest that some such provisional reconstruction is inevitable and need not be logically vicious. In one sense, we are beginning the editorial project ab ovo, editing each manuscript witness afresh with the goal of working inductively to construct from the corpus of variant lections the intermediate sub-archetypes, archetypes, and eventually the critical texts of the three canonic versions. A critical question at once poses itself: if we already know the readings of Bx sufficiently to cite them for understanding the relation of this manuscript to others, what is the point of laboring to transcribe documentary editions or of doing either elaborate collations or complex thinking about the relationships among the manuscripts? The fact of the matter, however, is that we are not starting at the beginning, that we come to this task in the middle of a long-standing scholarly project and after more than a century of editorial work on these poems and these manuscripts. A great deal of careful and reliable work by a variety of scholars has been done, and a number of theories about the work exist which, in the light of presently available evidence, we take to be valid. Though we must reconsider the most basic issues in the light of new evidence, we necessarily use those theories we have found persuasive until such time as we have reason to think them wrong. We begin our editorial project with fundamental hypotheses about the author, the number of versions, the relations among the manuscripts, the governing features of the poet's metrical rules, as well as a number of assumptions about what it is that editors ought to do. We are aware that all of these are contestable, some of them hotly. Like scholars in other fields, we can only attempt to revise our hypotheses in the light of the data, as fresh data becomes available. That is, the process of editing a textual archive such as this will consist of a series of provisional passes through the evidence, and we anticipate that at least some of the hypotheses we have formulated now will require revision at a later date when we have more precise and full data to bring to bear on the reconstruction of Bx. Therefore, our citations of Bx are offered as provisional, and we do not devote annotations in this edition to the still-to-be-constructed text of the archetype. Though for the most part we use the forms of L (our copy text for Bx) to represent the readings of Bx, we cite the spellings of other manuscripts when L's forms are improbable.

At the third level R contains the work of a scribe whom we, following Schmidt, will call "alpha." As is the case with Bx, our ad hoc reconstructions in our notes of the text of alpha are also provisional. They are based upon our detailed comparison of R's lections with those of its sister manuscript F, a task made easier by the careful work of Kane and Donaldson in their edition of the B text. See the lists and discussion of the relationships between F and R in George Kane and E. Talbot Donaldson, eds., , 2d ed. (London, 1988), pp. 16-69. We offer here neither systematic comparison of F with R nor a detailed reconstruction of alpha—both must come at a later stage in the construction of the Archive—but we have incorporated into our apparatus numerous textual notes calling attention to readings in the text of R that are owed to the efforts of alpha.

It seems clear that neither R nor F was copied from the other. R's having copied from F is simply impossible. F's pervasive eccentricities of text are nowhere mirrored in R. Moreover, F omits several sizeable passages of text that are accurately preserved in R. On the other hand, that F used R as his exemplar is merely very improbable. Because the text of F suggests extensive conflation, it is at least technically possible that R was sporadically consulted by the F scribe (or one of his predecessors). Sean Taylor, in a 1996 article in , argues that stylistic similarities between an altered rubric in R (at the beginning of Passus 8) and the formal hand of the Corpus scribe, indicate that the latter did, in fact, copy directly from R (534-36, 544-45).Sean Taylor, "The F Scribe and the R Manuscript of Piers Plowman B," 77 (1996): 530-548. The hands are indeed strikingly similar (although the sample for comparison is miniscule), and the immediate F scribe may well, at some unknown time, have had R in his possession and gone to the trouble of altering (or "repairing" as he might have thought) the aforementioned rubric. But this is a far remove from Taylor's inference that R was his exemplar for copying F.

The weakest point in this hypothesis is that it overlooks a huge number of unique readings in R not reproduced in F, a few of which were noted long ago: e.g., at KD10.307, the careless omission of a crucial negative; in 11.390b, a flat half-line representing an extremely rare effort on R's part to "correct" his copy; at 13.165b-66, a lapse of attention that generates a meaningless phrase, "erl kynge," followed by loss of an essential direct object pronoun, "þee" (cf. Blackman 1918, 502, with Donaldson 1955, 187). Of course the three aforementioned errors appear in sections of B where beta omits text, so there is no external confirmation for the authenticity of F's reading. Here F may simply be showing superior emendational skills. However, dozens upon dozens of other readings, many involving quite subtle errors, occur in every passus of R. They often represent no serious deviation in meaning but simply a slightly less felicitous phrasal variation from the text witnessed by beta; and yet repeatedly in these cases F agrees with beta against R, or offers its own unique reading.A selection of the relevant evidence (81 individual variants for Prologue-Passus 6) can be seen in Appendix II of my article, "The R/F manuscripts of Piers Plowman B and the Pattern of Alpha / Beta Complementary Omissions: Implications for Critical Editing," in [vol. 14 (2001), 131-135]. As for F's supposed borrowings of distinctive errors from R, all the ones mentioned by Taylor are more plausibly explained in other ways, most as problems bequeathed to both copies by alpha.One allegation in particular, that F must have derived the detail of "foure" nails in the scene describing Christ's crucifixion from a private obsession of his own and then inflicted it as a marginal correction in R—after all, the other B and C manuscripts agree on "thre"— is quite wide of the mark (542-43). This topic was at the heart of a famous, albeit trivial, medieval theological discussion (Skeat 1886, 2, 251) known to many literate people in the fourteenth and fifteenth centuries. Moreover, the hand and ink of R's marginal correction here is indistinguishable from that used by R's editor in at least eight other cases. In seven of these instances of editorial correction, Taylor apparently has no interest in claiming intervention by the F scribe. This is not surprising since in five of these seven cases (9.34, 10.348, 11.82a-83, 11.245, and 11.369), F's reading of the line or verse in question is unparalleled by any other B witness while R always mirrors beta more closely and is usually identical with it. Especially damaging for the hypothesis of the F scribe as R's mysterious occasional corrector is the example at 10.348. Here an interlinear addition by the hand in question has supplied reuth, bringing the text into line with the version preserved in beta ("Ther riche men [R = man] no riȝt may cleyme but of ruþe and grace"). But F's line shows no sign of reuth at all ("For þere may no man ryght cleyme but of goddis grace"). In the only other instance of editorial correction where Taylor wishes to claim the presence of the F scribe (9.1, in which R's grotesque dowelleth has been repaired by a combination of erasure, striking, and the interlinear supply of dwellis), Taylor thinks the inflected form of the correction better fits F's dialect than R's); but we are not told why F, while correcting R in this way, would have written dwelliþ in his own copy.

The theory that F copied directly from R also fails to address the Norfolk relict layer apparent in F but not present in R, which would imply, at the least, that some ancestor of F, rather than F itself, may have consulted R. And yet it is clear from the paleographical arguments offered by Taylor that his claims are wedded to the notion that the Corpus Christi scribe himself used R to produce F. Likewise, it is hard to grasp why the F scribe would have bothered to "correct" R's Passus 8 rubric to reflect the more conventional four-part segmentation of the poem seen in copies like W at the same time that he was creating from scratch, for his own copy (or duplicating from an unknown conflational source), an entirely distinctive set of passus divisions that bears no resemblance to R's pattern or W's.Taylor tries to avoid this problem by arguing (544-45) that R's altered rubric was intended to be read "Incipit Passus octauus de visione petri plowhman / Dowel . Dobet . & Dobest" and that there was no aim to echo (nor even any awareness of) the typical format of the W family of beta manuscripts. Such an argument stretches credulity since in fact R's altered rubric actually assumes the W family's ordinatio and certainly reads "Passus octauus de visione petri plowhman / Incipit Dowel . Dobet . & Dobest."

Owing to R/F's shared preservation of some 170 lines not found in the beta tradition, and their shared omission of an approximately equal amount of text witnessed by beta, some have wondered—Skeat, characteristically, was the first (1869, xii)— whether the R/F shared text may actually represent a discrete version of , either midway between A and B, or midway between B and C. The best-known argument for the A —> R/F —> B hypothesis was articulated fifty years ago by E. T. Donaldson, who thought that some distinctive A-version features preserved exclusively in alpha revealed its relatively earlier date (a position abandoned by the time he and George Kane published the Athlone B version in the mid 1970s). Recently, the opposite case (A B —> R/F —> C theory) has been argued by Ralph Hanna (supported by Sean Taylor in a second article, in the ). Hanna has proposed that all of alpha's omissions with respect to beta can be accounted for as simple scribal oversights, but that some of beta's omissions with respect to alpha are not omissions at all. Instead, these passages are missing from beta because they are later authorial insertions of coherent thematic units, freshly written for alpha, which may have been a sort of "rolling revision" (1996, 215-29). Either way, R/F would remain relevant to editing the B version, but their evidence would necessarily be applied differently to the task of recovering B's original readings and might carry somewhat less weight than it customarily has. In Hanna's opinion, the practical difference would be minimal since well over 95% of the alpha and beta traditions manifest, with regard to each other, no evidence of revision or distinctive authorial intentions.

Endorsing such a thesis would entail seeing the alpha pair of manuscripts as a slightly different authorial text "state" of B, rather than a completely distinct "version." The latter term implies major shifts of theme and plot as well as numerous small touchups of phrasing, and it seems obvious that no such extensive alterations distinguish R/F from the other B manuscripts. At this point, conservative editors might see the situation as theoretically hopeless: that is, if R/F were to be regarded as embodying a distinctive, somewhat later (or earlier), authorial moment than 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), and most of the rest present few practical display problems for a hypertext edition such as the one now in progress at the Piers Plowman Electronic Archive.

Such finespun theorizing about the relationship between alpha and beta may seem labored, or even fruitless, but how one goes about editing depends heavily on one's conclusions concerning such matters. George Kane and E. T. Donaldson certainly realized this when they argued in the prolegomena to their edition, that the alpha and beta sub-archetypes are on precisely equal footing, and that all instances of text loss in both are products of ordinary carelessness (1988, 66-69). Having no hypertext format available to them (their first edition appeared in 1975), it is hard to see how they could have edited the B-version at all—other than as parallel alpha and beta texts— had they come to any other conclusion. And in many cases, especially where short passages of text are concerned, the evidence for their view (that these complementary omissions result from typical lapses of scribal attention) is compelling. Unfortunately, as both Hanna and Taylor have noticed, the Athlone claims for simple mechanical error do not always seem so strong. Occasionally, an entire verse paragraph or the heart of a thematic unit has been omitted for no reason easily assignable to the usual slips of eye or memory.

On the other hand, Hanna's counterclaim (that such mysterious gaps are restricted to beta) also looks problematic. Several of the main textual anomalies in alpha and beta seem easier to explain on other grounds than those of "rolling revision," and evidence from the relevant C-version variants is less supportive than such a theory would lead us to expect. Alpha as well as beta omits several sizeable chunks of lines for no plausibly mechanical reason (e.g., cf. alpha's loss of 15.533-69, 17.39-49, 17.115-26, and 17.221-47 with beta's loss of 12.116-25a, 13. 436-53, 14. 228-38, and 15.511-28). Donaldson himself recognized this problem in 1955, when he invoked the improbable theory that both traditions have unusual omissions in their latter parts because the same erratic scribe may have copied both sub-archetypes (1955, 186). It now seems likely that any hypothesis attempting to account for these phenomena by positing a small, relatively arbitrary set of authorial paragraph revisions between the production of the two sub-archetypes will tend to produce an easily reversible argument, one that highlights some aberrant features and suppresses others, exaggerating the differences between alpha and beta.

As for smaller discrepancies between alpha and beta—word and phrasal variations—Hanna is careful to distance his argument from any claims that individual variants and half-lines from alpha consistently represent a later stage of Bx (i.e., closer to Cx) than that attested by beta. The lectional data, like Janus, points in both directions simultaneously.

A single instance which seems to support "rolling revision" on this level is, however, worth examining as a prophylactic against hasty generalizations: at 6.218, beta's b-verse shows the variant fortune in the phrase "þat Fortune haþ apeired." This is also the variant attested by the A version. By contrast, R has falshed and F reads False. At this same place, Cx reads fals men (8.228). An eagerness to discover some clear pattern of versional evolution, combined with a judicious neglect of conflicting evidence, could easily lead one to infer from a handful of such examples that alpha here represents exactly the sort of intermediate stage between B and C that we have been looking for. However, a glance at the unanimously attested Bx line following this one ("Or any manere false men, fonde þow swiche to knowe")—a line completely missing from C— makes it obvious that the reading attested in alpha's b-verse resulted from scribal anticipation of a key phrase in the next verse and created an absurd redundancy. No one believes that Cx used alpha itself as his exemplar (passages missing from alpha regularly appear in C), and no one can believe that alpha's version of these two lines (or of this variant) is authorial. In fact, the Bx a-verse of 6.219 is also metrically corrupt (x a | a y), and therein lies the best clue to the entire puzzle. The common parent of Bx / Cx presumably botched the phrase corresponding to 6.219a (it would originally have read, as A, "Wiþ fuyr or wiþ false men") and then later tried correcting it with a marginal insertion, "false men." When the scribe of Bx came to this passage, he probably couldn't read the original a-verse at all, but he could read the correction (which might have seemed to him a gloss), so he improvised a new a-verse. In addition, he may well have reproduced the exemplar's "false men" as a marginal note, an extra inducement to alpha's subsequent error (easy enough to commit in its own right). Later, when Langland worked from the same document in creating C, he was no more successful than the Bx scribe at reconstructing his original verse at 6.219 and chose instead to delete the garbled line entirely and rewrite the previous one, substituting the "false men" of the marginal correction for "fortune." As on many other occasions, Langland's attempt at rescuing C from a passage from hopeless textual corruption involved him in adopting a hint from an earlier scribal error, but that is a far cry from any suggestion that falshed / Fals in alpha represents "authorial revision" of the text found in beta.

The following scenario offers what seems to us (at this admittedly hazy stage of our knowledge) a more convincing explanation: Bx's exemplar, also the source for Cx, was probably a scribal fair copy (already replete with small errors). It was then subjected to extensive, direct authorial revision, especially over the second half of the poem (giving that section some of the appearance of foul papers). It had blocks of recently composed or revised text attached at various points as extra leaves or scraps of membrane. A limited number of these authorial revisions appear to have passed down into Bx in the same form—as marginalia and attached slips of parchment. Some of these cases presumably occurred because the Bx scribe overlooked them until after he had already copied his base text, or because he was uncertain about their placement. However, some of these passages may have come into Bx as addenda because Langland actually wrote them shortly after Bx was originally copied. Having begun life, like its exemplar, as a scribal "fair" copy, although a rather mediocre one, Bx ended up being garbled by some such set of late authorial revisions added in ramshackle fashion. Hence, Bx was very confusing to follow when, in its turn, it came to be used as an exemplar by alpha and beta.

While another limited revision process, such as the one we have just postulated, may also have occurred between the production of the two sub-archetypes (as Hanna believes), the actual pattern of complementary omissions bequeathed by alpha and beta to the extant B manuscripts simply cannot be accounted for in terms of some hypothetical temporal sequence of those two documents. (Although one of the two sub-archetypes must have been copied before the other, their order of production looks irrecoverable.) Nor is this pattern of lost text the product of merely mechanical oversights. Instead, some of these differences may result from scribal censorship of sensitive topics, but most of the larger omissions probably reflect the relative luck of the alpha and beta scribes in locating Langland's marginalia and addenda slips and tracking them to the correct points for insertion.

Whatever account of production history might best explain the subtly divergent textual states in which B survives (as alpha and beta), one fact remains staring us in the face, just as it stared at Skeat more than a hundred years ago: the two surviving manuscripts that most faithfully reflect Langland's original are R and L (Oxford Bodleian Laud Misc. 581). And although they derive from different textual traditions (R from alpha, L from beta), these two witnesses closely resemble each other in all sorts of surprising (albeit superficial) ways, including many details of layout as well as uncommon morphological and orthographic features. More importantly, in an extraordinary number of cases, manuscripts R and L will agree on a good variant (or even share an error of omission) against all, or nearly all, of the other copiesa fact that long ago misled Elsie Blackman into hypothesizing a sibling relationship for them.

What now seems apparent, however, is that when two unusually good copies have descended, as these have, from separate branches of a textual tradition as complex as that of and yet display such strong affinities with each otherthey likely must be positioned very near the original transmission point. Ralph Hanna has offered his opinion that both R and L date from the late fourteenth century, not the early fifteenth (as Kane believed); moreover, some six years ago Hanna noted that these two copies share a sufficient number of distinctive layout features with at least three other B manuscripts (manuscripts W, M, and Y) to suggest that they all may have been produced by a single loosely organized group working in London during the 1390s and early 1400s. Of the lesser three witnesses in this hypothetical production group, the most erratic is manuscript Y and the best is manuscript M; but a recent study by Thorlac Turville-Petre has demonstrated that M was extensively "corrected" after its completion, into forced agreement with the textual tradition of manuscript W (a good London-normalized copy whose readings are, nevertheless, often inferior to those witnessed by R and L).

Where has this gradual accumulation of evidence led us as editors? To the realization that it would be impossibly timid to hide behind an affected agnosticism about Langland's own linguistic forms (in the absence of a holograph) and thus embrace as our critical base a witness like manuscript W (used as copytext by both Schmidt and Kane-Donaldson). The choice we have made, instead, is to return to Skeat's usage of manuscript L as copytext, corrected and supplemented, where necessary, by readings from any other relevant manuscript (including those of other versions), but especially and continuously checked for authenticity against R. Hence the importance of R to the critical editing of the B-version of is that, more than any other surviving evidence, it validates the authority and primacy of L. The unusual word forms and inflections that it shares uniquely with L (some of clear Southwestern origin) cannot all be Langland's; some must have been original to the scribe of Bx. But manuscripts R and L, with only two documents probably separating them from a holograph, inevitably reflect better than any other B copies the mixed usage and dialect of William Langland himself.

Vowel length of <e> and <o> is occasionally marked by doubling in closed syllables, but on the whole spellings with doubled vowels are recessive. <e> is distinctly more common for /e:/ and /ε:/ than is <ee>. An exception to this is the word pees, where it would appear that the doubled vowel has become to this scribe a spelling convention, rather than a productive marker of length. This is true as well for word-final open syllables <a> is not doubled to indicate vowel length. The anomalous and phonologically suspect form haad 3.332 for had/ hadde occurs once; the (Biblical) loan proper noun Isaac 16.243 is the only other instance of <aa>. ME <ii> occurs only as word-final <ij> in hij "they" (3x). Final <-e> and <-es> are more usual signs of length. <ee> is not infrequently found in word-final open syllables, especially for personal pronouns, <oo> much less commonly so. The digraph <oe>, which elsewhere in ME often indicates the [ö] sound, is regularly used in R to represent the long vowel /o:/.

For /a:/:<a>

made (87x), mad "made" (3x); take; tale(s).ME word-final <-a> is extremely rare, for the long /a:/ will by this time have regularly shifted to /ɔ:/; na marks an exception to this general rule, but it is relatively scarce (18x) [cf. no (265x)] and occurs only in the collocation na more, which suggests that it is in fact a fossilized idiom serving to dissimilate; the collocation no more does not occur at all. Not even in open syllables [e.g. fareth (11x), taketh (24x) saued (27x) fader (26x)] does <a> necessarily mark /a:/; e.g., catel (25x), water (18x), maner(e) (39x).

For /e:/ and /ε:/:<e> ~ <ee>

bred(e) "bread" (27x); deth; erth(e) (34x) ~ eerth(e) (5x); ete (15x) ~ eet (6x); feet (4x) ~ fete (3x); hed(e)(s) "head(s)"; hede "heed"; pees (27x) ~ pes (8x); preste(s) (34x) ~ preest (1x); sen (3x) ~ seen (3x).<e> and <ee> to mark /e:/ and /ε:/ show similar distribution in word-final open syllables. <ee> is most common for the verb "see" (as well as the noun "sea"), and the personal pronouns: se "see" (57x) ~ see (7x) (= "see" (4x) + "sea" (3x)); ye (5x), ȝe (166x) ~ ȝee (9x); he (805x) ~ hee (3x); me (330x) ~ mee (3x), etc. Forms of the verb "be" are never written with <ee>: e.g. be (436x), ben (135x), beth (30x), best (1x).

For /o:/ and /ɔ:/:<o> ~ <oe>See the discussion of dialect above and III.2.1.1.1 below. ~ <oo>Word-final <oo> is extremely rare: foo (1x), pl. foes (2x); goo (5x) ~ go (49x); loo (1x) ~ lo (16x); þoo (4x) ~ þo (121x), tho (1x). Forms of "do," like those of "be," never show the doubled vowel: do (87x), doen (2x), don (22x), dos (1x), dost (2x), doth (44x).

blod(e) (18x) ~ bloed R9.144; bok(e)(s) (38x) ~ book(e)(s) (6x) ~ boek (6x); don(e); fode "food"; fote (5x) ~ foot (2x) ~ foet R17.45; gode (58x) ~ goed(e) (51x) ~ good(e) (11x); gost(e).

1. OE, ON /a:/:<o> ~ <oo> ~ <oe> ~ <a>

abrode; foo (1x), foes (2x); fro (34x) ~ froo (1x); hole "whole"; holy (98x) ~ holi (2x) ~ haly (2x); hote "hot"; lore; no (265x) ~ na (18x);The stressed form no (265x) is overwhelmingly dominant, but stressed na (18x) survives with the unchanged OE spelling in the collocation na more, where it appears to the complete exclusion of no. On the other hand, mo and mor(e) completely replace the comparative ma, suggesting that vocalic dissimilation may account for the preservation of this a: > a/o hybrid. non (57x) ~ none (22x) ~ noen (3x); roper(e); sore; stone(s); wrot(e).

2. OE, ON /a:/ + w:<ow> ~ <ou> ~ (<oo>) ~ (<o>)

blow(e); know(e)(n) (55x); low(e) (21x), lowh (1x) ~ loo (1x);The <ow> in lowe does not come from a native /a:/ + w. The /w/ sound in this case is a reflex of the voiced velar fricative following the long /a:/ in ON lagr "low," just as OE agan vb., agen adj. > ME ow(e)n. soule(s) (104x) ~ sole(1x).

3. OE, ON /a/:<a>

caste (8x) ~ cast (6x); hap (1x), happ(e)(s) (2x) sb. "luck"; happe (3x) vb. "happen"; lappe.

4. OE, ON /a/ + nasal:<a> ~ (<o>)

fram (44x) ~ from (4x); can(st); game; man; ran; wan "won"; name(s); s(c)hame.

5. OE, ON /a/ + lengthening consonant group:<a> ~ <o>

bonde "bound"; cold; hand(e)(s) (20x) ~ honde(s) (15x); hange(n) (12x), hangeth (3x), hanged (2x) ~ honged (1x); lond(e)(s) (37x) ~ land(e) (2x); longe (53x) ~ lange (1x); lombe (2x) ~ lamb (1x); stonde(n) (9x), stondeth (3x) ~ stande (5x).

6. OE, ON /a/ + <-nk>:<a> ~ <o>

banke(s); dronke;These three occurrences are 3rd sing. preterite, where OE /a/ > ME <o>. The OE plural has /u/. sanke; stanke; thanked (1x) ~ thonked (1x).

7. OE, ON /o:/:<o> ~ <oo> ~ <oe>

blod(e) (17x) ~ bloed (1x); bok(e)(s) (38x) ~ book(e)(s) (6x) ~ boek (6x); brother(e).

dom(e)(s); flod(e) (5x) ~ floed (5x); fode (14x); fote (5x) ~ foot (2x) ~ foet (1x); god(e) (62x)12.203 god friday and 12.274 god is so god are the only two occurrences of god "good." ~ goed(e) (51x) ~ good(e) (11x); loke; rote(s); schope (3x);R also has the weak preterite (for-)schupte (3x) < OE scieppan with <u>, but it is not possible to say whether the historical stem vowel /o:/ of the class VI strong preterite underlies the analogical weak form. toles; toþaches.

8. OE, ON, OF /o/:<o> ~ (<u>) ~ (<a>)

box; cros(se); folk(e); god; hulpe(n); mosse; pecok; sonner R10.495 ~ sannore R12.173 "sooner";Sannore is the comparative of OE sōna, and as such has experienced a shortening of the tonic vowel. spottes; wedlok(e).

9. OE, ON /o/ + lengthening consonant group:<o> ~ (<oo>)

bold(e); borde; gold(e) (14x) ~ goolde (1x); molde; word(e).

10. OE, ON /u:/:<ou> ~ <ow> ~ (<o>) ~ (<ouȝ>)

aboute; adoun(e) (10x) ~ adowne (2x); cloudes (1x) ~ clowde (1x); doun(e); hous~ (12x) ~ hows~ (4x) ~ hosewif R14.3; how; mous; now (x) ~ nouȝ R17.254; proud(e); þow (200x+) ~ þou (5x).

11. OE, ON, OF /u/:<o> ~ <u>The semi-vocalic glide consonant <w> not infrequently represents the combination of consonant + rounded (high) back vowel: /wu/ (or /wo/). Thus for example in wlueliche 15.131, a derivation from OE wulf, the historical short /u/ is rendered by the consonant graph alone. The scribe of R frequently uses <w> to represent Latin <vu> as well: wlt, wltis for vult, vultis. ~ (<i>)

by-swonke; buttere; kunnen (5x), cunne R20.315 ~ conne (3x); dronke(n);Both the indicative past plural and the past participle, whether verbal or adjectival in function, are included here. flix R5.182;The spelling flix for "flux" < L. fluxus by way of OF is attested elsewhere in ME sources, but does not in this manuscript represent a regular development of short /u/. ful; pulle; sone "son"; sonne "sun";The <n> vs. <nn> orthographic distinction to mark the two words "son" and "sun" respectively is scrupulously maintained in R. þoruȝ (88x), thoruȝ (31x), þorȝ (13x), thorȝ (10x) ~ þurȝ (3x), thurȝ (1x); woke (3x) "week";According to Campbell §218 the word "week," deriving from Primitive OE wicu, underwent "combinative back umlaut" in WS but not in Angl. This means that the back vowel /o/ or /u/ in the syllable following the affected vowel, working together with the back semivowel /w/ preceding the affected vowel, caused u-umlaut even in the environment of a back consonant /k/, which normally prevented back umlaut in both WS and Angl. See Campbell §210. A ME realization of the word with a front vowel is thus of Angl. origin from wicu, whereas a spelling with <o> or <u> is WS from wucu. wolle "wool"; (I-)wonne.

12. OE, ON, OF /u/ + lengthening consonant group:<ou> ~ <ow> ~ <o> ~ <u>

doumbe; grounde (14x) ~ grownde (1x); hound(es) (4x) ~ hownde (1x); morned(e) (3x) ~ mourned (2x) ~ murned (1x); mornynge (1x) ~ murny[n]ge (1x); turne (11x) ~ torne (2x); wonden "wound";This is the past plural and participle of "wind." wounde(s) n. & vb. "wound, injury/injure."

13. OE, ON /y:/:<uy> ~ <ui> ~ <u> ~ <i> ~ <y> ~ (<ue>) ~ (<ee>) ~ (<e>)

fuir(e) (3x) ~ feer (2x) ~ fuyr (1x) ~ fuer (1x); fust(e) (11x) ~ fist (1x); huyre (7x) ~ huire (4x) ~ hire (2x); ken 6.142 "kine"; lys 5.200; mys P.22 ~ muys P.71; pruyd(e) (13x) ~ pruid(e) (10x) ~ pride (3x); whi (20x), whies pl. 12.221 ~ why (2x); wisched(e) (3x), wischedun (1x), wischen (1x) ~ wysche 5.113.

14. OE, ON /y/:<u> ~ <i> ~ <y> ~ <e>

bugge(n) (9x) ~ a-byggen 2.89; brugge(s); dede (28x) ~ dide 9.85; fulfille(d) (7x), fille v. 5.346 ~ fulled 15.369; fille n.; furst(e) (43x) ~ first(e) (4x) ~ ferst (2x); gult(es) (8x) ~ gyltes (1x) ~ giltes (1x) "guilt, sin"; gulte (3x) ~ gilte (1x) "gilded"; hudden; hilles (3x) ~ hulles (2x); kyn(ne); left adj. "left"; lofte (6x) ~ lift (1x) "air, heaven"; muche (61x) ~ miche (1x) ~ myche (3x); muchel (3x) ~ michel (1x); merie (5x), mery (1x) ~ murie (3x) ~ myrie (2x) ~ murgur (1x) 1.108 "more merry, merrier"; synne; which(e).

15. OE, ON /y/ + lengthening consonant group:<e> ~ <y>~ (<uy>) ~ (<u>)

buyrde (1x) ~ burde (1x) ~ berde (1x);The a-verse of 9.186 has Ne schulde no berde a-bedde be. Following the A and C texts, K-D emend here to bedbourde. B-text witnesses agree overwhelmingly with one another in reading here bourde o(n) bed(de) or some slight variation of this. Although this reading may not be authorial, it seems clear that a strong scribal consensus understood the sense to call for the word meaning "girl" or "young woman." The <e> representing the reflex of OE /y/ + lengthening consonant group in R is thus unlikely to be a confused or unreliable spelling of a misunderstood word, for it is a usual sign for the unrounded vowel. kende (88x) ~ kynde (9x); mynde (9x) ~ mende (3x).

16. OE, ON /i:/:<i> ~ <y>

blithe (1x) ~ blythe (1x); chid(d)e(n) (8x), chidyng (1x) ~ chyde (2x), chydynge (1x); knyf (2x), knyues (1x) ~ kniues (1x); lif (+ compounds) (111x), liue(s) (2x) ~ lyf (+ compounds) (20x) ~ lyue(s) (14x); ride(n) (13x) ~ ryden (2x); tyme(s) (100+x) ~ time (3x); wide (5x) ~ wyde (3x); wis(e) (24x) ~ wyse (7x); wyn(e) "wine."

17. OE, ON /i/:<i> ~ <y> ~ (<o>)

bitter(e) (7x) ~ bytter 5.121; lyue(n) (37x) ~ liue 12.37; nyme; whider (3x) ~ whyderward 5.314 ~ whoder 16.12; wydewe(s) (9x) ~ widewe(s) (5x).

18. OE, ON /i/ + lengthening consonant group:<i> ~ <y>

bynde(n); blynd(e) (11x) ~ blinde (5x); child(e) (8x), childrenChildren is in fact only one of eight or nine forms, depending on how one normalizes expansions, orthographic/morphological realizations of the plural of "child" in R. Singular and plural taken together, there are in all 32 forms with the stem vowel <i> against the single instance with <y>. (24x) ~ chyldren 3.263; fynde; mylde (4x) ~ milde 10.159; wiȝt(e) (21x), wiȝth (1x) ~ wyȝt(e) (2x), wyȝth (2x); wild(e); wynd(e)(s) (12x) ~ winde 8.27.

19. OE, ON, OF /e:/:<e> ~ <ee> ~ <ey>In R <ey> appears more often than not as a spelling of French <ei>. As such it indicates length, though it is probably nearly always a diphthong. In the case of a plural where there is an open tonic vowel at the end of the singular form of the word, it clearly indicates syllable juncture. This may indeed be the more important function.The samples here show orthographic representations of the ME long vowel. In most environments where a long vowel was followed by two or more consonants in OE, even if these were a combination that caused lengthening, /e:/ will have become short in late ME. Cf. for example the infinitive fede(n) "feed" with the preterite fedde(n) "fed."

beches; brede(n); contre, contreyes pl.; crede; deme(n); fede(n); feet (4x) ~ fete (3x); grene; hede "heed"; kene; kepe(n); mede; seke; spede; swete.

20. OE, ON, OF /e/:<e> ~ <a>

dowel; federes; geste(s); peny (7x) sg. ~ pans (9x) pl.;The sg. of pans in R is peny (7x). rek(e)n(e); sarmon(s) (3x); web; wed(de); wel; wre(c)ched.

21. OE, ON, OF /e/ + lengthening consonant group:<e> ~ (<ee>)

best(e)(s); blende(th); elde (18x) ~ eelde 12.8; fest(e)(s);Feste "feast" is a borrowing from OF. While <st> is not regularly in OE a consonant cluster that led to vowel lengthening, when <st> occurred between syllables, the syllable division juncture tended not to separate them. Thus the entire cluster came to be perceived as syllable-initial and the preceding syllable became open, occasioning the lengthening of the short vowel. See Joseph Wright, (Oxford: Oxford U P, 1928), §97, and Richard Jordan, , trans. Eugene J. Crook (The Hague: Mouton, 1974), §§220, 225. feld(e); selde(n).

22. OE /æ:/ (1) & (2):<e> ~ <a> ~ (<ee>) ~ (<o>) ~ (<oe>)

breth; clene; dele(n); drede;Although the reflexes of OE class VII verbs generally show considerable vacillation between <a> and <e> in the preterite, R has only <e> for the present indicative, infinitive and imperative as well as for the substantive, while preterite verb forms are spelled exclusively with <a>. any (72x), eny (18x) ~ ani (1x); ar (43x) ~ er(e) (20x) ~ or 14.148 ~ oer 1.131; euen(e); euer(e); hele; lasse (18x) ~ lesse (5x); leste (18x)Lest spells "lest" (< OE [ðȳ] lǣs ðe) 9x, "least" (< OE lǣst) 3x, and "list" (< OE lystan) 2x. Leste spells "least" 6x but never spells "lest." It also spells "list" 2x. Of these only the reflexes of OE /æ:/ are included in the statistics recorded here. ~ last(e)(th) (10x);The recorded occurrences of last, laste and lasteth include various present tense or infinitive verbal forms (< OE lǣstan "last") together with reflexes of "least" (< OE lǣst) (6x). Not recorded here, moreover, is one instance of a past participle last, in which the assimilation of the <-st> cluster of the stem with a weak tense marker <-t> would likely already in OE have produced a short /æ/, of which this <a> is a reflex. Most common is the assimilated superlative form last(e) (< latest; pos. late) (21x), where <a> is not a reflex of OE /æ:/. Compare n.26 above. Note that although "least" may be spelled either with <a> or <e>, in R the verb forms are never spelled with <e> and "lest" is never spelled with <a>. This latter distribution not only conforms to modern English usage, but also suggests that the /æ/ vowel which characterizes the verb today persisted throughout the ME period. lat(e) (37x) ~ let(e)(n) (30x), letun P.54, lett 4.163; reden (3x) ~ reeden 11.100; see (4x) ~ se 11.378 "sea"; seed (6x) ~ sede 5.563; slepe;Both the noun (7x) and present forms of the verb (OE VII) (9x) are reflexes of OE /æ:/. The strong preterite slep 5.563, slepe 5.373 is a reflex of OE /e:/. swete "sweat" (5x) ~ by-swatte 13.419;Swete represents the infinitive and present tense of the verb (3x) as well as the noun (2x). The vowel of the preterite will likewise have been a reflex of OE /æ:/, though the double consonant caused shortening in late OE or early ME. According to Joseph Wright, (Oxford: Oxford U P, 1928), §89, the vowel in a word which had undergone this shortening early is /a/ (< /æ/ < OE /æ:/; see also Joseph Wright, (Oxford: Oxford U P, 1928), §43). In the language of R, however, <a> can be a regular reflex either of the shortened or the long OE /æ:/. teche(n); weet 14.47.

23. OE /æ/:<a> ~ <e> ~ (<ee>)

apple(s); bak(e), pl. backes "back"; blak; faire (39x) ~ fare (1) 5.485 This form is unique, and Kane-Donaldson treat it as a possibly substantive variant, but it is likelier to be merely an unusual spelling variation for standard faire found in the other B witnesses, as well as in Ax. Cf. , s. v. fair (adj.), where this spelling is documented, sporadically, for c. 13-15.; feelde 3.119 vb. pret. "felled"; had(de); masse(s) (15x) ~ (miȝhel)messe (3x); wasche(n) (6x) ~ wesche(n) (3x); water; what.

24. OE /éa/:<e> ~ (<a>) ~ (<ee>) ~ (<u>) ~ (<eu>)

bem; brede; chepe(d) (3x), chepyng(e) (2x) ~ chapmen (4x); ded(e) "dead, death"; def; deuh 5.636 "dew"; dreme; ere; grete; hed(e)(s), heued(es); hepe; lef(e) (5x) ~ leef 5.206; lepe (6x) ~ luppe 11.333.The infinitive luppe is recorded by the on-line MED as "early S."

25. OE /éa/ + voiceless velar fricative:While the range of graphic variation is not particularly remarkable, all the variants here indicate that the reflex of OE /éa/ + /x/, unlike that of simple /éa/ (see §24), is a diphthong.<ey> ~ <eyȝ> ~ <eiȝ> ~ <eȝ> ~ (<eyh>) ~ (<ei>)

hey(e) (12x) ~ heyȝ(e) (8x) ~ heiȝ (5x) ~ heȝ(e) (3x); heie 7.17;Although this might appear to be a vocalic trigraph, a preponderance of manuscripts reads a definite article þe preceding the adjective, which suggests that the R scribe may have inadvertently omitted the article but nonetheless copied the weak adjectival ending. Another possibility is that the scribe, misanalyzing singular dayes "dais" for the plural "days," regarded the weak <-e> ending as apt even where the article was lacking. neyȝ (4x), neyȝ(e)bore(s) (9x) ~ neiȝ (2x) ~ neȝ 20.174, neȝbore 13.380 ~ neyh 14.124; þeiȝ (6x) ~ þeyȝ (2x).

26. OE /ea/:<a> ~ <e> ~ <o>

al(l)(e); barn(e)(s); calf; fall(e)(n); flex 6.13; half(e); holde(n) (9x) ~ halde(n) (3x) ~ helde 4.21;These examples are all infinitives. Breaking, or the development of the /ea/ diphthong before an lC consonant cluster, does not normally occur in Anglian. One would thus regularly expect halden either if the vowel remains short or otherwise does not undergo raising and rounding - that is, does not become a monophthong until after the early /a:/ raising is completed. Holden, where the vowel is lengthened when followed by the lC cluster and then undergoes raising, is the regular form we have in PDE. The helde form is possibly Southern. salt; schafte; wex~ (21x) ~ wax~ (3x) ~ wox~ (2x).The verb waxen has a complicated phonology; the distribution of forms in R is likewise extremely complex. A comparison with its Gmc cognates makes it evident that it once belonged to the class VI strong verbs (a - ō - ō - a), but by the time of historical OE it generally conforms to the class VII gradation pattern (ea - éo - éo - ea), which naturally affects its development in ME. See Randolph Quirk and C. L. Wrenn. , 2nd ed. (London, 1957), §83(a), and Karl Brunner, , trans. G.K.W. Johnston (Cambridge, MA, 1965. §69, note 21. In some ME witnesses, moreover, at least some stem forms analogically follow the pattern of strong class IV verbs (e - a - o - o). Cf Brunner §69, note 21 and Joseph Wright, (Oxford, 1928), §412. The development of ME wex/wax as a class VII verb varies according to dialect because of differences in the monophthongization of OE diphthongs: wex < WS e + h < OE /ea/, but wax < Anglian æ + h < OE /ea/. See Joseph Wright, (Oxford, 1928), §28. Forms which fall under 2) and 3) below do not derive from the /ea/ diphthong. They do, however, provide evidence of the analogical shifting between strong verb classes, which in turn may serve in part to explain the forms found under 1) and 4) in cases where the phonological development seems otherwise anomalous for R.

A combination of Southern (West Saxon) and Midland (Anglian) variants of both classes VII and IV allows for the following possibilities: 1) present indic., infinitive: <e> = S VII, IV; <a> WM VII 2) pret. sing.: <e> S, WM VII; <a> S, WM IV 3) pret. pl.: <e> S, WM VII; <o> S, WM IV 4) participle: <e> S VII; <a> WM VII; <o> S, WM IV

In addition to the noun wex (2x) "wax," R has the following distribution of verbal forms: 1) wex(e) infin. (3x), wexeth pres. (7x) ~ wax infin. 4.176, waxeth pres. 13.364 2) wex(e) pret. sg. (8x) ~ wax pret. sg. 16.224 3) wexen pret pl. (2x) ~ woxen pret. pl. 16.58 4) wexed participle 5.357 ~ woxen participle 10.79

27. OE /éo/:<e> ~ <ee> ~ <eo> ~ (<u>)16.110 reads if any perel fulle. The sense is that of "befall" in the subjunctive, where the stem vowel of the strong verb corresponds to that of its preterite plural indicative. The <u> is thus a reflex of OE <ēo>. ~ (<eu>)

crepe; depe; felle(n); fende(s) (25x) ~ feend 20.38; frende; heo (25x) ~ he (12x) "she"; lede(s) (9x) ~ leedes 4.186 "man"; lef (4x) ~ leef (4x); lem; leep(e) (2x) ~ lepe 5.512 ~ leup 2.29;These forms are reflexes of the preterite hléop of the OE VIIa hléapan. The on-line records leup as "early SW." swerde; thef.

28. OE /eo/: <e> ~ <o> ~ <u> ~ <ee> ~ (<ui>) ~ (<uy>) ~ (<a>)

buyrn (2x) ~ buirn 16.188 ~ barnes 12.73 Kane and Donaldson, following W, have the correct reading burnes "men." Manuscripts CrHmYOC2CBR read barnes, normally "children," but the records barn as a late spelling for burn, in part owing to the conflation of the two words in cases where either term satisfies the general sense "human," in part owing to the adoption of AN baro(u)n (<OF <Frankish ber, cognate with the OE beorn), with a similar meaning, into ME. Although most of the manuscripts cited will have been following an exemplar with <a>, this variant likely came into existence in the first place as an error that needed no correction, producing adequate sense as it does, for the reasons given here. "man"; cherle; erl(es) (7x) ~ eerldam 2.44 "earldom"; erth(e) (34x) ~ eerth(e) (5x); fele "many"; fer(re) "far"; hert(e); heuen(e); lerne; -selue(n) (123x) ~ -sulue(n) (18x); siluer (25x) ~ seluer (10x) ~ suluer (4x); ster(r)e(s); sterueth; werk~ (72x),Of the 72 instances of the <werk-> spelling, sixteen are found in variants of the compounds "workman," "workmen," or "workmanship." No other spelling of the element "work" is used in these compounds. werch(e)~ (38x) ~ worche~ (8x) ~ wurche~ (4x); world(e)(s) (39x) ~ werld(e)(s) (5x).

29. OE /éo/ or /eo/ + /w/:<ew> ~ <eu> ~ <e> ~ (<euȝ>) ~ (<ee>) ~ (<ou>) ~ (<ow>) ~ (<we>) ~ (<u>)

blew; brewe; gleman (3x) ~ glwemannes 5.359; greu3 11.381; knowes 5.365; rewe; tre(s) (13x) ~ trees 11.376; treuth(e) (63x) ~ trewth(e) (34x) ~ trouthe 10.48 ~ truthe 5.619.

30. OF /ue/:<oe> ~ <e> ~ (<eu>) ~ (<o>)

doel (3x) ~ deul 6.122 ~ delfol 15.562 "doleful"; mebles; meue; poeple (70x) ~ peple (9x) ~ pople 15.92.

It is not always possible to be certain precisely when, or if, a borrowing from a Romance language has adopted the Germanic "left-handed" stress. See Lass §2.6 for a description of stress shift. The examples provided here all occur in lines where the alliteration provides evidence that the primary stress falls on the initial syllable. In such words the spelling of the vowel of the syllable with secondary stress, usually a digraph, may thus indicate a non-central quality. In native compounds the second element very often retains secondary stress, in which case the retention of non-centralized vowel quality is to be expected: e.g. -full, -lich, etc.

barefoet 18.11; bethleem, bischopes 15.550; citee P.35 ([s]-alliteration); contee 2.47 ([k]-alliteration); monee 1.46; patroen 12.231; pharaoes 7.181 ([f]-alliteration).

III.2.2.1. OE /hw/:<wh> ~ (<w>) ~ (<h>)

whan, when; what; wher(e); which(e) (39x) ~ wich(e) (2x); while; white; whete; whistlyng; ho(o)-so (45x) ~ who-so (8x), who (18x).

The OE <hw> is very consistently spelled <wh>, only rarely <w>. There seem, moreover, to be only two single word-initial instances of the reverse spelling of <wh> for /w/. One at R13.337 is orthographically suspect and suggests possible scribal confusion: 13.337 whitus weyus "wit's men." The spelling whithliche (16.286) for wiȝtliche occurs at this point uniquely in R. A word-final <wh> for /w/ occurs in 20.11 lowh herted. OE hwā becomes both who (10x) and ho (18x) as well as in collocations such as ho-so (43x) ~ hoo-so (2x) ~ who-so (8x), where it is recessive.

III.2.2.2. Loss of initial aspirate /h/:

Of the words beginning with inorganic <h> in this sample, several alliterate in lines marked by vocalic alliteration, frequently realized as an alternation of words beginning with vowels and words beginning with <h>: hacre, hasked, heir, herly, holde.

The evidence of inverse spellings and the consistent use of the n-final possessives (e.g. myn, þin, etc.) before words beginning with <h-> as well as before words beginning with vowels suggest that initial aspirate /h/ has been lost in the dialect of the immediate scribe.Following his useful discussion of "orderly variation in spelling" (134-136), James Milroy, (Oxford, 1992), devotes considerable space to the relationship of orthographic <h> to the loss or dropping of the aspirate sound in Middle English pronunciation (137-145). Particularly valuable is his reasoning as to what use it is possible to make of manuscript evidence in this regard, and the pitfalls of disregarding such evidence. The indefinite article an regularly precedes words with initial <h->.

hacre "acre" 6.108; hasked 1.75, 10.167; heir "air" P.4; heres "ears" 10.282; herly "early" 5.332; hers 5.178 "arse"; hesily "easily" 20.352; his; holde "old" 6.86.

Note also the loss of <h-> in the compound falsed 1.66 "falsehood."

III.2.2.3. Word-final <h>:<ȝ> ~ <uȝ> ~ <w> ~ <yȝ> ~ <iȝ> ~ (<gh>) ~ (<wȝ>) ~ (<wh>) ~ (<y>) ~ nil

Word-final /x/, for the most part spelled <h> in OE, though sometimes by analogy with oblique forms <g>, followed either a vowel or a liquid.

burgh 2.59 ~ borewe 6.311 "burg, borough"; fee 4.133; heiȝ (5x) ~ heie 7.17; ynowe, Inow(e), I-now(e), a-now 13.274Richard Jordan, , trans. Eugene J. Crook (The Hague, 1974), §133, finds that this prefix takes on a more open sound between /i/ and /e/ in the fifteenth century; today's spelling "enough" is the result of this development. The spelling <a-> is attested as "scattered" by Jordan. The produces many variants with the <a-> prefix under inough (adv.), though none with <-w> as a reflex of the word-final fricative. (17x); lawhe 5.114 "laugh"; neyȝ(e)bore(s) (9x) ~ neiȝ (2x) ~ neȝbore 13.380; plow~ (47x) ~ plowȝ 14.32; seiȝ (14x) ~ say (3x) ~ saw 5.9 "saw"; þouȝ (58x), thouȝ 14.2 ~ þeiȝ (6x) ~ þowe (2x), þow ȝ.349 ~ þei 5.622; þoruȝ (88x), thoruȝ (31x) ~ þorȝ (13x), thorȝ (10x), þurȝ (3x), thurȝ 15.79.

III.2.2.4. OE /xt/:<ȝt> ~ (<ȝth>) ~ (<ght>) ~ (<ȝct>) ~ (<wt>) ~ (<th>)

abouȝte < OE abycgan "redeem"; almiȝti, almiȝty; auȝt; bouȝte; brouȝt(e); cauȝt(e); douȝt(e)r(es) (6x) ~ dowtres 1.28; eyȝte(the) (2x) ~ eyghte 6.332; hiȝt(e), hyȝte (30x) ~ hiȝcte 1.17; kniȝt(e)(s); lauȝte; liȝt(e); miȝt(e); nauȝt(e) (227x), nouȝt(e) (102x) ~ nauȝth 10.295; niȝt(e)(s); riȝt(e); siȝthe 17.52 "sight"; sle(y)ȝte(s) (3x) ~ slithes 13.424 "sleight(s)"; thouȝthe 2.13 "thought."But cf. deliȝt 10.393 < AN/OF delit < L delict-. Deliȝt 10.393 derives from AN/OF delit, which in turn derives from L delict-. The yogh here is as likely to be otiose or the product of scribal error as to be a "learned" respelling based on the scribe's knowledge of the Latin etymon. Luick § 769.2 considers the modern spelling a reversed spelling (cf. spright(ly) < OF esprit) owing to the falling together of the long /i:/ + <t> in <-it(e)> and <-ight> once the /x/ was lost. This fluctuation shows up in earlier spellings of <-ite> for present day <-ight> and vice versa.

III.2.2.5. OE /š/:<sch> ~ (<sh>) ~ (<ssh>) ~ (<ch>) ~ (<ssch>) ~ (<s>)

bischop(p)(e)(s) (15x) ~ bisshop~ (8x) ~ bishop(es) (2x); childisch; englisch(e) (9x) ~ englich (2x) ~ englys 7.118; felaschip~ , felaschypp (4x) ~ felachipp(e) 11.455; fisch(e)(s) (10x) ~ fissches 10.318 ~ fissh 5.180; flesch(e); lord(e)schip(p)(e); punische(n); schafte; schal; schame~ (14x) ~ shame 20.256; schip(p)(e)(9x); schrift(e)(s) (13x) ~ shrifte (2x); schul~ (178x) ~ shulde 18.388; worschip~ (3x) ~ worchip~ (3x), wurchip 3.332.

III.2.2.6. OE, ON /sk/:<sc> ~ <sk>The forms aschaped 6.80 ~ asckapen 2.164 "escape(d)" reflect normal developments from OF and AN.

ask~; buske(d); descreue~; scarlet; scathe; scolde; scole; score; scorne(d); scribes; skil(l)(e)(s); skynnes.

III.2.2.7. OE, ON <þ> and <ð>:<þ> ~ <th> ~ <Th> ~ (<tȝ>) ~ <tth> ~ <d>

The R scribe's distribution of line initial <Th> and <Þ> manifests fewer than twenty instances of the former in line-initial positionR1.43; R5.1; R6.1, 329; R10.251, R16.189, 238; etc. Just over eight hundred lines begin with <Þ>, though inside the line, <þ> and <th> appear to be in free variation, with about five instances of <þ> to four of <th>.

quatz (20x) = "quoth" < OE cwæð ~ q(uo)d (200+x).See Joseph Wright, (Oxford, 1928), §408. Verner's law produces plosive /d/from the old preterite plural; early rounding of /a/ < OE /æ/ produces ME /o/. Richard Jordan, , trans. Eugene J. Crook (The Hague: Mouton, 1974), § 32 remark 4, suggests that the /w/ (i.e. <qu> = /kw/) had an influence in the rounding of the vowel, once /æ/ had become /a/.; sitth 17.18; hatz 8.85 ; out 1.121 = "out."Cf. ouȝt (4x) "out"; Note that all but one of these appear in a seven-line section at 1.117, 124, 127; 17.189. See also 13.312 ouȝt "aught."

The graphies <th>, <-the>, and <ȝth> occasionally are written for /t/; e.g., a-thachud 2.198 "attached"; nauȝth 10.295; siȝthe 17.52 "sight"; thouȝthe 2.13 "thought"; thouche 17.150 "touch"; wyȝth 10.295; etc.

III.2.2.8. <ȝ>

Yogh appears in the usual ME contexts, word-initially as the on-glide [j], e.g., ȝates, ȝeue, ȝif, etc., and inside or closing a syllable as the voiceless velar [x]. The spelling of OF quand as quantz (or perhaps quantȝ?) suggests that at least in some contexts no phonetic value is attached to word-final <z> (or <ȝ>). Compare also <souȝd>, "wages, stipend" (3.350) from OF soudee. In this case, yogh appears to function as a lengthening graph.

III.2.2.9. <-g->

In one instance, <g> appears to be a relict form representing the on-glide /j/: murgur, "more merry" (1.108).

III.2.2.10. <-t> ~ <-th>

A serious and systemic problem with R's spelling system (influencing our understanding of R's morphology and potentially the authority of the B archetype) must here be addressed: in a wide variety of circumstances, especially evident in his finite verb suffixes, R seems to use word-final <-th> to signify something completely unexpected. This feature first came to our attention when we noticed that R shares an apparently nonsensical past-participle inflection (engendreth for engendred) with beta witnesses LCY. Any RL shared form, however odd, is intrinsically likely to be archetypal, albeit perhaps non-authorial, because of their extraordinary accuracy as well as their definitive stemmatic positions. If this lection is not merely a blatant archetypal error (one "corrected" by most later copyists to the expected form), it may be that the R and L scribes (or the Bx scribe) understood the <-eth> suffix in this word as allomorphic with the past participle suffix <-ed / -et> attested in other B copies. The final phone of engendreth would then probably have been construed by L and R as /t/ or /d/ (not the /θ/ which the spelling would suggest to us). A few pieces of evidence scattered throughout manuscripts L and R may support such a conclusion. One wonders, for example, whether the strong preference in manuscripts L and R for the ON-derived spelling of the cardinal number 100 (= hundreth) over the OE-derived form (= hundred) indicates that these scribes, or their models, would have pronounced that word with /θ/ as the final phoneme, rather than /d/. Such a conclusion seems doubtful. Rather, this spelling preference for the number 100 probably attests the same trivial orthographic anomaly hypothesized above concerning engendreth. A similar phenomenon seems illustrated by L's unique, anomalous spelling of a proper noun at KD13.266 (stretforth) in place of the usual Stratford. Another odd <th> verb ending occurs at R20.28, where all the other manuscripts, including L, read a preterite (most of them attesting ouertilte) while R has ouertilth. Yet this is almost certainly intended by R as a preterite (terminal phone to be articulated as /t/) since the line begins with another preterite in all copies, including R (Torned).

A similar oddity shows up in R's unique spelling of a Latin verb form at R18.166: fallereth versus falleret (which is found in the other manuscripts). As with hundreth and ouertilth, it seems implausible to regard the terminal <h> for this Latin verb form as having any phonological value at all. Finally, a similar circumstance occurs at R11.390, where manuscripts Bo and Cot join R in what seems to be an obvious misreading: R attests misfeith at a point where the context clearly endorses the beta reading, mysfait (= ModEng "misdeed"). But the implication of lections such as the aforementioned (e.g., hundreth and fallereth) is that R's apparent disagreement with beta at 11.390 may be illusory.

More important than a few isolated anomalies such as these, two complementary sets of apparently contrasting tense inflections (between R and beta) seem central to the question of a potentially variable role for the <-eth> tense marker in R:

(1) In ten cases involving verbs other than be, R deploys a seemingly present-tense <-eth> morpheme where beta (and usually F) shows a preterite. In addition to R2.80, these include R2.76teneth versus tenedR3.120appayreth versus (ap)peiredR5.226bummeth versus bummedR5.228vsethHere an alpha reading. versus vsedR5.489hath versus hadR6.134apayreth versus appeyredR12.147repreueth versus repreuedenR14.68wexeth versus wexenR17.30bileueth versus bileued

2) In seventeen cases of this sort (finite verbs other than be), R uniquely deploys a past-tense morpheme where the other witnesses show a present tense (with one or more important copies using the -eth inflection). These include the following examples: R2.88nuȝet versus noyeth / noyenR3.129lat versus letethR3.213techet versus teche(n) / techethR3.247taked versus taketh (L) / take(n)R3.247songen versus synge[th]R7.88bit versus biddethR9.42semed versus semethR9.96bit versus biddethR9.102drad versus dredethR11.298seide versus seithR12.101dampned versus dampnethR12.291lyued versus lyuethR14.279folwed versus folwethR15.465called versus callethR15.580nedede versus nedethR16.195hat versus hathR17.194clensede versus clenseth

By comparison, there are only eight instances where R uniquely attests a preterite but no beta copy's present-tense form is marked by <-eth>. These include R4.21helde versus holdeR6.122deyede versus deye(n)R13.116prentede versus preynteR13.344Auenged versus AuengeR14.76rett versus redeR15.232semed versus semenR20.230kepten versus kepen

Also of great importance is the fact that there is only one tense disagreement in the entire text between R (as unique witness) and the other manuscripts where a be verb is at issue (at R16.142).

Collectively, these examples may reveal nothing more than R's oddly systematic dissent from its cousins. But, on the other hand, they may indicate that the R scribe is reflecting a written language (that of alpha?) in which a terminal inflection <-eth> normally used for third-person present-tense tagging had encroached on the usual role of -<et> / <-ed> in marking preterites. Moreover, documents a few early instances (cf. note at R3.213) of the <-et> suffix moving in the opposite direction, being used as a present-tense marker. Embracing such an explanation for R's behavior would not require believing that the present-tense <-eth> was vocalized identically as the same suffix when it occasionally denoted a past tense, nor that they were equally and randomly distributed markers. All it would assume is that the <-eth> suffix could serve either tense-marking purpose (as may also have been the case with <-et>).

While the X family of the C-version agrees somewhat more often with B's alpha than with beta in situations where the two versions share textual variants, in most of these cases outlined above, the Cx reading agrees with beta, indicating either that R (or alpha) was very prone to tense errors - or that we have not been reading some of R's ambiguous tense cues accurately. Hence there may be somewhat less substantive textual variation in the B tradition than we have usually thought.

Although adjectives with the <-e> ending are overwhelmingly more common than endingless forms in constructions where the historical weak adjectival declension is to be expected, adjectives marked with <-e> are also decidedly more common where the unmarked form would be historical. See § III.3.4 below.

III.3.2.1 Nominative / Accusativenil

At this period in the development of ME there is no longer any significant distinction between the accusative and the dative. The two cases had long since coalesced into a single "object" case, which itself is formally distinct from the subject case only for the personal pronouns. It is thus necessary to understand "accusative" here as "oblique, not genitive," and not necessarily as accusative in the historical sense.

III.3.2.2 Genitive Singular:<-es> ~ (<-ys>) ~ (<-us>)This genitive <-us> is the expansion of a suspension and occurs but a single time in a reading otherwise not consonant with the usual scribal practices of R: 13.337 whit(us) wey(us) "wit's ways." ~ nil

bernes 4.59 "barn's"; cattes; craftys 5.25;Although it is possible that the form craftys 5.25 is plural, in that it follows som manere it is likely that this is an instance of the partitive genitive. cristes (24x) ~ criste 14.81; fendes; herlotes 13.433; kynges; kniȝtes; lordes; mannes; prestes; rigges 5.355; sapiences 3.330.

Some nouns, especially proper names ending with <-s>, remain unchanged in the genitive:

Iudas felawes 9.89; with mammonas mone 8.85; in seynt fraunceys tyme 15.258.

Reflexes of certain declensions in OE that had no <-s> genitive and nouns of romance origin occasionally occur in R in an unmarked form:

of an addre tonge 5.89; in þi brother eyȝe 10.282; of adames issue and eue 11.211; of a frere frokke 5.83; euery man(er) goed man 17.236; Anomalous beggere chambres at 4.126 is probably textual error rather than analogy.

III.3.2.3 Nominative / Accusative Plural:<-es> ~ <-s> ~ (<-us>)See the note to III.3.2.2 above. ~ <-z>Most likely following French scribal practice (cf. e.g. filtz 7.179), the <-z> plural marker regularly follows <t> (22x). The three exceptions are mynstralz 20.67; eu(au)ngeliez 10.259; pestilensez 5.13. ~ <-e> ~ (<-(V)n(e)>) ~ (<-re>) ~ (<-(V)r- + -(V)n(e)>) ~ (<-r- + -es>) ~ nilIncluded are the mutated endingless plurals such as men, teth(e), gees, muys, as well as reflexes of old neuter endingless plurals: scheppe, hors (2x), with gris (3x) "pigs" (< ON gríss, pl. grísir) probably falling into this group by analogy. Gris occurs all three times in alliterative collocation with gees: gode gris and gees P.100, bothe my gees and my gris 4.53, ne noyther gees ne gris 6.286. In the example from the prologue the word gris might best be construed to mean "pork," despite its being paired with the plural gees. The and Schmidt agree in this regard in reading "pork," but J. A. W. Bennett, ed. (Oxford, 1972), p.240, cites this passage and gives only the plural "pigs, piglings."

The regular plural in R is <-es> and it predominates almost to the exclusion of other <-s> variants:

armes; clerkes; creat(o)ures; dayes (34x) ~ daȝes 1.103; eyȝes (21x) ~ eyes (6x) ~ eiȝes 2.51; fis(s)ches; fyng(e)res; kynges; ladyes (8x) ~ ladies (2x); prestes; ratones; synnes; soules; wyues

There are no examples of the <-is> plural; there is but a single suspect instance of <-us> (see note to section III.3.2.2). Other plurals, including bare <-s> unpreceded by an unstressed linking vowel, are recorded below:R omits the a which is common to most manuscripts in many place 15.17. G, one of the two manuscripts with which it shares this distinction, reads pl. places. A scribal omission seems the most likely explanation for place, which is not likely to have been understood as an unmarked plural.

argumentz (2x); brether(e)n (3x); burgeys (4x) ~ burgeis P.90; child(e)r(e)n(e) (16x), chyldren 3.263 ~ childurne (2x) ~ childrun 15.277 ~ childre 16.126 ~ childres 6.99; cristene;Cristen is not only a noun but also an adjective, and as such may have a plural or weak adjectival <-e> ending. That the noun has the plural ending <-e> rather than the regular <-(e)s> probably reflects its origin as a substantivized adjective. dys 6.73 "dice"; enemys (4x); foes (2x) ~ fon 5.98; gees (3x); gris (3x); hors 6.216, 11.358;The two occurrences of the singular are both spelled with final <-e>: horse 17.95, 17.96. knes (4x); leopartz; lys 5.200; marcha(u)ntz (3x) ~ marchauntes 7.18; mareys 11.368 "marshes"; minstrales 3.211, mynstrales 10.55 ~ mynstralz 20.67; mys P.22, muys P.71; pans (9x) (never pens) "pence";The sg. of pans in R is peny (7x). religious(s)es (3x) ~ religiouse 10.314, 10.336 ~ religious 15.346, 15.373; scheppe 15.395 "sheep"; shoes 20.193 ~ schone 14.357; teth(e) (3x); þing(e) 9.30, 10.470, 13.317In OE þing was a long-stemmed monosyllabic neuter, and consequently had no ending in the nom. and acc. pl. Perhaps because it falls into the category of "collectives" (see Joseph Wright, (Oxford: Oxford U P, 1928), §331.2; Fernand Mossé, , trans. James A. Walker (Baltimore: Johns Hopkins U P, 1968), § 60.1) it preserves the older inflection longer in ME than many other historical neuters. It is not always easy to tell whether the s-less form is singular or plural in R, for the indefinite article does not always precede the singular, e.g.: Ac þing þat alle þe worlde wote wherefore schuldestow spare / And reeden it in rethorike 11.99-100; astronomye is hard þinge and euel for to knowe 10.221; Ac þinge þat wikkedliche is wonne and with fals slei3tes / Walde neuere wit of witty god but wikked men it maked 15.151-2. For unmarked plurals see the examples cited in III.3.2.3 above. There are also, though less frequently, regular <-s> plurals, e.g.: clergie þat can many þinges 10.179.

III.3.2.4 Genitive Plural:There are ambiguous constructions where adjectives may be construed as genitive plural substantives. Although there is no reason to rule out cristene as an adjective, on cristene kyng 3.284 has the characteristic indefinite pronoun governing a possible genitive plural noun, which is particularly common when a people is referred to. oen of godes chosene 11.117 has oen possibly governing the plural substantivized deverbative adjectival participle chosene: "one + GEN [of the chosen (ones)] of God." Corresponding to modern usage, godes chosene may on the other hand together constitute the object of the preposition: "one of + OPREP [God's chosen].<-es> ~ <-s> ~ <-(e)ne> ~ (<-erne>) ~ (<-ren>)

bischopes; bones 5.355;Although most manuscripts have a genitive plural compound such as ruggebones "backbones" here, C and R preserve two independent genitives, a singular and a plural: his rigges bones ende 5.355. childerne 4.119 ~ children 9.175;The phrase non children 9.175 would seem to be an instance of the partitive genitive "none of children," for the indefinite pronoun + genitive plural noun construction is very common in Old English and its Germanic relatives. By late ME, however, it is possible that such a construction could be interpreted much in the way of today's Adj + N pl., e.g. "no children." kyn(n)e;As with modern English "kind of," kynne regularly occurs as the governing element of the partitive genitive: none kyne riche 11.193; of alle kynne landes 13.216. lechores; lordes; mennes; names; seyntes; tailoures 15.490; þinges 9.2

With the possible exception of iewen, iewene,Tribal and ethnic proper names, often adjectival in origin, are commonly declined weak. none of the following words would have had the weak <-ena> genitive plural in OE:

clerkene 4.121; iewen 15.590; iewene 18.259;R reads here alle þe iewene ioye 18.259. Throughout R Iew- is nonetheless regularly declined as a noun with <-es> plural and genitive singular. There is, however, another probable, though ambiguous, example of the same genitive plural form (with loss of final <-e>) of this word: is parfit iewen lawe 15.590. F's þe iewes confirms this sense, although it is also possible to understand iewen here as an adjectival formation in <-en>, following the pattern of "wooden," "golden," "leaden," etc., where the derivational suffix makes an adjective out of a noun. Likewise offering circumstantial support for this interpretation, W conversely reads Iewen 1.67 where R has the regular genitive pl. with Iewes siluer 1.69. kyngene 1.106; menne 6.104; wyuene 5.29

The samples below only take into account historical monosyllables, of which "black," "broad," "false," "glad," "good," "great," "long" and "small" are the present-day reflexes. Although the <-e> ending is dominant where it would be expected, namely in definite and plural constructions, it is also distinctly dominant for the indefinite usage. Endingless adjectival forms occur, moreover, for the definite singular and plural. It would appear that the scribe of R tends more often than not to mark any and all monosyllabic adjectives with final <-e>, though not fastidiously.