The Canterbury Tales

by Geoffrey Chaucer

  Object–Verb–Subject God loved he best

  Object–Subject–Verb Frenssh she spak

  Adverb–Verb–Subject soore wepte she

  Subject–Past Participle–Auxiliary Verb whan she dronken hadde

  Complement–Verb–Subject Ful fetis was hir cloke

  Complement–Subject–Verb Curteis he was

  Prepositions often follow the nouns they govern: him aboute = about him, him bifore = before him. They are often placed directly before or after a verb: ‘That I of woot’ (ML 1021) = that I know of, ‘to shorte with oure weye’ (GP 791) = to shorten our journey with/with which to shorten our journey.

  A Note on the Text

  Previous Editions

  There are some 55 surviving manuscripts of the (more or less) complete Canterbury Tales, and a further 28 or so excerpted or fragmentary copies. 1 The eight-volume edition of the Canterbury Tales by John M. Manly and Edith Rickert published in 1940, which occupied a large portion of the working life of both scholars, is the only full-scale attempt to record all the readings of all these manuscripts, to try to establish their relation to an original common ancestor (which they dubbed O1), and so to produce a fully critical text of the work. Unfortunately, the defects of Manly and Rickert’s editorial assumptions and methods, which have been analysed with devastating thoroughness by George Kane, 2 deprive it of the authoritative status to which it might seem to be entitled. As Manly and Rickert themselves recognized, a vernacular text such as the Canterbury Tales is subject both to contamination between manuscripts (which may be the result of deliberate comparison between texts by scribes, or the unintentional and unconscious effects of scribal memory), and to coincident variation (the independent occurrence of the same scribal errors in manuscripts unrelated by descent). The operations of these two processes blur the lines of textual affiliation beyond recovery, and make it impossible to construct a satisfactory stemma (a ‘family tree’ representing the relations between the manuscripts) which could be used as a guide in determining the most probable readings of the author’s original. In these circumstances, editors have returned to the practice of selecting a good manuscript as base text and emending it eclectically after comparison with selected others. The two most favoured candidates for such a base text are the manuscripts known as Hengwrt (Hg), now Aberystwyth, National Library of Wales, MS Peniarth 392, and Ellesmere (El), now San Marino, California, Huntington Library, MS EL 26 C 9, which are generally considered the earliest and best copies of the work. 3 Most student editions of the Canterbury Tales – and also more ambitious editions, such as the Variorum Chaucer 4 – are based on one of these two manuscripts, borrowing readings from other manuscripts only in the case of manifest errors of sense or gross lapses in metre. The Riverside Chaucer, like the second edition of F. N. Robin-son, 5 of which it is a revised version, takes El as its base text, although it admits an even larger number of Hg readings than Robinson did, and also takes into account the readings of manuscripts which Manly and Rickert’s analysis established as important either in their own right or as representative of larger groups. 6

  The Present Edition

  The choice of either El or Hg as base text has the disadvantage of allowing the selected manuscript to take over the work of editorial decision in the multitude of smaller, humdrum instances of variant readings. For this reason, the present edition does not privilege either manuscript against the other (whether on the basis of a presumption of their position in a hypothetical stemma or on the basis of a presumption as to their relative accuracy in reproducing the original), but is squarely based on both. I have compared all their readings afresh, weighing each case of manuscript variance on its own merits, in the light of sense, metrical quality, knowledge of common types of scribal error and probable direction of scribal error (that is, which reading is more likely to have been the source, and which the result, of miscopying). George Kane’s pioneering work on the habits of Middle English scribes has been of great help to me in the last two respects. 7 The collation of El and Hg has been accomplished with the aid of the Facsimile and Transcription of the Hengwrt Manuscript, edited by Paul G. Ruggiers (Norman, OK, 1979), which prints all Ellesmere variants as marginal annotations to the transcription of the Hengwrt text, and the Working Facsimile of the Ellesmere manuscript (Cambridge, 1989). In order to identify and emend places where both El and Hg are likely to be wrong, I have also taken into account the wider manuscript tradition as it is represented in the Riverside Chaucer, and in Manly and Rickert’s three-volume Corpus of Variants. 8 Restrictions of space have prohibited complete documentation of every emendation made as a result of this comparison, but textual matters of particular importance or interest are discussed in the Notes. If an emendation is supported by the readings in other manuscripts of the Canterbury Tales, this is indicated by an asterisk against the headword (lemma); full details of these readings may be found in the Corpus of Variants contained in vols. V–VIII of Manly and Rickert’s edition.

  On numerous occasions I have also emended the text on metrical grounds. For discussion of the kind of quasi-unconscious scribal variations in morphology, spelling and grammar that frequently impair the metre in El and Hg, see my article on ‘Chaucer’s Meter and the Myth of the Ellesmere Editor’. 9 The following is a sample list of commonly emended forms, illustrated by selected examples (my emended form of the text is followed after the square bracket by the El/Hg readings):

  1. Presence or absence of final -n in verbal inflexions and in such words as bitwixe(n), ofte(n), sithe(n), withoute(n). E.g. GP 40: weren] were El; weere Hg. GP 326: pinche] pinchen El Hg.

  2. Presence or absence of final -e in the past tense endings of weak verbs (touched/e). E.g. GP 756: lakkede] lakked El Hg.

  3. Presence or absence of a y-prefix in the past participle. E.g. GP 396: ydrawe] drawe El Hg.

  4. Presence or absence of ‘ne’ or ‘nat’. E.g. GP 320: ne mighte] mighte El Hg.

  5. Contraction or expansion of endings in d/t/th (depeinted/ depeint). E.g. Kn 2049: depeint] depeinted El Hg.

  6. Variation between syncopated and unsyncopated forms of the verb (preyed/preyd). E.g. Kn 2108: preyed] preyd El; prayd Hg.

  7. Variation between ‘muche’ and ‘muchel’. E.g. GP 132: muche] ful muchel El; muchel Hg.

  8. Variation between ‘made’ and ‘maked’. E.g. Fri 1642: maked] made El Hg.

  9. Presence or absence of a final -e in certain nouns and adjectives (hoost/e, Seint/e, 10 Oxenford/e). E.g. GP 509: Seinte] Seint El Hg.

  10. Presence or absence of medial -e- in such words as Brome-holm, chaunterye, chivetein, cloisterer, faierye, giltelees, lichewake, mottelee. E.g. GP 259: cloisterer] cloistrer El Hg.

  11. Omission or addition of ‘that’ after conjunctions or relatives. E.g. Kn 1848: wher that] where El Hg.

  12. Repetition or omission of prepositions or articles in parallel constructions (‘for gold or [for] silver’). E.g. GP 558: A swerd and bokeler] a swerd and a bokeler El Hg.

  13. Simple changes in word order. E.g. GP 516: to sinful men noght] nat to sinful men El; noght to sinful men Hg.

  It should also be noted that the definite article ‘the’ or the negative ‘ne’ may be written out in full by the scribes even though the metre requires them to be elided with a following word beginning with a vowel or h; e.g. GP 110: th’usage] the usage El Hg. Similarly, verbs ending in -eth, such as bereth, cometh, falleth, maketh, speketh, taketh, thinketh (whether third person singular present indicative or imperative plural), are frequently pronounced as monosyllables (e.g., ‘Maketh’ at ML 898), and scribes will not necessarily indicate this by using a contracted form (berth, comth). Conversely, they may use a contracted form where a dissyllable is called for by the metre.

  In emending the El/Hg text on metrical grounds, I have made no attempt to indicate manuscript support for the adopted readings, not only because Manly and Rickert’s Corpus of Variants avowedly excludes this sort of detail, but also because these are among the feature
s most subject to coincidental variation, and the readings of other manuscripts have in consequence no confirmatory value even if they happen to agree with the reading preferred. Even if the emendations I have made do not restore exactly what Chaucer wrote (given the possibility of alternative emendations that would fit the metrical requirements equally well), they produce what Peter Shillingsburg has called a ‘less unsatisfactory text’, 11 in that it at least respects Chaucer’s ability to write good metre.

  Versification

  Emendations on metrical grounds are based on the assumption, confirmed by my own experience of reading Chaucer and working with the text, that Chaucer uses a five-stress (rather than ten-syllable) line. 12 The stress pattern is normally rising (that is, a weak stress is followed by a strong one). (The term ‘iambic’ is often used but is misleading, since it is borrowed from classical metre, which is constructed in a quite different way from English verse. The classical notion of metrical ‘feet’ is particularly inappropriate to English prosody.) Strong stress falls on syllables that would bear stress in normal speech, but verse also makes use of the fact that in a succession of syllables, the balance of strong and weak stress is determined by the relative strength of neighbouring syllables. So, in the following line of the Merchant’s Tale: ‘He háth it pút, and léid it át his hérte’, the word ‘at’ stands in a stressed position, but in the subsequent line, ‘The mó one, thát at nóon was thílke dáy’, it does not. Final -e is sometimes pronounced (as in the words ‘moone’ and ‘thilke’ in the line just quoted), and sometimes not, as the metre requires; 13 it is elided before a following vowel or h. Final -e is always pronounced in the following instances: (1) at the end of monosyllabic adjectives in a weak position (that is, following a definite article, demonstrative adjective, possessive adjective or noun in the vocative case), unless the following noun is stressed on the second syllable; 14 (2) in monosyllabic strong adjectives in the plural. 15 Variations in the general stress pattern which occur with some frequency are: the inversion of weak and strong stress at the opening of a line or after the mid-line break; the absence of a weak stress at the opening of a line (a so-called ‘headless’ line); and the absence of a weak stress between two strong stresses, especially at the mid-line break, where the perception of a tiny pause between the two strong stresses takes its place.

  Spelling

  Since neither El nor Hg is used as the base text of this edition, I have followed the spelling of the Manly–Rickert edition, very occasionally altering it to an El or Hg form for the sake of greater accessibility (e.g., ‘saugh’ instead of ‘say’). 16 However, with the aim of making the text more accessible to modern readers, I have regularized use of i/j, i/y and u/v so that the spelling of a word is always as close as possible to modern practice. 17 In recording readings in El and Hg in the Notes, differences between them involving capital letters and variable word division in words such as ‘upon’ are ignored, and the spelling of i/j, i/y and u/v follows the rules adopted for the edited text.

  Tale Order

  Despite having been written by the same scribe, El and Hg present very different pictures as far as the ordering of the Canterbury Tales is concerned. In Hengwrt, the order of the tales is confused and uncertain, and there are clear indications (spaces left to be filled later) that the scribe did not have all the material in front of him as he copied, but was receiving it – from whatever source – piecemeal. This meant that he occasionally had to alter the names of tellers in the links, since he had already copied the tales in the wrong order, leaving space for the link to be inserted later. The Canon’s Yeoman’s Prologue and Tale never reached him at all. Ellesmere, in contrast, confidently presents the tales in a smooth and plausible order. Whatever the problems with which the scribe was initially dealing, it looks as if, by the time he copied Ellesmere, they had been authoritatively cleared away. The present edition follows the Ellesmere order, as being the nearest we can get to Chaucer’s likely intentions. 18

  Line-numbering

  Those early editors who accepted the validity of the ‘Bradshaw shift’ (see note to the Wife of Bath’s Tale 847), carried the line-numbering of the tales straight through from the end of the Man of Law’s Tale to the end of the Nun’s Priest’s Tale (that is, through the fragment that they labelled B2). This numbering is used, for example, in the editions of Skeat and of Manly and Rickert (see Abbreviated References), in the Tatlock and Kennedy Concordance, 19 and in the Chaucer Glossary edited by Norman Davis et al. 20 The Riverside Chaucer, which leaves fragment B2 in its Ellesmere position, starts the main line-numbering afresh with the Shipman’s Tale, but gives the alternative numbering as well (indicated by an asterisk after the line-number). The present edition follows the Ellesmere order, and therefore likewise starts the numbering afresh with the Shipman’s Tale; anyone who needs to convert these line-numbers to their counterparts in the ‘Bradshaw shift’ system should add 1190 to the number in question.

  Headings and Verse Paragraphs

  The headings that mark the beginning and ending of tales are taken from El, and printed in italics. Latin headings marking internal divisions in certain tales have been translated into English, and are likewise printed in italics. The small indentations marking minor narrative breaks are editorial.

  NOTES

  1.

  For a complete list, see Derek Pearsall, The Canterbury Tales (London, 1985), pp. 321–5.

  2.

  George Kane, ‘John M. Manly (1865–1940) and Edith Rickert (1871–1938)’, in Editing Chaucer: The Great Tradition, ed. Paul G. Ruggiers (Norman, OK, 1984), pp. 207–29. For a defence of the Manly and Rickert edition, and a moving account of the difficulties they faced, see Roy Vance Ramsey, The Manly– Rickert Text of the Canterbury Tales (Lewiston, NY, 1994).

  3.

  Modern palaeographers agree that El and Hg were written by the same scribe; because of the scribe’s problems with tale-order in Hg (see below), it is generally assumed that it was copied after Chaucer’s death, and because these problems have been resolved in El, it is generally assumed that El was copied later than Hg, but both manuscripts date from the first decade of the fifteenth century.

  4.

  This series aims to provide an analysis of the textual history of Chaucer’s individual works, and to offer a comprehensive survey of all aspects of critical commentary on his work. For individual volumes cited in the Notes to this edition, see notes to GP 164, Sq 110, Pri 518 and NP 2771–90. For a full list of published volumes and an update on the project, see the Variorum website: http://www.ou.edu/variorum/

  5.

  F. N. Robinson, ed., The Works of Geoffrey Chaucer, 2nd edn (Boston, 1957).

  6.

  See the Riverside Chaucer, Textual Notes, pp. 1120–21.

  7.

  For examples and discussion, see George Kane, ed., Piers Plowman: The A Version (London, 1960; rev. edn 1988), pp. 115–65; The Legend of Good Women, ed. Janet Cowen and George Kane (East Lansing, 1995), pp. 43–111.

  8.

  Examples include (1) cases where El and Hg have inaccurate forms of proper names (e.g. ‘Risus/Rusus’ for ‘Rufus’ at GP 430; ‘Pavik’ for ‘Panik’ at Cl 590; ‘Neptimus’ for ‘Neptunus’ at Fkl 1047; ‘Onedake’ for ‘Odenake’ at Mk 2272); (2) lines present in other manuscripts but absent from El and Hg (see, e.g., the notes to Kn 2681–2; ML 1163–90; WB 44a–f; Mel 1335–6, 1777); (3) cases where deficiencies of sense, often in conjunction with deficiencies of metre, suggest scribal corruption in El and Hg (see, e.g., the notes to Kn 1252, 1376, 1906, 1945, 2202; ML 289; Mch 2229–33, 2240; Sh 214, 228; Pri 564, 676; Mk 2467; NP 2855, 3422, 3444–6).

  9.

  Studies in the Age of Chaucer, 23 (2001), 71–107.

  10.

  For an explanation of the use of ‘Seinte’ before masculine names, see E. Talbot Donaldson, Studia Neophilologica, 21 (1948–9), 222–30.

  11.

  P. L. Shillingsburg, Scholarly Editing in the Computer Age: Theory and Pra
ctice, 3rd edn (Ann Arbor, MI, 1996), p. 63.

  12.

  The most helpful account of Chaucer’s metre is Morris Halle and Samuel Jay Keyser, ‘Chaucer and the Study of Prosody’, College English, 28 (1966), 187–219. A refinement of their theory proposed by Dudley L. Hascall, ‘Some Contributions to the Halle-Keyser Theory of Prosody’, College English, 30 (1968), 357–65, was accepted by Halle and Keyser in the section on iambic pentameter in their book English Stress: Its Form, Its Growth, and Its Role in Verse (New York, 1971); see p. 172, n. 15. Further modifications to the Halle and Keyser theory are discussed in Gilbert Youmans, ‘Reconsidering Chaucer’s Prosody’, in English Historical Metrics, ed. C. B. McCully and J. J. Anderson (Cambridge, 1996), pp. 185–209. For a helpful survey of theories of English metre, which includes a measured assessment of the advantages and disadvantages of the Halle–Keyser theory (pp. 34–46), see Derek Attridge, The Rhythms of English Poetry (London, 1982).

  13.