"The Fift Booke" of Richard Hooker's treatise, Of the Lawes
of Ecclesiasticall Politie, is a folio in sixes, collating
A8
B-Z6 2A4. Printed by John Windet on
commission from Edwin
Sandys for its author, it was published in late 1597.[1] Since 1888, when the Church and
Paget
revision of John Keble's edition of 1836—itself frequently
reprinted—was published, it has been known that Bodleian MS Add.
C.165 served as printer's copy for the 1597 edition.[2] The printer's markings in the
margin and
within the text of this manuscript were described by Percy Simpson in his
work on proofreading of early printed books.[3] The scribe, Benjamin Pullen, a
clerk in
the employ of John Churchman, Hooker's father-in-law, was identified by
C. J. Sisson in 1938.[4] But the Pullen
manuscript has never been systematically compared with the folio Windet
produced
from it. In particular, the volume's composition, imposition, and presswork
have never been described in detail, perhaps because of the complexity of
the process itself, perhaps because only recently has it been recognized that
many Elizabethan books, especially folios, were set by formes, not
seriatim, and that copy necessarily had to be cast off to make this
possible.[5] With the Pullen
manuscript before us, however, it is possible for the first time to view the
process of casting off copy from over the compositor's own shoulder, rather
than—as a bibliographer after the fact—inferring what
happened from
the evidence of the printed book alone. As Simpson commented, 'Tis noble
manuscript, with its literary
value, its accuracy and completeness of text, and the range and variety of
evidence which it yields to the bibliographer, must be unrivalled'
(2.23).
Casting off copy is simple enough to describe in the abstract: it is the
estimating of how much manuscript copy a given printed page requires so
that intervening pages can be 'cast off'—that is, marked but not set
in
type—to enable the compositor to set copy for a quired folio from
the
inside of a quire out, thereby conserving type and matching composition to
presswork. In practice, however, the process can be dauntingly complex to
reconstruct, as evidenced by Charlton Hinman's work on the First
Folio.
The Pullen manuscript consists of 225 folios; four of the original
leaves are missing, and two of these have been replaced; of these, the first
(f. 140) seems to have been used for setting type, the second (f. 228)
not.[6] Pullen, a professional scribe,
wrote in an eminently legible secretary hand for Hooker's text and an
equally clear italic for marginal notes and for quotations within the text. No
compositor could have wished for clearer copy, and in many respects,
Windet's folio is simply a reprint of that manuscript. However, Hooker
himself worked over Pullen's transcript with great care, adding marginal
notes, interlining additions to or deleting material from the text, correcting
both the spelling and pointing of his scribe's transcription, and underlining
words, phrases, or whole passages as an instruction to the printer to set
these in italics.[7] Where Hooker's
intervention was substantial, casting off was
obviously more difficult. The text contains a number of marginal notes, and
while Pullen allows for the lengthier ones by cut-outs into the text, others
added marginally by Hooker made accurate estimates of the amount of type
required more difficult.
For purposes of exposition, I will first outline separately the various
stages of the process, the evidence for each that the manuscript provides,
and the logic by which that evidence has been interpreted.[8] I will then analyze
Opposite:
Folio 60r, Bodleian MS. Add. C.165 (reduced). The
cancelled
page break at line 14 ('expresseth, | declaring'), 78/H6,
is a casting-off
break, marking the point from which setting H3v ('H6')
began after
H3r had been cast off. The actual break, corresponding to
H3r|3v in 1597 ('was | they'), 'H6',
marks the point the
compositor reached when the page lengths within sheet 3 of quire H had
been adjusted. The second series of numbers, '38 . . .
<42><8>4', tallies the number of lines the compositor had
set on his way back to H3v. These begin again at line 16,
below
'God,', where '9' is visible, '13 . . . 17 . . . 21 . . . 25', are not.
Reproduced by permission of the Bodleian Library.
sample gatherings to show how, in the event, these processes interact in
actual composition. Finally, I will draw what inferences seem appropriate
for bibliographers (or editors) who would study the order of composition
of a text known or suspected to have been set from cast-off copy.
The principal manuscript evidence that copy was cast off is a series
of marks in its right-hand margins.[9]
These are in the form of a dash, drawn between the lines, with the number
20 or 40 following, counting the number of lines of manuscript copy cast
off. Not all folios are so marked, for not every page was cast off. In the
analysis that follows, I begin by assuming that if a page was marked for
casting off, it was in fact cast off in actual composition unless unambiguous
evidence demonstrates the contrary. There is a division of copy between the
compositors in the manuscript between ff. 1-136 (= $B-Q1) and ff.
137-227 (= Q2-$2A).[10] I will be
discussing only the work of the first compositor (Compositor A), whose
marks in ff. 1-136 are more revealing of the process itself. The following
table indicates both the frequency and distribution of casting-off marks
within the folios of the MS set in $B-P in 1597:
Because the normal sequence of printing for a folio in sixes was from
the inside of the quire out, and because printing could not begin until the
conjugate pages of the inner forme of sheet 3
(3v.4r) were in type
and imposed, the compositor typically cast off the earlier pages in the forme
(1r-3r) twice as frequently as the
remaining pages (3v-6v):
61 out of 83 pages cast off, or 73%. But in the thirteen gatherings
Compositor A set ($B is a special case), every page of the twelve was cast
off at least once.
The pages in the Pullen manuscript corresponding to quire B, it will
be noticed, do not bear the characteristic markings. As the initial gathering
to be set (quire A was reserved for the preliminaries, whose manuscript
does not survive), the compositor—or more likely, the master, John
Windet—was obliged to work out in setting it a correspondence
between
his copy and the text as set to measure for use throughout the volume.
Moxon gives three methods for calculating the ratio between copy and type:
a) line against line (that is, how many lines of copy are needed per single
printed line); b) measuring with a pair of compasses the number of whole
lines of copy necessary to set a given number of whole lines of print; and
c) adding the number of characters in the copy and dividing by the known
number of characters on
Casting-off Marks for Signatures B-P in Bodleian MS Add.
C.165
| page: |
1r
|
1v
|
2r
|
2v
|
3r
|
3v
|
4r
|
4v
|
5r
|
5v
|
6r
|
6v
|
total |
| $B |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
0[*]
|
| $C |
-- |
20/40 |
20/40 |
20/40 |
20/40 |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
4 |
| $D |
20/40 |
20/40 |
20/40 |
20/40 |
20/[40] |
20/40 |
20/40 |
20/40 |
20/40 |
-- |
-- |
-- |
9 |
| $E |
20/[40] |
20/40 |
20/40 |
[2]0/40 |
[20/40] |
-- |
-- |
-- |
[20?] |
-- |
-- |
20/40 |
6 |
| $F |
[20]/40 |
20/40 |
20/40 |
20/40 |
[20]/40 |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
5 |
| $G |
20/40 |
20/40 |
[20/40] |
20/[40] |
20/40 |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
5 |
| $H |
[20/40] |
20/40 |
20/40 |
-- |
20/40 |
20/40 |
20/40 |
20/40 |
-- |
-- |
-- |
-- |
7 |
| $I |
20/40 |
20/40 |
20/[40?] |
20/40 |
20/40 |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
5 |
| $K |
20/40 |
20/40 |
20/[40?] |
20/40 |
20/40 |
-- |
-- |
-- |
20/40 |
-- |
-- |
-- |
6 |
| $L |
20/40 |
20/40 |
20/40 |
20/40 |
-- |
20/40 |
20/-- |
-- |
20/40 |
20/40 |
-- |
-- |
8 |
| $M |
20/40 |
20/40 |
20/40 |
20/40 |
20/40 |
-- |
-- |
-- |
-- |
-- |
-- |
-- |
5 |
| $N |
[20]/40 |
20/40 |
20/40 |
-- |
20/[40] |
20/40 |
20/40 |
20/--? |
-- |
-- |
-- |
-- |
7 |
| $O |
20/[40] |
20/[40] |
20/40 |
20/40 |
20/40 |
20/40 |
20/40 |
-- |
-- |
-- |
20/40 |
-- |
8 |
| $P |
20/40 |
20/40 |
20/40 |
20/40 |
[20]/40 |
-- |
-- |
-- |
20/40 |
20/40 |
[20]/40 |
-- |
8 |
| total: |
12 |
13 |
13 |
11 |
12 |
5 |
5 |
3 |
4 |
2 |
2 |
1 |
83 |
a page of any given format.
[11] Method
b seems likeliest. In setting the 24 lines of type appearing on
B1
r in the
1597 folio, and ignoring the 10 set to short measure opposite the initial
factotum, the compositor would have reckoned that 18 lines of manuscript
supplied him 14 whole lines of type. To complete a 47-line page, the
standard with which he began the volume, he would need 47 x 18/14 =
60.6 lines of copy. Having calculated a working ratio between his copy and
his type, he verified it by setting a full page (B1
v) in
which 60 lines of
manuscript copy were set in 47 lines of type. He could now count ahead
with some confidence by units of 60 whole lines, ignoring part lines, in
order to estimate how much copy would be needed per printed page. The
divisions marked at 20 and 40 lines within a given page-unit of copy
supplied convenient subdivisions of the basic page unit of 60 manuscript
lines, made line-counting easier, and told him where he
was in the midst of composing that page.
Although the corresponding folios in Pullen are not so marked, there
is evidence that pages 3-7 of quire B were also cast off. If we count ahead
in the manuscript from the end of B1v,
B4v is marked to begin
exactly 300 manuscript lines ahead. In addition, the break between
B4r
and 4v is marked at '|worke' (f. 5v),
flush with the left-hand
margin. The mark is the usual one: an extended bracket that looks like an
elongated 'L' with the vertical stroke crossed, |thus. This
break, labelled 'B8' in the left-hand margin, is marked three lines above in
the manuscript at 'of | true' (cf. 1597, B4r.-3; that is, 3
lines from the
bottom of the page). The copy bracketed by the two breaks as marked in
the manuscript is equivalent to exactly two printed lines in the 1597 text.
Although the setting of quire B was to a degree anomalous and to analyze
its composition here is to anticipate certain points in the argument presented
later, it seems best to clarify the problems
its composition presents before proceeding.
After the compositor had calculated the ratio of manuscript copy to
type in setting B1r, verified it in 1v, and
cast off 2r-4r
(without marking individual pages within this block), he began
4v at a
break marked flush with the left-hand margin of f. 5v
('|worke', line
11), the beginning of the line. Once B4v was in type, its
forme-mate
3r was set, beginning at a punctuation break, mid-line:
'religion; | that'
(f. 3v.9). B3r ends mid-word,
'possibi|litie' (f. 4r.29),
evidence that 3r was set to this point in the
manuscript, not
from it. It is also noticeably loose in its type-setting at the
bottom of the page, as the compositor sensed that he needed to stretch his
copy to fill out the 47 lines allotted to his page. The outer forme of sheet
3 was now ready for the press. The inner forme
(3v.4r) was then
set, in that order: the page break at 3v|4r
is not noted marginally
but is bracketed in the text.
Turning to sheet 2, the compositor began B5r at '|
be
exemplified', the beginning of a line in the manuscript (f.
6v.11) and
continued through 5v: both pages break in the middle of
a line. To
supply forme-mates, B2r and 2v were set,
in that order, for the
break at 2r|2v is in the middle of a word
('a|dorned', f. 2
r.10; cf. 1597, B2
r.1).
In imposition, one line of
type was taken from B2
r and given to 2
v.
That actual page break
in 1597 falls at 'Romans | had' (B2
r.-1), marked in
pencil; there is a
second mark, in ink, at line 29, 'conjecture | before' (cf. 1597,
B2
v.1), exactly one line of type beyond the actual break
in 1597. The
ink mark represents the point to which the compositor set
B2
r, using
48 lines of type; the pencil mark, where the break fell when the 47-line
standard was applied.
Finally, B6r and 6v were set, in
that order, to supply
forme-mates for 1v and 1r, which had
been left standing. The
evidence is in the manuscript markings: both pages begin in the middle of
a line; that is, from where the previous page had ended, not from a point
to which the compositor had counted when he cast off copy.
If we put the subsequences together into a single skein (I find no
evidence of two compositors at work throughout the quire, although a
second workman certainly assisted in the imposition and probably set
1r and 1v), we have:
1r→1v→(4v
→3r→3v→4r)
→(5r→5v→2r
→2v)→6r→6v.
If we
examine the pages as printed, we find by optical inspection that the outer
forme in the inner two sheets was machined first,[12] confirming the compositorial
sequence
above. Three headlines were set for the verso (X, Y, Z) and two (x, y) for
the recto; these pair up as follows:
| conjugate pages |
forme |
headlines |
| 1. 3r.4v
|
outer 3 |
y.Y (missing 'of') |
| 2. 4r.3v
|
inner 3 |
x.X |
| 3. 2r.5v
|
outer 2 |
y.Y (has 'of') |
| 4. 5r.2v
|
inner 2 |
x.X |
| 5. 6r.1v
|
inner 1 |
x.Z |
| 6. 1r.6v
|
outer 1 |
--.Y |
In interpreting page breaks as marked within the manuscript text, we
have to distinguish between those made when copy was being cast off
initially and those made subsequently, in the course of actual composition
or, later, of imposition. If all went well, these would coincide; but if there
were miscalculations, or if the copy was especially irregular, a single page
(or line) might have as many as two cancelled breaks in addition to the
actual break.[13] Two axioms,
therefore, are useful in distinguishing between the two:
(1) If the break falls within a line of the manuscript,
and
not at a punctuation break, it marks the point to
which actual composition had come and stopped. Similarly, if the
break is marked within a word, the same logic holds: no
compositor is going to estimate to an arbitrary point within a line, much
less within a word, of his copy. Conversely, breaks marked at the
beginning of a line in the left-hand margin, or at
punctuation breaks within a
line, are normally casting-off marks and indicate the point
from
which composition of a given page would have begun, after it had
been cast off.
(2) If a page break is marked twice within the manuscript and if one
of these is cancelled and the other corresponds to the break as it appears in
the printed text, the first will usually (but not invariably) be the casting-off
mark made prior to setting type, the second, the one made in the course of
actual composition to fit copy in when the page ahead is already in type, or
later at imposition, as page lengths are juggled to achieve pages of uniform
length on facing pages.
Of the two, the first is more reliable. Applying it, together with the 20/40
marks in the margin, makes it possible to identify which pages were cast
off and to compute how many lines of copy were cast off for each
page.
To return to quire B, the ratio of 60 lines of copy to 47 lines of type
holds up quite well. B1v-6v (that is, the
full size pages) are each
set to a 47-line standard, and 661.5 lines of manuscript copy, or an average
of 60.14 lines of copy per printed page, were set in type. After the first two
pages of quire C, the page-unit drops to 46 lines and remains
there—except for variations caused by misestimating copy being cast
off—for the rest of the volume. Such variations, however, may
require
as many as 48 lines per page (e.g., on D3r,
D3v, and D5r,
within one gathering) or as few as 45 (N3v,
N4r), or even 44
(O1v). The corresponding number of lines of copy, which
average 60
for pages 3-7 in quire B (2r-4r) and quire
C (3 x 60, 1 x 59),
begins to drop to 59 in quire D (1 x 62, 1 x 60, 3 x 59, 1 x 58, 2 x 57, 1
x 53), and to 57-58 in quire F (1 x 56, 3 x 57, 1 x 58) and subsequent
gatherings. Naturally, when the copy demanded it, there were
variations in the number of pages the compositor elected to cast off, as well
as the number of lines of copy he allotted each page. In quire H, for
example, which has a number of marginal notes cut into the text (its
composition is analyzed below), the counts run as follows:
H1r:62
lines, 1v:58, 2r:58, 2v
(not cast off), 3r:57, 3v:55,
4r:57, 4v:56. The next gathering,
however, was quite regular:
I1r-3r were cast off in page-units of 58,
59, 58, 58, and 59 lines.
As the compositor became accustomed to Pullen's copy, he was able to
vary, ad hoc, the two halves of his basic ratio: the number of lines of
printed text per page (46 ± 2) and the number of lines of manuscript
copy cast off (59 ± 3).[14]
Similarly, the number of pages cast off within the 13 gatherings for
which we have unambiguous markings varies from 4 to 9 (median, 6).
Within a given sheet, the inner forme was composed first in 32 out of 42
sheets,[15] and in the 14 quires
(including B), the normal sequence, 3→2→1, from the inside
out, was followed in 10 cases; in two, the order was 2→3→1, and
in two it was 3→1→2. The table below summarizes the presswork
in quires B-P, based on an analysis of the MS markings and checked by
optical inspection of the printed sheets.
| order through press |
1st |
2nd |
3rd |
4th |
5th |
6th |
| $B |
outer 3 |
inner 3 |
outer 2 |
inner 2 |
inner 1 |
outer 1 |
| $C |
inner 3 |
outer 3 |
inner 2 |
outer 2 |
inner 1 |
outer 1 |
| $D |
outer 2 |
inner 2 |
inner 3 |
outer 3 |
inner 1 |
outer 1 |
| $E |
inner 3 |
outer 3 |
inner 2 |
outer 2 |
inner 1 |
outer 1 |
| $F |
inner 3 |
outer 3 |
inner 2 |
outer 2 |
inner 1 |
outer 1 |
| $G |
inner 3 |
outer 3 |
inner 2 |
outer 2 |
inner 1 |
outer 1 |
| $H |
inner 3 |
outer 3 |
inner 2 |
outer 2 |
inner 1 |
outer 1 |
| $I |
inner 3 |
outer 3 |
inner 2 |
outer 2 |
inner 1 |
outer 1 |
| $K |
inner 3 |
outer 3 |
outer 2 |
inner 2 |
inner 1 |
outer 1 |
| $L |
outer 3 |
inner 3 |
inner 1 |
outer 1 |
outer 2 |
inner 2 |
| $M |
inner 3 |
outer 3 |
inner 2 |
outer 2 |
inner 1 |
outer 1 |
| $N |
inner 2 |
outer 2 |
outer 3 |
inner 3 |
inner 1 |
outer 1 |
| $O |
outer 3 |
inner 3 |
inner 2 |
outer 2 |
outer 1 |
inner 1 |
| $P |
inner 3 |
outer 3 |
outer 1 |
inner 1 |
outer 2 |
inner 2 |
From the point of view of the compositor setting from cast-off copy,
the reason why the inner forme would normally be composed first is that
the maximum number of pages remains
(1r-3r) in which to
apportion the intervening cast-off copy. Were he to begin with the outer
forme of sheet 3 (that is, with 3r), only four pages would
remain
(1r-2v), and if the copy was irregular or
if a miscalculation took
place, the margin of error was one page less. The same logic holds,
obviously, were he to set sheet 2 before sheet 3, for once 2 is in type, only
two pages are available for cast-off copy in sheet 1 (1r,
1v).
To take a comparatively simple example, the copy for quire E
contains no marginal notes cut into the text by Pullen or added by Hooker.
The compositor cast off the first five pages (59 lines X 4 + 58 X 1 = 294
MS lines). He began setting E3v at '| no ordinarie
meane', at the
left-hand margin of folio 32r.13 (that is, line 13). The
'—20' and
'—40' are not visible on the preceding folio because the right-hand
margin is so tightly bound that the 20 and 40 are in the gutter; the dashes,
however, are visible. The break at 'no ordinarie meane' is not marked: the
compositor simply counted to this point and began setting copy. He could
not have begun setting at E4r, for the page break at
3v|4r falls
in the middle of the word 'nourish|ment' (f. 32v.33), so
that in this
instance 3v must have been set before 4r.
Thus, the inner forme of
sheet 3 was set in normal order:
3v→4r. To set the outer forme,
he needed only start E4v where 4r left
off—at the '|ment' of 'nourish|ment'—and to set
3r he returned
to the cast-off break on f. 31r.30, which is marked in the
left-hand
margin and at a punctuation break: '| And because'. In the printed volume,
the telltale signs of cast-off copy are variations in the spacing of the type
on the page, especially towards the bottom, as the compositor squeezes in
a
given amount of copy into the space previously allotted, or, alternatively,
stretches it out to get back to a page already in type. In the 1597 folio, the
bottom of E3
r is noticeably loose in its type-setting. The
entire sheet,
then, was set in the sequence
3
v→4
r→4
v→3
r (or
3
r→4
v, for there is no way to
determine priority between
E4
v and 3
r, nor is it significant: once we
know that 4
v was set
after 4
r, we know that inner 3 was set before outer
3).
A slightly more complicated example illustrates the flexibility of the
basic process. In quire L, 1r, 2r, and
2v were cast off in units
of 58 manuscript lines (1v, which had a Hooker insertion,
was cast off
for 50 lines), and setting began with 3r, marked in the
manuscript at
f. 88v.18. A glance at the opening at ff.
88v-89r of the
manuscript shows why (cf. 1597, sig. L3r): there are nine
marginal
notes cut into Hooker's text, and the compositor did not want to estimate
the type/manuscript ratio here. Having set L3r, he cast off
3v (58
lines) and 4r (59) in order to complete outer 3 with
3r's
forme-mate, 4v, which he began at a punctuation mark (f.
91r.15).
He then went back to set inner 3 (3v.4r)
in the normal order,
beginning L3v where 3r had left off, in
the middle of a line (f.
89v.1). If we look to the printed book for evidence of
casting off, we
find only that at the bottom of L4r, where one would
expect
trouble, the compositor has squeezed in a note by cutting in below the
catchword—a very awkward solution (cf. a similar example at
K5r).
Otherwise L3r-4v have 46 lines each, and
the type-setting is even
and regular throughout. However, the sequence within the sheet was the
reverse of normal
(3r→4v→3v→4r), and outer
3 was printed first.
Glancing back now over the pages remaining to be set, he saw that
L1v contained a lengthy insertion by Hooker at the end of
chapter 52
(f. 86v). Only 50 lines of copy had been cast off in order
to leave room
for it, but Pullen's and Hooker's hands differed so much that such a guess
was likely to be inexact. Starting therefore at the cast-off break marked at
L1r|1v, a line-end (the initial word of the
line is itself divided at
the end of the line before, 'to-| gether', f. 86r.26), he set
1v.
Casting off L5r and 5v from the end of
4v, he set 6r next,
the forme-mate of 1v, completing inner 1, which was
printed first. He
then set L6v and 1r to complete the
sheet.
Returning now to sheet 2, he set L2r from the end
of 1v, but
he required 47 lines to reach the casting-off mark at
L2r|2v. The
60 lines of copy allotted L2v could be set in 46 lines of
type so long
as an extensive note was set entirely in the margin and did not encroach
upon the body of the text. In imposition, the extra line in
L2r was
assigned 2v, for the break in 1597 ('nature
| do') falls one
set line earlier than the break for 2r|2v
marked in the manuscript,
'|cannot naturallie' (f. 87v, —10,
—8; cf. 1597,
L2v.2). To supply forme-mates for L2r and
2v, he returned to
5r and 5v, which he had cast off. These
he set seriatim, for
L5r ends mid-line. However, the pattern of headlines
suggests that
outer 2 was printed first. The order of the formes through the press was:
| formes |
O.3 |
| |
I.3 |
∥ |
I.1 |
| |
O.1 |
∥ |
O.2 |
| |
I.2 |
| ------------------------------ |
| headlines |
z.X |
| |
x.Y |
∥ |
z.X |
| |
y.Y |
∥ |
x.X |
| |
z.Y |
In quires, D, H, and N, sheet 2 was composed before sheet 3. The
reason, as in quire L, was the complexity of the underlying copy. In D, the
compositor encountered the first extensive marginal notes cut into the text
block in the manuscript. He cast off the first five pages (299 lines),
stopping at the left-hand margin (f. 22v.25). But instead
of beginning
with sheet 3, he elected to begin with sheet 2 and started with
D2r,
casting off the intervening copy to find its conjugate, 5v.
He then set
D2v where 2r had ended, for the break
between the two occurs in
the middle of a line, at 'beholding | them' (f. 21r.19), and
no 'D4'
appears in the left-hand margin. Because D5v was in type,
5r
needed two extra lines, resulting in a 48-line page, in order to avoid turning
over a marginal note onto 5v. The entire sheet, then, was
set in the
sequence
2r→5v→2v→5r.
Turning now to sheet 3, he could begin either with
D3r—that
is, where 2v ended—or with 3v,
at the casting-off mark on f.
22v. Although one would expect him to begin with
D3r, the
manuscript markings favor 3v. If we remember that when
the actual
break in the printed text either coincides with the indicated break in the
manuscript, or is one or more full lines of type discrepant, setting
began at this point, we can see why. The compositor began
at
a casting-off mark, '| which in both' (f. 22v.25), the
beginning of a
line, and set 46 lines of type, through 'which | the matter' (f.
23v.13;
cf. 1597, D3v, last line), four words beyond the
casting-off mark.
D4r and 4v were set seriatim:
4r fell short eight words of its
cast-off page break (see illustration 2, FLE, vol. 2, of f.
24r, 8 lines from bottom), but 4v reached
the beginning of
5r—already in type—with some stretching.
But returning to set
D3r, 4v's
formemate, he needed 50 lines of type to set the 60 lines of
cast-off MS copy. Extensive cut-outs into the printed text page were
necessary to prevent the carry-over of marginal notes into the margin
opposite the next chapter, for the chapter heading had to be placed in the
margin too, opposite the beginning of the chapter (cf. sig.
D3r,
1597).[16] In imposition, the
page-break at 3r|3v was moved back two
set lines to 'un|to' (f.
22v.22; cf. 1597, D3r, last line), creating
a non-standard 48-line
page, and the two extra lines of type were added to D3v,
originally set
as a 46-line page. Thus, to accommodate the vagaries of his copy, the
compositor needed 48 lines for D3r and
3v and 47 lines for 4r
and 4v.[17]
Faced with irregular copy or a major miscalculation, the compositor
had
but two choices: he could expand (or crowd) the type on a single page to
get back to copy already in type and imposed, or he could, in imposition,
shift lines from one page to another. Expedients such as variant spellings
or contractions, or tinkering with the wording of the text itself, were used
but sparingly. It was apparently a matter of the workman's pride—or
the
master's rule—that orthographic variation or textual emendation was
to
be used only as a last resort. In a prose text, there is no flexibility in
converting prose to verse, or vice-versa, or of creating or avoiding
turnovers of individual lines of verse. As Hooker's chapter breaks were
infrequent, the compositor had perforce to set large blocks of unbroken
text.
To return to quire D, it will be remembered that I started by
assuming that 'if a page was marked for casting off, it was, in fact, cast
off'; that is, that the compositor did not set it at once, seriatim, but skipped
it to begin setting within the quire at a later page. But the operative unit is
not the quire, but the sheet, and there is no way to determine by simple
inspection whether sheet 2 in quire D was wrought off before or after sheet
3. Outer 2 was printed before inner 2 and inner 3 before outer 3, but both
formes printed first have the same pair of headlines, as do both formes
printed second. But in D, 9 out of 12 pages are marked in the manuscript
for casting off. Had composition started with sheet 3, the compositor would
not have had to resort to 47-line pages for D4r and
4v or 48-line
pages for 3r and 3v; rather, setting from
the inside out, only 5
pages would have been cast off, 3v-4v
would have been set to the
normal page-unit of 46 lines, and
there would have been no need for pages of non-standard length. In this
instance, then, the evidence of non-standard page lengths within the quire
confirms the evidence of the manuscript markings.
In sheet 1, D6r and 6v posed no
problems: they were set
seriatim from the end of 5v to the 46-line standard.
D1r and
1v, set from cast-off copy, did. Setting
D1v first, to go with
6r, the compositor apparently started at the casting-off
mark, 'right, |
when' (f. 19v.14; cf. 1597, D1r, last
line) and set to the break
marked in the manuscript for 1v|2r, 'in.
| For there' (f.
20r.35; cf. 1597, D1v, last line). A
cancelled break eight words
earlier, at 'deprived | of', suggests a miscalculation in the actual
composition of D1v. When he returned to set
D1r, however, he ran
into more substantial trouble. He originally set the 60 lines of cast-off copy
in 47 lines of type, through 'preeminence' (f. 19v.15; cf.
1597,
D1v.1), but this left only 45 lines for
D1v. In imposition, one line
was given D1v, evening the pages at 46 lines each. In
setting D1r,
a turnover of a word and a half ('hono-|rable
worke.') at the
end of chapter 15 wasted most of a line, and he had to squeeze his
remaining copy into the six lines at the bottom of 1r. Line
14 of the
manuscript, 'injuried and | defrauded of theire right, ⌿ when ⌿
places not sancti-', has two cancelled breaks in addition to the actual one,
evidence of the difficulty the compositor had in apportioning his copy
between the bottom of D1r and the top of
1v.
In quire N, the second in which sheet 2 was composed before sheet
3, the compositor elected to postpone composition of a series of insertions
in the
manuscript in Hooker's hand on f. 111
r, which fell in
N4
v.
[18] Casting off N1
r
and 1
v (58 lines
each), he began with 2
r, at a punctuation break in his copy
(f.
106
v. 17), 'propagation, | or', and set 2
r
and 2
v, in that
order. Although N2
r is marked for casting off, setting did
not begin
with 2
v, for it begins mid-line ('Christ | belongeth'; f.
107
v.5); at
the top of f. 107
r, a '—20' has been can-celled. He
then cast off
N3
r through 4
r (3 X 58 lines) and
4
v (44 lines), stopping at a
punctuation mark just beyond the Hooker insertions. He then set
N5
r,
the forme-mate of 2
v, beginning
'
qTertullian' (cf. 1597,
N4
v.—9). He continued by setting
N5
v through f. 113
r.4,
marking the page break at 'of | theire'. The sequence
2
r→2
v→5
r→5
v would have readied inner 2 for
the press, had not substantial complications arisen in the composition of the
intervening copy that had been
cast off for sheet 3.
N3r—4r having already been
cast off, sheet 3 was begun
from the end of 2v at 'only | for theire' (f.
108r.26) in order to set
the Hooker insertions first. Starting N4v at '| gather' (f.
110v.22),
a line-end, he reached the break marked for 5r ('voyde.
|
qTertullian') in only 39 lines of type, 7 short of the 46
needed. The
inner forme of sheet 3 was likewise troublesome. As set,
N3v overshot
its cast-off break by 5.5 lines (the indicated break is at 'obedience | to', f.
109v.—3; the actual break, not marked, is at
'gratious | and', f.
110r.5; cf. 1597, N3v.—6,
—1). Had he set N4r to the
standard 46-line page, it would have overshot its cast-off page break by 4.2
lines, encroaching further upon the already short 39-line
N4v. (The
indicated break between N4r and 4v is at
'|gather,' f. 110v.22;
the actual break falls mid-line at 'theire | evidence', line 24, and is marked
in ink.)
Such major discrepancies could only be resolved in imposition, and
the presence of ink markings in the manuscript suggests a second hand. The
page size for inner 3 (3v.4r) was reduced
to 45 lines. The two lines
thus released narrowed the gap in N4v from —11
(7 from the
original setting, 4 from the spillover of 4r) to —9.
These 9 lines,
lifted from N5r, are found at the bottom of
4v in the 1597 volume:
they are marked in the manuscript at f. 111r.29 ('voyde.
|
qTertullian') and f. 111v.5 ('faultes |
wherewith'). A
corresponding block, unmarked in the manuscript, was taken from
N5v
to make good 5r. Despite these adjustments within the
gathering, the
only variation in the order of the formes through the press was that outer
3 preceded inner 3 and sheet 2 preceded sheet 3 through the press.
By all odds the most complicated signature was H. It is anomalous
because sheet 3 seems to have preceded sheet 2 through the press although
sheet 2 was certainly composed before sheet 3. Its difficulties are instructive
as to the limits to which a compositor was willing to go to compose by
formes.
He first cast off three pages (62, 58, and 58 lines) and began to set
H2v. He was thus able to accommodate two large cut-outs
into the text
for marginal notes on f. 59r (cf. H2v,
1597) on a single page and
to begin chapter 41 at the top of H5r.
H2v was begun at 'mindes,
| doth' (f. 58v.22), a punctuation
break. Although the manuscript is not marked, I suspect that
H2
v was
originally set through 'magnifie; or when', the last line of f.
59
r; only
the actual break, at line 33, 'petiti|ons,', is marked. The composition of
sheet 2 proceeded as follows:
2
v→5
r→5
v→2
r.
H3
r—4
v were necessarily cast off
(see Table 1). But in casting
off copy
within sheet 3, the compositor realized he was short
of copy and moved the page break at
H2
v|3
r back 6 set lines,
releasing these for 3
r. H3
r was cast off,
and the 20/40 marks are
counted from 'petiti|ons,'. Still, the page breaks within sheet 3 fell at 58,
55, 57, and
48 lines after
H2
v|3
r. But with sheet 2
already in type, he had no choice but to follow the normal sequence. He
began setting H3
v, then 4
r (completing
inner 3), 4
v (noticeably
loose in its type-setting) back to 5
r, and finally
3
r, loosest of all,
to complete outer 3. But as set from the cast-off break
marked at f. 61
v.20 ('| for the minister'),
H4
v was only 41 lines
long, five short of the 46 needed (cf. 1597, H4
v.6). Five
lines, marked
in the manuscript at f. 61
v.14, 21, were subtracted from
H4
r to
make good 4
v and lesser amounts adjusted backwards in
3
v and
3
r. As originally set, the page-units were:
- 3v: 46.9 lines of type
- 4r: 46.8 " " "
- 4v: 40.2 " " "
As adjusted, these became:
- 4v: 40.2 + 5.0 = 45.2 (equivalent to 46 because
at the end
of a chapter)
- 4r: 46.8 (est.) - 5.0 + 4.2 = 46.0
- 3v: 46.9 (est.) - 4.2 + 3.3 = 46.0
- 3r: 46 - 3.3 = (expanded to 46)
For the purpose of making such adjustments, I estimate that the
compositor had, by manipulating how tightly or loosely he set his type, an
average of 6 sorts' margin per 83-sort (average) line. In a 46-line page,
therefore, the maximum number of sorts available to him was 46 x 6 = 276
÷ 83 = 3.325, which is exactly what he was obliged to make good by
the conspicuously loose type-setting of H3r evident in the
1597
folio.
The compositor was now obliged to return to sheet 2, for what
H2v lost at its end, it had to make up at its beginning.
Thus five set
lines were taken from H2r (from 'remaine, | must', line
16, to
'mindes, | doth', line 22, f. 58v) to make
2v good, and a similar
series of adjustments backwards from these five lines was effected. When
the page break between H1v and 2r was
moved back 4.5 lines of
type (to 'scripture | besides', f. 57v.—5, from
'|and so pleasinge',
f. 58v.2), it solved another problem as well, for in
attempting to set
1v with 2r in type, there was no way he
could set the first three
lines of chapter 38—that is, between the beginning of the chapter at
the
bottom of f. 57v (2 lines of copy) and the casting-off mark
at the top
of f. 58r (1 line of copy), from which 2r,
already in type, had
begun—in complete lines of type. Recall that the original proportion
was
14 lines of type for 18 lines of copy, and 14/18 X 3 = 2.33.
So the cast-off break at H1v|2r, '|and
so pleasinge', was
cancelled.
Within sheet 1, H1r, set from the end of
G6v, undershot its
casting-off mark on f. 57r.15 by 3.5 lines of type,
probably because a
marginal cut-out into the text on f. 56v, with its 10 short
lines, threw
off the compositor's estimates. Had the initial three pages of quire H been
set as they were cast off, the page lengths would have been:
- 2r: 47.5 lines of type
- 1v: 47 " " "
- 1r: 49.5 " " "
As revised, these became:
- 2r: 47.5 - 5.0 + 4.2 = 46.7 lines of
type
- 1v: 47.0 - 4.2 + 3.5 = 46.3 " " "
- 1r: 49.5 - 3.5 = 46.0 " " "
These were manageable discrepancies within the 46-line page norm.
Just as the unit of printing is the forme, not the page, so the unit of
composition of cast-off copy is the sheet, not the forme, and insofar as
possible all adjustments of page lengths to compensate for errors in casting
off copy were made within the sheet. In quire H, unless considerably more
than six pages of type were standing at one time, sheet 3 must have been
machined before sheet 2, although sheet 2 was composed before sheet 3.
The evidence for the latter is the fact that the casting off marks for
H3r
are counted from the break at 'petiti|ons,', in the middle of a word:
3r,
therefore, must have been set after 2v. The evidence for
the former is
inferential, as it allows for the shifting of blocks of type in imposition
between adjacent sheets. In imposition, then, sheet 3 was readied first (by
the subtraction of 6 lines from H2v) and the necessary
adjustments
backwards from 2v held sheet 2 up. The resulting order
of the formes
through the press, then,was
absolutely regular.
| formes |
I.3 |
| |
O.3 |
∥ |
I.2 |
| |
O.2 |
∥ |
I.1 |
| |
O.1 |
| ------------------------------- |
| headlines |
y.Y |
| |
x.X |
∥ |
y.Y |
| |
x.X |
∥ |
y.Y |
| |
x.X |
The difficulties generated by setting cast-off copy were thus resolved in
advance of the actual printing.
Two additional series of marks appear in the margins of the Pullen
manuscript. In order for the compositor to keep track of where he was in
a gathering, and to set the pagination accurately, he marked each page in
the form 'C4' (= C2v; see illustration 4,
FLE, vol. 2) or,
more frequently, in the form frac78-H6 (that is, page 78, or sig.
H3v;
see illustration). Not infrequently, these have had to be cancelled and the
text marked again at the page break in the final printed form, whose
pagination is noted in the margin.
A second series of marks appears in quire H for the first time (see
illustration 7, FLE, vol. 2, of f. 107r).
These tally the lines
of copy as they are set, in the ratio of 5 lines of copy to 4
lines
of type, as an aid to estimating exactly how much copy has been set and
how much was left before reaching a cast-off break. They may start with
any number, arbitrarily, but they rise by 4's and appear every fifth line in
the copy. For example, in the right-hand
margin of f. 59
v, beginning at line 20 with '—22',
they continue
with '—26' (l. 25), '—30' (l. 30), '—34' (l. 35), and
on to the top
of f. 60
r where '—38' is visible, '—42' and
'8' have been
cancelled, and '4' (that is, 44), corresponding to 44 lines of
H3
r,
remains. Were the original series to have continued, '—46' would
have
fallen at line 13, corresponding to the page break marked in line 14,
'expresseth, | declaringe'. But this break has been cancelled and moved
back to line 10, 'was | they' (cf. H3
r|3
v,
1597, and see
illustration of f. 60
r).
Two conclusions come from this analysis. First, the nature of the
underlying copy and of the difficulty the compositor had in setting it are by
no means necessarily reflected in the bibliographical data of the printed
volume. More particularly, without the manuscript, an analytical
bibliographer could not reconstruct the exact order of the formes and sheets
in most quires from the standard evidence of the order of running-titles and
tests for printing on recto or verso of a sheet, and he could not even suspect
the sort of adjustments that took place to even out pages.[19] This inference may seem a modest
return
on the very considerable effort required to reconstruct the composition of
$B—P in Windet's 1597 folio, but its caution is nothwithstanding
useful
to analytical bibliographers and editors alike.
For example, the sheets of quires G and F were each printed in the
same sequence—from the inside out and inner forme first—but
they
were composed in different sequences:
The evidence that G1v was set before G2r
(although sheet 3
presumably was machined before sheet 2) is that G2r
begins in the
middle of a line—that is, at a break made by the end of
G1v, not
from a cast-off break. I cannot account for the variation.
In quire H, sheet 2 was composed before sheet 3, as we have seen,
but this would not be evident from headline analysis: the inner and outer
formes in each sheet have identical pairs of headlines, and none of the
difficulties in composition are reflected in the printing of the volume.
Similarly, the composition of quire N was complicated by Hooker's
interlinear and marginal additions on f. 111r, but the only
bibliographic
evidence that major adjustments had to be made is that sheet 2 preceded
sheet 3 through the press. The evidence here is that, given the priority of
inner 2, outer 3, and inner 1, which we can establish by inspection of the
printed sheets, the following sequence of headlines is the only possible one,
if we assume that no single
headline can appear in consecutive formes within the same quire in standard
two- or three-skeleton printing:
| formes |
I.2 |
| |
O.2 |
∥ |
O.3 |
| |
I.3 |
∥ |
I.1 |
| |
O.1 |
| ---------------------------- |
| headlines |
x.X |
| |
y.Y |
∥ |
z.X |
| |
x.Y |
∥ |
y.X |
| |
z.Y |
Lacking the manuscript, the underlying textual causes for the variant order
would be simply irrecoverable.
Secondly, setting by formes from cast-off copy had little effect on the
integrity of the text. In six instances the compositors have added words not
found in the copy and unlikely to have been added by Hooker in
proof;[20] of these, four are attributable
to setting from cast-off copy:
| reference[21]
|
MS reading |
1597 reading |
1597 reference |
| 19.24 |
Infidels |
and Infidels |
3.21 (B2r) |
| 157.22 |
same |
selfesame |
78.26 (H3v) |
| 259.1 |
think the |
think that the |
133.-1 (N1r; in errata) |
| 469.9 |
Church |
Church of God |
269.-4 (2A3r) |
The criteria for selecting these additions as attributable to the compositor
are that they were added toward the bottom of a page in which he was
trying to get back to a page already in type (259.1, 496.9), or in pages
where the extra word was an obvious aid, whether added consciously or
not, in filling out a page allotted to cast-off copy (19.24, 157.22). In
twenty-six instances the compositors dropped words or omitted marginal
notes found in the manuscript, but of these only five are attributable to
cast-off copy, and two of these fall in the same line:
| reference |
MS reading |
1597 reading |
1597 reference |
| 89.19 |
dailie |
omitted
|
40.-9 (E2v) |
| 198.18 |
yeat, that |
" |
100.-4 (K2v) |
| 368.11 |
likewise |
" |
195.29 (S2r) |
| 433.9 |
O God
|
" |
232.-1 (X2v) |
Finally, one word was omitted from the text ('
sheepe', at
V6
r) but retained as the catchword at the bottom of
V5
v. If we
consider that the text of Book V runs to some 182,000 words, nine changes,
none crucial to the meaning of the passages altered, are trivial indeed.
Given the complexity of the underlying copy, with its cut-outs for marginal
notes, infrequent chapter breaks within long stretches of prose, interlinear
insertions and additions difficult to estimate, the fidelity to copy of Windet's
two compositors is extraordinary. Granted, Hooker did read
proof—he
caught one of the compositors' additions in the errata—and Windet,
Hooker's 'cousin,' (Sisson, p. 53) could be expected to take a more than
commercial interest in the volume. The 1597 folio of Book V, then, may
well be an exception to the general run of Elizabethan books. As compared
to the 1593 folio of Books I—IV, which Windet also printed, the
composition of 1597 was certainly more
leisurely.
[22] There
is only one compositor at work at a time, and there was no way—nor
any
need—for him to keep up with the press, assuming that both were
simultaneously at work on this volume alone, of which there is no
evidence.
[23] Why, then, did he go to
the trouble of casting off copy if not to allow simultaneous setting by two
or more compositors, as in 1593? The answer seems to be simply to avoid
too much standing type. Even at its most complicated, setting cast-off copy
in $B—P of 1597 did not require that more than the usual six pages
of
type be standing at any one time: one forme on the press or being
distributed, one sheet being composed and imposed. The only possible
exception is quire H, but here the printing of sheets 2 and 3 in the reverse
order of their composition may well have been to prevent having more than
six pages in type at once.
