The usual method of editing seventeenth-century poems has been to
choose the earliest printed edition as a copy-text and to depart from it only
when that printed text is manifestly corrupt. Variant manuscripts are
relegated to the notes or, if too radically variant, to the appendix. Thus
Grierson put his faith in the 1633 edition of Donne's poems; Rhodes
Dunlap virtually reprinted Carew's 1640 edition; F. E. Hutchinson based
his text on Herbert's 1633 edition, even though he had access to a
manuscript fair copy containing the licence to print and another manuscript
seen and corrected by the author; L. C. Martin used Herrick's
Hesperides (1648); and G. Moore Smith followed Lord
Herbert's Occasional Poems (1665) in preference to a
manuscript corrected by the author. Such reverence for printed texts is
hardly justified by the circumstances, for in most instances the authors were
dead before publication, and the character of copy that lay behind the book
is obscure.
We know little about how or where the publisher got his poems and less
about the textual history of the extant manuscripts. Only Helen Gardner has
seriously questioned the relative value of various witnesses for a poetical
text of the period. In her edition of Donne's The Divine
Poems
(1952), after an elaborate analysis and construction of manuscript trees, she
found reason to base her text on the 1633 edition, although she did not
follow it slavishly.
It is only sensible that, when a number of independent manuscripts
survive alongside a printed edition, an editor ought to take them all
seriously until he has reason to value them otherwise, and when analysis
suggests that one witness stands closest to an author's original, whether it
be print or manuscript, that witness should be the basis of the edition. Such
a procedure has difficulties when applied to poems.
For a single poem is usually too short to yield a sizeable number of variants
for analysis, publishers want a uniform texture of accidentals for all poems,
and editors do not have confidence in the old method of recension. The
doubts about traditional textual analysis reached a climax when George
Kane roundly challenged the method of recension, in his edition of the A
version of
Piers Plowman (1960), and there is a good deal of
skepticism about Vinton Dearing's large claims in his
Manual of
Textual Analysis (1959). Nevertheless the outlines of a more reliable
procedure are visible in W. W. Greg's
A Calculus of
Variants
(1927), combined with Dearing's early chapters, especially his remarks on
the ambiguity of distributional formulas.
As an example of what can be done with multiple texts of poems, I
shall suggest that at least two manuscripts of "The Witts," better known
under the title "A Sessions of the Poets," probably stand closer to the
author's papers than any of the printed editions, and consequently that a
critical editor should base his text of this poem on a manuscript and treat
the printed versions in the same way he would the other less authoritative
manuscripts. If my reasoning is correct, it has serious consequences for the
editorial method applied to Donne, Herbert, Carew, Suckling, Lord
Herbert, and perhaps other poets of the time. Need scholars use such
ingenuity to defend readings in printed texts against a number of
independent manuscripts? Should an editor base his text on the printed
version "for the sake of uniformity" rather than depart to a manuscript for
this or that poem?
"A sessions was held the other day" exists in seven manuscripts and
eight printed editions which appeared before 1700. The analysis of them is
in five steps: 1) brief description of each witness with relevant
bibliographical evidence, 2) the elimination of the obviously derivative
states, on textual and bibliographical evidence, 3) a distributional study of
the type two variants, 4) a directional study of the type one and type two
variants, 5) construction of a genealogical tree. The reason for such a
rigorous, step by step method, first brought forward by Archibald Hill in
"Postulates for the Distributional Study of Texts," Studies in
Bibliography, III (1950), 63-96, is partly to avoid confusion but
more
to avoid bringing literary assumptions into the study before bibliographical,
statistical and other forms of objective evidence have been exhausted.
Otherwise we may beg the question by assuming that we know what the
corruptions are before we have established the lines
of transmission.[1]
