University of Virginia Library

The order of the editions, the textual transmission, and the authority of the various texts of Christopher Marlowe's translation of Ovid's Amores have not been completely studied. The problems are complex but, fortunately, not insoluble. In all, six editions have been preserved,[1] each undated and purporting to have been printed in Middleburgh in the Low

Countries.[2] Two of these editions — obviously the earliest — contain a selection of only ten elegies[3] which are printed as the second section of a book that begins with the Epigrams of Sir John Davies and includes as well three amatory poems headed 'Ignoto'. The third and later editions present all of the elegies;[4] in these editions the Ignoto poems are dropped and Davies's epigrams follow the elegies, distinguished merely by a head-title. The titlepages reflect this different order, obviously an indication of the relative popularity of the two parts.

The earliest two editions, those containing the selection of the elegies, are listed in the Short-Title Catalogue under number 6350 as one edition, erroneously as duodecimos, and with a ghost copy in the Bodleian Library. The cross-entry 18930 under Ovid corrects this entry (although misspelling the imprint as 'Middlebourgh'), lists the two properly as octavos, drops the Bodleian copy, and assigns the order as the Huntington copy (1595?) and the British Museum copy (1598?).[5] Strictly bibliographical evidence, not previously presented, confirms this order.

The earliest known edition, then, is O1, preserved in the Henry E. Huntington Library, an octavo collating A-G4 (A1, G4 blank and wanting), unpaged, with the titlepage on A2 reading 'EPIGRAMMES | and | ELEGIES.

| By I. D. and | C. M. | [two blocks of type-orn. arranged horizontally] | At Middleborough.' The text of Davies begins on sig. A3 and ends on D3v with a 'FINIS. I. D.' The second section begins on sig. E1 with the title 'CERTAINE | OF OVIDS | ELEGIES | By C. Marlow. | [two blocks of type-orn. as on A2] | At Middleborough.' The type of the imprint and the type-ornaments on the two titlepages are identical. Sig. E1v is blank. Marlowe's text begins on sig. E2 and ends on G3v. The first signed leaf is signed 'A3'. Thereafter only the first leaf of a gathering is signed, with the letter alone, as 'B', 'C', and so on, and the second and third leaves are signed only with the arabic numbers '2' and '3' respectively. The fourth leaves are unsigned. The first text leaf in Marlowe's section is signed 'E2', the third is simply '3', and thereafter the same system obtains (F1 is in italic). The general system for catchwords is to place a catchword only on sig. $4v of the gatherings. Catchwords in Davies's section occur on sigs. B4v and C4v. Sig. D4v precedes the title to Marlowe's poems and thus has no catchword. In the Elegies sig. E4v lacks a catchword, probably because Marlowe's signature at the foot of the elegy ending on E4v left little room. However, F4v has the catchword 'Amorum' for Elegy III.vi beginning on G1. The first three elegies, all in sheet E, are signed at the foot 'C. Marlowe.' and 'C. Marlow.' (twice). Thereafter, starting with II.xv on F1r-v the elegies are unsigned. The principle of selection seems to have been to start Book I and thereupon to make a selection from the more erotic of the poems, arranged in no perceptible order.

The system of signing the leaves is a characteristic of the Edinburgh printer Robert Waldegrave, but some doubt may be cast on his having printed this book, not so much by the Middleburgh imprint (which is a convention and need not be taken seriously), but instead by the system of placing catchwords only on the last page of gatherings (as in some early manuscripts and the books that imitated them) — not one of his habits. Firm establishment of the unknown printer could be made only by identification of the mix of specific broken and damaged types in his font corresponding to the mix in this book. It may be that Waldegrave would prove to be the printer, or a printer abroad, or even Thomas Scarlet in London who — the late Mr. John Crow confirmed — also used this signing convention on occasion. The date is equally uncertain.

The second edition (O2), another octavo, has been known only in the British Museum copy (C.34.a.28), which is imperfect, wanting sig. A4 of the text as well as the blanks A1 and G4. However, an unrecorded copy exists in the Carl H. Pforzheimer Library[6] which, though wanting A1, G4, does preserve A4, which contains a substantive variant from O1 in line 4 of "In Rufam" that is perpetuated in later editions. This O2 is a close imitation of the Huntington edition and collates the same, with the same

system of signing the leaves. The titlepage repeats that of O1 except that the imprint reads 'At Middleborugh.' both in the general and the section title; however, although the same design of type-ornament is employed, only one block, made up from two halves, decorates the titles. The section title reads, 'CERTAINE | OF OVIDS | ELEGIES. | By C Marlowe, | [block of type-orn.] | At Middleborugh.'

When the two earliest editions of a book are paginal reprints of each other with the same date or undated as in the present example, general typographical characteristics and the corruption of readings may offer some small hints as to their order, but the case can seldom be really proved when strict bibliographical evidence is absent. Fortunately, such evidence can be identified in these two editions to demonstrate with certainty the priority of the Huntington (CSmH) edition and the fact that the British Museum and Pforzheimer edition is the reprint. For example, on sig. D3, in line 10 of Davies's Epigram 47, the error 'starres' occurs in O2, which can be explained by reference to CSmH 'States' with a broken second 't' that resembles an 'r'. A better example comes on sig. E3v in Marlowe's Elegy I.v, line 10, where the O2 nonsense error 'trells' is explicable only by reason of the use in CSmH 'tresses' of a broken ligatured pair of long s's that closely resemble a double 'll'. Similarly, on sig. F4v in Elegy II.x.36 the O2 error 'let' was caused by a broken long 's' in CSmH 'set' that is very like an 'l'. Less obvious corroborative evidence of the same kind appears on sig. E3 when an O2 period in I.iii.18 can perhaps be explained as resulting from a CSmH comma so broken as to resemble a full stop. Four lines down better evidence appears in the commonplace error of O2 'loue' where CSmH 'Ioue' is set with a damaged 'I' that is difficult to distinguish from an 'l'. A really anomalous period in O2 on sig. G3 appears in I.ii.34 after 'Io' as the result of a damaged comma in CSmH that could readily be mistaken.

Although the compositor of the BM edition imitated the unusual signing of the Huntington edition, he was less successful in restraining his normal impulse to set catchwords. Hence he omitted the catchword only on sigs. B2r-v, B4, C2v, C3, D1, D2, D3, D4, E2, E2v, F1v, G1 (and on E3v, E4v where the Marlowe signatures left no room). This irregularity suggests copying CSmH rather than an original most eccentric practice. Another small piece of evidence is useful for the same conclusion. The British Museum edition is a line-for-line and page-for-page reprint up to sig. G1, where Huntington has 23 lines but BM only 21. On G1v and G2 CSmH has the usual 29 lines, whereas BM adds an extra line 30 in order to catch up. As a result, on G2 both editions end with the same line and the paginal reprinting is resumed. The anomaly in BM seems to have been created by two extra lines of white space set under the title of III.vi heading sig. G1. It would seem that when he discovered the error the compositor preferred to tie up the page and to add the extra lines to G1v and G2 instead of disturbing the title arrangement.

The text of O2 is marked by many literal errors, more usual perhaps in a reprint than in an original, and by a considerable amount of corruption in the readings. In the Epigrams O1 prints two lines in no. 40, "In Afrum," not found in any other edition. One cause of the omission — the eyeskip caused by the repetition 'No sooner' — suggests, further, that O2 was the reprint. In O1 lines 9-12 read

In O2 and in all other editions lines 11-12 are wanting. (In the Elegies O2 omits I.xiii.14.) The Elegies contain a few variants that ordinarily would seem to go beyond reprint corruption. For example, on sig. E3 line 3 of I.iii O1 'I aske too much' becomes 'I craue too much', and in line 7 O1 'If loftie titles cannot make me thine' becomes the hypermetrical 'Yf loftiest titles cannot cause me to be thine'. Yet this apparently is true corruption (picking up Marlovian readings elsewhere), for this page has other examples of manifestly unauthoritative variants. In line 2 the sense is sophisticated by the substitution in O2 of 'euer' for O1's 'neuer' in the line 'Either loue, or cause that I may neuer hate'; in line 6 a hypermetrical 'thee' is added after O1 'loue' in "Accept him that will loue with spotlesse truth', an error picked up by contamination from the preceding line, 'Accept him that will serue thee all his youth'. In line 19 O2 has the dittographic 'the the' for O1 'the'. More of this sort of variation follows on sig. E3v where in I.v.23 O1 'likt' is altered to O2 'pleasde' and in the next line O1 'naked bodie' becomes 'faire white body'. The variants continue sporadically, as in III.xiii on sig. E4 where in line 2 O1 'But let not mee poore soule know of thy straying' is changed to O2 'wit of thy straying'; in line 16 'folke' becomes hypermetrical 'people', and in line 18 'tricks' is altered to 'toyes'. In the third edition where the text of the first edition O1 has in some part been collated against another manuscript, all such cases of O2 variation are ignored and the readings remain those found in O1. This fact appears to confirm the belief that the altered words in O2 when they are deliberate and not merely mechanical errors or memorial lapses have no authority, that their nature is such as to make it highly doubtful that they derived from consultation of another manuscript, and thus that they probably originated with the publisher looking over the copy and 'improving' it for the worse, even though a few necessary corrections were made in the process.

The first of the complete editions, which substitutes 'All' for 'Certaine' in the title and provides the full roster of elegies, is O3, represented by the Bodleian Library copy, Mason. AA.207. Other copies are found in the Dyce Collection of the Victoria and Albert Museum and in the Huntington Library. This is STC 18931a, but the STC attribution of two issues to the

Bodleian is in error. The titlepage reads, 'ALL | OVIDS ELEGIES: | 3. BOOKES. | By C. M. | Epigrams by I. D. | [double row of three asterisk orns., each row within square brackets] | At Middlebourgh.' This octavo collates A-F8G4; Marlowe's elegies are now arrayed in order (as distinct from their O1 jumble), starting on sig. A2 and ending with 'FINIS.' on sig. F3v. Davies's Epigrams start on F4 with a headtitle and end on G4v with 'FINIS. I. D.' The elegies not previously printed must, of course, have come from some manuscript. The possible effects of this manuscript on the text of the reprinted Certaine Elegies will be touched on below; but, first things coming first, one must see whether these ten elegies were printed in O3 from this manuscript. Or, on the other hand, whether in the standard Elizabethan manner — as has been demonstrated in the copy for such texts as the Folio version of Shakespeare's Othello, or of Richard III, or Troilus and Cressida, or King Lear — the publisher or the printer of O3 may have manufactured what was regarded as superior copy for the compositor by annotating the ten poems of O1 or O2 to bring them into general conformity with the readings of the manuscript. Until one or other procedure is established, no critic (or editor) is in a position to evaluate the authority of specific readings in the texts, not alone where they disagree but even where they agree with each other.[7]

As a source of the Elegies in O3, O2 must be ruled out because O3 consistently retains O1 readings in all but four indifferent cases of variation between O1 and O2.[8] The best evidence that O1 served as printer's copy for the ten selected elegies would be mechanical and bibliographical, like that utilized to show that O2 must derive from O1 because it misread damaged O1 types. As is usually found when annotated copy is in question, this evidence is too sparse to take the whole weight of the case, but a few examples may be observed. In O1 I.iii.9 the final 's' of 'lands' is damaged so that in the Huntington edition the top part barely prints. Given a lighter impression than in this copy the 's' might not have been legible, and such a fault could explain the O3 reading 'land'.[9] More trustworthy evidence

comes at I.iii.25 where a comma after 'rung' in O1 is damaged and looks very like a period, and a period (a real anomaly) is found in O3.[10] Then in III.vi.38 an O3 misprint 'moned' can perhaps be explained by reference to the typography of O1. In O1 the correct word 'loued' is turned under with a parenthesis so that the lines appear as It would seem more than possible that the O3 compositor's eye, coming to the end of the line with 'to haue', was confused and by a slip, visual or memorial, he set 'moued' (with a turned 'u') from the preceding line, as if it had been turned up, instead of the turned-under 'loued'.[11] Finally, a standard Elizabethan confusion in reprints (already observed in O1-2) appears in O3 at I.iii.4 where O1 reads 'Loue', O2 'loue', and O3, incorrectly, 'Ioue'. Although it is true that the confusion usually results from lower-case 'l' as in O2, in this example the O2 'I' is clearly inked but the Huntington's 'L' is damaged and might in haste be mistaken for an 'I', especially since at first sight the context might seem to call for Jove.[12]

These pieces of evidence do not add up to anything like demonstration but they may at least suggest the possibility that O3 had a printed source for the ten reprinted elegies. Less mechanical textual evidence may thus be brought up in support of the hypothesis. Of all parts of a text's accidentals, the punctuation is most likely to be compositorial owing to the usual light and sometimes almost non-existent punctuation of Elizabethan poetical manuscript whether dramatic or not. Thus identical anomalies in punctuation between two editions are more likely to reflect derivation of one from the other than independent compositorial faithfulness to faulty manuscript pointing. For example, in I.v.13-14 O1 (and O2) reads

where the comma after 'thin' is clearly wrong and should have been placed after 'gowne'. Although end-stopping the line with a comma after 'small', O3 repeats the mistaken modification offered by the comma after 'thin' instead of 'gowne'. Similarly, in II.iv.33-36 O1 (and O2) reads where O3 prints the last couplet as The removal of the O1-2 comma after 'troth' is only a partial improvement, for O3 still follows the lack of a comma or a stronger stop after 'rounder', an error that may well have originated in the O1 printed copy. One of the best examples of O1,3 concurrence in unusual or erroneous punctuation comes in I.iii.10. In O1 lines 7-12 read This passage is so contorted in its translation as to cause considerable difficulty. Modern editors treat lines 9-10 as parenthetical, but they are not so in Ovid nor, properly, in Marlowe. The difficulty comes in the failure of the translation to distinguish a number of parallel statements each prefaced by if, which are (1) if I am only of knightly birth, (2) if my land is so small that numberless ploughshares are not needed to till it, (3) if my parents are penurious and not generous, yet at least I shall have on my side Apollo, etc. The O3 editor or compositor seemingly could make little of these lines, and followed the punctuation exactly save for substituting a comma after 'line' for the O1 period. In the crucial and gnomic line 10, however, O3 — one would suggest not by accident — duplicates not only the wrong O1 comma after 'giuen' but also the wrong period after 'saue'.

Other punctuation anomalies are less striking, but several unusual placements of question marks in O3 seem to relate it to O1. An example occurs at I.i.11-12 in which O1-2 read

Here O3 follows the query after the first line instead of after the second, something of a characteristic of O1 typesetting. (O3, however, corrects the O1 error 'That' to 'What'.) Another case appears in III.vi.15-18, where O1 reads In O3 this is printed as This substitutes the more conventional spelling 'shade' for what is perhaps an O1 misprint, and corrects the O1 misprint 'Io'. Of more concern is the intelligent repunctuating that transfers the question mark after 'I' to substitute for O1's comma after 'say', the lack of a stop after 'I' perhaps being the result of this mending, and particularly the anomaly of a comma after 'shade' as in O1 but — with the new punctuation — no stop after 'body'. The real misunderstanding in O1, however, lay in the compositor's failure to recognize 'age I cannot shunne' as a parenthesis within the total query 'What . . . spent and done' which should thus have had no question mark after 'do' but instead one after 'done'. This same error is repeated faithfully in O3 save for the substitution of a closing period for O1's comma. The O3 pointing here does not seem to be an independent interpretation of manuscript punctuation, or lack of it.[13] One may also note the lack of capitalization of 'age' in both texts.

The attempt in the above passage on the part of the O3 compositor to mark the line 16 query by an appropriate question mark may emphasize the various times that O3 follows O1 in commas or periods when question marks of the same sort are definitely called for, as in I.ii.5-6 where O3 repeats the O1 period after 'him' in the second line of the couplet

Another example occurs in I.xiii.33-34. O1 prints Here O3 agrees with the final period where a query is required. A third case is in III.xiii.11-12, where O1 reads Here O3 placed a question mark after 'name' but substituted an anomalous period for the O1 colon after 'same', where the question mark should ordinarily have come. Again, O3 seems under the influence of O1, although attempting an improvement. (O2 here reads with O3.) Exactly the same situation occurs again in I.xiii.41-42. In O1 the couplet runs Once more O3 sees the need for a question mark but mistakenly follows the O1 comma after 'waine' and places the query, incorrectly, after 'swaine'. (Incidentally, in line 41 O3 prints 'Doest punish me' for O1 'Punish ye me'.)

To suggest that one text derives from another because it duplicates certain spellings is always difficult without such a compositorial analysis as will indicate the percentages of spelling preferences and of tolerances in the compositor conjecturally coming under the influence of certain copy. Therefore not a great deal of weight can be placed on an example like the following without an analysis of the O3 compositor in other books (if the printer could be identified[14]) as well as in this. But the spelling 'yong-young' is at least suggestive. The O1 compositor never varies from 'yong', whereas the O3 compositor shows a preference for 'young'. For instance, in Ben Jonson's version of I.xv, in line 4 the compositor violates visual rhyme (against his strong custom of spelling rhyme words alike) by matching 'young' with 'sprong' as a rhyme. In I.ii.13 he spells O1 'Yong' as 'Young' but appears to fall under the influence of O1 copy at line 27 when he follows O1 'Yong', as also in I.xv.4, II.iv.41, and III.vi.53 where O2 had 'young' in all three places but O1,3 print 'yong'. Yet the value of this evidence may be called in question by the appearance in elegies set from manuscript of 'yong' in II.iii.3 and of 'yong-mens' in II.xvi.17, although these examples are all.

The spelling evidence for O1 copy, then, will concentrate not on preferential spellings being interrupted by copy but instead on suggesting the influence of O1 on O3 either where the coincidence is unique or where the spelling is distinctly unusual. A striking example of O3 following O1 in a rhyme spelling comes in III.xiii.3-4

where 'chaste-paste' is also the spelling in O3 although O2 and O4-6 all read 'chast-past'. Another striking example, this time of O3 following what may perhaps be a transposition misprint in O1, or else an eccentric spelling, occurs in I.xiii.13 where O1 and O3 read 'Poore trauailers though tierd' (O2: 'tired'). Elsewhere, O3 in copy set from manuscript has 'tire' in I.vi.44 (although as a rhyme word with 'hire') and 'tyred' in III.x.13, and it follows O1 'tir'd' in II.x.34. Because of the slight possibility of a misprint in O1 (although see O1 'fier' for 'fire'), this is particularly strong evidence. In I.v.4 O1-3 read 'twincles' but in copy set from manuscript the other occurrences in O3 are represented by 'twinckles' twice in I.viii (and 'wrinkles' in II.iv.7, 'anckles' in III.v.6). Significant, perhaps, is the fact that O3 invariably spells 'Oh' as 'O', this happening at least a dozen times in copy set from manuscript, and that the only appearance in its pages of 'Oh' comes when it repeats O1-2 'Oh' at II.iv.6. No weight at all can be placed on O3 and O1 agreeing at III.vi.23 in 'boorded' in the light of O3 (from manuscript) 'boord' in I.iv and 'foord' in III.v. However, although these next examples are not in a class with 'paste' and 'tierd' above, nevertheless it is interesting to see O3 agree with O1 in 'wooddie groues' (O2: 'woodie') in I.i.13, in 'fauorit' (I.i.23), 'easly' (I.ii.10), 'glimps' (I.v.5), and 'touldst' (O2: 'toldst') in II.x.1, this last the only time that 'told' is so spelled in O3. The agreement in 'doote' for 'do't in II.x.22 is interesting. In I.xiii O3 spells 'Cœphalus' in line 33 but, more correctly, 'Cephalus' in line 39. Possibly the first spelling was under the influence of O1 'Cæphalus' found in both lines.

Finally, one can call on the evidence of common errors that are more likely to be O1's compositorial mistakes than mistranslations. The first comes in I.i.33-34, where O1 reads

Here editors have adopted the Dyce emendation 'sprays' for 'praise' in view of the memorial confusion possible from the sound of the two, of the fact that myrtle is not a wreath that honors the wearer and thus could not be said to praise him (but is it praising Venus?), and of the Latin which gives no hint of 'praise', cingere litorea flaventia tempora myrto — surround with shore-loving myrtle your shining temples. A similar case involving necessary emendation comes in I.xv.3-4 which reads in O1-2 where 'come' is repeated in O3, as is the spelling 'yong', although editors since Dyce, on the example of Jonson's version in O3, rhyming 'sprong-young', have emended Marlowe. Jonson cannot be an authority here since it is probable that he was influenced in his translation by O1 or O2, not by a manuscript; nevertheless, his word may very well be what Marlowe also wrote, for such a marked assonance and non-rhyme are unknown elsewhere in the elegies. If 'sprong' is correct, therefore, and 'come' is an O1 error, then the appearance in O3 of 'come' derives from its use of O1. However, a much better and less arguable example appears in I.ii.51-52 where an O1 misreading or sophistication seems to have been transferred to O3: Here the Latin adspice cognati felicia Caesaris arma — | qua vicit, victos protegit ille manu (Look but on the fortunate arms of thy kinsman Caesar — the hand that has made him victor, he uses to shield the vanquished) gives no warrant for the O1-3 reading which makes Cupid the conquered guarded by Caesar. It would appear, then, that O1 'thee' is a misprint or misreading transferred to O3, and the reading should be 'the conquered' as emended by Dyce. The third example seems equally persuasive. In I.xv.17-18 O1 reads and 'hoord' is the reading in O3. The Latin here is durus pater, which in the succeeding version in O3 Jonson translated as It is difficult to take it that 'hoord' is Marlowe's archetypal word transferred independently from two manuscripts to O1 and O3. In the first place, the O1-3 line limps badly and must be corrupt. The reason seems to be that the original word 'hard' was mistaken by an a:o confusion for the verb 'hoard', perhaps under the influence of the preceding verb 'cheat' and the following verb 'flatter', and the intervening word 'be' ('fathers be hard' is the usual emendation) — which also is required to modify 'hoorish' — was overlooked or ignored. As a result, a grammatical mess and an imperfect line resulted in O1, to be repeated faithfully by O3,[15] even to the spelling 'hoorded' although 'hoorish' was modernized.

In conjunction with the misprint or misreading 'thee' followed in I.ii.52, this misreading 'hoord' may help to clinch the case for the reprinting of the ten elegies by O3 influenced by O1 copy, and this conclusion is aided by the various pieces of slight bibliographical evidence and by other examples of O1 influence on O3, most notably in the misprint or odd

spelling 'tierd'. If more proof were needed, it may be found in I.ii.31-32, which read in O1, faithfully followed by O3 (as by O2): It would seem that O3 has repeated an O1 printer's error in omitting the necessary comma after 'meaning'. From the Latin Mens Bona ducetur manibus post terge retortia | et Pudor, et castris quidquid Amoris obest (Conscience shall be led along, with hands tied fast behind her back, and Modesty, and all who are foes to the camp of Love), it is clear that 'Good meaning' is not a compound adjective modifying 'shame' but instead an independent noun and personification in series with 'shame', and with 'such as seeke'. Wrongly punctuated O1 copy, not independent error from manuscript, most satisfactorily explains the O3 reading.

One more example of O3 following O1 in error occurs in II.x.33-34. O1 reads

which is repeated in O3 except for a semicolon after 'lips', no comma after 'wrackt', and a period after 'ships'. The first line is imperfect unless one wishes to make 'periured' into three syllables, an arguable matter. More important, the second line as it reads in O1,3 is an unexampled hexameter. The odds strongly favor the Cunningham emendation which transfers the 'And' of line 34 to line 33, placed after 'wealth'. This has the virtue of making the metre of both lines regular; but in addition it brings the lines into conformity with Ovid's Latin, quaerat avarus opes et, quae lassarit arando, | aequora periuro naufragas ore bibat, translated in the Loeb classics as Let the grasping trader's quest be wealth, and his perjured mouth drink in when he is wrecked the billow his ploughing keel has tired. Here one has a complex of error, at least in major part presumably originating with O1, transferred verbatim to O3. The chief point is the absence of the 'and' from line 33, which in Cunningham's version would read Without the 'and' in line 33 the merchants seek wealth with perjured lips; but the metre is off both here and in line 34. When the 'and' is transferred to line 33, the English agrees with the Latin and the metre is mended. In this error it is possible, of course, that O3 is independently following a faulty manuscript that also influenced the manuscript behind O1. But in conjunction with the various other common errors in O1 and O3 it is more probable that O3 is merely repeating an error originating in O1 or in its manuscript.

That O1 was the copy used by the O3 compositor — not O2 — would seem to be established by the readings, for in all but four indifferent forms O3 follows O1 in the various examples of O2 disagreement with O1 (see footnote 8 above). Any hypothesis that the annotation of O2 copy (not of O1) was so thorough as to bring every O1-2 substantive variant into conformity with the manuscript and thus to blot out the O2 departures from O1 lacks probability — somewhere there should have been a real slip — and is quite definitely discouraged by a not unimportant range of evidence for the derivation of O3 from printed copy that applies only to O1 and not to O2, with especial reference to the significant O1,3 'paste' but O2 'past'; O1,3 'tierd' but O2 'tired'; O1,3 'wooddie' but O2 'woodie'; O1,3 'yong' (at I.ii.27, II.iv.41, III.vi.53) but O2 'young'; O1,3 'Cæphalus' (at I.xiii.33) but O2 'Cephalus' (although this is reversible because of O3 'Cephalus' in line 39); O1,3 'touldst' but O2 'toldst'; and the damaged ambiguous sort in O1 that produced the anomalous period at I.iii.25 in O3 although O2 has a full comma.

On the other hand, it is a curious fact that Davies's Epigrams in O3 were set from a copy of O2, and not from O1. The case is demonstrable. Of at least 38 substantive variants between O1 and O2, the O3 text follows O2 in all but 13 readings. The qualitative evidence is better than the quantitative, or statistical. All of the thirteen O1,3 concurrences against O2 represent no more than O3 corrections of relatively obvious O2 errors that hit upon the O1 readings, such as O1,3 'requite' but the O2 misprint 'require' (no. 24.9), or 'woorthy' (O3: worthy') for O2 'worthlie' (no. 25.4), or the obvious omission of 'Dacus' in O2 (no. 30.14), or 'After' for the O2 misprint 'after' (no. 40.1), or 'his' for O2 misprint 'hig' (no. 45.4), and so on. Some required a certain amount of luck, as well as ingenuity, on the part of the O3 editor, like the return to O1 'reprooue' from O2 'approue' (Ad Musam, 7), or to 'this' from O2 'his' (no. 27.3), or 'do owe' from O2 'drawe' (no. 46.3), or 'his' from O2 'hie' (no. 48.7). That the changes were not made by reference to the O1 text of the Epigrams may be suggested by the various times that O3 repaired difficulties caused by O2 departures from O1 by different readings that clearly resulted from guesswork. Examples are O1 'Geron whose mouldie', O2 'Geron mouldie', O3 'Gerons mouldie' (no. 20.1); O1 'which did in Epigrams excell', O2 'which in Epigrams did excell', O3 'that did in Epigrams excell' (no. 29.1);[16] O1 'which', O2 'with', O3 'must' (no. 36.3); and O1 'so ill', O2 'ill', O3 'my ill' (no. 48.4). No evidence that can be trusted suggests consultation of O1 at any time by O3 in these Epigrams. The case is positive for O2 as the copy. Not only does O3 repeat such sense-destroying O2 variants as O1 'Gentleman' but O2-3 'lawyer' (no. 38.1), or acceptable enough O1 'Then is he' but O2-3 'When he is' (No. 49.9), or O1 'so smooth' but O2-3 'forsooth' (no. 36.39); but O3's derivation from O2 is pretty well demonstrated by other

evidence, such as its following the O2 misprint 'loue' for O1 'Ioue' (no. 17.2) and 'qd' for O1 'quoth' (no. 14.6), as well as its omission, as in O2, of the two lines in no. 40. Nor will some intermediate edition either before or after O2 that could have served as O3 copy satisfy the difficulties inherent in the O3 derivation from the O1 tradition for the Elegies but from the O2 tradition for the Epigrams. The bibliographical construction of the book whereby O3 starts the Epigrams on sig. F4r does not assist a hypothesis that the two sections were simultaneously set, which might have made convenient the use of two separate copies for the whole. Indeed, although the present writer admittedly has not made a computer analysis of the compositorial characteristics of O3, yet certain relatively obvious compositorial tricks like the hyphenation of suffixes are found in both sections, as in 'money-lesse' on F6 (no. 16.2) of the Epigrams and in the Elegies 'luck-lesse' (I.xi.7) on B4, 'loue-liest' (II.10.6) on C7 or, on the same page, 'lust-full' (II.10.25). In these noted cases the printed copy gives no warrant for the hyphens. Other practices of hyphenation are found in both sections, like 'a-round' (no. 26.10), which may be compared with Elegies 'some-times' (I.vi.21, I.viii.79), 'hus-band' (I.ix.25), 'her-self' (I.x.31), 'neck-lace' (I.x.52), or 'al-fowles' (II.vi.2). The odd 'heau-ns' (no. 46.2) on G4 may not be a misprint, therefore, but could align itself with 'run-ning' in the Elegies (I.vi.42) on sig. A6. Close reading, but not detailed analysis, does not seem to disclose evidence for more than one compositor throughout O3.[17] The cause of this difference in the copy for setting the two sections of O3 remains a puzzle to which the present writer has no answer except the printer may somehow have secured a made-up copy. The Elegies section begins with its own title-page on sig. E1 in O1 and O2, a possible indication that the two sections could have been sold separately as well as together. However, see footnotes 18, 26 below for a suggested explanation based on the hypothesis that annotated O2 was given to the printer of O3 for the Epigrams but only the manuscript for the Elegies and that it was he who secured a copy of O1 to assist in the setting of the Elegies.

The matter of the O3 editor having been raised requires some further remarks, for when his operations produce variants in the ten reprinted elegies a question must always be asked about the status of their authority. If the Elegies had been printed alone in O3 we should have little evidence,

although it would be difficult to prevent a certain uneasiness from creeping in as to the authority of some variants which when they are not outright sophistications like O3 'Temple' for O1 'tempe' (Heliconia tempé) in I.i.19 or explicable misreadings like 'slackt' for 'shakt' (vidi ego iactatas mota face crescere flammas) in I.ii.12, may occasionally be suspicious in their bland smoothing out of normal roughness, like 'Doest punish me' for 'Punish ye me' (I.xiii.41). This is not the place for a detailed reckoning of the authority of these variants, but it is appropriate to compare them both in frequency and in kind with the changes that O3 made in the O2 control text of the Epigrams where fresh authority could not have entered from any manuscript source. In the relatively brief text of the Epigrams O3 makes 32 substantive changes. These must have come in large part from an editor, probably the same person who prepared the copy for the Elegies.[18] Of these 32 variants in the Epigrams at least two are manifest errors, such as the omission of 'the' in no. 40.11, and the printing of 'shall' for 'hall' in no. 43.9, but these may perhaps be assigned to the printer. Nine are necessary corrections of O2 errors, such as the return of 'Dacus' which had dropped out of O2 (no. 30.14) or the correction of O2 'with' to 'which' (no. 30.11), or of O2 'require' to 'requite' (no. 24.9). In no. 48.4 an attempted correction of 'for my' for O2 'for' produced a fresh error. Interestingly, nine represent tinkering with the metre by smoothing the syntax. Typical are O3 'doth him so often' for O2 'so often doth him' (no. 7.7), 'seauen years in towne' for 'in towne 7 yeeres' (no. 9.2), 'one another' for 'eyther' (no. 24.13), or 'George Gascoines' for 'Gascoines' (no. 22.10). In no. 28.3 an attempted improvement of the metre by the addition of 'and' between O2 'braue, most' came to grief when the 'and' was inserted but the 'most' was thereupon inadvertently omitted. The remaining improvements alter individual words. Sometimes the motive seems to be modernization, as in O3 'an hundred' for O2 'a hundreth' (no. 13.8), and possibly this reason is behind the change of O2 'forsooke' to 'refus'd' (no. 14.2), of 'eke' to 'then' (no. 14.3), of 'come on' for 'come a' (no. 21.2), and even of 'listening' for 'harkning' (no. 38.14) and of 'heed' for 'mark' (no. 38.14). Such a change as 'newest' for 'flying' (no. 40.2) seems to have been motivated by a desire to reduce a bold metaphor to a conventional one.[19] The motives for alterations like 'spies' for 'lookes' (no. 23.2) or 'as a Pharasie' for 'like a Pharisie' (no. 13.6), or 'debitor' for 'debtor poor' (no. 38.15) cannot be determined with any certainty. In short, the history in O3 of the text of the Epigrams may serve to cast serious doubt on the authority of many of the readings that editors have conventionally accepted in the Elegies from their usual O3 copy-text. That the annotator of O3 consulted some authority other than O1 for the texts of the ten reprinted elegies is clear enough from its restoration, or insertion, of two sets of lines wanting in O1.[20] But that its general run of variants from O1 has manuscript authority is a doubtful proposition, on the evidence of the changes made by the editor in the Epigrams.

This problem is, of course, intimately related to the vital question about the manuscript behind the Certain Elegies in O1 and that behind the complete set in O3. One hypothesis would be that the Certain Elegies represent Marlowe's first attempts at translation in which he selected those that most interested him. It would follow that he thereupon, whether or not after an interval, engaged himself to the complete translation. If so, did he then give something of a revision to the earlier translated ten when he incorporated them in the full series? Does the O3 text then represent the revised and the O1 text the earlier version? Attractive as such a hypothesis may seem, what evidence there is runs contrary to it. A considerable number of the approximately 75 substantive variants between O3 and O1 represent the relatively easy correction by O3 of O1 misprints or misreadings as well as new misprints and misreadings perpetrated by O3 itself.[21] When these matters are set aside, we find that O3 makes four changes that can be imputed to metrical improvement in smoothness,[22] and thirteen alterations that are nothing more than modernizations,[23] all these to be classed as

sophistications that do not necessarily have any authority.

The real case for revision (or for purer readings) in O3 must rest with the twenty instances when in substantive variation O3 is closer to the Latin than O1, in contrast to the four cases in which O1 is more faithful in its translation.[24] These examples range from the O3 restoration of the Latin preterite for the O1 present in I.i.5,8 or of the present for the preterite (III. vi.47), through the restoration of correct Latin meaning from O1's faulty punctuation that had distorted it, as in the O3 comma after 'tho' (I.ii.3) and after 'so' (I.ii.7) for O1 omission,[25] to more accurate translation of number, as in O3 'triumph' for O1 'triumphs' (I.ii.28) or 'my' for O1 'our' (I.xv.2), perhaps an O1 slip; of transposed pronouns, perhaps another O1 slip (possibly in the manuscript) in O3 'yours euer mine' for O1 'mine euer yours' (III.xiii.22). Relatively few, but they are the most interesting, alter O1's words in the interests of exactness to the Latin. Certain of these are changes in quite minor words, as in O3 'All' for O1 'This' (I.xiii.25), 'into' for 'to the' (I.xv.10), 'And' for 'The' (I.xv.34), 'say' for 'speake' (II.xiii.35), 'that being' for 'and being' (III.vi.19), or 'And' for 'Or' (III. xiii.8).[26] Only a handful concern more concrete words, like O3 'palme' for

O1 'garland' (III.xiii.47) or 'heldst' for O1 'hadst' (I.xiii.39). The case of O3 'drugges' for O1 'droughs' is moot (III.vi.28). That is, since O1's word seems to represent either an eccentric spelling, or even a compositorial misreading, of 'draughs' or 'draughts', the possibility exists that the O3 variant is a sophistication or rationalization, particularly since the Latin herba nocent is at least as close to 'draughts' as to 'drugges'. The cluster in III.xiii. 40-47 may be suggestive. First, in line 38 O3 mends the metre by reading 'thorough' for O1 'through'. The change of O1 'dying' to 'dead' in line 40, 'And would be dead, but dying, with thee remaine' may be as much a metrical smoothing as a return to the Latin tunc ego, sed tecum, mortuus esse velim. In line 47 comes the correct word (and also metrically regular) 'palme' (palma) for O1 paraphrase 'garland'; but in between, in line 45, O3 changes O1 'yeeld not' to 'deny' (quae bene visa mihi fuerint, bene visa negato). There is no question that here O3 is simply adopting a more direct word for a slightly old-fashioned and less literal but certainly appropriate enough translation 'yeeld not'. These illustrations do not necessarily suggest authorial revision of early work (particularly in the last example) but instead the attentions of an editor.[27] In most of them, given the choice, one would be more likely to assign the O1 readings to Marlowe than the O3 versions. With Jonson's 'Accius high-reard straine' (I.xv.19) in mind as a gloss, we might place among these O1's 'With muse vpreard I meane to sing of armes' (I.i.5) instead of O3's bland 'With Muse prepar'd' despite its more literal translation of Arma gravi numero violentaque bella parabam | edere. (O3's change of 'meane' to 'meant' and probably of 'take' in line 8 to 'tooke' seem to be necessary, however, to mend the sense in O1.) And O1 'louers' is certainly better for Lais amata viris than O3 'wooers' (I.v.12). Nothing much can be made of O1 'deep vast sea' but O3 'vast deepe' (II.x. 14),[28] or of the neutral O3 'strugling' (an subitum luctando accendimus ignem) for O1 'striuing'.[29]

The real puzzle is that although fairly numerous these O3 variants are in fact, given the amount of text, not much more frequent than the demonstrably unauthoritative changes made in the Davies Epigrams nor do they differ markedly in their kind. In no sense is there such a reworking as might be expected of a revising author and, especially in their occasional correction of loose or mistaken translation, they often seem to bear the mark of an editor, who also was concerned with exact metrical regularity and with some modernizing of what had become old-fashioned loose grammar and syntax as well as language. Indeed, if it were not for the addition of the couplet missing in I.xv and of the four lines in II.iv[30] there is little solid evidence that the O1 printed text was actually compared with the manuscript that furnished the full set of the elegies, since editorial tinkering would account satisfactorily for most of the changes made in O3.[31] Perhaps

there is, in fact, not much connection between these additions and the question of collation. The kind of editor who was reading over the O1 text and altering it to suit his purposes would naturally notice the two hiatuses and supply them by reference to the manuscript; he need not, thus, have had his attention called to them in comparing the manuscript with O1. Indeed, if he was comparing O1 with anything, it would have been with the Latin text, on the evidence of the alterations made in the direction of greater exactness to the Latin.[32]

Yet there are difficulties with this editorial theory if it is narrowly applied to the annotation of O1 to produce printer's copy. That an editor was present behind some features of the O3 text may be readily assumed from the attentions given to Davies's Epigrams. But if this editor applied himself to O1 instead of to the manuscript itself, the result is a curious combination of minute correction and coarse oversight. The evidence suggests the possibility, instead, that whereas a marked copy of the Epigrams was given to the printer, the manuscript alone was furnished him for the Elegies; and the compositor set those texts present in the Certaine Elegies either from the manuscript with consultation of the print or largely from

O1 with consultation of the manuscript. In some obscure way this situation may have something to do with the fact that O2 was utilized as copy for the Epigrams but it was O1 that influenced the setting of the elegies in O3. If one wanted to engage in idle speculation one could even guess that the anomaly occurred because it was the printer himself who privately secured the copy of O1 to speed his labors.

This matter of the mixed readings has an intimate relation to speculations about the nature of the manuscript behind O1 and that behind O3, and their relationship. Not only does the nature of the variants in O3 (even though not all perhaps can be attributed to the O3 editor) discourage a belief that the Certain Elegies represent early trials; the composite nature of O1's book of Epigrammes and Elegies makes a hypothesis quite plausible that it derived from someone's collection or commonplace book of satiric and amatory poetry, a possibility encouraged by the signing 'C. Marlow(e)' of the first three poems[33] and by the appearance between the epigrams and elegies of the three Ignoto amorous poems not to be attributed either to Davies or to Marlowe. If this is so, the manuscript behind O1 is a copy that stems from the manuscript behind O3 or from an ancestor. The printer's copy for O3 could be Marlowe's own papers worked over by a publisher's editor, or a transcript made of these, at what distance is impossible to guess. Who introduced Jonson's version of I.xv, and whether this has a bearing on the nature or derivation of the manuscript, are questions not to be answered. In short, no assumptions can be made on any evidential basis about the relative authority of each manuscript, including its distance from Marlowe's papers.[34] It is a fair conjecture, however, on the evidence of the treatment of the text of Davies's Epigrams that the larger number of variants found in the O3 text from Certain Elegies were inserted by a publisher's editor and were not present in the manuscript behind O3. This hypothesis helps to place the O3 manuscript as perhaps relatively close to the original if it is not itself the original, which we cannot know. Of one

thing we may be sure, however. Given the degree of sophistication of the Epigrams and of the Certain Elegies in O3, we may expect the same treatment to have been given to the text of the elegies published there for the first time. Hence whatever authority might be credited hypothetically to the O3 manuscript because of a possible closeness to the original is vitiated for its substantives, in some part, by the editorial tinkering to which it has been subjected.

The family tree of the remaining editions presents no difficulty. The Bodleian Library preserves another octavo edition (O4) collating A-F8 G4 like O3. This is Douce O.31 (STC 18931). Its title omits 'ALL' and reads '[row of type-orn., same as in O1-2] | Ouids Elegies: | Three Bookes. | By C. M. | Epigrames by I. D. | [narrow orn. of fish on each side of a man's head] | At Middlebourgh.' As in O3, the Epigrams begin on sig. F4r with a headtitle. Since O3-4 are paginal reprints, the question of priority arises but is readily solved, for where Douce (O4) departs in the Epigrams or the Certain Elegies from Mason (O3), it is O3 that agrees with O1 (O2 for the Epigrams) and O4 that disagrees. Examples may be cited from II.iv: line 24, O1-3 She would] O4 She will; line 41, O1-3 blacke haire] O4 browne haire; line 46, O1-3 looks, that] O4 lookes and that. In III.vi.66, O1-3 more then] O4 more like.

O5 and O6 are also paginal reprints, each collating 8, A-F8. The O5 edition, which should be STC 18932, may be represented by British Museum C.57.i.42.[35] Its title-page reads 'ALL | Ovids Elegies: | 3. Bookes. | By C. M. | Epigrams by I. D. | [row of three vertical leaf type-orns. above two fists, the right inverted] | AT MIDDLEBOVRGH.' The elegies end on E8v with 'FINIS.' and a rule, and beneath the Epigrams start with a headtitle, ending on F8v with 'FINIS. I. D.' In the readings cited above of variants between O3 and O4, this O5 edition agrees with O3, as in other variants between Douce and Mason. In turn, when O5 disagrees with O3, in these readings O3 agrees with O1. Examples are I.i.28 O1-3 Saying] O5 Saving; I.ii.14 O1-3 which] O5 that; I.iii.21 O1-3 horned] O5 honored. Other readings following the same pattern demonstrate that O5 derives from O3, since when O3 varies from O1 then O5 follows O3, as in I.i.5 O1 vpreard] O3,5 prepar'd; I.i.19 O1 tempe] O3,5 Temple.

The last observed edition is O6 (STC 18933), represented by the British Museum copy, shelf mark 1068.6.20(2).[36] The octavo also collates A-F8 and is a paginal reprint of O5. When O5 departs from O3, O6 follows O5 as in I.i.28, I.ii.14, and I.iii.21. On the other hand, it has various unique readings

that show it to be a terminal edition from which O5 could not derive, as I.iii.i O3,5 he] O6 him; I.iii.13 O3,5 gives] O6 give; I.v.7 O3,5 shamefast] O6 shamefac'd. O6 also uses medial 'v' for 'u' and in other respects betrays a later origin. It employs a curious typographical device by which alternate elegies are set in italic. The title-page of O6 reads 'ALL | OVIDS ELEGIES: | 3. Bookes. | By C. M. | Epigrams by I. D. | [block of four lace type-orn.] | AT MIDDLEBOVRGH.'

The Short-Title Catalogue's queried dates of 1595 for O1, 1598 for O2, 1635 for O5, and 1640 for O6 are, of course, purely conjectural and the evidence is not known on which these estimates were based. The O3 printing of Jonson's Elegy I.xv takes its text from the 1602 Quarto of Poetaster on the evidence of the accidentals, including the Q-O3 spelling 'Æney' in line 25 for the 1616 Folio 'Ænee', as well as other spellings, supported by the Q-O3 'The frost-drad mirtle' in line 37 for the Folio 'Frost-fearing myrtle'. The first complete edition of the Elegies, therefore, was certainly printed after 1602. Whether its use of the Quarto instead of the 1616 Folio for its text of I.xv means that it appeared before the Folio is not subject to demonstration and would be a dangerous conjecture. Jonson himself seems to have consulted either O1 or O2 for his translation; if, instead, he used a manuscript, certain of the O1 readings would be authenticated as against the O3 variants.