The "Envoy to Alison," a ballade of 27 lines, is extant in two
fifteenth-century manuscripts and in seven of the black-letter
editions of Chaucer's works. I shall attempt to work out the
relationship among these nine texts, trying to limit the problem by
studying the distributional and geneological evidence that is
available.[1]
The two manuscripts containing "Alison" are Fairfax 16 and
Tanner 346, both in the Bodleian Library. The handwriting of both
has been dated as being of the fifteenth century, and the date 1450
is written on the flyleaf of Fairfax 16 in an apparently
contemporary hand.[2] In Fairfax 16
"Alison" is quite legible. In Tanner 346 a number of graphemes in
"Alison" are badly faded or blurred; these were illegible in 1898,
when the oldest extant transcription, that of Vollmer, was
made.[3]
Black-letter editions of Chaucer which contain "Alison" are
Thynne's 1532 and 1542 editions, his undated edition, Stow's
edition of 1561, and Speght's three editions—1598, 1602, and
1687.
I shall refer hereafter to the nine texts of "Alison" by the
following abbreviations:
- Fairfax 16: F
- Tanner 346: T
- Thynne 1532: Th1
- Thynne, 1542: Th2
- Thynne, no date: Th3
- Stow: St
- Speght, 1598: Sp1
- Speght, 1602: Sp2
- Speght, 1687: Sp3
In studying the relation of these texts to each other I shall
first consider the distributional evidence derived from a
compilation of all variants. This compilation will make possible a
statistical statement of the
resemblances of the texts to one another. As an accessory to such
statistics I shall where possible use the method of looking for
runs of identical readings in pairs of texts; where two texts show
such a run of considerable length, the likelihood of kinship is
increased, for a number of consecutive agreements is much less
likely to be due to mere chance than an equal number of
non-consecutive ones.
Distributional evidence may be expected at least to limit the
number of possible trees. When it has been exhausted, I shall turn
to geneological evidence, if it can be found, for further aid.
In the texts of "Alison" the number of substantive variants
(i. e., those involving meaning) is exceedingly small. It is
therefore necessary to consider also minute differences in spelling
and orthography. In order to assign proper value to variants of
different degrees of minuteness, I arranged them according to the
following classification. Beside each type I have indicated the
number of variants of that type that occur.
I. Substantive variants (7)
- A. Lexically different words (6)
- B. Add-omissions (1)
II. Non-substantive variants (242)
- A. Non-significant variation in morphemes (3) (All of these
involve actual difference in pronunciation, but without lexical
difference.)
- B. Presence or absence of final —e
(50)
- C. Significant difference in graphemes without morphemic
difference (91) (These do not include words differing only in final
—e.)
- D. Difference in allographs of the same grapheme, viz.,
v:u, þ:th, lower-case letter: capital letter, and
ff:F (43)
- E. Difference in suprasegmental phonemes, viz.,
punctuation marks: slant, period, and comma. (55)
In compiling the specimens of the several types I made no
attempt to avoid overlapping. Therefore a word whose readings
contain differences in final -e, in i and
y,
and in þ and th appears three times. Ordinarily, however,
each word appears only once or twice. The illegible graphemes in
Tanner were recorded as blanks. The occurrence of such a blank was
not considered a variant in itself.
The relation of the seven black-letter texts to each other was
considered first. Within this group there are only three
substantive variants. The first of these is in line 2, where Th3
and St read thus and other texts read thou. The
minority reading makes little sense and could hardly have occurred
without direct influence of one of the two dissenting texts on the
other. Since St is the later, this variant is evidence that St was
derived from Th3. The fact that the usual reading
thou
reappears in Sp1 is of no evidential value, for Speght not only
might have used one of the texts earlier than St to correct the
suspicious-looking
thus but also had the verb
haste
calling for
thou as subject.
The second substantive variant is in line 6, where St and the
three Speghts have abiyng (spelled abying in
Sp2 and
Sp3) and the three earlier texts have various spellings of
obeying.[4] This seems
to
indicate that Sp1 is derived from St.
The third substantive variant is brought in Sp1 against
through in all other texts. This has no value whatever.
All that can be extracted from these three variants, then, is
their evidence that St is from Th3 and that Sp1 is from St.
Turning to the non-substantive variants, we find that in Class
IIA there are only three and that these indicate merely a gradual
modernizing of spelling in successive editions.
The non-substantive variants of the other four classes are
numerous enough to give significant statistical results.
With regard to all four of these classes, each of the Speght
editions resembles its immediate chronological predecessor more
than it resembles any previous edition.[5] The substantive variant
obeying:
abiyng, indicating derivation of Sp1 from St, does not
contradict these findings. We may place the four latest texts,
then, in straight-line derivational arrangement in order of
date.
When agreements among the three Thynne editions are examined, it
is found that Th3 agrees with Th1 a total of 224 times (out of a
possible 239) and with Th2 only 201 times. This indicates that Th2
and Th3 must have been derived independently from Th1. This
relationship is indicated not only by the total for the four
classes but also by the figures for each of the individual classes,
except for IIE, which consists of differences in punctuation.
However, if slants in Th1 are equated with commas in Th3, these two
texts are found to agree perfectly; that is, they are punctuated in
exactly the same places, whereas between Th2 and Th3 this
correspondence is not quite perfect.
A check on the statement that Th3 resembles Th1 more than it
resembles Th2 was made by looking for long continuous passages in
which two of these texts were identical. It was found that Th1 and
Th3 are exactly alike for 71 consecutive words, including 29
variants (from haste, line 2, through O, line
12).
Also there is a run of 43 consecutive
words, including 12 variants, in these two texts from
for,
line 14, through
is, line 19. When Th2 and Th3 were
compared
thus, however, no run longer than 17 words, including 3 variants,
was found.
The probability of a derivational relationship between two texts
is enormously increased by the occurrence of such runs as occur in
Th1 and Th3. The principle involved may be illustrated simply. If
a given variant occurs in three forms throughout a set of texts,
the chance that a certain two texts will agree thereon is one in
three. If the next variant occurs in four forms, the chance that
the two texts will agree on it is one in four. The chance that they
will agree on both variants is one in three times four, or one in
twelve. Thus, in a run, the chance of continuous agreement is one
in a number equal to the product of the number of forms for each
variant in the run. This product reaches astronomical size in a run
of considerable length, and the presence of such a run is very
strong evidence of derivational relationship between the two
texts.
Therefore the occurrence of the runs described removes all doubt
that Th3 is more closely related to Th1 than to Th2. In the face of
such evidence it is impossible for Th2 to be intermediary between
Th1 and Th3. This instance demonstrates clearly the value of
looking for runs as a method in distributional study. Where
extensive runs are found, they add substantially to the evidence
from statistics of number of agreements and may resolve cases
undecided by such statistics.
Looking for runs is most likely to prove valuable when the
student is dealing with minute variants, which occur consecutively
or close together, rather than substantive ones, which are usually
fairly far apart. The frequency of minute variants of course makes
it more likely that runs will occur and that significant ones may
be detected in passages only a few lines in length.
The question may be raised as to whether Th2 might have been
derived from T or F rather than from Th1. But the table of
agreements gives no support to such a conclusion; Th2 is more like
Th1 than it is like any other text.
We may, then, arrange the Thynne texts thus:[6]
To complete the tree for the black-letters, we now need to know
from what text Stow was derived.
It will be recalled that the sharing of the reading thus
in line 2 by St and Th3 was taken as evidence that St was derived
from Th3. Let us see how this evidence compares with the testimony
of the non-substantive variants. With which of the Thynne editions
does St agree most often?
On the three IIA variants the three Thynne texts agree
perfectly, so that those variants are worthless.
For IIB-IIE variants, agreements of Stow with the Thynne texts
are as follows:
Class
|
Total
|
Th1-St
|
Th2-St
|
Th3-St
|
IIB |
50 |
44 |
41 |
44 |
IIC |
91 |
45 |
51 |
49 |
IID |
43 |
40 |
40 |
42 |
IIE |
55 |
47 |
49 |
51 |
|
---- |
---- |
---- |
---- |
|
239 |
176 |
181 |
186 |
These figures show that St agrees most often with Th3; thus they
corroborate the evidence of the thus: thou variant. When the
method of looking for runs of consecutive identical words was
applied to these four texts, no significant results emerged.
Therefore, we may conclude that St is from Th3,[7] and the complete tree for the
black-letter editions is as follows:
The major problem in establishing the text of "Alison" is to
determine the relationships among the three oldest versions of the
poem — F, T, and Th1.
Some important evidence emerges when the arrangement of the poem
with regard to what precedes it and what follows it in each of the
three texts is observed. In F it is preceded by "The Book of the
Duchess," from which it is separated only by the line "Explicit the
boke of the Duchesse," and at the end of "Alison" is no "Explicit,"
as the heading of the next poem, "Balade upon the Chaunse of the
Dyse," follows immediately.
In T "Alison" is immediately preceded by "The Cuckoo and the
Nightingale," with only the line "Explicit" separating the two
poems. At the end of "Alison" is the line "Explicit þe Cuck. &
þe Nighting." in a hand different from the rest of the text, and
the remainder of the page is left blank. On the next page begins
"The Book of the Duchess" (Hammond, op. cit., p.
337).
The Thynne text, like T, is preceded by "The Cuckoo and the
Nightingale" with only "Explicit" between the poems. Following
"Alison" is "Explicit" and the heading and beginning of the next
poem, "Scogan unto the lordes . . ."
It should also be noted that between the third stanza and the
fourth is the heading "The lenvoye" in T and "¶ Lenuoye." in
Th1, whereas in F there is no heading at this point.
It appears that since both T and Th1 are preceded by the same
poem and share the "Lenvoy" heading there must be some sort of
kinship between them.
The variant readings in the three texts must now be examined for
distributional evidence.
Within these texts there are only three substantive variants:
Line
|
F
|
T
|
Th1
|
1 |
foole |
foole |
foule[8]
|
12 |
me |
ne |
ne |
24 |
and of goodnesse |
and goodnes |
and goodnesse |
In these three variants, F and T agree against Th1 once, and T
and Th1 agree against F twice. So the odds are two to one in favor
of a closer relationship between T and Th1 than between either of
them and F. This conclusion is consistent with the testimony of the
order of material and the envoy headings, as mentioned above.
Agreements among the three texts with regard to non-substantive
variants are summarized in the following figures:[9]
Class
|
Total
|
F & T
|
F & Th1
|
T & Th1
|
IIA |
3 |
0 |
2 |
0 |
IIB |
39 |
8 |
17 |
13 |
IIC |
71 |
19 |
26 |
13 |
IID |
40 |
7 |
27 |
4 |
IIE |
28 |
5 |
2 |
20 |
|
---- |
---- |
---- |
---- |
|
181 |
39 |
74 |
50 |
These figures agree with the other evidence in showing a
divergence between F and T, but they contradict the evidence for
kinship between T and Th1, for T and Th1 agree only 50 times,
whereas F and Th1 agree 74 times. Further, this similarity between
F and Th1 is attested by four out of five of the classes of small
variants.
However, the value of these statistics should be carefully
weighed before they are allowed to counterbalance the evidence of
the connection with "The Cuckoo and the Nightingale," the envoy
headings, and the substantive variants.
The non-substantive variants on which F and Th1 agree can be
explained as due to coincidence and do not necessarily indicate
kinship between these two texts. There are 74 such agreements,
constituting only 41 per cent of the total of 181 non-substantive
variants in the three texts. Further, in the IID group, 19 of the
27 agreements of F and Th1 against T are due to the use of the
thorn (þthe three texts. Further, in the IID group, 19 of the 27
agreements of F and Th1 against T are due to the use of the thorn
(þ) in T; and three such agreements in IIC variants are due to
the use of the yogh (ȝ) in T. But fifteenth-century printers did
not use þ and ȝ, and Th1 would not have shown them even if its
printer had been copying directly from T. These 22 agreements
involving þ and ȝ are therefore valueless, and if they are
disregarded there are only 52 agreements of F and Th1 in 159
variants, or 33 per cent.
These remaining agreements might be of some value if they
revealed the two texts sharing even a few unusual spellings. But
the NED shows all the shared spellings to be commonplace
ones, occurring frequently in the texts cited by the dictionary.
That F and Th1 should coincide in about one out of three common
spellings is no more than could be expected from chance.[10]
In these three texts no help is to be had from runs of identical
readings, as there are no such runs of appreciable length.
It seems best, then, to reject the evidence of the small
variants and, on the basis of the larger considerations, to call T
and Th1 a subfamily apart from F. A tentative tree can be drawn
thus:
Can this tree be simplified by identifying either T or Th1 as x?
To answer this we must return to the substantive variants. In these
F and Th1 never agree, and T agrees with F once and with Th1 twice.
Th1 has one unique reading (foule), F has two unique
readings (me and of goodnesse), and T has no
unique
readings. Therefore only T can be placed as an intermediary. T,
then, could be x in the above tree, but Th1 could not, for if Th1
were x, it would be difficult to explain the appearance of
foole in F and T, since Th1 has foule. But no
such
difficulty arises from making T the ancestor of Th1, and simplicity
is gained. The tree can therefore be changed to the following,
which, however, is still tentative:
We now have a two-forked tree and a state of affairs in which
the text of O cannot satisfactorily be reconstructed. Obviously,
for any reading on which F and T disagree, each text has equal
authority with the other. It would be desirable, in order to
resolve this conflict and achieve greater simplicity, to make
either F or T the archetype if possible.
In the light of the evidence of the substantive variants, the
envoy headings, and the relation to "The Cuckoo and the
Nightingale," it is possible for either T or F to be archetype.
This possibility is not disturbed by the fact that only T can be
intermediary. These two trees are equally consistent with the
distributional evidence:
Further, since the tentative tree with x was based on the
kinship of T and Th1 and the necessity that T be intermediary, a
third alternative is also theoretically possible:
Thus, with T as intermediary, any one of the three texts could
be the archetype as far as the evidence so far shown is
concerned.
This is as far as distributional evidence can take us.
Geneological evidence should next be sought, but in the texts of
"Alison" such evidence is found to be so meager that it is of no
value.
If further evidence is to be produced, it must come from a
comparative study of the entire Fairfax and Tanner manuscripts, not
from examination of "Alison" alone.
This study of the texts of "Alison" has not produced evidence
sufficient to establish an authoritative text. However, it has
partially solved the problem of the relationship among the nine
versions, and these results may be stated:
- 1. The relationship of the black-letter texts may be
established by means of distributional evidence.
- 2. T and Th1 are more closely related to each other than
either is related to F; therefore the probable relation among the
three texts is not simple radiation from an archetype.
- 3. Any tree not using a hypothetical manuscript must show T
as intermediary.
Of greater value than these specific results concerning "Alison"
are the conclusions as to methods of textual study of manuscripts
that have emerged. These are as follows:
- 1. When the texts being studied show so few substantive
variants that it is necessary to attach great importance to
non-substantive variants, it is useful to classify the variants
according to minuteness so that appropriate evidential value may be
assigned to them. The classification used in the present study is
a workable one.
- 2. As an auxiliary to statistics of agreements, a
consideration of runs of identical readings is valuable.
Consecutive agreements are of more significance than scattered
ones, and the presence of identical runs of considerable length is
strong evidence of kinship between texts. The
value of such runs was demonstrated in the case of the Thynne
texts, where the derivation of the Th3 directly from Th1 was firmly
established by the discovery of long runs linking these two
texts.
- 3. The need for studying the entire texts of the Fairfax and
Tanner manuscripts, in order to establish decisively the text of a
short poem contained in those manuscripts, suggests an important
principle for editors. The case of "Alison" demonstrates the
paucity of variation that is frequently encountered in short texts,
even when they exist in a considerable number of versions. When the
piece under consideration is surrounded by material which may also
be compared and studied, the evidence of this material should be
used. Minor poems may not be edited safely in isolation from the
larger units of text that contain them.