University of Virginia Library

Shakespeare's Merchant Of Venice in Sixty-Three Editions
by
Christopher Spencer

With textual material collected for the New Variorum edition of The Merchant of Venice and with the assistance of a computer, the following study analyzes the readings of sixty-three editions of The Merchant from the First Quarto (1600) to the text prepared by G. Blakemore Evans for the forthcoming New Riverside edition.[1] It shows the degree to which each of the sixty-three editions was dependent upon its various predecessors in substantive readings and in accidentals, and it also shows the quantity and kind of contribution that has been made by each edition to the text of the play in terms both of its new readings and of its contributions to the modern text. The study also lays out the pattern of the historical development of the text with sixty-three points of reference and in the different terms provided by several sets of statistics. The results supplement existing opinion, and the study gives an account of the various editions and a history of the text in much more specific terms than has heretofore been practicable.

The Merchant seems a fairly satisfactory play for a study of this kind. Since there are three quarto editions, two of which precede the First Folio, it has a more varied textual history during the seventeenth century than the eighteen of Shakespeare's plays that were not published before the Folio. On the other hand, the early texts of The Merchant are not independent: the differences among them are minor compared with the variations among quarto and First Folio texts of

such plays as Hamlet or Romeo and Juliet, and The Merchant presents few special and no great problems for the editor.[2]

The sixty-three editions on which this study is based consist of a group that have generally been collated by New Variorum editors (H. H. Furness, H. H. Furness Jr., S. B. Hemingway, M. A. Shaaber, H. N. Hillebrand, and M. W. Black), with additions and a few omissions resulting from the experience of G. B. Evans with I Henry VI and Richard Knowles with As You Like It, and from my own investigations.[3] Seven of the sixty-three are from the seventeenth century, 18 from the eighteenth, 24 from the nineteenth, and 14 from the twentieth. Of the 56 critical editions beginning with Rowe, 43 were prepared by British editors, 12 by American editors, and one by a German editor.

The readings of the sixty-three editions are recorded for 2583 passages in The Merchant. These are the variants remaining in the second stage of my preparation of the textual annotation for the New Variorum edition.[4] Originally I recorded readings for about five thousand variant passages and included almost all punctuation and many spelling differences. In revising this material I omitted most of the spelling and many of the punctuation variants and excluded also non-substantive changes in stage directions, thereby reducing the number of passages by about half. There remained 2583 variants in a play of

2658 lines.[5] For the purposes of this study the variants are classified in ten categories and two values. The categories are: verbal differences (621 variants); characters' names (42);[6] punctuation (1000);[7] lineation (59); stage directions, scene divisions, and locales (196); transpositions (37); non-metrical word forms (125); metrical variations (388); spelling (106); and miscellaneous (9). Of the 2583 variants, 457 were considered major (substantive) and the remaining 2126 minor.[8]

The computer produced two series of tables. One series consisted of a table for each of the sixty-two editions after Q1, listing the number of times the edition agreed uniquely with each of its predecessors and the total number of times the edition agreed with each of its predecessors; the tables recorded this information for each of the values. The second series of eight tables tabulated the number of new readings initiated by each of the sixty-three editions in each of the ten categories for each of the values (it also gave totals for each of the editions, categories, and values). Furthermore, this series gave the number of these new readings (for each edition in each category in each value, and with totals) that appear in Evans' edition and in Peter Alexander's edition of 1951.[9] The first series of tables showed the dependence of each edition upon its predecessors; the second series showed the contribution of each edition to the developing and to the modern text of the play. The three tables given below are composed of information from the seventy tables supplied by the computer and are designed to give the information in compact and emphatic form.

TABLE I

                                                                                                         

Edition	Total new	Evans	Alex.	Major	Verbal
1 Q1 (Heyes, 1600)	2583	1479	1354	457 (370)	621 (510)
2 Q2 (Roberts, [1619])	592	228	253	101 (14)	144 (31)
3 F1 (1623)	314	85	93	75 (8)	90 (10)
4 F2 (1632)	119	20	20	41 (3)	58 (7)
5 Q3 (1637)	176	41	49	23 (4)	37 (3)
6 F3 (3rd Imp., 1664)	81	12	14	22	32
7 F4 (1685)	225	79	111	13 (2)	37 (6)
8 Rowe I (1709)	321	100	143	54 (5)	70 (11)
9 Rowe II (1709)	16	2	3	5	7
10 Rowe III (1714)	64	13	18	8	18 (2)
11 Pope I (1723)	286	59	63	82 (10)	78 (8)
12 Pope II (1728)	12	2	3	4 (1)	8 (2)
13 Theobald I (1733)	195	72	85	22 (7)	18 (2)
14 Theobald II (1740)	33	10	12	7	6
15 Hanmer I (1743)	64	9	7	29	33
16 Warburton (1747)	24	1	1	13 (1)	13 (1)
17 Theobald III (1757)	11	2	4	1	3 (1)
18 Johnson (Corbet, 1765)	116	34	28	8 (3)	11 (2)
19 Capell (1768)	330	102	126	33 (7)	42 (8)
20 Hanmer II (1770)	11	1	2	3	2
21 Johnson-Steevens I (1773)	59	5	5	10 (1)	10
22 Johnson-Steevens II (1778)	19	0	0	4	3
23 Rann (1787)	18	3	3	7 (1)	6 (1)
24 Malone (1790)	34	8	10	8 (1)	7 (1)
25 Steevens-Reed I (1793)	15	2	2	4	5
26 Steevens-Reed II (1803)	2	1	1	1	1
27 Eccles (1805)	35	2	3	6	10 (1)
28 Boswell-Malone (1821)	4	0	0	0	2
29 Singer I (1826)	16	2	1	5 (1)	3 (1)
30 Knight (1841)	39	8	9	6 (2)	4 (1)
31 Collier I (1842)	90	27	25	12 (3)	6 (1)
32 Hudson I (1851)	25	4	7	4 (2)	2
33 Singer II (1856)	8	1	0	2	1
34 Halliwell (1856)	44	4	4	6 (1)	4 (1)
35 Dyce I (1857)	101	20	19	7 (1)	10
36 Collier II (1858)	38	1	1	18	24
37 Staunton (1858)	18	2	2	0	1
38 White I (1858)	21	2	2	5	2
39 Old Cambridge I (1863)	89	23	22	3 (2)	4 (3)
40 Globe (1864)	10	3	2	1	1
41 Keightley (1864)	81	12	11	13	14
42 Dyce II (1866)	20	1	1	7	6
43 Rolfe I (1870)	5	1	2	2	1
44 Delius (1872)	7	0	1	2	2
45 Hudson II (1880)	17	0	0	4	5
46 White II (1883)	9	2	1	3	2
47 Rolfe II (1883)	5	0	0	1	2
48 Old Cambridge II (1891)	0	0	0	0	0
49 Oxford (1891)	11	0	0	2	3
50 Rolfe III (1903)	11	5	1	1 (1)	1 (1)
51 Old Arden (1905)	6	0	0	0	0
52 Neilson (1906)	57	29	16	0	0

                                                                                                         

	Character	Punctuation	Lineation	SDD etc.	Tr. & Misc.	Form & Sp.	Metrics
Q1	42 (16)	1000 (315)	59 (31)	196 (59)	46 (38)	231 (153)	388 (357)
Q2	6 (3)	274 (151)	13 (2)	18 (6)	11 (2)	75 (26)	51 (7)
F1	5	116 (41)	15 (4)	23 (11)	12 (3)	34 (13)	19 (3)
F2	3	36 (10)	2	1	1	14 (3)	4
Q3	15 (3)	96 (31)	2 (1)	2	1	16 (3)	7
F3	0	24 (8)	0	1	2	18 (4)	4
F4	0	162 (65)	1	5 (2)	1	10 (4)	9 (2)
R1	16 (10)	105 (30)	4 (1)	65 (37)	6 (2)	22 (6)	33 (3)
R2	0	8 (2)	0	0	1	0	0
R3	0	33 (10)	1	1 (1)	0	6	5
P1	0	77 (34)	12 (9)	16 (1)	4	21 (6)	78 (1)
P2	0	3	0	0	0	1	0
T1	2 (2)	140 (54)	1	17 (9)	3 (1)	8 (2)	6 (2)
T2	0	19 (10)	1	0	0	3	4
H1	1	15 (7)	2	3 (2)	3	3	4
W	0	9	0	0	0	1	1
T3	0	0	0	0	0	3	5 (1)
J0	3	91 (27)	2 (2)	7 (3)	1	1	0
Ca	15 (5)	170 (48)	6 (4)	65 (33)	0	6 (1)	26 (3)
H2	0	5 (1)	0	1	0	1	2
S1	0	45 (4)	1	1	0	0	2 (1)
S2	1	13	0	0	2	0	0
Ra	2	5 (2)	1	1	0	2	1
Ma	1 (1)	10 (2)	2	6 (4)	0	2	6
R1	0	4 (2)	1	0	0	2	3
R2	0	0	0	1 (1)	0	0	0
Ec	0	10 (1)	2	8	1	1	3
BM	0	1	0	0	0	0	1
S1	0	8	0	0	0	0	5 (1)
Kn	2 (1)	29 (4)	1 (1)	0	0	0	3 (1)
C1	2	70 (23)	3 (1)	3 (1)	0	5 (1)	1
H1	2	13 (2)	2 (2)	1	0	2	3
S2	0	5 (1)	0	1	0	0	1
Ha	0	34 (3)	0	4	0	1	1
D1	1	33 (11)	0	14 (9)	0	0	4312
C2	1	10	0	1 (1)	0	0	2
St	1	13 (1)	0	2 (1)	0	1	0
W1	0	14 (1)	1	1	1	1 (1)	1
C1	1 (1)	27 (11)	0	8 (8)	0	5	4412
Gl	0	7 (3)	1	0	0	0	1
Ke	0	52 (12)	1	4	1	4	5
D2	2	6 (1)	0	1	0	1	4
R1	0	3 (1)	1	0	0	0	0
De	0	3	0	2	0	0	0
H2	0	5	0	0	1	0	6
W2	0	5 (1)	0	1 (1)	0	1	0
R2	0	3	0	0	0	0	0
C2	0	0	0	0	0	0	0
Ox	0	6	0	0	0	0	2
R3	0	10 (4)	0	0	0	0	0
OA	0	4	0	0	0	1	1
Ne	0	54 (27)	0	0	0	0	3 (2)

                       

Edition	Total new	Evans	Alex.	Major	Verbal
53 New Cambridge I (1926)	196	18	10	13 (4)	12 (2)
54 Ridley (1935)	11	2	0	1	0
55 Kittredge (1936)	71	11	15	3	1
56 Neilson-Hill (1942)	1	1	1	1 (1)	1
57 Alexander (1951)	17	1	17	2	3
58 Sisson (1953)	47	1	0	7	11
59 New Arden (1955)	175	13	0	2	4 (1)
60 London (1958)	19	0	0	2	1
61 New Yale (1960)	22	0	0	1	1
62 New Cambridge II (1962)	4	0	0	0	0
63 Evans	17	17	0	1 (1)	2 (2)

The foregoing table lists the sixty-three editions with the number of new readings originated by each edition in toto and in the different categories. The second column lists the total number of new readings, the third column the number of those new readings that appear in Evans, and the fourth column the number that appear in Alexander. The fifth column records the new readings in the 457 variant passages, and the figure in parentheses indicates how many of those major new readings appear in Evans' edition. The other eight columns tabulate the new readings in each category (followed, in parentheses, by the number of those readings that appear in Evans). Where there is no figure in parentheses in columns five through thirteen, there are no new readings in Evans. Because the numbers for transposition,[10] miscellaneous, word form, and spelling variants are small and comparatively lacking in significance, these four categories have been combined into two.[11]

                       

	Character	Punctuation	Lineation	SDD etc.	Tr. & Misc.	Form & Sp.	Metrics
C1	1	114 (11)	0	9 (2)	1	13 (2)	46 (1)[12]
Ri	0	10 (2)	0	1	0	0	0
Ki	0	58 (6)	1	5	0	4 (3)	2 (2)
NH	0	0	0	0	0	0	0
Al	0	12 (1)	0	1	0	1	0
Si	1	26	0	3	2	2 (1)	2
NA	0	167 (10)	1	2 (1)	0	1 (1)	0
Lo	0	14	0	1	0	0	3
NY	0	15	1	1	0	3	1
C2	0	1	0	0	0	0	0
Ev	0	9 (9)	1 (1)	3 (3)	0	1 (1)	1 (1)

The table shows the heavy dependence of the modern text on editions before 1800. Thus, of Evans' readings in the 2583 passages recorded, 1479 come from the authoritative text (Q1), 465 from other seventeenth-century editions, 425 from eighteenth-century editions, 116 from nineteenth-century editions, 81 from other twentieth-century editions, and 17 from no previous text collated. Alexander follows the authoritative text less often (1354 times), uses more from other seventeenth-century editions (540), more from the eighteenth century (515), about the same amount from the nineteenth century (114), and less from his twentieth-century predecessors (43); he too seems to originate 17 readings. In 457 major variants, Evans follows Q1 370 times, other seventeenth-century texts 31 times, eighteenth-century texts 37 times, nineteenth-century texts 12 times, previous twentieth-century texts 6 times, and no text collated once.

After 1700, the largest number of new readings was originated by Rowe in his three editions (401), followed by Capell (330), Pope (298), Theobald (239), New Cambridge (200), New Arden (175), Collier (128), and Dyce (121). The largest number adopted by Evans was originated by Rowe (115), followed by Capell (102), Theobald (84), Pope (61), and Johnson (34); the same five editors supplied the largest number used by Alexander. One edition originated no new readings (Old Cambridge II), and nine more editions contributed no new readings used by either Evans or Alexander (Johnson-Steevens II, Boswell-Malone, Hudson II, Rolfe II, Oxford, Old Arden, London, New Yale, and New Cambridge II). Only sixteen editions supplied more than one major reading (in 457 passages) adopted by Evans: Q1, Q2, F1, F2, Q3, F4, Rowe I, Pope I, Theobald I, Johnson, Capell, Knight, Collier I, Hudson I, Old Cambridge I, and New Cambridge I. (The list for Alexander would add Malone and exclude Hudson I and

New Cambridge I.) F3 contributed no reading in a major passage to either Evans or Alexander.

Of eighteenth-century editors, Rowe in his three editions supplied the most new readings in verbal variants (95) and the most adopted by Evans (13); Pope in his two editions contributed 86 (10 adopted); and Capell added 42 (8 adopted). In the nineteenth century Collier originated 30 in two editions (most of the 24 in Collier II are from the Perkins Folio) with one adopted, and Eccles, Dyce, and Keightley invented ten or more; the largest number adopted (three) was supplied by the Old Cambridge editors. The New Cambridge edition and Sisson have added the most new verbal readings in the twentieth century. In characters' names Rowe (16) and Capell (15) have added the most, with little contribution from others. In punctuation, Capell (170), Theobald (159), and Rowe (146) originated the most readings, and the same three in a different order (Theobald, Capell, Rowe) contributed the most to Evans' text. In lineation, Pope and Capell supplied the largest number of new readings and the largest number of contributions to the modern text. In stage directions, scene divisions, and locales, Rowe, Capell, Theobald, Pope, and Dyce have originated most; and Rowe, Capell, Theobald, Dyce, and Old Cambridge have contributed most to modern texts.[13] In every category, the quartos, folios, and eighteenth-century editors supplied the basic material, with fewer additions during the nineteenth century; new readings have appeared in the twentieth century but have not yet become established in the text.

II

By recording wholly new readings, Table I emphasizes the originality of editors; it favors those who were early enough not to be anticipated, and it looks forward to their lasting contributions to the text of the play. Table II notices the reintroduction or unconventional use of an old reading, thereby giving a broader base for evaluating an edition's individuality. It looks backward as well as forward, showing where an editor may have gotten most of the readings he did not invent and, at the same time, which editions supplied the most readings for their successors. The second and third columns below list the

TABLE II

                                                                 

Edition	Highest unique	2nd highest	No. eds.	Highest total	2nd highest
Total variants	2583	2583		2583	2583
Q2	1991 Q1	--	1	1991 Q1	--
F1	364 Q1	186 Q2	2	2083 Q1	1905 Q2
F2	285 F1	27 Q2	3	2412 F1	1974 Q1
Q3	60 Q1	57 Q2	4	2149 Q1	2047 F1
F3	74 F2	17 Q2	5	2414 F2	2327 F1
F4	65 F3	27 Q2	5	2241 F3	2139 F2
Rowe I	181 F4	29 Q3	6	2121 F4	1871 F3
Rowe II	320 Rowe I	6 Q2	3	2558 Rowe I	2098 F4
Rowe III	7 Rowe II	6 Q3	8	2459 Rowe II	2456 Rowe I
Pope I	49 Rowe III	36 Q2	7	2111 Rowe III	2026 Rowe I
Pope II	281 Pope I	1 3 eds.	4	2555 Pope I	2102 Rowe III
Theobald I	10 Q3	7 Q2, Pope II	6	2284 Pope II	2267 Pope I
Theobald II	192 Theobald I	2 Q3	3	2539 Theo. I	2250 Pope II
Hanmer I	11 Theo. II	3 Q3	4	2386 Pope I	2384 Pope II
Warburton	22 Theo. II	4 Hanmer I	4	2501 Theo. II	2463 Theo. I
Theobald III	2 Q3, Warburton		4	2549 Theo. II	2509 Theo. I
Johnson	8 Q2	6 Warb.	5	2372 Warb.	2355 Theo. III
Capell	17 Q2, Johnson		9	1776 Warb.	1775 Theo. III
Hanmer II	55 Capell	48 Hanmer I	5	2418 Hanmer I	2235 Pope I, II
Johnson-Steevens I	80 Capell	38 Johnson	4	2162 Johnson	2075 Theo. III
Johnson-Steevens II	75 Capell	24 J-S I	4	2290 J-S I	2281 Capell
Rann	12 J-S II	7 Capell	5	2525 J-S II	2273 Capell
Malone	29 Capell	2 Q2, Rann	8	2400 J-S II	2389 Rann
Steevens-Reed I	27 Malone	3 Capell	4	2516 Malone	2386 J-S II
Steevens-Reed II	14 S-R I	None	1	2569 S-R I	2516 Malone
Eccles	9 Capell	2 3 eds.	8	2431 S-R II	2429 S-R I
Boswell-Malone	2 S-R II	1 4 eds.	5	2545 S-R II	2537 S-R I
Singer I	3 Q2, Rann		3	2518 S-R II	2515 B-M
Knight	3 Capell	2 Q2, Singer I	8	2396 B-M	2386 S-R II
Collier I	6 Knight	3 Johnson	10	2281 B-M	2261 S-R II
Hudson I	33 Collier I	4 Knight	9	2355 Collier I	2334 B-M

                                                               

Edition	Highest unique	2nd highest	No. eds.	Highest total	2nd highest
Singer II	8 Singer I, Collier I		5	2471 Singer I	2433 S-R II
Halliwell	16 Knight	7 Collier I	9	2303 Knight	2288 Hudson I
Dyce I	8 Collier I	4 Halliwell	9	2261 Hudson I	2234 B-M
Collier II	29 Collier I	6 Dyce I	7	2469 Collier I	2292 Hudson I
Staunton	7 Dyce I	6 Knight	7	2374 Knight	2319 Singer II
White I	2 Collier I, Dyce I		7	2407 Collier I	2345 Collier II
Old Cambridge I	15 Dyce I	14 Q2	12	2179 Dyce I	2131 Hudson I
Globe	81 OCam. I	1 3 eds.	4	2520 OCam. I	2164 Dyce I
Keightley	4 Halliwell	3 Johnson	10	2254 Singer II	2215 Singer I
Dyce II	66 Dyce I	4 Collier II, Keightley	6	2393 Dyce I	2144 Hudson I
Rolfe I	4 White I, Globe		5	2313 White I	2248 Globe
Delius	3 White I	1 3 eds.	4	2389 Collier I	2334 White I
Hudson II	19 Dyce II	4 OCam. I	12	2295 Dyce II	2226 Dyce I
White II	5 Globe	2 Q1, White I	5	2396 Globe	2354 OCam. I
Rolfe II	4 Rolfe I	None	1	2496 Rolfe I	2295 Delius
Old Cambridge II	1 Q1, OCam. I		2	2541 OCam. I	2503 Globe
Oxford	1 4 eds.		4	2310 White II	2305 Globe
Rolfe III	5 Rolfe II	2 Keightley	3	2526 Rolfe II	2450 Rolfe I
Old Arden	2 White II, Oxford		3	2390 Oxford	2345 White II
Neilson	8 Keightley	7 Rolfe III	11	2242 White II	2223 OArden
New Cambridge I	10 Neilson	7 Johnson	20	1856 Globe	1855 OCam. II
Ridley	10 Q1	4 Q3, NCam. I	8	2143 OArden	2114 OCam. II
Kittredge	26 Neilson	15 NCam. I	9	2143 Neilson	2060 Rolfe III
Neilson-Hill	19 Neilson	1 Kittredge	2	2496 Neilson	2210 OArden
Alexander	14 Kittredge	2 NCam. I	9	2214 Kittredge	2207 N-H
Sisson	5 Kittredge	3 Halliwell, Keightley	11	1963 NCam. I	1906 Alexander
New Arden	5 Johnson, Sisson		13	1971 Ridley	1939 Alexander
London	11 Kittredge	5 Sisson	5	2103 N-H	2098 OCam. II
New Yale	5 Q1	3 Q3	6	2100 NArden	1985 Ridley
New Cambridge II	163 NCam. I	1 Johnson	2	2528 NCam. I	2084 London
Evans	5 Kittredge	1 3 eds.	4	2181 N-H, Alexander

highest and second highest number of unique agreements that each edition has with its predecessors (no other previous edition collated has the reading); and the fourth column gives the number of preceding editions with which there is unique agreement. The fifth and sixth columns give the highest and second highest number of total agreements (unique and shared with one or more earlier or intervening editions) with any previous edition.

The columns of unique agreement suggest both the extent to which an editor used readings that were not yet established and the degree to which he influenced his successors. Editions from Hanmer II to Eccles borrowed from Capell, some of them quite heavily, although none was actually based upon Capell. Pope, Johnson, Capell, Malone, Singer I, Knight, and the Old Cambridge editors went back to Q2 for readings not used since 1619 (as they thought, since 1600); and Ridley and New Yale adopted readings not used since Q1. Many later editors have borrowed from Johnson: unique agreements are shown in the table for Capell, Johnson-Steevens I, Collier I, Keightley, New Cambridge I and II, and New Arden; and unique agreements not listed appear in Hanmer II (2), Johnson-Steevens II (3), Rann (1), Steevens-Reed I (1), Hudson I (2), Neilson (6), Sisson (2), and London (1). In the fourth column — the number of editions with which there is unique agreement — Capell (9), Rowe (8), and Malone (8) are higher than their predecessors or contemporaries; Old Cambridge I and Hudson II with 12 and Collier I and Keightley with 10 are high for the nineteenth century; and New Cambridge I is most eclectic by far with unique agreements with 20 editions, followed by New Arden (13) and Sisson and Neilson (11).

For the non-critical texts of the seventeenth century and for many critical editions (especially those that are early and those for which the figures in the table are high), the largest number of shared readings shows the edition used as base text.[14] However, as the text of the play becomes more established during the nineteenth century and there are more editions with less variation among them, this figure is a less reliable guide. The texts showing the least variation are often the second and third editions. Rowe II agrees with Rowe I in 2558 of

2583 instances, and there are, therefore, only 25 differences among the variants included in the tabulation. Similarly, Pope II differs from Pope I 28 times, Hanmer II from Hanmer I 165 times, Globe from Old Cambridge I 63 times, New Cambridge II from New Cambridge I 55 times, and so on.[15] However, Hudson II and White II are markedly independent of their first editions: the former agrees more frequently with four other editions than with Hudson I (2112 agreements), and the latter more frequently with six other editions than with White I (2159 agreements). Only four critical editions have fewer than 2000 agreements with any single predecessor: in the eighteenth century Capell with 1776 and in the twentieth New Cambridge I with 1856, Sisson with 1963, and New Arden with 1971. In Capell 807 readings (31 percent) had not appeared in the closest text (Warburton); and, according to Table I, 330 readings (13 percent) had not appeared anywhere previously.

III

Table II provides insight into an edition's relationship to those of its predecessors from which it borrowed most, but only occasionally does the table suggest anything of the edition's relationship to the authoritative text (Q1) and the two principal early derivative editions (Q2 and F1). Table III, which records agreement for sixty-two editions with the three earliest texts, gives data on these relationships and, at the same time, outlines the historical development of the text of The Merchant, showing its departure from and then its return, to a limited degree, toward the early editions.

TABLE III

                         

Edition	Agree with Q1	Agree with Q2	Agree with F1
	Major	Minor	Major	Minor	Major	Minor
Total variants	457	2126	457	2126	457	2126
Q2	356	1635	--	--	--	--
F1	365	1718	303	1602	--	--
F2	331	1643	271	1574	413	1999
Q3	417	1732	348	1639	348	1699
F3	315	1609	259	1573	395	1932
F4	304	1374	255	1448	385	1684
Rowe I	278	1151	229	1259	340	1407
Rowe II	271	1139	227	1255	333	1395
Rowe III	277	1112	235	1235	342	1363
Pope I	254	998	238	1152	246	1201

                                                                                                         

Edition	Agree with Q1	Agree with Q2	Agree with F1
Pope II	251	1000	237	1152	242	1196
Theobald I	255	885	241	1048	239	1078
Theobald II	252	860	238	1021	237	1054
Hanmer I	229	928	218	1087	216	1127
Warburton	248	849	232	1014	233	1038
Theobald III	252	864	238	1025	237	1053
Johnson	259	831	251	1004	242	1009
Capell	324	875	280	1011	289	1039
Hanmer II	229	842	218	1009	213	1037
Johnson-Steevens I	283	862	267	1025	266	1036
Johnson-Steevens II	306	865	281	1017	288	1043
Rann	301	866	273	1016	283	1042
Malone	333	862	301	1009	302	1030
Steevens-Reed I	323	863	291	1009	295	1037
Steevens-Reed II	322	864	290	1012	292	1037
Eccles	306	833	275	995	280	1014
Boswell-Malone	331	862	298	1011	306	1038
Singer I	326	852	295	1004	300	1028
Knight	327	897	286	1043	325	1067
Collier I	345	883	302	1048	316	1041
Hudson I	343	878	296	1035	309	1051
Singer II	325	852	285	1014	299	1027
Halliwell	345	898	305	1035	302	1055
Dyce I	343	831	299	970	308	986
Collier II	324	892	282	1044	289	1041
Staunton	339	886	295	1031	310	1058
White I	327	900	281	1044	330	1064
Old Cambridge I	338	837	326	1002	291	982
Globe	339	842	325	1005	294	991
Keightley	322	836	282	1000	289	992
Dyce II	321	816	277	954	284	960
Rolfe I	330	909	294	1048	330	1065
Delius	343	890	299	1036	308	1039
Hudson II	317	728	275	873	282	878
White II	337	906	320	1063	294	1054
Rolfe II	338	909	305	1049	313	1054
Old Cambridge II	339	850	326	1015	294	993
Oxford	330	887	295	1039	294	1037
Rolfe III	337	908	305	1049	312	1049
Old Arden	339	921	303	1067	304	1071
Neilson	345	914	323	1074	298	1056
New Cambridge I	351	910	306	977	305	989
Ridley	371	1138	321	1200	316	1222
Kittredge	353	1010	299	1084	305	1090
Neilson-Hill	360	955	319	1079	311	1078
Alexander	360	994	309	1088	310	1093
Sisson	355	1022	320	1111	311	1103
New Arden	386	1223	321	1211	326	1229
London	353	806	312	924	304	932
New Yale	392	1354	325	1322	336	1355
New Cambridge II	354	918	305	985	306	1001
Evans	370	1109	313	1161	316	1175

The second column shows the gradual departure from Q1 in major readings through the folios and the eighteenth-century editors to Johnson, the sudden check to this trend given by Capell and by Malone (who borrowed in part from Capell), the static quality of the nineteenth century, and the efforts of some twentieth-century editors to use much from the seventeenth century — a trend most marked in Ridley, New Arden, New Yale, and Evans. The closest agreement with Q1 in the seventeenth century is in Q3 with only 40 disagreements in major readings (457 minus 417), in the eighteenth century Malone with 124 disagreements, in the nineteenth century Collier I and Halliwell with 112 disagreements, and in the twentieth century New Yale with 65 disagreements. The low marks of agreement with Q1 by century are F4 (153 disagreements), Hanmer I and II (228 disagreements and 229 agreements), Eccles (151 disagreements), and Rolfe III (120 disagreements). The third column shows that after Rowe III and with the exception of only four modern editions, no text agrees with Q1 in even half the minor variants recorded. The same may be said for Q2 from Theobald I to Rolfe III (except two editions) and for F1 from Warburton to Rolfe III (except three editions).

This table also shows that textual theory had only occasional influence upon editorial practice before the twentieth century, and that it has had some effect on both major and minor readings in modern editions. All editions since Pope have more major variant readings in common with Q1 than with Q2, despite the fact that before 1909 it was usually thought that Q2 was the earlier and the better text.[16] Even the Old Cambridge editors, with their strong support of Q2, actually used more major readings from Q1, though their respect for Q2 exceeded that of their contemporaries.[17] In recent years there has been a significant gain for Q1 over Q2 and F1, presumably because modern editors are certain that Q1 is the substantive text. In minor variants agreement with Q2 more frequently than with Q1 has been the rule since 1685, with the exceptions of only the New Arden and New Yale editions. Doubtless this situation results from the more modern accidentals in Q2 and the fact that the modernizing editor

usually is not obligated to preserve the accidentals of an early text, as he is to preserve its substantive readings. Nevertheless, New Arden and New Yale have used more from Q1 than from Q2, and most modern editors have preserved substantially more minor readings from the early editions than did their nineteenth-century predecessors. The data support Fredson Bowers' generalization that among modern editions "a steady move has been observable towards the purification of the text, less in the direction of independent emendation, or the introduction of brand-new readings, than towards the restoration of original readings wrongly emended by eighteenth-century editors and subsequently established as traditional."[18]

IV

The data confirm the accepted view that Q2, Q3, and F1 are based upon Q1. In 2583 readings (Table II) Q2 departs from Q1 592 times (2583 minus 1991), whereas F1 departs from Q1 500 times and Q3 departs from Q1 434 times. The larger number of variants in Q2 may be accounted for as the work of Jaggard's Compositor B, who, as D. F. McKenzie has shown, seems to have been responsible for the quarto.[19] In major variants (Table III) Q2 introduces differences in only 101 of the 457 passages where the full textual history of the play shows editorial disagreement (457 minus 356), whereas F1 has variants in only 92 such passages. Since there are 17 unique agreements in major readings between Q2 and F1,[20] it follows that F1 adds 75 new major disagreements to the 101 originated by Q2: 176, then, of the 457 major variant passages originated in the second and third editions of the play, whereas 281 were added by later editors. According to Table III every edition before Ridley (except F1 and Q3) departed from Q1 more often in major variants than did Q2.

A few words might be said for Q3, which was published in 1637 by the son of the man who had published Q1, Laurence Hayes. Furness

called Q3 "merely a careless reprint of a careless book,"[21] and other editors have had little to say about it. Although Q3 does have a number of typographical errors and some lines seem to have shifted and some letters to have dropped out, Furness' judgment needs qualification. From Table III it appears that Q3 is a more faithful reprint of Q1 than either Q2 or F1 — especially in major variants, where Q3 departs from Q1 40 times, in contrast with Q2's 101 and F1's 92 departures. Although Q3 originated fewer new readings than Q2, F1, or F4 (Table I) and although Evans and Alexander retain only 41 and 49 (respectively) readings that originated in Q3, editors have frequently come into agreement with Q3. Thus, Rowe I has 29 unique agreements with Q3, Rowe III has 6, Theobald I 10, Theobald II and III 2 each,[22] Hanmer I 3, Ridley 4, and New Yale 3. Sixteen other editions, have unique agreements with Q3 that are not listed in Table II, including F3 (6), F4 (24), Pope I (8), Capell (6), and New Cambridge (3). Perhaps editors have actually borrowed from Q3 on occasion; but the agreements are more likely to be accidental, and the large number of unique agreements probably results from Q3's being outside the main stream of development (from Q1 through the Folios to Rowe, etc.), so that revival of a Q3 reading is possible when an editor makes a fresh departure from Q1. Although Q3 has no authority, it is still an early reprint that should be of interest to editors.[23]

In general, the data given above confirm the conclusions of M. W. Black and M. A. Shaaber in their study of Ff2-4.[24] In F2 there are 119

new readings, 58 of which (49 percent) are verbal; in F3 there are 81 new readings, 32 of which (40 percent) are verbal; and in F4 there are more new readings (225) without a corresponding increase in verbal changes (37; 16 percent), whereas there are many more punctuation variants (162; 72 percent). According to Table II, F2 departs from F1 only 171 times (2583 minus 2412), F3 from F2 only 169 times, and F4 from F3 342 times. The very large proportion of verbal changes in F2 and F3, which otherwise follow their base texts closely, suggests deliberate editorial attention as found by Black and Shaaber; but the small total number of changes (compare Q2's 592 and F1's 500 departures from Q1 and Rowe I's 462 departures from F4) indicates that the editorial attention was on a limited scale. Black and Shaaber found fewer verbal and more punctuation variants in F4 than in F2 and F3, although not as many of the latter as my data suggest. The emphasis on metrical changes that Black and Shaaber found in F2 is not reflected in my data, largely because I counted as verbal a number of alterations that they considered metrical.[25]

Since 1700, Rowe and Capell have contributed most to the text of The Merchant. As the first critical editor, Rowe had the opportunity and made good use of it; Capell, coming after the material had been reworked several times, created his own opportunity and made even better use of it.[26] Next after these two in contributions come Pope, Theobald, and Johnson. As the reasonable possibilities have been tried out, the opportunity to find acceptable new readings has shrunk, but in the nineteenth century the Old Cambridge editors made some additions to the text, seconded by Collier and Dyce and then by Knight and Keightley. More eclecticism, originality, and respect for early seventeenth-century readings are characteristic of twentieth-century editors. The first two flower in John Dover Wilson's New Cambridge edition and all three in John Russell Brown's New Arden edition. Neilson, Ridley, Kittredge, Sisson, and A. D. Richardson III's New

Yale edition show some of these characteristics to a lesser degree. Alexander and Evans are moderately conservative texts for their age, and London is extremely conservative, its second-highest number of agreements being with the Old Cambridge edition.

General conclusions based upon the data for only one play are tentative. However, many of the relationships among folios and among critical editions pointed out above must apply to other plays as well. One might speculate that where there are considerable differences among early texts of a play with a large number of important alternative readings, a study such as this might set off much more sharply one editor's dependence on another for minor variants and at the same time show a greater cross-fertilization among editors in major variants. Thus, many of the same characteristics might appear, but in exaggerated form. Chief among the factors that seem likely to influence the data for other plays are the existence of two or more independent or partially independent early editions; the nature and quality of the authoritative text(s); and the presence of special problems (such as the "Sallies" in The Merchant). The data for individual editions seem likely to vary significantly from one play to another as a result of these conditions or such others as different editors for different plays (e.g., New Arden), different base texts for different plays (e.g., Johnson), an editor's bestowing more care upon one play than upon another (e.g., F3 and F4),[27] or his publishing his plays over a long period of time (e.g., New Cambridge).

Notes