Bo/etíll Mí//ares Car/o
1005-1006,14-25: 379-394
ISSN: 0211-2140
Orthography, Codicology, and Textual Studies:
The Cambridge University Library, Gg.4.27
"Canterbury Tales" 1
Jacob TIIA1SEN
Adam MickiewicL University
SU:\lMARY
An analysis 01' thc distrihution 01' orthographic variants in the Cambridge University Library,
MS Gg.4.27 copy ofChauccr's CUlllahur\' Ta/es suggests that longer tranches ofthe source text
were together, and partially ordered, already when it reached the scribe. The cvidencc 01' this
manuscript's quiring, inks, miniatures, and ordinatio supports this finding. Studying linguistic
and (other) codicological aspects 01' manuscripts may thus he 01' use to textual scholars and
editors.
Key words: Chaucer - Call1erhur\' Ta/cs - manuscript studies - Middle English - scribes - illumination
-- dialectology - graphemic and orthographic variation - textual studies
RESUMEN
El análisis de la distribución de las variantes gralcmicas en la copia de los ClIclllos dc
Call1crhlllT de Chaucer en Cambridge University Lihrary, MS Gg.4.27 sugierc que fi'agmentos
largos del texto ya estahan juntos y parcialmente ordenados cuando éste llego al escriba. Esta
idea tamhién sc sustenta en evidencias paleográficas en el manuscrito, como las tintas, las
miniaturas, los cuadernos, y la disposición en la página. El estudio de los aspectos lingüisticos
y eodieológicos de los manusritos pueden serie úti les a editores y critieos textuales,
Palabras clavc: Chaucer, Cllclllos dc Call1erhllr\,, manuscritos, inglés medio, escrihas, illuminación,
dialectología, variación gralcmica y ortográfica, estudios textuales
INTRODUCTION
The dialecto10gist Angus McIntosh has noted that one type of exemp1ar
influenee has the interesting characteristic that "jt tends to assert itsc1f less and
1ess as the scribe proceeds with his work" (1975 [1989: 44 n. 10]). He distin-
I The aulhor was lhe rccipient 01' a doctoral bursary rrom De Vlonlforl University that cnabled the
research leadíng to the lindings hel"C presented. lle ís gratcful lo hís eolleagucs and supervísors at the
Canterbury Tales Project, Profcssors NOrlnan 13lake and Peler Robínson in particular, ror sharíng their materíals
ami views wílh hím: ami lo Dr .Ioanna Nykiellór her íneisíve eommenls on eadier dralis orthe present
papel'. The maín jindíngs were presentcd lo the \Jth lnlcrnatíonal Confcrence on English Hislorical
Linguistícs, University 01' Vienna, August 2004, as well as lo lhe 7th Confcrenec 01' the Europcan Socicty
¡<Jr the Study 01' Englísh, Uníversity 01' Zaragoza, September 10()4.
379
Jaco/J Thaisen Orthography. Codicology. and Textual Studies: The Cambridge University...
guishes graphetic variants from those graphemic ones which imply no phonologieal
difference ("written-Ianguage ones"), and the latter in tum from those
graphemic ones whieh do ("spoken-Ianguage ones"), though noting that clear
boundaries cannot be set along this eline. The type is that which makes itself
felt toward the former pole of the cline. Gradual change may be observed at
the opposite pole too, and has more often been so in the literature since we
tend to record the various spellings of whole words in our profiles and treat
them as the unit of variation. Drawing on the electronic transcripts made available
by the Canterbury Tales Projeet, two of its researchers, Norman Blake
and Jacob Thaisen (2004), traced variations at all three levels during the complete
text of two manuscripts of that Chaucer poem with early content,
London, British Library, MS Harlcy 7334 and Oxford, Christ Church MS 152.
We concluded that these variations in a scribe's orthographic practices can signal
how many exemplars were used in the production of a manuscript and thus
can be of use in textual studies, and our conclusion was supported by the evidence
of these manuscripts' codicology. In what follows, 1 discuss another
scribal copy of the poem, that found in Cambridge University Library, MS
GgA.27, part 1, considering tirst its linguistic aspects and then its codicologicalones
before discussing the evidence they provide.
Part I of the manuscript today consists of 43 vellum quires totalling 516
folios, and it contains al! Chaucer's longer poetical works and many ofhis shortel'
items in addition to Lydgate's Temple olGlas. The shorter items fill quire 1,
which constitutes an independent unít (Elanor Hammond 1908: 190; Malcolm
Parkes and Richard Beadle 1979-80, 3: 2; Daniel Mosser 1996), and the nonChaucerian
text come last. Canterbwy Tales [Gg], which sits between Troilus
and Criseyde and Legend olGood rVomen. now begins imperfectly at A37 on the
verso of quire lI 's twelth folio and ends with the explicit to the Parson's Tale on
quire 37's first verso; the next folio, which presumably contained Chaucer's
Retraction, is no longer present. The poem is arranged in the same order as is
found in the San Marino, Califomia, Huntington Library, MS EI.26.C.9 copy of
it. [Ellesmere] and includes only material which is also present there except as
individual lines natural!y became slightly modified during their scribal transmission.
Textual scholars and editors have long agreed that Gg and the other
Chaucerian texts all are high in authority, and some have given the manuscript
special attention because it includes many unique readings. For example, Legend
olGood TVomen has a unique prologue which scholars generally accept as being
genuinely Chaucerian2, and the manuscript may be the earliest extant witness to
" What one eritie correctly described as "entcrlaining speculations" (Schl11idt 1975: 391) Icd F. W.
l3atcson (1975) to sllggest that lhe best extant witness to Chaucer's own spelling is the later version of lhe
prologue to Legena olGood Women which survives lIniquely in Gg. The logic of l3ateson 's arglll11ent is hard
10 follow. It appears to be that becallse Manly and E. Riekert held that lhe Gg scribe had "access to speeial
sources" (1940. 1: 179) c10se lo the CUl1lerhlllT Tules arehelype and beeause the poet intended his poel11 to
Bo/elín Mi//ares Car/o
2005-2006.24-25: 379-394
380
JlIcob T/llIi"en Orl!lOgrtll'h)'. Codico!og)'. am! Texlua! Sludie,,: The Cambridge Univer"il)'...
"Lenvoy de Chaucer a Scogan". A facsimile ofboth parts ofthe manuscript was
cdited by Parkes and Beadlc (1979-80). In their commentary, whieh is inc1uded
in the third volume, thcy eonfínn the customary production date of c. 1420 for
part 1, although they note that celiain features of the anglicana formata scribal
hand and of the illumination are slightly old-fashioned for that date3; they suggest
that the illumination is based on an early model (1979-80, 3: 6-7, 60; cf.
Hammond 1908: 189; Miehacl Seymour 1997: 51).
The scribe also wrote Oxford, Bodlcian Library, MS e Musaeo 116, part 1
(Treatise 011 the Astro/ahe. MOl1devi//e:\' Trove/s. and a treatise on arboriculture),
and Columbia, Univcrsity of Missouri, Fragmcnta Manuscripta 150 (one
folio of Chauccr's Boeee). Parkes and Beadle propose that he was commissioned
to produce CigA.27, part I in an East Anglian country house 01' vicarage
(1979-80,3: 56,63-64). Thc illumination supports a place ofproduction in this
part of the country, for Kathleen Scott considers it to be a precursor of what is
found in the later London, British Library, MS Harley 2778 and San Marino,
Huntington Library, MS 268, both of which are Lydgate manuscripts associated
with Bury St Edmunds (1996, 2: 145; cf. Parkes and Beadle 1979-80,3:
60). Parkcs and Beadlc conc1ude from their study of threc longer dittographies,
two 01' which occur within Gg, that the scribe generally adapted the usage 01'
what was before him into his own dialcct as he transferred it to his copy (197980,
3: 56). Further support 1'01' the placing comes from this and other linguistic
cvidence, for the manuscript and e Musaeo 116, part 1 both contain far-eastern
features, and the former was localised to Cambridgeshire by A Lil1guL<;tic Atlas
oj'Lute Mediaeval El1glish [LALME] (McIntosh et al. 1986,1: 67). This rcvised
an earlier, unpublishcd, localisation to the borders of Suffolk and
Cambridgeshire by Michacl Samue1s (Andrcw Ooyle and George Pace 1968,
25 n. 25), who additionally noted thc presence of London-Westminster features
also found in Aberystwyth, National Library of Walcs, MS Peniarth 3920
[Hengwrt] and Ellesmere and of a scatter of western spellings4 . He suggested
from this that a copy ofthe poem akin to those two manuscripts passcd through
at least one stage of copying before it reached thc Gg scribe (1983 [1988: 3 IJ).
The presence of eastern spellings is a characteristie of other copies of
Cal1terh/llY Tales with the same, 01' a related, order of tales (.Iohn Manly and
Edith Rickert 1940, 1: 555; Simon Horobin 2003: 64-70), including the
London-Westminster Cambridge University Library, MS OdA.24 and
Ellesmere itself5. That the Gg scribe may thus alternatively have had his work-be
read a10ud. the unusual spellings \()und in Gg are phonetie ones that go back lo Chaucer himse1f. For
anothcr dismissive criticism 01' this view. see Sal11uC!s (19X3 l \9XX: 36 n. 27) l.
) The hand indudes occasional secrelary !Catures such as singlc-compartmenl o.
.j In addilion to the westcrn spellings (e.g .. In'/h, heg/¡, hlll'e). some northcrn ones (e.g" agu\'I1.1'. 01'1")
are scattcrcd in Cig.
, For (he linguistic evidence for (his plaeing of the Cambridge Dd manuscript. see Thaisen and Da
Rold (in press).
381 BolelÍn Millares Cario
2005-2006. 24-25: 379-394
Jacob Tllaisen Ortllo/(raplly. CodicoloKJ!. and Textual Sludies: Tlle Cambridge University...
place in the metropolitan area is suggested by Blake (19SS: 76) from the known
availability of exemplars for the manuscript's various Chaucer texts there
earlier in the fifteenth century6.
The scribe did not work alone, for Gg.4.27, part l is the product of a
team. A supervisor marked corrections by supplying a cross or an omitted
word in the margin, and some of his directions for rubrics and catchwords
survive. A second scribe is responsible for fols. SOSr-S l Ov and SI 4 recto and
verso in the final quire, which contains the Lydgate item. Fols. SOS and SlS
are conjugate as are S09 and SI 4 but SI Ois a singleton, for its conjugate, fol.
SI 3, is no longer present. This means that the two scribes appear together on
one bifolium, indicating that they collaborated (Parkes and Beadle 1979-S0,
3: 44). This second scribe is ignored in the present discussion unless otherwisc
st~ted. Lastly, one or more artists contributed decorative initials with
leaf borders, miniatures of the Canterbwy Tales pilgrims, and sets of the
Vices and Virtues, the latter as illustrations to the Parson 's Tale. Most of
these decorations are lost through thc excision of folios. Part 2, of 3S folios,
contains passages copied from Thomas Speght's l S9S-edition of Chaucer 's
works to repair this extensive mutilation and was bound in with the mediaeval
text until the late nineteenth ccntury.
!> The language 01' Gg has reeeived far greater seholarly attention than what is here reeounted. Manly
and E. Riekert"s assistant Dean. who is responsiblc for the commentaries on dialeet alld spclling included in
their edition. deseribed the Gg dialeet as "East Midland. with suftieient traces 01' Northern to suggest Norfolk".
noting its resemblanee to that fi.llllld in the Paston letters (Manly and E. Rickert 1940. 1: 176-77). Samuels
(19X3 [19XX: 31]) applies the labe! "very pronounced East Anglian". In addition. the prescnce ofthc unusual
spellings along with the many eorreetions and lhe thrce longer repeated passages have !ed some past scholars
to suggest that tbe scribe was a foreigncr who pcrhaps had a limited undcrstanding of what he was copying.
For cxamplc. Skeat (1 X99-19(2) held that hc was influcnccd by Anglo-French usagc. whilc Dcan (Manly and
E. Rickcrt 1940. 1: I77-7X) suggested he possibly was Flcmish or Outch. Furnivall had previously assigned
the manuscript to thc borders 01' the Midlands and the North (1 X6X: 59). which might have meant a location
ncarcr the Ilul11ber in his days though mediaeval East /\nglian dialcct contained northcrn features. Caldwcll
(1944) f"und resemblanccs betwecn the spelling 01' ccrtain lemmata in Gg and thcir Duteh cognates and
argued. uncollvincingly, far a natively Dutch-speaking scribc on this basis. Thc manuscript contains examples
of.l"('h- in words likc ESCAPE. SCHOOL, and SIIAPE (see further p. 3X5 below). ofwhat might be described
as an epenthetic vowel in the unstressed final syllablc in. ¡"r examp!e. reso/1ahe!e and \'/1 stahe!e, of he fi.,r BY.
and 01' a plosivc rathcr than a t;-icative in spcllings ofTHITHER and WHETIIER (fJed\'/; H'hed)'l). Caldwell
based his argul11ent on thcse and certain other fcaturcs that Gg also contains bcing common in Middle Dutch.
although hc noted that "parallcls... can bc f"und in othcr Fnglish texts t;-om the first hall' ofthe fifteenth eentury.
notably cCl1ain 01' the Paston !etters and various pieces frol11 the Fast Anglian area" (t 944: 34). More
recent scholarship has favoured this seribe being no f"reigner. For examplc. Seymour first proposed and later
rejeeted that Gg.4.27. part 1 (and e Musaeo 116) are the work ofan Englishman who had lived in Ilolland t"r
a long time (196X). now suggesting that a west Suf1"lk seribe wrote it in Cambridge. probably on the basis 01'
the findings 01' dialcctologists (1997. 5¡). Ral11scy (1994: 359-77) suggested this scribe was an Englishman
who was trained lo copy Latin and thereforc unfamiliar with secing the vernacular in writing. Parkcs and
13eadle (1979-XO. 3: 46. 54) conclude f;-om his handwriting that he was no f"reigncr and t;-om a comparativc
study ofthe spclling j"und in the three rcpeated passages within the manuscript that he "was a reasonably carefÍJI
and conscicntious copyist [who] understood what he was copying [but was] prone to marked lapses ofconcentration,
perhaps becausc 01' infinnity or old age" cr also Ilammond (190X: 191).
Bo/etín Mi//ares Cario
2005-2006.24-25: 379-394
382
Jacob Tllai"en Ortllograplly. Codicology. and Textual Sludie,,: Tlle Cambridge Univer"ity. ..
THE EVIDENCE OF GG'S LANGUAGE
The present research is based upon an electronic transcript of Gg prepared
by the Canterbury Tales Project. An index of this transcript was semi-automatically
compiled by me from the forms registered in the published Spelling
Database for all fifteenth-century witnesses to the Miller's Tale (Peter
Robinson 2004). A comprehensive spelling profile for Gg was extracted from
this index, answering to the questionnaire used for LALME. This profilc gives
the distribution of the variant spellings of each questionnaire item in separate
tales and the Wife of Bath's Prologue, but not other links because of their
brevity; the Cook's Tale is absent from the data set because the text of the tale
is lost due to the mutilation to the manuscript.
The methodology 1 have employed for the analysis ofthe linguistic aspect of
Gg is that described by Blake and Thaisen (2004). It focuses on the variations in
the rclative usage of functiona! equivalents at aH three levels defined aboye
during the course of the text. A scribal copy written throughout in a single hand
will contain some random variation at these lcvels. This is because the orthography
of a scribe is variable in itself in addition to being subject to many and
varied int1uences, prominent among whieh are the number of exemplars and the
order of copying. The criterion for c1aiming that a non-random pattern is present
in the movements in the rclative use of diverse femns of a single lemma during
the course of a given text must consequently be that other forms for other lemmata
are found to show coincident movements. Alternative!y, the movements in
the use of a single form relative to the lcngth of each text unit must be paralleled
in other, unrelated fonns. Some individua! fonns and even lemmata will be too
rarely attested to produce a elcar pattern. It is a reasonable expeetation that the
variable feature in those fonns will be found to be similarly distributed in rclated
forms eontaining it, since the context is stable. If similar variation was found in
the representation ofthe initial consonant in the forms recorded for, say, the items
YOU, YOUR, and YOUNG in the Gg spelling profile, the distributions ofthose
spellings were therefore cont1ated so as to increase the reliability of the pattern.
This procedure amounted to positing the representation of that consonant as a
\cmma in itself The profile derived from the original profile through the application
of this procedure thus came to \can more toward the graphetic/'"writtenlanguage"
pole of the cline than the original one did, the unit of variation being
smallcr than the word in many cases. To iIIustrate the findings 1 have selectcd
those fonns which most clcarly show thc pattern that 1see in the data.
My research indieates that Gg contains two co-ordinate usages, their
boundary falling at the junetion of the tales of the Summoner and the Clerk.
The two usages can be distinguished because, for a number of unrelated lemmata,
the greatest shift in the relative use of one femn to another between any
two consecutive tales consistently occurs at this point. Especially prominent
383 Bole/ín Millares Cario
2005-200(,.24-25: 379-394
.Iacob Thaisen Orthogruphy, Codicolo!.:y, amI Textual Studies: The Cambridge University...
among the spellings which differ significantly in their proportions between the
two usages are those with,b and those with eCe/oCe characteristic of the tales
that come before this junction, against those with th and those with eeC/ooC
characteristic ofthe tales that come after it. This finding may be illustrated by
the variation in the occurrence of,b and ee/oo in the following table:
6
-----------71
LLr Ir J h- rr ~ i rr 1. 1 1~ 1I~ 1
I 16
14
12
10
8
3
I : ~
l °O"z:::::(..r..)....1O"¡:.....~~....1(..r..)o,~::r::Q::r::~¡:.....""oo..~>--« O
v~~~~~~~~u~~~o..o..~o..¡:.....~~ZZu~o.. I
~:$:
________________________~ ._ _________l
Table l.
Occurrence, by talc, ofee (white: 2,677 oecurrences), 00 (grey: 1,327 oecurrences), ano /J (black:
3,263 occurrcnees) in Gg (per thousano charaetcrs). ee ano 00 plot on lhe right vertical axis
Note. Al! links except for the Wife 01' Bath's Prologue are omitteo. The Cook's Tale is lost through
mutilation to Gg
The tales up to the junction, which coincides with a quire boundary,
account for around the first third of the Gg text, and where it is possible for
the spellings of a given lexical item to exhibit this variation, they invariably
do. Of the total of 155 examples of,be as many as 120 are thus found in this
third, along with 576 out ofthe 647 examples of,be. The figures for ,bu, ,bow.
and,bu collectively are 238 in the first third of the Gg text but just 29 in its
final two-thirds, against thou, thow, or thu with 73 and 422 respective exampies,
reversing the proportions. The corresponding figures for OTHER are
75 and 8 for fonns with,b but 26 and 186 for forms with th. For theer(e) they
are none and 83, for vvheer(e) none and 29, for weer(e) none and 74, and for
been 24 and 536. These forms may be comparcd with ther(e), wher(e). and
were being evenly distributed in the text and with there being 181 examples
of ben in the Gg first third against 536 in its final two-thirds. Among other
functional equivalents that are used in significantly different proportions
between the two usages and so confirm their cxistence are be- in lexical
items such as BEFORE and BETWEEN, her(e). hey(e), /et(e) , and
sen(e)/sens preferred in the first usage, against bi-/by-, hir(e)/hyr(e) ,
high(e)/hygh(e), Jat(e) , and sith-!~J'th- preferred in the second. Their occurrence
is given in the following table:
Bo/etín Mi//ares Cario
200S-2006, 24-25: 379-394
384
Jaeoh Thaisen Orthography, Codieology, and Textual Studies: The Cambridge University...
Oeeurrenee Oeeurrenee
Lemma Spelling Quires Quires Lemma Spelling Quires Quires
11-20 21-37 11-20 21-37
be- 30 30 nal 42 261
BEFORE NOT
bi-/bv- 15 57 nol 271 377
be- 18 17 peple/peph~ 4 62
BETWEEN PEOPLE
hi-/bv- - 17 plIple/pllp/is 8 15
hesi/besy(e) 7 7 sen(e)/,el1.\· 39 8
SINCE
BUSY hisv/bvsy - 5 silh-/svlh- 29 47
hllsr - 2 sH't!che 4 53
SUCH
he,.(e) 423 295 'swieh/swych' 92 191
HER/HIR
hir(e)/hyr(e) 201 581 wil 12 67
WILL
hev(e) 23 19 wol 11 14
HIGH
high(e)/hygh(e) 11 45 wilh/wVlh 272 453
lal(e) 25 57 WITII
LET lel(e) 46 35 wl 189 211
Table 2. Absolute oeeurrenec, by seleeted groups of quires, of seleeted spellings in Gg
Note. AH links exeept for the Wife of Bath's Prologue are omitted. The Cook's Tale is lost through
mutilation to Gg
In addition, Gg contains a number of unusual minor spel1ings which have
received considerable scholarly attention7. Frederick Furnivall was the first
to note their presence (1869-77: 6-7, 51-59), and it is these spellings that are
especially diagnostic of East Anglian dialect. They are not among the scatter
of easternisms that regularly crop up in metropoJitan varieties of English at
this date. A preliminary investigation into those among them Jike schastite
and scherche which have sch -where one would expect to find chreveals
that they are evenly distributed in the text8. They also occur in the
other texts inc1uded in Gg.4.27, part 1 as well as in e Musaeo 116 (Seymour
1968: 169).
7 See fn. 6 above.
K The forms and distributions considcred as evidence for the unusual seh- spellings bcing found
throughout Gg are: schangede (F370), schanou[n] (G573), scharge (1363), schaslile (A2055), schau[mJhyr
(A4143), schaufnJge (C734, B22246, B22453, B24264, (368), schau[nJged (A1400), schau[nJgede
(A2809), schau[n}lilh (A3367), schall[nJsel (A3656), schally[nJge (G 1239), scheke (D433), schekes
(A633), scherche (A2760, A3429), and schide (1632).
385 Boletín Millares CarIo
2005-2006, 24-25: 379-394
Jacob Thaisen Orthography, Codicology, and Textual Studies: The Cambridge University...
THE EVIDENCE OF GG'S CODICOLOGY
Scholars have noted that, except for the independent quire 1, the bulk of
Gg.4.27, part 1 appears to be consecutively copied, for most new items begin in
the middle of a quire (Mosser 1996); but coincident textual and quire boundaries
occur within Gg. The Cook's Prologue, the Man of Law's Tale, and the
Clerk's Tale are thus marked in the quiring, inks, and ruling, and seem each to
signal the beginning of a stint of copying. This can be inferred from the following
features, sorne of which 1 have developed from the commentary by
Parkes and Beadle (1978-79). (1) Before the manuscript was mutilated, the Man
of Law's Tale began on the first verso of quire 17, and the Clerk's Tale commenced
on the last folio of quire 209. (2) Quire 11, on whose eleventh folio the
General Prologue began before the mutilation, is signed a with the folio sequence
indicated in Arabic numerals. Signatures in this format, be they contemporary
with the scribal work or not (Mosser 1996; Parkes and Beadle 1979-80,3: 62;
cf. Manly and E. Rickert 1940, 1: 170; Seymour 1997: 47), continue up to 1 in
quire 20, although quire 16 is signed k 1 between the quire 15 e and the quire 17
f Sorne spacing separates the two letters in the three surviving quire 16 signatures,
which appear on fols. l83r, 184r, and 186r respectively. Of these, the
numeral that follows the letters is easily missed in the case of the latter two
because the relevant folios are worn at the outer edge. It may be this that led
Parkes and Beadle to read the signature as k followed by a Roman numeral" 1"
although 1 here has the emphatic form ofthe letter. AlI signatures after quire 20,
if present at all, consist of an arbitrary symbol followed by a Roman numeral.
(3) The ink has changed from a lighter to a darker shade ofbrown by the Cook's
Prologue towards the end ofquire 16 (Manly and E. Rickert 1940, 1: ]73). (4)
Quires ]1-]6 are ruled to receive 36lines to the page, while quires ]7-37 are
ruled for 38 or 39lines; quire 37 concludes the poem. (5) Ofthese, quires 17-19
and 2] -23 were each ruled as a batch, the latter twice because the original ink
flaked off (Parkes and Beadle 1979-80, 3: 39-40). Scholars have attributed the
new ruling of38/39 lines to the page to the requirements ofthe ordinatio, for the
rime royal Man ofLaw's Tale is the first stanzaic tale in the Gg tale order, which,
as has been mentioned, is the a-Ellesmere one (Parkes and Beadle 1979-80, 3:
42). Five seven-line stanzas separated by a blank line are accommodated on
every page (5x7 lines+4xl line=39 lines). This may mean either that the Gg
makers had not yet received this tale when they copied the tranche ending at the
Cook's Prologue, or that they copied that tranche without looking ahead to the
Man ofLaw's Tale in planning the ordinatio.
9 Two additional link/tale boundarics in Gg coincide, probably accidentalIy, with a quirc boundary.
These are the Prioress's Tale which began on the now-Iost first folio of quire 27 (fol. 319r) and the
Manciple's Tale which begins on the first recto of quire 33 (fol. 395r).
Boletín Millares Carla
2005-2006,24-25: 379-394
386
Jacob Thaisen Orthography, Codicology, and Textual Studies: The Cambridge University...
In addition, the placing of the miniatures upon the page indicates that the
scribe had already completed certain parts of his work when the Gg makers
decided to include them. The extant miniatures, which, like those found in
Ellesmere, are of pilgrims on horseback (Margaret Rickert 1940: 590-604;
Parkes and Beadle 1979-80,3: 59-60), are those ofthe Reeve (fol. 186r), Cook
(fol. 192V), Wife of Bath (fol. 222r), Pardoner (fol. 306r), Monk (fol. 352r),
and Manciple (fol. 395r). Parts of a penwork border surrounding what was that
of the Miller survive on fol. 174v, and the tail of a horse is visible on the stub
of fol. 243r which carried that of the Clerk before line 57 of his tale where its
narrative part begins. Every miniature with its border is aligned with the text
and slightly exceeds halfthe ruled area in size, occasionally extending slightly
beyond that area in the vertical dimensiono The regular placing of one is within
that area either on the same page as, and immediately before, the beginning of
the tale to which it relates or of this tale's narrative part, or on the page preceding
the start ofthat tale so as to fill it, for the Gg makers were always reluctant
to start a new link or tale on the bottom half of a page and so regularly left
a blank there. The miniatures ofthe Monk, Clerk, and Canon's Yeoman illustrate
these respective placings. The presence of both miniature and decorative
initial on the same page always resulted in the omission of those parts of the
miniature penwork border which were to occupy the same space as the 3/4 leaf
border growing out of the initial. The penwork borders are therefore most
probably later than both the initials and the miniatures. If they are in the hand
of the scribe, as Parkes and Beadle suggest (1979-80, 3: 42-43, 59-60), this
means that he retumed to Gg.4.27, part I after it had been decorated.
However, exceptions to this pattem are found within the consecutive
tranche ofthe text containing the General Prologue and the prologues and tales
of the Knight, Miller, and Reeve, as follows. The text missing from the absent
fols. 142-144 is A757-858 (=101 lines), and a rubric followed by the Latín
Thebaid epigraph to the Knight's Tale plus A859-964 (=2+ 106 lines). Since
these six pages were almost certainly intended to receive a total of 216 lines
of text given that 36 lines to the page is the standard ruling in this section of
the manuscript, a space ofjust seven lines was available for receiving the lost
miniature of the Knight after the concluding lines to the General Prologue on
fol. 143r. It must thus have extended to the very bottom edge of that page, far
exceeding the ruled area. The miniature of the Miller, the next one in the
tranche, is unexpectedly supplied before his prologue, on fol. 174v. This is
because a space of sixteen ruled lines was available there after the Knight's
Tale due to the makers' regular practice of leaving the bottom half of the page
blank after a tale. Thc space betwecn the Miller's Prologue and Tale on fol.
176r-thc expected placing-was of an insufficient sizc for taking the miniature
as they are separated by just one line. The last miniature in the tranche,
that of the Reeve, survives and extends to the vcry bottom edge of the page,
387 Boletín Millares Cario
2005-2006,24-25: 379-394
Jacob Thaisen Orthography, Codicology, and Textual Studies: The Cambridge Vniversity...
like that of the Knight. This page, fol. 186r , which is mled for the usual 36
lines, additionalIy contains the concluding 30 lines of the Reeve's Prologue,
and cropping has caused the loss of parts of the lower penwork border
associated with the miniature. These three exceptions indicate that the conflict
between miniature and initial ultimately arose because the Gg makers
modified the ordinatio to set aside space for taking the former only when the
tranche of the text up to the Cook's Prologue was already copied. In addition,
a mbric in red ink, written in the regular hand ofthe scribe, is cramped into the
one-line space between the Miller's Prologue and Tale touching the text aboye
as welI as below, and an unfilIed three-line space appears before the Cook's
Prologue on fol. 192r in quire 16; no mbric introduces this link. The Gg
scribe adopts a more formal script indicated by the more upright slant and the
absence of a lobe on the ascender of the b and h graphs for most other mbrics
except for those occurring next to a miniature, though not that of the Wife. So
not only the change of ink and the lack ofplanning for the incorporation ofthe
miniatures, but also irregularities in the mbrication show that there occurred
sorne interruption to the scribal and decorative procedures at the junction of
the Reeve's Tale and the Cook's Prologue.
Reconstmction strongly suggests that the first nine lines of the Man of
Law's Prologue [B 11-9] were accommodated in the space after the last 21 lines
ofthe Cook's Tale [A4401-22] on the lost fol. 193v. This arrangement ofthe
text, which can be calculated with confidence from the surviving fols. 192 and
194, ran counter to the normal practice of the Gg makers since it involved the
starting of a link on the bottom half of a page. It also produced a gap between
tale and link that is of an unusual size for Gg, and it made it the sole earIy
Canterbury Tales copy which both excludes the Tale of Gamelyn and has no
gap to the end ofthe page or, more usualIy, quire after the incomplete Cook's
Tale. A possible account mns as folIows: the Gg makers abandoned the copying
of quire 16 at the Cook's Prologue because this link and his tale would
fail to filI this quire, or there may have been uncertainty about what was to
folIow next. So they started the Man of Law's Tale on what is now quire 17 in
the new, darker ink and left the first recto of that quire, fol. 195r, now lost,
blank to permit this tale to be joined to whatever would eventualIy precede it.
When it later became clear that no conclusion to the Cook's Tale was forthcoming
and that the Man ofLaw's Tale would be the next tale after it, the team
resumed quire 16, now in the new ink, by writing the Cook's Prologue and
Tale in and linking them to the already started Man ofLaw's Tale through the
Man of Law's Prologue. They had already taken the decision to include the
miniatures at this time, for they were able to arrange the text so that those of
these two pilgrims fitted exactIy within the mled area on their respective
pages. This account gives the possibility that the irregular k 1 signature signals
that quire 16 was completed last, or just possibly penultimately, in the series
Boletín Millares Carla
2005-2006,24-25: 379-394
388
Jacob Thaisen Orthography, Codicology, and Textual Studies: The Cambridge University...
of alphabetieal1y signed quires, for the present quire 20 is signed 1and follows
the bateh-ruled quires 17-19 as we have seenlO. In other words, the progress of
the Gg scribe's copying was (l) the General Prologue to the Reeve's Tale; (2)
the Man of Law's Tale to the Summoner's Tale; (3) the Cook's Prologue and
Tale, and the Man of Law's Prologue; and (4) the Clerk's Tale to the Parson's
Tale, with the proviso that the third stint may have been started before the
second was completed.
CONCLUDING DISCUSSION
The analysis of the linguistic evidence indicated that co-ordinate usages
exist within Gg. The first usage is characteristic of those tales which come
before the junction of the tales of the Summoner and the Clerk and the second
usage is characteristic of those tales which come after it. The existence of the
two usages indicates, 1 feel, that a change of exemplar took place at this junction,
for they show the Gg scribe adopting features from what lay before him
to copy from. That jJ peters out gradually after the Summoner's Tale reveals
him abandoning the policy of transcription he had followed hitherto in favour
of one of translation, by which he sought to continue the previous usage with
respect to the graphemic representation of the dental fricative. He soon abandoned
this new policy due to the new exemplar inereasingly activating the
functional equivalent th in him. The distribution of eeC and ooC shows them
likewise becoming increasingly activated in the scribe only after the
Summoner's Tale, at the expense of eCe and oCe which had been dominant up
until this point. This means that the Gg seribe fluctuated between adopting and
adapting exemplar features during their transfer to his eopy. This characterisation
ofhim calls for an affirmation ofhis dialectal consistency given the inclusion
of GgA.27, part 1 as a survey point in LALME.
Among the fe atures signalling the two usages, the pairs jJ:th and
eeC/ooC:eCe/oCe are almost eertainly phonologieally identical but graphemically
distinct as are with/wyth:wt, whereas other pairs sueh as be-:bi-/by-,
nat:not, and wil:wol differ both phonologically and graphemically (and dialectally).
Still other pairs such as schau[nJge:chau[nJge may have been phonologically
identical to the Gg scribe but phonologically distinct to most contemporary
readers, at ¡east at first sight. Mid-fourteenth eentury northem and
10 The Cook's Prologue and Tale do not follow the Reeve's Tale in the pair London, British Library,
MSS Additional 35286 and Harley 7335. lfthis was the case also in the Gg exemplars, there seems to be no
reason why the rest ofthe quire would be left blank after the Reeve's Tale. Nor is it Iikely that the Gg makers
originally planned on omitting the Cook's Prologue and Tale altogether or alloeating them a position
later in the poem, for the same reason.
389 Boletín Millares Cario
2005-2006,24-25: 379-394
Jacob Thaisen Orthography, Codicology, and Textual Studies: The Cambridge University••.
southem scribes may similarly have disagreed about whether an adjectival
final -e was a feature of spoken or written language, and about whether a ylike
figura used in both jJ and y contexts represented one or two fitterae. It
appears from the present evidence that the Gg scribe adapted what he encountered
in his exemplars to his own graphemic system as a process separate from
that of him adapting it to his own phonological system, and that the rules
goveming these processes themselves evolved as his copying progressed. Both
processes can be described in the familiar terms of"transcription" or "translation"
of exemplar features introduced by McIntosh (1963: 28); but because the
graphemic and phonological systems of scribes and the relationship of these
systems to one another differ, there will be a fresh answer to what constitutes
movement along which of the two clines for every scribe and exemplarl1 .
From the narrative part of the Gg Clerk's Tale starting on the first recto of
quire 21, the first in a series of batch-ruled quires, and from the alphabetical
series of signatures ending in the immediately preceding quire, it may be surmised
that a stint of copying starts at the junction ofthe Summoner's Tale and
this tale. This congruence of spelling with quiring, ruling, and textual arrangement
strengthens the hypothesis that the scribe began copying from a second
exemplar there. The use of th to represent the dental fricative and of eeC/ooC
to represent a long or lengthened vowel characterises the usage found in the
earlier Hengwrt and Ellesmere manuscripts of Canterhury Tales as does the
spelling nat for the negation, which likewise increases after the Gg
Summoner's Tale. Manly and E. Rickert (1940, 1: 175-76) found the Gg tcxt
to be independent of the other major manuscripts up to and inc!uding that tale
but related to those two manuscripts as far as the remaining tales are concemedl2
. The Gg rubrics resemble those present in Ellesmere (Blake 1985: 7778;
Seymour 1997: 51), and the Gg miniatures may be associated with "the
idea or debased copying" of those found in that manuscript (Seymour 1997:
51)13. In other words, the second exemplar was Iikely a manuscript akin to
Ellesmere or was perhaps this manuscript itself. Unlike that of nat, the distribution
of other forms typical of Hengwrt and Ellesmere, 'swich/swych',
'yeue-', and wol, conf1icts with this inference since none of these forms exhibits
the expected increase after the Summoner's Tale, and still other forms
typical of them such as 'muche(l)', 'sholde', and 'thurgh' are practically
absent. But it is distinctly possible that all six were present also in the Gg first
exemplar given its likely London-Westminster origin that follows from its
[1 These taxonomical issues are further discussed by Mclntosh (1974, 1975) and Laing (1999).
12 Manly and E. Rickert describe many additional changes in textual affiliation during the course of
Gg, none of them between altogether separate branches of their stemma.
13 Beside Gg, only the earlier Ellesmere and four later extant witnesses to Canterbwy Tales contain
one or more miniatures ofthe pilgrims. M. Rickert (1940) does not suggest any link between Ellesmere and
Gg in terms of their miniatures.
Boletín Millares Carla
2005-2006,24-25: 379-394
390
Jacob Thaisen Orthography, Codicology. and Textual Studies: The Cambridge University...
early date. The Gg scribe, then, perpetuated the fonner three from both exemplars
('yeue-' as '3eue-') but rejected the latter three in all but a few instances.
The Gg ordinatio appears to have been developed ad hoc due to the mode
of reception of the two exemplars. The first was probably received as two
batches of text, their boundary falling at the junction of the Cook's Tale and
the Man of Law's Prologue, as for example indicated by the change in ruling
at quire 17. That those two batches derive from the same source-and so constitute
one exemplar as the tenn is here employed-is shown by the linguistic
evidence. The tales from that of the Clerk up to and including that of the
Parson, which concludes the poem, were copied from the second exemplar.
This exemplar certainly contained the complete text of Canterbury Tales since
it provided the model upon which all the Gg miniatures and sorne of its rubrics
are based. Its coming to hand may have confinned that no conclusion to the
Cook's Tale was forthcoming, leading to the completion, in the new, darker
ink, of the abandoned quire 16. This reception of the two exemplars along with
the availability of other exemplars, likewise high in textual authority, for most
of Chaucer's shorter and all his other longer works, the presence of the supplementary
hand in one of those other texts, and the likely return of the Gg
scribe to GgA.27, part 1 after it was illuminated combine to suggest a population
centre as its place of production. The possibility arising from the handwriting
and the illumination that the manuscript dates earlier than scholars
have customarily conceded strengthens the suggestion of Blake that this
centre was London-Westminster. This is becausc exemplars for Canterbury
Tales in unbound fonn may have been available nowhere else in the years
immediately following Chaucer's demise. But both Ellesmere and other a or
a-related manuscripts have associations with East Anglia through their dialect
and marks of early ownership or readership, and so the possibility can by
no means be excluded that exemplars in bound or unbound fonn had
disseminated to an urban locality there by the customary production date for
GgA.27, part 1 of c. 1420 or even a decade earlier.
A temporal hiatus occurred at the junction of the tales of the Summoner
and the Clerk during the copying of another manuscript of the poem with an
a-rclated text and early content, Oxford, Christ Church MS 152, presumably
because its scribe needcd to devise an order for the remaining tales. The junction
is there marked by an ink change and absent folio that is coincident with
a shift in spelling preferences (Blake and Thaisen 2004). One border artist
worked on the Ellesmere tranche containing the General Prologue and the prologues
and tales of the Knight, Miller, Reeve, and Cook. This tranche terminates
in a blank of two and a half pages to the end of this manuscript's quire 6
and has a separate style of paraphing. These tales make up the stable "fragment
A" in the traditional lineation system that goes back to the nineteenth-century
Chaucer Society prints of the pocm. A second artist is singlehandedly respon-
391 Boletín Millares Cario
2005-2006.24-25: 379-394
Jacob Thaisen Orthography, Codicology, and Textual Studies: The Cambridge University...
sible for all borders found in that tranche ofEllesmere which contains the prologues
and tales of the Man of Law, Wife of Bath, Friar, and Summoner; or
Man ofLaw followed by the likewise stable "fragment D". No such cohesion
characterises the allocation of borders to the total of three artists in any other
longer consecutive tranche of the Ellesmere text (Scott 1995: 92-94). It is,
therefore, possible to argue that fragments A and D may have existed in
separate booklets but nonetheless come to be transmitted together earIy in the
manuscript tradition, as ifChaucer himselfhad finished the poem up to the end
of the Summoner's Tale with the exception of him conclusively solving
the special problems surrounding the incomplete Cook's Tale. It is even
conceivable that the Gg first exemplar contained the complete text of the
poem. If so it was ordered only up to the end of the Summoner's Tale, leading
to greater uncertainty about its completeness than what the Cook's Tale had
already prompted and so to its eventual rejection; for it is unlikely that the Gg
makers would have initiated the copying of the poem at all if they had no
realistic expectation of obtaining its entire text. Scholars have devoted much
attention to the textual study of Canterbury Tales. But studying the linguistic
and (other) codicological aspects of Gg and other manuscripts can help us to
arrive at a more complete picture of their genetic relationships and of earIy
stages in the poem's composition and scribal transmission, for much recent
scholarship in those areas suggests that other manuscripts than Hengwrt and
Ellesmere bear testimony to those stages.
BIBLIOGRAPHY
BATESON, F.W. 1975. Could Chaucer SpeIl? Essays in Criticism 25: 2-24.
BLAKE, N.F. 1985. The Textual Tradition of the "Canterbury Tales ". London: E.
Amo1d.
BLAKE, N.F. and 1. THAISEN. 2004. Spelling's Sign{ficancefor Textual Studies. In C.
DOLLERup, ed., 2004: 93-107.
CALOWELL, R.A. 1944. The Scribe of the Chaucer MS, Cambridge University Library
GgA.27. Modem Language Quarterly 5: 33-44.
DA RoLO, O., ed. In press. A Digital Facsimile ofCambridge University Library, MS
Dd.4.24 of Chaucer 's "Canterbury Tales ". Leicester: Scho1arly Digital Editions.
[CD-ROM.]
DEAN, J.M. and C.K. ZACHER, eds. 1992. The Idea ofMedieval Literature: New Essays
on Chaucer and Medieval Culture in Honor of Donald R. Howard. Newark:
University of Delaware Press.
DOLLERup, c., ed. 2004. Worlds of Words: A Festschriftlor Ame Zettersten (=Nordic
Joumal oIEnglish Studies 3(1)).
DOYLE, A.I. and G.B. PACE. 1968. A New Chaucer Manuscript. Publications o( the
Modem Language Association oIAmerica 83( 1): 22-34.
Boletín Millares CarIo
2005-2006,24-25: 379-394
392
Jacob Thai"en Orthography, Codicology, and Textual Studies: The Cumbridge University...
FURNIVALL, F.J. 1868. A Tempormy Prelace to the Six-Text Edition al Chaucer's
"Canterbllry Tales": Part 1, Attempting to Show the True Order olthe Tales, and
the Days and Stages ofthe Pilgrimage, etc., etc. London: Trübner.
---, ed. 1869-77. A Six-Text Print olChaucer:~ "Canterbury Tales" in Parallel
Columnsf;'om thefollowing MSS: l. The Ellesmere. 2. The Hengwrt 154.3. The
Camhridge Univ. Lib¡: Gg.4.27. 4. The COIjJUS Christi Col!., Oxford. 5. The
Petworth. 6. The Lansdowne. (Chaucer Society, 1st ser.) London: Trübner.
GRAY, O. and E.G. STANLEY, eds. 1983. Middle English Studies Presented to Norman
Davis in Honour ofhis Seventieth Birthday. Oxford: Oxford University Press.
HAMMOND, E.P. 1908. Chaucer: A Bibliographical Manual. New York: Peter Smith.
HOROBIN, S.c.P. 2003. The Language of the Chaucer Tradition. (=Chaucer Studies
32). Cambridge: O.S. Brewer.
LAING, M. 1999. Confusion "wrs" Confounded: Litteral Substitution Sets in Early
Middlc English Writing Systems. Neuphilologische Mitteilungen lOO: 251-70.
--, ed. 1989. Middle English DialectolobTY: Essavs on Some PrincipIes and
Prohlems. Aberdeen: Aberdeen University Press.
MANLY, 1.M. and E. RICKERT, eds. 1940. The Text olthe "CanterbUlY Tales": Studied
on the Basis ol AII Known Manuscripts. 8 vols. Chicago: Chicago University
Press.
MCINTOSII, A, 1963. A New Approach to Middle English Oialectology. English Studies
44: 1-11.
__o - 1974. Towards an Inventory of Middle English Scribes. Neuphilologische
Mitteilungen 75: 602-24. Reprint in: M. LAING, ed. 1989: 32-45.
1975. Scribal Profiles from Middle English Texts. Neuphilologische
Mitteilzll1gen 76: 218-35. Reprint in: M. LAING, ed. 1989: 46-63.
MCINTOSlI, A., M.L. SAMUELS, and M. BENSKIN, with the assistance of M. LAING and
K. WILLlAMSON, eds. 1986. A Lingllistic Atlas olLate Mediaeval English. 4 vols.
Aberdeen: Aberdeen University Press: Edinburgh: Mercat Press.
MOSSLR, O.W. 1996. Witness Deseriptions. In: P. ROBINSON, ed. 1996: n.p. Reprint,
with slight revisions, in: P. ROHINSON, ed. 2004: n.p.
PARKlS, M.B. and R. BEADLE, eds. 1979-80. The Poetical vVorks olGeoff;'ey Chaucer:
A Facsimile ofCamhridge Universitv Lihrmy MS Gg.4.27. 3 vols. Norman, Okla.:
Pilgrim Books, in association with O.S. Brewer, Cambridge.
RAMSFY, R.V. 1994. The Manly-Rickert Text al the "Canterhury Tales". Lcwiston,
N.Y.: Edwin Mellen Press.
RICKERT, M. 1940. "lllumination". In: 1. MANLY and E. RICKERT, eds., 1940, 1: 561-605.
ROBINSON, P., ed. 1996. The Wife of Bath :\. Prologue on CD-ROM Cambridge:
Cambridge University Prcss. [CO-ROM.]
ROBINSON, P., ed. 2004. The Milla's Tale on CD-ROM Lcicester: Scholarly Digital
Editions. [CO-ROM.]
SAMUELS, M.L. 1983. Chaucer's Spelling. In: D. GRAY and E.G. STANLEY, eds., 1983:
17-37. Reprint in: J.J. SMITH, ed. 1988: 23-37.
SCIIMIDT, A.Y.C. 1975. Could Chaucer Spell? Essays in Criticism 25: 391-92.
393 Boletin Millares Carlo
2005-2006,24-25: 379-394
Jacob Thaisen Orthography, Codicology, and Textual Studies: The Cambridge University...
Seon, K.L. 1995. An Hours and Psalter by Two Ellesmere I1Iuminators. In: M.
STEVENS and D.H. WOODWARD, eds. 1995: 87-119.
--- 1996. Later Gothic Manuscripts 1390-1490. 2 vols. (=A Survey of
Manuscripts llluminated in the British Isles 6). London: Harvey Miller.
SEYMOUR, M.e. 1968. A Fifteenth-Century East Anglian Scribe. Medium Aevum 37:
166-73.
--- 1997. A Catalogue of Chaucer Manuscripts: Vol. 2: The Canterbury Tales.
Aldershot: Scolar Press.
---, ed. 1963. The Bodley Version of Mandeville s Travels, from Bodleian MS. E
Musaeo 116 with Paralle! Extraets from the Latin Text of British Museum MS.
Royal 13 E. IX (=Early English Text Society, O.S., No. 253.) London: Oxford
University Press.
SKEAT, W.W. 1899-1902. The Influence of Anglo-French Pronunciation on Modern
English. Transaetions ofthe Philologieal Society, 433-48.
SMITH, J.1., ed. 1988. The English ofChaucerand his Contemporaries: Essays by MI.
Samue!s and1.1. Smith. Aberdeen: Aberdeen University Press.
STEVENS, M. and D.H. WOODWARD, eds. 1995. The Ellesmere Chaucer: Essays in
Interpretation. San Marino, California: Huntington Library; Tokyo: Yushodo.
THAISEN, 1. and O. DA RoLO. In press. Linguistic Commentary. In: O. DA RoLO, ed. In
press: n.p.
Boletín Millares Carla
2005-2006,24-25: 379-394
394