University of Virginia Library

The First Edition of John Lyly's Sappho and Phao (1584)
by
David Bevington [*]

John Lyly's Sappho and Phao has not been edited often since its first publication in 1584. Nor, this essay will argue, has it ever been critically edited on the basis of the actual first quarto that should serve in this case as the proper copy-text. The quartos of 1584 and 1591, incorrectly identified (as we shall see) by R. Warwick Bond as Q1 and Q2, were followed by Edward Blount's edition of Six Court Comedies in 1632, essentially a reprint of 1591 with some added songs, and then by F. W. Fairholt's The Dramatic Works of John Lilly (London: John Russell Smith, 1858), derived from Blount. R. Warwick Bond undertook to reedit all of Lyly from original texts in The Complete Works of John Lyly, 3 vols. (Oxford: Clarendon, 1902), but he based his text of Sappho and Phao on a single copy of a 1584 quarto in the British Museum (now the British Library). That copy-text now appears to have been the wrong choice.

The extent of Bond's error has only slowly been discovered. The first edition of A Short Title Catalogue of Books Printed in England, Scotland, and Ireland . . . 1475-1640 by Pollard and Redgrave, 1926, implicitly compounded Bond's mistake by assigning only one number, 17086, to the 1584 "Q1" of Bond's edition. W. W. Greg made an important discovery in his A Bibliography of the English Printed Drama to the Restoration (1939) when he recognized that not one but two quartos were published in 1584. He gave the number 82a to the British Library copy (C. 34. d. 17) and the number 82b to another quarto, existing (as far as Greg knew) in two copies, one at the Huntington Library and one at the Library of Worcester College, Oxford (C2 def.). Greg listed a few sample variants, especially in the catchwords. Noting moreover that the speech prefixes are not indented in 82a except in the A gathering, whereas 82b extends the indenting of speech prefixes to the outer forme of B, Greg concluded that 82a (STC 17086) was the earlier quarto.

Thus, although Greg identified two quartos in 1584, he implicitly confirmed Bond's choice of copy-text for his edition, even though Bond had consulted the British Library copy only.

The second edition of the STC Catalogue, published in 1976, updated its information on the basis of Greg's discovery by assigning a new STC number, 17086.5, to Greg's 82b, by this time known also to exist in a third copy in the National Library of Scotland. (A supposed fourth copy of this quarto, reported in the National Union Catalogue for the United States to be located in the City University of New York, turns out only to be a photographic copy of the Huntington quarto.) Yet this edition of the STC catalogue once again accepted the assumed priority of 82a or 17086, even though Greg's discussion of the merits of the case had been necessarily very brief and assertive; Greg speaks only of the indented speech prefixes and of "other details" which he does not specify. This present essay will argue that Greg's ordering of the two 1584 quartos is incorrect, and that his 82b, or STC 17086.5, should be regarded as the first quarto and the appropriate copy-text for a critical edition.

The Stationers' Register entry for Sappho and Phao in 1584 is as follows:

Having undertaken to publish Campaspe in the same year without previous entry in the Stationers' Register, Thomas Cadman evidently decided to guarantee legal protection in his second venture with a Lyly play. The inclusion of Lyly's name in the entry, and the conditional nature of Cadman's obtaining license to publish the play in question, suggest (as with Campaspe) a close involvement of the author in the publishing of his own play and a wish on Lyly's part to obtain a text of literary integrity.

Certainly the play's first publication in 1584 resulted in a reliable text. This is true whether we regard 17086 or 17086.5 as the first quarto, for they are both carefully printed, and one is simply a reprint of the other. STC 17086.5 bears the following title page and colophon:

The other quarto, STC 17086, bears an identical title page (except, as Greg notes, that 17086.5 uses a swash capital in Boyes and a thick lead above the date 1584) but varies the colophon as follows:

An examination of the headlines or running-titles ("Sapho and Phao"), in which I have been very substantially assisted by Paul Werstine, suggests that in 17086.5 one skeleton-forme was used for the inner and outer formes

of BDF; there is, for example, a recurring swash P on BDF 2,3. A single skeleton-forme also appears to have been used for the inner and outer formes of ACE, but was rotated 180 degrees between the imposition of inner C and outer C; moreover, two of the headlines in it (one found on C3 and C4, another on C3v and 4v) were interchanged between the imposition of outer E and inner E (where the first of these headlines appears on E1v and E3, the other on E3 and E4v). The case for a single skeleton in 17086 for all formes is clear throughout A to D, after which there is considerable disruption. A swash capital P recurs on A3v and A4v and on BCD 1,2; a distinctive p in Sapho recurs on A4 and BCD 1v,2v; a spaced period is visible on BCD 3,4; another distinctive p in Sapho appears on B4v and CD 3v,4v; and another headline, distinguished by yet another p in Sapho, appears on B3v alone. This last unique headline strongly suggests that inner B was composed before outer B in 17086, since it appears singly and does not recur whereas the headline on B4v is repeated in later gatherings.

The three extant copies of 17086.5 appear to be identical, except for a torn page (C2-C2v) in the Worcester copy that obliterates the page signature, the catchword on C2, and several words of the text, and occasional variations among the copies in the darkness of the inking. A comma after "blind" in line 22 of E3 and a period after "common" in lines 14-15 of D2 that are present in the other copies appear to be missing in photographic copies of the Huntington quarto, but examination of the original shows the punctuation marks to be present though extremely faint. Similarly, the Scotland copy appears to lack a comma after "arowes" in line 9 of F2 in photographic copies, but the original shows a faint impression of what may well be a punctuation mark (and there are several other faint commas in the surrounding text). Such infrequent and isolated changes in punctuation would be unlikely in press correction in any case. The 17086.5 quartos are substantially identical.

The physical similarities and differences between the two quartos plainly suggest that sales of the first quarto were sufficient to support a quick reprinting, one in which almost nothing essential was changed. Both quartos are printed in fours from A through G2, A1v and G2v being blank. The reprinting is page for page with only one slight variation in page content and catchword, as noted by Greg. (There are other occasional variants in the catchwords—e.g., "Phao" and "Sapho" on E1v and E2 respectively are followed by a period in 17086, not in 17086.5, and others are noted by Greg— but only once does variation affect the contents of the page.) On A3, the catchword in 17086.5 appears to be "forge. What," although in fact only "What" is carried over to A3v. 17086's catchword is "his." 17086.5 thus prints "with fire in his forge" as part of A3, whereas 17086 carries over "his forge" to A3v. It is not easy to understand why 17086.5 would provide a two-word inaccurate catchword as it does if it were the reprint, since the printer's copy (17086) would have been perfectly clear and correct, and the 17086.5 compositor, printing A3 and A3v on opposite formes rather than seriatim (as we have seen), could not have known in printing the A3 catchword that "forge. What" would follow, whereas this would be an easy thing to do in

following manuscript copy. 17086's better version can thus be seen as a compositor's correction of a natural error in his 17086.5 copy.

The reprint is generally line for line, except for the following passages: lines 4-10 of the spoken text on A3; the last line of that page and the first 4 lines on A3v; lines 3-4, 6-8, 11-12, and 17-18 on B1; lines 4-5 on B2v; the last four lines on C2; lines 24-25 on C4; lines 16-19 on F1; lines 9-10 on F2; the first two lines on F2v; lines 22-23 on F4; lines 6-8 and 10-11 on F4v; and lines 2-3 and 10-13 on G2, the final page containing "The Epilogue." This is a small percentage of the total. It is clear that the printer of the reprinted text, whether it be 17086 or 17086.5, was following the first quarto as his copy and wished generally to keep to its arrangement of lines, for, even when crowding requires him to depart from his copy in this respect, he repeatedly maneuvers his type to rejoin the lineation of his model and avoid relining a whole paragraph.

The two quartos are similarly alike, as Greg notes of 17086, in generally indenting the second line of a short prose speech whenever the prose turnover is short enough not to result in a third line of text for the speech in question. Otherwise, speeches of two substantial lines or more are returned to the left margin below the speech prefix. The quartos are identical in this feature, and priority cannot be established on this basis.

In a number of formal and nonsubstantive respects the two quartos are physically different. The printer of the second quarto uses different initials in every instance where they appear, on A2, A2v, A3, and G2. Although the lace border decoration and groups of type ornaments on the title page are basically identical in the two quartos, 17086 provides a border along the bottom of A2v to help fill a blank space below "The Prologue at the Court," and fills the space at the bottom of G2, after "The Epilogue," with a decorated border different from that on G2 in 17086.5 (made up, in fact, of units used in the center of the title page in both quartos). These differences do not help greatly in deciding which quarto was printed first. One might wonder why, if 17086.5 were printed second, it would choose to omit a decorated border on A2v and why it would choose a decoration on G2 that differs in style from that of the title page; however, it is also possible to construct a scenario in which 17086.5, printed to meet unanticipated demand for a second edition, simply omitted some ornaments. 17086's decorated borders are of a piece and suggest an attempt to impose stylistic unity where previously it had been more lacking, but the typographical ornaments used to build the decorative border on G2 in 17086.5, though not exactly the same as those in the decorative border on the title page, are in fact identical with those of the units within the border on the title page. Each quarto has its own consistency of style in the use of ornaments.

Some other differences are of no use in determining priority: 17086 is missing a page signature on B4 that is present in 17086.5; B4's catchword, "Phao," is on the final line of text in 17086 whereas it is below the text and on the same line as the signature in 17086.5; on B2, line 21 and following,

17086.5 prints an Exeunt on the same line as the end of the scene's last speech whereas 17086 drops the stage direction to a new line; and so on.

Other formal differences do tend, on the other hand, to support the argument that will be advanced here that the first quarto of 1584 is in fact 17086.5 and not the British Library's 17086. One such difference, having to do with indented speech prefixes, is, as we have seen, noted by Greg but is used by him to advance the opposite theory of the priority of 17086. The facts are as follows. Both compositors begin with indented speech prefixes in the first gathering, on A3v, A4, and A4v. (There are no speech prefixes previously in the text except for one on A3 that, because of an initial, cannot readily be indented.) The 17086.5 compositor then carries the practice of indentation of speech prefixes on into what Greg assumed was the first-set forme (the outer forme) of his second gathering, on B1, B4v, and B2v-B3, all of these pages lying on one side of a single printing sheet, and then chooses not to indent speech prefixes on B1v, B2, B3v, B4, and throughout the rest of the play. The 17086 compositor's practice is identical with that of 17086.5 for the first gathering, but he then regularizes to unindented speech prefixes for the rest of the play. Greg implies that the 17086.5 compositor resisted the change he found in his copy for a time but then yielded to its example in the second (inner) forme of B.

Yet Greg's argument is eminently reversible. If the 17086.5 compositor went first, inconsistently setting one forme with indented speech prefixes as he had done in gathering A and the other with unindented speech prefixes (perhaps because of something unclear or inconsistent in his manuscript copy), then the printer's copy for 17086 would have appeared to be a muddle. The B gatherings of both quartos appear to have been set by formes, but the printer's copy for the second quarto was presumably the earlier quarto, its pages arranged in numerical sequence. It is easy to suppose that the 17086 compositor, seeing that his copy for the B gathering inconsistently interspersed indented and unindented pages, proceeded to remedy an inconsistency he could not explain. Especially if he started with the inner forme in setting B—and Werstine's analysis of headlines and skeleton-formes above strongly suggests that he did—this compositor may have followed his copy (17086.5), noting that the speech prefixes were now unindented, and then, moving on to the outer forme and finding the speech prefixes in his copy indented, decided in favor of unindented consistency. The 17086.5 compositor, printing first, may have composed his outer forme first, continuing in it the practice of indented speech prefixes, after which, for some reason, he shifted to unindented speech prefixes. This hypothesis is more logical than the reverse, to suppose that the 17086.5 compositor might have followed indented speech prefixes from his copy in the A gathering and then, faced with consistently indented speech prefixes throughout the B gathering in his copy, chose to indent on one side of the sheet and not on the other. (Compare the first two quartos of Hamlet, where we can see that at the beginning Q2 [sigs. B1-B1v] must have been set from an annotated copy of Q1 since both texts begin with

speech prefixes that are flush with the margin and then shift to indented speech prefixes at the same point in the text ["Hora. Tush, tush, twill not appeare"] but at a point where Q1 begins a new page [B1v], whereas the longer Q2 text is a third of the way down the second page when the shift occurs.)[1]

A comparison of the two quartos in their use of periods after speech prefixes is easier to explain if 17086 is the reprinted text. In the A gathering, 17086.5 is somewhat inconsistent but tends toward distinguishing between periods after abbreviated speech prefixes and none after unabbreviated ones. 17086 meantime consistently prefers a period in all cases. Gathering B continues 17086.5's inconsistency, so much so that it is hard to say whether the text is distinguishing between abbreviated and unabbreviated speech prefixes (as on B2v and B3) or preferring periods in all cases (as on B3v and B4). Some pages in gathering B (1v, 2) mix the use of periods and no periods after unabbreviated speech prefixes. 17086 meantime also hesitates and mixes forms on the same page (e.g., B1v, B2, B3, and B4). In gatherings C and D, 17086.5's preference is by now clear in favor of periods after all speech prefixes, while 17086 opts in most cases, especially on the inner or second forme, for distinguishing between abbreviated (with periods) and unabbreviated (without periods) speech prefixes (though there are exceptions on C1, C1v, D2, and D3). Beginning with E1 and continuing to the end of the play, 17086.5 is entirely consistent in affixing periods after both abbreviated and unabbreviated speech prefixes, while 17086 follows a similar practice of affixing periods in all cases (but with exceptions on E1 and F3). Since the practice that is ultimately adopted by both, to provide periods in all instances, is consistently adopted by 17086.5 nearly two whole gatherings earlier than 17086, it appears more likely that 17086 yielded to following copy at the start of gathering E rather than that 17086.5, having preferred periods in all cases, suddenly found his copy conforming to the practice he had already settled on. The pattern in both quartos militates against a theory of two compositors, since there is no discernible shifting back and forth between the A, C, E, G gatherings and the B, D, F gatherings.

The forms of abbreviation in speech prefixes suggest a sporadic effort on the 17086 compositor's part to regularize the greater variety in 17086.5. (The text is a "literary" one throughout, with characters' names grouped at the heading of each scene, with sparse stage directions, and with little evidence of playhouse practice or prompt-book regularization.) The useful speech prefixes to watch in this regard are Trachinus (Trach., Trachi.), Mileta (Mile., Milet., Mileta.), Criticus or Cryticus (Criti., Cryti.), and Calypho (Caly., Cali.). Sybilla is sometimes abbreviated (Syl.), and Lamia is also spelled Lamya. In Act 3, scene 3, for example, the 17086.5 compositor generally uses Milet. as he has done previously for Mileta but prints Mil. at ll. 54 and 77 and Mile. at l. 79; 17086 prints Milet. or Milet in these instances, though he also spells out Mileta once at l. 89.[2] In the following scene as well, Act 3, scene 4 (mislabeled "Actus terius, Schaena prima" in 17086.5 and "Actus tertius, Schaena prima" in 17086), 17086 again offers a consistent reading of Milet. at ll. 38

and 40 where 17086.5 reads Mile. None of these instances shows evidence of crowding as a constraint on the use of a longer form; the lines are normally set, even loose. It would be hard to explain why the 17086.5 compositor arbitrarily switched to Mil. and Mile. if the more consistent text of 17086 were his copy. Of course the author's papers may have been irregular to begin with, but that possibility renders even less likely the hypothesis that 17086 applied a consistency to his copy which 17086.5 then proceeded haphazardly to undo.

In Act 1, scene 2, the 17086.5 compositor begins with Trachi. in line 1 but immediately switches to Trach. for the remainder of the scene except for another Trachi. at l. 70; the 17086 compositor prints Trachi. at l. 6, Trachinus's second speech, as though following the precedent established in l. 1, and only thereafter switches to Trach. Later, at l. 70, he prints Trachi. as does 17086.5. The hypothesis of 17086 as the second printer supposes that the compositor was attempting to regularize at first but then gave it up and followed his copy, whereas the reverse hypothesis must suppose that the 17086.5 compositor saw that his copy was going to shift to Trach. and so decided to anticipate that move at l. 6—a less plausible supposition.

Similarly, in Act 1, scene 4, 17086.5 inconsistently shifts from Lamia in the opening stage direction to Lamya for her first speech prefix, whereas the 17086 compositor corrects the speech prefix on the basis of the stage direction. Such a change is easier to explain by postulating the 17086 compositor as the printer of the second quarto than by the reverse hypothesis. The 17086.5 compositor could have ignored a consistent spelling of Lamia in his copy and produced Lamya out of nowhere, but it would have been arbitrary for him to do so whereas the 17086 compositor's revision is entirely plausible and consistent with the other attempts on his part to regularize.

Spelling preferences reinforce the hypothesis that the 17086 compositor printed the second quarto. When we consider the spelling variants, that is, words spelled one way in 17086.5 and another way in 17086, we can see that one compositor set each text throughout and that the two compositors had distinctive spelling preferences. The 17086.5 compositor has a marked tendency to add a final -e to words like whome, sweete, howe, woulde, paye, indeede, thinke, witte, foule, and so on. All these words are printed by the 17086 compositor with no final -e, and they appear in roughly equal distribution in outer and inner formes. There are, by a rough count, 42 occurrences of a term thus spelled without the final -e in 17086, and only 16 instances of contrary evidence in which the 17086 compositor chooses to spell with a final -e a word that the 17086.5 spelled without. When we consult the text for spellings common to both editions, we find the added -e everywhere: stinge, behinde, finde, heale, and so on. The likeliest hypothesis, then, is that the 17086 compositor, finding the final -e in his copy as a general rule and staying close to his copy in most matters, imposed his spelling preference sporadically but in enough instances to be statistically significant. The reverse hypothesis has to allow for a more random procedure on the part of the 17086.5 compositor or the author, one that generated a text with no consistency in this kind of spelling. Such an inconsistency is possible, and indeed the 17086

compositor occasionally chooses the final -e spelling against his usual habit, but probability inclines to the hypothesis that the author and/or the first compositor had one kind of consistency (favoring final -e) whereas the second compositor's spelling habits (favoring no final -e) ran counter to what he found in his copy.

The case is even stronger for 17086.5's preference for a final -ie in place of -y: readie, enemie, pollicie, studie, bewtie, ferrie, etc., all changed to ready, enemy, and the like. Here again the common spellings show a marked preference for the -ie ending and suggest that the 17086 compositor is at times imposing his preference rather than the reverse. There are 25 instances where -ie in 17086.5 reads -y in 17086, none in the other direction. Spelling preferences are not as conclusive evidence as other sorts offered in this essay, and one can construct a scenario in which 17086.5 tried to regularize the mixed spellings he found in 17086, which could in turn have been found in the manuscript copy, but such a hypothesis begins by assuming that the 17086 compositor haphazardly vacillated between -y and -ie or was following arbitrary spellings in his manuscript copy, whereas the hypothesis that 17086 came second assumes that the 17086.5 compositor and/or author had a predilection for one spelling (-ie) whereas the 17086 compositor had a predilection for another spelling (-y), one at variance with his copy.

The statistics on the spelling of stinges, lockes, kissinges, thinges, wordes (the common spelling in many instances, and 17086.5's preference when the two texts vary) as opposed to stings, locks, kissings, and the like (appearing almost exclusively in 17086) are virtually as onesided: 22 cases where 17086 prefers -s to -es, and only two in the other direction. Because the common spellings in the Prologue (for example) show a marked preference for meaninges, mirthes, fooles, habites, eares, reportes, cheekes, etc., the spelling evidence tends to support the hypothesis of 17086 as the reviser on the grounds already cited in the previous paragraph, granted again that spelling preferences are not as conclusive as some other kinds of evidence. In any case, the spelling preferences of both quartos are common to inner and outer formes and to all gatherings, strongly suggesting a single compositor for each quarto.

Another onesided statistic has to do with ampersands in place of "and." The 17086.5 compositor prefers "and" where 17086 chooses the ampersand 21 times; the reverse is true only three times. Ampersands are fairly common in both texts and are found jointly in 24 instances, but 17086's augmentation of the number is still large as a percentage. Examination of individual instances points to a conclusion that is borne out in other ways as well: if we assume that the 17086 compositor is printing the second quarto, then we can hypothesize that he often finds his lines crowded when he attempts to follow his 17086.5 copy virtually line by line but in a more generous sized type (albeit in a wider column width). Reasonably clear examples are at 1.3.30, 1.4.48, 2.1.134, and 2.3.87; and there are others. When on the other hand the 17086 compositor changes from ampersand to "and," as at 1.1.2, he does so where the line in 17086.5 is loosely composed or where (as in this instance) the providing of a new initial has given the compositor a wider column width

beside the capital. This evidence is reversible in individual cases (i.e., to suppose that the 17086.5 compositor had to work with more cramped space beside initials), but it is less reversible in the numerous cases not involving initials, where we would have to assume that the 17086.5 compositor, finding ampersands in his copy text, repeatedly chose to spell out "and" despite his working with a narrower column width (and smaller type, to be sure); the lines hypothetically thus produced in 17086.5 are all quite crowded and seem like odd places for the compositor to have insisted on spelling out a word that both texts are content to abbreviate elsewhere. Cumulatively, this evidence is consistent with a hypothesis of 17086 as the reviser.

The 17086.5 compositor is somewhat less likely than the 17086 compositor to use the tilde over a vowel for the nasal, as in vppõ, the͂, cã, heaue͂, and the like, though 17086 also spells out some words that are abbreviated thus in 17086.5. Individual cases show space considerations at work, as in 1.3.30 (me͂), 2.1.119 (cõsume), and elsewhere.

Spellings of certain distinctive words confirm 17086's role as printer of the second quarto. In Act 2, scene 3, for example, 17086.5 is generally consistent in the spelling of deuill and deuilles; it appears thus eight times, although once as deuil and once as deuils. 17086 clearly prefers diuell, for he spells it thus in the first three instances, and subsequently has diuelles, diuels, and diuell. Beginning with the word's fourth appearance, however, he vacillates, printing deuill twice and deuil once, mingled among his preferred spellings. Each of these departures from his preference exactly reproduces the spelling in 17086.5. 17086 is influenced by 17086.5 in this regard not only in the matter of eui - iue but also in whether to double the consonantal ending or not, capitulating to 17086.5's somewhat random choice of -l or -ll. This last point is reversible as evidence, since the two texts are identical in their choice of doubling the consonantal endings, but the hypothesis of 17086.5 as a second printer does not explain why 17086 would have been so random in the first place and why 17086.5 would then have regularized to deuill in regard to the vowels but followed 17086 in the matter of final consonants.

One small but telling bit of evidence, pointed out to me by Eric Rasmussen, is that an apparently stray mark on C2v of 17086.5 between lines 14 and 15, caused perhaps by a space that has worked up, is converted by the 17086 compositor into an unnecessary and intrusive comma. The evidence is of the mechanical sort that enabled W. W. Greg to determine the priority of nearly identical editions of The Elder Brother.[3]

Concluding then that the 17086 compositor set the second edition, Q2 (not to be confused with Bond's Q2, dating from 1591), we have in the Q2 compositor a workman who makes relatively few errors. "Teech" for "teeth" in 2.1.101 is an obvious misprint; "Kanope" for "Canope" in the opening stage direction of 3.3 is an unexplained vagary; a comma at the end of "Actus secundus, Schaena prima," is inconsistent with his usual format; punctuation omissions are sometimes a clear lack of improvement (as in 2.1.28, "Lady nothing . . ." or 2.1.35, "No I may not obtein but . . ."); and the like. On the whole, though, the Q2 compositor is conscientious and accurate. Bond's

reliance on Q2 as though it were Q1 for his copy-text has hitherto not introduced many difficulties for a modernized text, though it does to be sure make for very numerous inaccuracies in his old-spelling edition.

There are finally some readings in which a substantive difference occurs, and so we come to the central question as to whether the Q2 compositor had access to any reliable authority for the few such changes that do occur. A survey of the substantive variants in the order in which they appear reveals evidence of several errors (quite a tolerable percentage) but also of some persuasive new readings. As an example of manifest error, the Q2 compositor fails to correct Q1's "needelesse" in the Prologue at the Court, l. 9; a needed change to "needle's" had to await Q3 of 1591. (This is of course not a Q1/Q2 variant.) On the other hand, the correction at 1.2.73 of Q1's "And you sir boyes" to Q2's "And you sir boy" is clearly justified, though "boy" is not, as Bond supposed, a Q1 reading that was misprinted as "boyes" in what he calls Q2 (1591); instead, it looks like a sensible guess that any good compositor could have made. At 1.3.2, Q2's "the diet" for Q1's "thy diet" looks like either a misprint or an astute guess based on the assumption that there ought to be a Lylyan antithesis between "thy commons in Athens" and "the diet at court," one representing a scholar's life and one a page's; "the" is followed by Bond in the mistaken impression that it represents the original reading. A closer examination reveals, however, that Q1's reading is right after all, since in his reply Molus refers to both diets as his own: "there of a little I had somewhat, here of a great deal nothing."

Q2's "wordlings" for "worldlings" at 1.3.22 is a plain error, mistakenly preserved by Bond. So is the omission of the word "it" at 1.4.6 (silently supplied by Bond without a textual note). Q2 repeats an incorrect speech prefix, Sapho instead of Phao, at 2.1.60. Q2's "vine" for Q1's "vines" at 2.1.142 is clearly an error, though preserved by Bond. So is "in" for "is" at 2.2.2, treated by Bond as an emendation. Q2's "hearken" for "yearken" at 2.1.30, on the other hand, is the simple correction of a misprint (missed as such by Bond, since he assumed "hearken" to be the original reading). Q2's "Wilt thou forsake the ferrie" for Q1's "Wilt thou forsake thy ferrie" at 2.2.25, though preserved by Bond as the presumed Q1 reading even if "thy" in the later texts must have seemed attractive to him, is clearly a misprint in Q2. Q2's printing of "a" for Q1's "an" at 2.3.24 cannot be justified, though it is followed by Bond. All the instances thus far plainly suggest, moreover, that Q3 (1591) follows Q1 for its copy, since it retains the error of "boyes" along with the correct original reading of "worldlings," "vines," "is," and "an," and the two correct original readings of "thy."

Q2's substitution of "lesse" for Q1's "losse" at 2.4.74 is plausible, though Bond is wrong in ascribing it to the original text; it reads like a compositor's commonsense emendation, especially in view of the fact that Q1 has a typographical error two words earlier in the same line ("requirerh"). Paleographically, the confusion from similarity of letter formation is an evidence for the error "losse" having been set. Again, Q3 follows the error of Q1. In 3.1.30, Q2's "weane" is an astute emendation of Q1's "weaue" (not recorded by

Bond), but not beyond a compositor's capability in that the metaphor is plainly discernible in "noursed" in the same line. At 3.2.47, Bond's choice of Q2's "being holde" is clearly misinformed; the phrase is "being holden," not only more idiomatic but the authentic reading of Q1 and followed by Q3.

Some Q2 readings are good enough to raise momentarily the possibility of independent authority. "Boy," "losse," and "weane" are fine emendations, if commonsensical. The emendation of Q1's "crooked" to Q2's "croked" at 3.3.64 is also perhaps self-explanatory since it follows the word "Rauen." (Bond does not record this, though Q3 follows Q1's "crooked.") The Q2 reading of "reare me vp, my bed, my head" at 3.3.84-5 for Q1's "reare me vp, my bead, my head" is interesting and probably correct, using "me" to mean "for me," but bead is a most unusual form for bed (not in OED) while it is an easy misprint for head, and the sentence as a whole could possibly be intended to read, "reare me vp, my head, my head lies too low," as indeed its punctuation in Q1 and Q2 suggests. More likely, the compositor of Q1, memorizing his line as he set, was influenced by "head" and produced the unusual spelling "bead" accordingly. The Q2 reading of "hearte" for Q1's "heate" at 3.3.107 is, on the other hand, more complex. The whole phrase, "and thou Phoebus that in the pride of thy heate shinest all daye in our Horizon" (so printed in Q1) might support "heate" or "hearte." Is the Q2 compositor guessing, and if so do we go along with him or print "heate" as did Q3 and subsequent texts (except Bond, who believes he is following the original copy in "hearte")? In any case, as we shall see, this and "my bed" at l. 85 are the only serious candidates for independent authority in all the substantive variants of this play, and since "heate" can so adequately be explained as a compositor's guess (probably unnecessary), there is finally no reason to suppose that Q2 had access to any other authority than Q1. Conceivably Q2's guess is right here, since "pride of thy heart" is such a recognizable phrase and since "heate" is such an easy misprint for "hearte," but a conservative choice by an editor would be understandable. Q2's emendation of "binde" to "blinde" at 3.4.11 is possible but unnecessary; it seems at best a plausible word in a speech about the delight of beauty to the senses, and "binde" could be an easy misprint for "blinde," but the change may instead be a printer's sophistication, since "binde" looks forward in the same sentence to the image of being rocked to sleep as if bound up in swaddling bands.

Q2 fails to correct an error in the speech prefix at 3.4.39: Q1's Sapho should clearly read Phao (though unrecorded by Bond). Nothing can support Q2's "sight" (twice) for "sighe" and "sigh" at 3.4.76 and 80, despite Bond's attempts to rationalize what he takes to be the original reading; possibly "sight" is Q2's spelling variant (OED allows it as such); since Q2 also prints "sighe" in this same sentence, it appears that "sight" in the first instance is either a simple misprint for "sighe," after which the Q2 compositor changed "sigh" to "sight" in an attempt to be consistent with what he had begun, or the sporadic result of his choosing a somewhat idiosyncratic spelling against the preference of his copy. Q2's "one" for "on" at 4.3.13 is clearly a misprint, erroneously preserved by Bond. At 4.3.43, Q2's "the best friend with the

tatling" makes things worse than Q1's "the best friend with thy tatling"; the phrase should read "thy best friend with thy tatling," and Q2's "correction" may represent a bungled attempt to emend what he found in his copy. Q2's "ouerwatching" for Q1's "ourwatching" at 4.3.58 is, on the other hand, a plausible emendation, even if "ourwatching" can be made to work by dividing it into two words or regarding it as a variant spelling of or printing error for "o'erwatching." But Q2's "though" for Q1's "thought" at 4.3.72 is a simple misprint. So is Q2's "women" for Q1's "womens" at 4.3.96. At 4.4.28, Q2's "one his heart," followed mistakenly by Bond, is a simple misprint for Q1's "on his heart."

In 5.1, Q2 is probably intrusive in emending Q1's "mistake it not" to "mistake me not" at line 21; the need for the change is not obvious. Q2 normalizes "Swite" (cf. OED "swiete" etc.) to "Sweete" earlier in line 21 (an excellent example to confirm the order of the Qq.). In the following scene he misprints Q1's "Mee thinkes" as "My thinkes" at line 3. His rendition of Q1's "withdrawing" in line 4 as "withstanding" is interesting, and might merit serious consideration if this were the original text (and of course Bond prints "withstanding"); but Q1's phrase, "as it were a withdrawing in my self," is plainly so intelligible and right that Q2's attempt has to be regarded as some kind of aberration and not based on any reliable authority.

In summary, all the worthy emendations in Q2, of which there are several, can be explained as the work of a conscientious compositor, such as the Q2 compositor appears to be from other evidence presented here. The number of errors in Q2, including such entirely characteristic errors as the for thy (twice), tends to support the overall conclusion of this essay that 17086 is in fact the second quarto, since texts continuously degenerate in transmission, even if it must be emphasized that the discussion here of editorial decisions does not bear primarily on the earlier question of establishing priority; the evidence presented earlier is of a more conclusive sort. The compositor who set Q2 worked intelligently and has left no implausible emendations directly traceable to error. The possibility that he was working with a marked copy or corrected proofs seems minimized by the sparsity of the positive corrections and their commonsense quality. We have no reason to follow Q2 except where the emendations are self-justifying.

The subsequent printings of Sappho and Phao produced no new textual authority. We have already sampled above the plentiful evidence that Q3 (STC 17087; Bond's Q2), printed by William Broome in 1591 (although with no record of transfer in the Stationers' Register), is a reprint of Q1 without recourse to Q2. The pagination is identical with that of Q1. A colophon is lacking. Other changes are confined to routine printer's errors in a page by page and line by line reprint from Q1, along with some arbitrary editorial changes such as the following: "the Theater of Athens" for "the Theater at Athens" in "The Prologue at the Blackfriars," l. 12, "which" for "that" in "The Prologue at Court," l. 10, "glaunce at others faces" for "glaunce on others faces" at 2.1.99, "will I" for "I will" at 2.1.127, "a deuill" for "the deuill" at 2.3.50, "Be they all deuilles that haue hornes" for "Be they all

deuilles haue hornes" at 2.3.70, "it is" for "is it" at 3.2.11, "gyue your selfe leaue to slumber" for "gyue your selfe to slumber" at 3.3.32, and "to be imaginatiue" for "to be the imaginatiue" at 4.3.49. None of these emendations is more than uninspired guesswork or stylistic "improvement," and cumulatively they provide no evidence of authorial intervention. As in the case of Q4 of Campaspe in the same year, the transfer of the play to Broome seems not to have involved Lyly.

Edward Blount's Six Court Comedies of 1632 (STC 17088) is simply a reprint from Q3 with minor changes except for the addition of the songs. Typical of readings that show the dependency of the 1632 reprint on Q3 are the following: Prologue at the Blackfriars, l. 12, at Q1-2, of Q3-Bl.; l. 17, bee from Q1-2, bee free from Q3, be free from Bl.; II.i.65, read Q1-2, red Q3-Bl.; II.i.67, waxed Q1-2, wexed Q3-Bl.; II.i.77, suckered Q1-2, succored Q3, succoured Bl.; II.i.99, on Q1-2, at Q3-Bl., etc. Blount's major contribution was to provide songs to the plays where in quarto they had only been indicated by stage directions. The issue as to whether to include the songs in a critical edition of Lyly's play has been much debated by G. K. Hunter, W. W. Greg, John Robert Moore, and M. R. Best, among others,[4] and is in any case another story. Apart from that issue, the editorial implications of this textual history of Sappho and Phao seem clear. Q1 is the authoritative text. Even if one should normally be on the lookout for priority in the latest edition to appear in the author's lifetime, Lyly seems to have had nothing to do with the minor changes successively introduced into Q2 and Q3. The Blount edition of 1632, appearing after Lyly's death in 1606, seems also to record no corrections or additions (other than the songs) that might have been left by him, though it, like Q3 and especially Q2, does introduce a few sensible emendations that commend themselves to the editor.

Notes