University of Virginia Library

The Empire of Lucius Iberius
by
P. J. C. Field

For centuries, the Arthurian legend included a story of war between King Arthur and the apocryphal Roman Emperor Lucius Iberius. It appears first in Geoffrey of Monmouth's Historia regum Britanniae,[1] and was included in the reworkings of the Historia by Wace, Layamon, and Robert Mannyng of Brunne.[2] Among those who retold it later were the anonymous author of the Middle English alliterative Morte Arthure and Sir Thomas Malory. They told it very differently, the poet giving an autonomous account of the downfall of a great king and Malory making the story (with a new outcome) a triumphant end to the beginning of King Arthur's reign. Nevertheless, the fact that Malory's story is based on the alliterative Morte and the alliterative Morte on a version of Geoffrey's story makes possible some discoveries about them that have been overlooked until now.

This appears most clearly from what may be textually the most corrupt episode in either work, their accounts of Lucius's summons to war. Those accounts derive ultimately from a short list of provinces and peoples in the Historia which looks like a typical stroke of Galfridian historical verisimilitude.[3] Wace re-ordered the list and brought it up to date by replacing Parthia with Turkey, and Layamon and Mannyng used the revised list with a few small changes.[4] The alliterative poet reworked and expanded it, turning it into an elaborate account of the imperial summons and response. He names Turkey rather than Parthia, suggesting that he worked from Wace or a derivative rather than Geoffrey, and adds further material drawn mostly from Mandeville's Travels and a short passage in the Acts of the Apostles.[5] Apart from Arthur's conquests, which he listed in an earlier passage (lines 26 — 47), the places he names embrace what was then almost the whole known world.

The result is a visionary tour de force that presents dozens of recognisably real localities in a Miltonic sweep across three continents. Malory seems to have attempted to abbreviate this while keeping as many of the place-names as possible.

Both original manuscripts, the alliterative poet's and Malory's, are lost, and what they wrote has to be deduced from three early texts: the unique manuscripts of the alliterative Morte (the Thornton manuscript: T) and Malory's Morte Darthur (the Winchester manuscript: W), and Caxton's printed edition of Malory (C).[6] The names of the provinces in the three texts and the ultimate chronicle sources are shown in Table I.[7]

All three texts show signs of corruption in their presentation of the names: some are simply incomprehensible as they stand. Editors, however, have been distinctly cautious in emendation. Both James Spisak's edition of Malory based on the Caxton text and Eugène Vinaver's editions based on the Winchester manuscript print their base-texts without emendation.[8] Mary Hamel's edition of the alliterative Morte changes only two names: she emends Orcage to Arcage and inserts into her text from W a complete alliterative line containing Calabe and Catelonde, emending the former to Calabre, its form in C.[9] This is a normal spelling of Calabria in Malory's time,[10] and when I revised Vinaver's edition, I too emended Calabe to Calabre (Works [1990], p. 193).

Closer consideration however suggests that dozens of other superior readings may be recoverable. This is partly because of the very factor that generated the corruption — the strangeness of exotic names. When scribes copy exotic and unfamiliar originals, they almost inevitably commit errors, often creating even stranger forms. Strange forms, however, may be the product of very familiar processes of error, and when a plausible lost original is inferred from such a process, it may be apparent that it alone could have generated the extant reading. Passages that contain more errors than usual may therefore be capable of more correction than usual, and so may throw particular light on the mentalities of those who created and transmitted what are (if the Caxton Malory is taken as a distinct literary entity) three important literary texts.

Not every conjecture will approach certainty, but even those that are merely probable should be implemented, because it is an editor's duty to correct all corrigible error. As W. W. Greg said, in the most important textual essay of the past half-century:

Table I

                                                                                                           

	GWLR	T	W	C
1.		Ambyganye	Ambage	ambage
2.		Orcage	Arrage	arrage
3.		Alysaundyre	Alysundir	Alysaundrye
4.		Inde	Ynde	Ynde
5.		Ermonye	Ermony	hermonye
6.		Asye	Assy	Asye
7.	Africa	Affrike	Aufryke	Auffryke
8.		Ewrope the large	Europe the large	Europe the large
9.		Irrttayne	Ertayne	ertayne
10.		Elamet	Elamye	Elamye
11.		owte ilez	oute yles
12.		Arraby	Arrabe	Arabye
13.	Egypt	Egipt	Egypte	Egypte
14.		Damaske	Damaske	damaske
15.		Damyat	Damyake	damyete
16.	Crete	Crete
17.				Cayer
18.		Capados	Capydos	Capadoce
19.		Tartary	Tars	tarce
20.	Turkey[11]	Turky	Turke	Turkye
21.		Thebay
22.		Amazonnes landes
23.	Babylon	Babyloyn
24.		Baldake
25.		Perce
26.			Pounce	pounce
27.		Pamphile	Pampoyle	pampoylle
28.		Preter Johne landes	Preter Johanes londe
29.	Syria	Surrye	Surre	Surrye
30.		Nylus
31.			Nero
32.		Nazarethe	Nazareth
33.		Garyere	Garese
34.		Galele	Galely
35.				gallacye
36.				Grece
37.		Cyprys	Cypres	Cypres
38.		Roodes
39.	Greece	Grekes	Grekis
40.		Macedone	Macidony	Macydone
41.		Pulle
42.		Pruyslande
43.		Lettow
44.			Calabe	Calabre
45.			Catelonde	Cateland
46.			Portyngale	portyngale
47.	Spain		Spaynardis	spaynardys
48.	Boeotia
49.	Phrygia
50.	Media
51.	Libya
52.	Bitunia
53.	Ituria

Recovering authentic readings that were not apparent to previous editors, some them distinguished practitioners of their art, may not be easy. It will be assisted in this case, however, by two special factors in the relationships of the texts to one another and their sources. Those relationships may be set out in the following stemma:

For clarity's sake, the stemma omits a number of lost intermediate manuscripts posited by Professor Vinaver and Dr Hamel. It includes only (because they are necessary to what follows) the three lost manuscripts assumed to stand in the line of descent from the original manuscript of Morte Arthure to the archetype of the two Malory texts. These are: X, that archetype; E, the Morte Arthure manuscript Malory used; and O, the archetype of the Thornton manuscript and E.

The first special factor that may assist correction is that the three texts descend independently from their lost original.[13] Although scribes are much inclined to corrupt unfamiliar names, different scribes often corrupt them in different ways. New possibilities may therefore become apparent if the three versions are used in concert, even when all of them are manifestly corrupt.

The second factor is what may be called, since Professor Vinaver set it out with exemplary clarity in his exposition of textual-critical principles and used it to great effect in his editions, the Vinaver Principle (see Works, pp. cxiv — cxvi). That principle is that, when contamination and coincidence can be ruled out, what appears in a writer's source and in an extant version of his text should be assumed to have been in his original. T is not the manuscript of the alliterative Morte that Malory used,[14] but it is often close enough to it to make it possible to apply the Vinaver Principle. An edition of Malory based on the Winchester manuscript can therefore be corrected from agreements between T and the Caxton text, and vice versa. More succinctly, MA + C = M and MA + W = M. On that assumption, I emended another reading in the Winchester manuscript's list of Roman provinces and peoples, the river-name Eufrate, "Euphrates" (Works, p. 193.6). Although the form Eufrate is attested in Middle English,[15] the simplest and therefore most probable explanation of its appearance in the Winchester manuscript, given Thornton's Ewfrates and Caxton's Eufrates, is that the Winchester scribe committed a common copying error, omitting the last letter of a word in his original.

The Vinaver Principle can also be applied to the alliterative Morte: a text based on T can be corrected in the same way from agreements between the poem's sources and Malory. More succinctly, S + M = MA. S is not a single text. It must include not only the place-names in the chronicle sources, but also Mandeville, the passage from Acts, and any other passage the poet used that can be identified and defined with sufficient probability. The need to define the passage has a surprising consequence: the few verses from Acts form a more reliable component of S than Mandeville's Travels because, although the poet took many more names from Mandeville, those names are so scattered through the Travels that it is difficult to be sure which passages he had in mind.

Finally, it should be noted that the Vinaver Principle can be applied to the Morte Arthure and the Morte Darthur in series. If contamination and coincidence can be ruled out, what appears in the source of the alliterative poem and in either extant version of Malory's book must be assumed to have been in both the Morte Arthure and the Morte Darthur.

These preliminaries disposed of, we can now turn to the place-names themselves.

No. I. The first name raises several complex issues. The peculiar readings exhibited by the texts do not much resemble the name of any province, district, or country that could have been thought of as subject or allied to Rome; but a fine stroke of induction led Dr Hamel to a suitable name in a form that could have produced those readings. Observing that several names

in the list had a French cast, she suggested that the first name might have been a French form of Albania,[16] perhaps Aubaigne, a possible though unattested form,[17] which could have been corrupted first (by minim error) to Ambaigne in O, and then by further degrees to the readings in each of the extant texts. If this is the Albania in Mandeville rather than Scotland or the modern country on the Adriatic, it is a country near Scythia, fitting in with the overall east-west movement of the passage.

This is a plausible suggestion, but it can perhaps be made more plausible still. Since, for phonologically straightforward reasons, initial Alb- sometimes became Amb- in other place-names,[18] we can postulate, without implying error of any kind, that it did so in this one, and that what stood in S was an original French Ambaigne; and, since fifteenth-century English orthography often prefers y where French would have an i (as in Almaygne "Germany" in T, and Grete Bretaygne in W),[19] that that Ambaigne perfectly regularly became Ambaygne in MA and O, before corrupting by way of Ambygane (anagramatism) to T's Ambyganye (normalization of final vowel: cf. Asye, Ermonye, Surrye), and by way of Ambaygne (omission of macron) to Ambage (simplification of second vowel) in X and the two Malory texts.

It should be noticed that although each of the extant forms would be two stages of corruption removed from O, each double corruption could have been perpetrated on a single occasion by a scribe who misread his exemplar and then changed what he thought he read. In the line of descent to the Malory texts, for instance, a scribe might have failed to notice a macron in his exemplar and have corrected an Ambayge that existed only in his own mind. That possibility reduces the number of intermediate manuscripts that would have to be assumed to explain the extant readings, and, for any given number of stages of transmission, correspondingly increases the number of points at which the corruption might have taken place. On the assumptions above, it could have happened at E or M or X or any two of them, and Malory's own manuscript could therefore have contained any of the three readings Ambaygne : Ambayge : Ambage.

The practical consequences of this depend on what is accepted as the aim of textual criticism. It is universally agreed that all textual transmission tends to corrupt and that textual criticism exists to purge corruption as far as possible, but scholars have disagreed about what is possible, and in particular about how far an editor should attempt to work his way back along the line of transmission. Any text may of course contain incorrigible errors,

but an older school of critics reacted to arbitrary and wilful emendation by their predecessors by proposing various self-denying ordinances restricting what should be attempted at the limits of possibility.[20] Among them, Professor Vinaver, whose Malory editions were among the most distinguished products of that school, asserted that an editor should only aim to reconstruct the earliest written form of his text (Works, p. cvi). Objections to this have become more apparent as textual criticism has come to be applied systematically to post-Renaissance texts, where information about the publication process frequently survives in quantity. The restriction would require us, for instance, to reproduce spelling mistakes in an author's manuscript, even if a letter from the author to his publisher survived listing the mistakes and asking for them to be corrected in a future edition. This and other considerations have produced wide although not universal agreement that an editor should aim to recover, as far as the evidence permits, what his author finally intended to write.[21]

For the Morte Arthure, these disputes have no practical consequences. The hypothesis that manuscript O read Ambaygne provides an acceptable sense and elegantly explains the readings of the three surviving texts. There is no reason to assume that the author either wrote or intended to write anything other than Ambaygne. Since no other hypothesis has been put forward that explains as much as this one,[22] and since it is not in itself a virtue to exhibit the paradosis within the text of a critical edition,[23] such an edition of Morte Arthure must emend to Ambaygne.

The implications for Malory's Morte Darthur are more complicated. We do not know what Malory wrote or what he intended to write. Because, as we have seen, the textual tradition below the archetype could have been corrupted at E or M or X or any two of them, Malory's own manuscript could have read Ambaygne or Ambayge or Ambage. It could be argued that we should emend to Ambaygne, because Malory intended that word, even if he never saw it: he was trying to reproduce the reading of the Morte Arthure, which was Ambaygne. There is some force in this: Malory did have a general intention throughout the Morte Darthur to retell the "authorised" Arthurian story, and although that frequently implied intention was compatible with a good deal of invention of his own at many points, even

points at which he asserts his explicit dependence on his sources,[24] it is clear that here he is attempting high fidelity rather than re-creation. However, that argument must surely be rejected, not so much to avoid complex argument or editorial self-delusion — individual textual critics may propose, but the scholarly world will dispose — as because it strains the meaning of "authorial intention" to have an author "intend" something dependent on an entity that he may not have known to have existed. We have no reason to suppose that Malory had any notion of the Morte Arthure other than as the sum of the readings of the manuscript he used.

It may, however, be possible to resolve the uncertainty in another way. Ambaygne turned into Ambage either in three stages or in two. If it happened in three stages, then assuming the relationships set out in stemma above, the manuscripts involved must have been O → E → M (Case 1), E → M → X (Case 2), or O → EM → X (Case 3). If it happened in two stages, they must have been O → E (Case 4), E → M (Case 5), or M → X (Case 6). In Cases 1 and 3, Malory would have been trying to reproduce Ambayge, in Case 4, Ambage, and in the other three cases, Ambaygne. Ambaygne is therefore the most probable reading, and we should emend to it. Since the stemma is an hypothesis, we cannot be certain that the texts descended in the way it sets out, but as long as the textual relationships it expresses are the most probable, an editor is obliged to act on them.[25] Moreover, even if the stemma is inaccurate, a variant of the argument proposed will hold unless the ratio of the number of stages of transmission before M to those after it was lower in reality than in the stemma. There is no reason to suppose that that was so: there was much more time for multiple copyings during the process that brought the Morte Arthure to Malory than between Malory's composition and the copying of the archetype of W and C.[26]

No. 2. This name presents similar problems, to which Dr Hamel offered a similar and equally satisfying solution, Arcage, a French form of Arcadia that the alliterative poet would have been familiar with from one of his minor sources, and which she assumes to have been transmitted accurately to the archetype of the extant texts, below which T or an ancestor confused the first vowel, and X or an ancestor mistook the second r for c (Hamel, line 572n). This is entirely plausible, and in the absence of any other hypothesis at all, she very properly emended T's Orcage to Arcage.

As in the previous case, this creates complicated problems for the text of Malory, but a solution can be provided on similar lines. The error postulated here must be assumed to have occurred in a single copying process, either O → E (Case 1), or E → M (Case 2), or M → X (Case 3). In Case 1 Malory would have been trying to copy Arrage, and in Cases 2 and 3, Arcage: probability therefore requires us to emend to Arcage.

No. 3. There are two points at issue here. First, the third vowel in W's Alysundir is suspect because T and C agree against it, because it could easily have been created by the momentary scribal hesitation that Vinaver called arrhythmia,[27] and because it is an unlikely vowel in the name Alexander or any of its derivatives. Alexander and names derived from it are not uncommon in Middle English and French Arthurian literature, but no other form with u alone as the third vowel is recorded anywhere in W or any other text.[28] The only other possible example of the place-name Alexandria in W, when Priamus later in this tale declares himself the heir (apparently) to Alexandria, has the same third vowel as the eighty occurrences of the personal name Alexander: in all cases -au-, the vowel found at the corresponding point in T and C.[29] W's reading should therefore be emended to -au-.

The end of the name also presents problems. The name Alexandria, whatever its other variations, rarely has a vowel between d and r, but it does so in all three texts we are considering, both here and in that later incident involving Priamus.[30] Dr Hamel suggests that the spelling may derive ultimately from a work that she believes to be a minor source of the alliterative poem, The Parliament of the Three Ages, which at one point gives the place-name as Alexaunder (Hamel, 2607n), but The Parliament of the Three Ages is not the only possibility: at least one other English text of Malory's time and from Malory's part of England gives the name as Alisander.[31] Our three texts give it as follows:

     

	T	W	C
I.	Alysaundyre	Alysundir	Alysaundrye
II.	Alexandere	Alysaundir	Alysaunder

It must be kept in mind that in all three texts the sentence before the second occurrence includes the name of Alexander the Great, and that a grammatical ambiguity makes it possible, at some cost to a parallelism with next two phrases, to read the putative place-name as a personal name, making Priamus declare himself the heir to Alexander the Great. The simplest explanation of the six readings is that in both passages the poet wrote Alysaundyre, which T reproduced in the first passage but corrupted in the second by contamination from the personal name five lines earlier; that in both cases Malory simplified Alysaundyre to Alysaundir, which W reproduced imperfectly in the first passage and accurately in the second; and that C deliberately corrected Alysaundir in the first passage, but failed to do so in the second, perhaps misled by the factors that had misled Thornton. It follows that an edition of Malory, whichever text it is based on, should read Alysaundir in both passages.

That, however, is not the only kind of edition that could be based on the texts we are considering. Mediaeval romance does not always have a single identifiable author. Scribes produced rewritings, revisions, recensions, and remaniements of many kinds, and with some romances — the Old French prose Tristan is a notable case in point — it is difficult to know when a scribe ceases to be an unauthorized tamperer with someone else's composition and becomes instead the author of a new work. The two works under consideration here do have clearly identifiable authors, the anonymous alliterative poet in the one case and Sir Thomas Malory in the other, but it would nevertheless be perfectly proper to edit Thornton's Morte Arthure not as, in Greg's phrase, a particular authority for Morte Arthure, but as Thornton's work for its own sake. Similarly, the Caxton text of the Morte Darthur could be edited as Caxton's work.

There is no apparent reason for an edition of Thornton's work, but an edition of Caxton's could be justified by his fame as a publisher, translator, and man of letters, and the fact that it was his Morte Darthur that became an English prose classic in the sixteenth century and again in the nineteenth. The text he published, by a mixture of unconscious error and deliberate alteration, was significantly different from the one that Malory wrote. It was even he who (by mistake) gave Malory's book its memorable and misleading name. To edit this passage as Caxton's prose would produce different conclusions about a good many of the names in it.

Re-running the probabilities for the first two names suggests that Caxton was trying to reproduce ambage and arrage: an editor should therefore keep those forms, glossing them perhaps as "unknown (? Eastern) countries." Alysaundrye, in contrast, looks like a deliberate alteration. Many of the differences between Caxton's Morte Darthur and Malory's, particularly in the Roman War story, are the product of such alterations by Caxton himself,[32] which makes it reasonable, when there is no evidence to the contrary, to attribute to Caxton any apparent conscious correction in C. Alysaundrye is an entirely plausible conscious correction: it looks like a form that would have been recognised as authentic in the merchant community in which Caxton lived. An editor editing C as Caxton's own prose should therefore keep it.

In the later passage, the editor would have to choose whichever of Alysaunder, Alysaundir, and Alysaundrye he thought Caxton intended to write.[33] The choice would not be easy,[34] and it is fortunately not necessary to decide it here, but if Caxton's change to Alysaunder was deliberate, he presumably meant it to mean "Alexander." It is more difficult to be sure what Malory understood by Alysaundir: at least in the first passage he presumably thought of it as a place-name, but he may have meant only "whatever obscure place in the Near East was intended by the knowledgeable author of my source."

Nos 5 — 7, 12, 19, 20, 27, 29, 40 and 47. In no. 5, the Vinaver Principle suggests that the initial h- in C's hermonye is scribal, and it looks likely to have been added by conscious choice. The change might have been made in confusion by a scribe as muddled about Armenia as Malory may have been about Alexandria, or it might be a normalization to the scribe's own orthography, something that is the more likely in that Caxton seems to have systematically modified the Morte Darthur towards his own linguistic norms.[35] In either case, an edition of Caxton's prose should keep it, because Caxton may be the "scribe" who made the change or chose to let it stand. Such an edition would reverse the change only in unlikely circumstances — if, for instance, it were demonstrable that it was primarily caused by distraction by the word harmony.

If C were edited as Malory's work, however, things might be different. If the change were shown to be the product of confusion, it would simply be reversed, but "translation" into another dialect would raise the complicated and important issue of restoring authors' linguistic norms. To do so at this point in this work might imply millions of other changes elsewhere. Such restoration is rarely attempted with early texts, because the surviving witnesses are often the product of long copying traditions, in the course of which it is natural to assume that the author's orthography and other accidentals (the presentational features of the text) have been irrecoverably lost, leaving the substantives (the words themselves) as the only subject for textual criticism. In mediaeval vernacular texts, moreover, orthography is often notably uncertain, and rarely shows any sign that a scribe felt bound by the spelling of his exemplar. In the formative years of modern textual criticism, some pioneers suggested that an editor's duty of restoring as far as possible what the author intended to write extended to systematic restoration of the author's dialect and orthography: some heroic souls even attempted

to put this into practice, but their editions were widely felt to be unreliable.[36] Since then editorial practice has varied. Some editors have chosen to standardize readings imported from other texts and any spelling errors they identified to the norms of their base-texts.[37] Others have preferred to reproduce their authorities except when the evidence seemed to warrant emending individual scribal spellings to authorial ones, as both Professor Vinaver and Dr Hamel do on occasion.[38]

The problem may be soluble, at least in principle. If textual criticism exists to recover what the author intended to write, a critical edition must attempt to recover the author's accidentals unless the evidence for doing so is insufficient, or unless there is reason to suppose the author was indifferent to a particular feature. Although there is rarely any evidence of authorial preferences for the accidentals of an early text, it may sometimes be reasonable to suppose from the practice of the period and genre that an author would probably have been content to leave some of them to his scribes to decide. Both parties may have been as well aware as any modern scholar that part of the meaning of a work will be conveyed by the mise en page of the manuscript in which it is contained; but both the variety of presentation found in surviving mediaeval manuscripts and the analogy with modern publishing suggest that mediaeval authors would often have been willing, perhaps even eager, to delegate, for instance, the number of columns and lines per page, the scripts used, the amount of abbreviation, and other "publication details" to professionals. That might be less likely with (say) a multi-level commentary on a book of the Bible or a multi-coloured genealogical scroll setting out the royal claim to the throne, where fundamental decisions about what was to be said might require a grasp of the physical possibilities of manuscripts, a grasp that might normally be accompanied by preferences about the use of those possibilities; but it would be much more likely with a vernacular prose romance composed by a manifestly nonprofessional author.

Orthography, however, is a skill that — dictation apart — an author must possess. Of all scribal skills, it is the one on which an author is most likely to have had preferences, and for that reason, and also because of its implications for the semantic ambiguities and sound patterns of a work, it is the one that would generally be most desirable to recover. There are, however, problems in this with Malory's work. No other written material survives against which linguistic norms suggested by the surviving texts of the Morte Darthur could be cross-checked. I have argued that Malory wrote a second romance, The Wedding of Sir Gawain and Dame Ragnell, but it is very

difficult to detect beneath the dialect of the scribe of the only surviving manuscript of that work any feature of the dialect of its author.[39] Moreover, Malory's identity has been disputed,[40] so an editor cannot infer features of Malory's language not evident in his base-text from a knowledge of his author's origins. Any attempt to do so might need a good deal of biographical knowledge: one of the men suggested as author of the Morte Darthur, for instance, was born near Shrewsbury in the far West Midlands but apparently spent his adult life at Papworth St Agnes in the East Midlands;[41] since nothing indisputably written by him survives, there is no telling whether his English was West Midland, East Midland, a mixture of the two, or something entirely different. Another candidate, Thomas Malory of Hutton Conyers in Yorkshire, in contrast, is associated only with that place.[42] An editor who accepted him as author would have to face the argument that an edition of the Morte Darthur, whatever its base-text, should in principle be "translated" into the orthography of north Yorkshire. One who accepted Sir Thomas Malory of Newbold Revel as author would similarly have to explain why an edition based on C should not be "translated" into the orthography of the Newbold Revel area. One of the reasons for basing editions of the Morte Darthur on W is that both its scribes came from near that place.[43]

It may, however, be felt with Malory, as has often been felt with other authors, that, desirable as it might be to recover his orthography as a whole, the benefits of doing so would be exceeded by the risk of getting it wrong. If so, an editor must surely do with orthography as he does with substantives: recover as many individual earlier spellings as far back in the tradition as he can. Because the available evidence is likely to be both limited and patchy, he may have to accept the readings of his copy-text almost all the time, and his few improvements may reproduce the orthography of different stages of the tradition in different places. That is no reason for not emending: exactly the same will have happened with the individual words of his text. Where earlier spellings cannot be recovered, the editor may properly correct misspellings and imported readings to the norms of his base-text.

On these principles, the lexicographical evidence for early spellings of Armenia seems to suggest omitting the initial h- in hermonye from an edition of C as Malory's work. The same principles also provide a yardstick for six other small sets of variants in nos 7, 12, 19, 27, 40, and 47. Capitalising the

differences, C has one letter more in aufFryke and pampoylLe, one fewer in arRabye, and different letters in tarCE, macYdone, and spaynardYs. Comparison with T sheds no light on tarCE, macYdone, or spaynardYs, but raises the possibility that C might have preserved the reading of the archetype in aufFryke, and W in pampoylLe and arRabye. It is tempting, for instance, in the light of T and C's -ff-, to emend W's Aufryke to Auffryke. These variants, however, might be spellings preferred by Malory and Caxton respectively: W gives the name on one other occasion, as Aufrike, and Caxton uses it once elsewhere in his own prose, as Auffryke.[44] Even if they are not, f/ff variations in the spelling of Africa are so common elsewhere that coincidence cannot be ruled out. The other variants too could have been produced by a scribe or compositor's need to adjust the length of his line, or by scribal whims of the most ephemeral kind. It seems best therefore to retain the reading of the copy in all these cases.

Although the original accidentals may not be recoverable in the cases above, applying the Vinaver Principle to nos 5, 6, and 29 suggests the scribe of W or a predecessor simplified the ending of the name to -y in two cases and to -e in the other from an original -ye, representing the Latin -ia. That conclusion makes three other readings suspect. An original -ye may also have been reduced in nos 12 and 20 to -e in W and to -y in T, and in no 40 to -y in W and to -e in T and C. There is some support for this in Mandeville's Travels: one good edition of the Middle English text gives the six names in forms found in these three texts, except that they all end in -ye.[45] This, I suggest, justifies emendation to -ye throughout: in three cases in T, six in W, and one in C; but in W, the emendation for Asia should be to Asye, the full form in both the other texts. An edition of Caxton's prose, however, should retain C's Macydone: it is similar to the form Caxton uses elsewhere and might well be a compromise between that form and the spelling in his copy-text (see Mizobata, s.v. Macedone).

No. 9. This is Hyrcania, in three forms, since Irrttayne is a correction: Thornton first wrote Irrttanye, then corrected it to Irrttayne (Hamel, line 575n). That reading is apparently the product of a series of separate changes to different parts of the name that need separate consideration.

The -ayne ending must in one sense be a corruption, since the -ia of Hyrcania ought also to be represented by -ye, as its counterpart is in two of the four previous names in T. Thornton seems first to have normalized it (much as someone normalized the first name of all to Ambyganye) and then corrected what he had written. His careful correction gives us confidence that in that respect the corrected form exactly reproduces his exemplar; and the reading of the Malory texts confirms that the -ayne ending goes back to the archetype, if (as Dr Hamel argues on other grounds) the exemplar was not the archetype. The preceding consonant is both right and wrong in a similar way but at an even longer remove: the c/t error is common in late

mediaeval English and French scripts, but the existence of a form Hirtanye in a number of manuscripts of Mandeville[46] strongly suggests that in this case the error comes from the poet's source. In the first syllable, the dropping of initial h- and the y → i change may be not a copying error but a deliberate alteration by the author or one of the scribes of the poem or its source: such changes are common in Middle English. Finally there is Thornton's -rrtt-. This seems to be a kind of dittography, since the Mandeville and Malory texts agree in -rt-, and the occasional doubling of letters is a recognised although irregular feature of Thornton's work, in other texts as well as in this one (Hamel, pp. 66 — 67). One has to say "a kind of" dittography, since the term implies unconscious error, and this phenomenon looks like the product of conscious although inconsistent preference. We can hardly doubt that the archetype read -rt-.

The most economical explanation of the evidence is to postulate Hirtanye in S and Irtayne in O, consciously changed to Irrttayne in T and to Ertayne in the Malory texts. The latter change must be presumed to be by Malory, and so should be preserved in any edition of the Morte Darthur, but T's Irrttayne is a more complicated matter. The initial H- might be restored, but dropping it was probably a conscious change, which may well be by the author: it should therefore stand. The -anye / -ayne variation, on the other hand, looks like a case of difficilior lectio, the unconscious substitution of a more familiar sequence of letters for a less familiar one: it is probable that the poet intended -anye, and that therefore is the form we should give. Finally, the medial -rrtt- is plainly an orthographic variant, and almost certainly (as we have seen) Robert Thornton's. It follows from what was argued above that the only kind of edition of Morte Arthure that ought to keep such eccentric letter-duplications would be one edited as Thornton's rather than the original poet's; but although there is a case for a Caxton Morte Darthur, none has yet been made for a Thornton Morte Arthure. We should therefore sacrifice the scribe to the author, and emend to Irtanye.[47] On the same grounds, we should emend Amazonnes (no. 22) to Amazones,[48] and, among the common nouns, tythynnges (line 582) to tythynges.

No. II. This is not a name at all. The Caxton text omits it, but neither T nor W gives owte ilez / oute yles the initial capitals they give to all the other items on the list above. Presumably both scribes and both authors intended the phrase to be a description, not a name. Professor Vinaver and I failed to see this. He supplied and I kept initial capitals that imply the phrase is a name: these should be removed. Presenting the phrase as a description should help to avoid confusion between these islands tributary to Rome and other sets of "out islands" in the Morte Darthur, such as those under the lordship of Ywayne and Ider, which Malory mentions in a passage a little earlier in this tale, in his account of Arthur's forces.[49]

No. 15. This again illustrates the c/t error. The concurrence of T and C in the final consonant suggests a reading Damyat both in O and in X. The scribe of W or a predecessor will have misread the final -t as a -c which he deliberately changed to a preferred -ke, and the compositor of C or a predecessor will have read the word rightly but deliberately varied -at to -ete, perhaps because whoever recognized Alexandria earlier also recognized Damietta here. Morte Arthure's reading Damyat should therefore stand, and any Malory edition be emended to it, but an edition of Caxton's prose should keep the reading of C.

Nos 16 — 17. The problems presented by these names need to be set in context. The Morte Arthure, having given thirteen places in the East Marches to which the emperor has sent messengers, begins a new sequence of places from which troops come to the emperor with

Malory, who has been following his source very closely, gives all thirteen names in the first sequence and then effectively reproduces (apart from interjecting his favourite adjective) the first line of the second sequence: He continues with three of the poem's next four names, before omitting the four after that. The most natural explanation of the absence of this latter quartet of names is that his copy of the poem lacked the lines of the poem in which those names appeared, lines 583 — 586. The absence of other names later (numbers 38 and 41 — 43, from lines 597 and 604 — 605 respectively) might also be due to the loss of whole lines. The end-stopped style of alliterative poetry makes line-loss particularly easy. Athough Robert Thornton was a careful copyist, his manuscript has lost both individual lines and groups of lines (Hamel, p. 4). Those lines may have been lost by an earlier copyist less conscientious than Thornton himself, but there is no reason to suppose that the manuscript Malory used was copied in fewer stages or by more careful scribes. We cannot therefore suppose that it would have been exempt from such accidents.

Crete, however, is not likely to have been lost by the omission of a line, since Cappadocia, the other name in the same line of the Thornton manuscript, appears in both Malory texts. It would be natural to assume instead that Malory's eye skipped from one similar-looking form to another, were it not for C's Cayer. The most conspicuous features of the entire Caxton Morte Darthur are the shortening of this tale and the reduction of its alliteration, and it would be surprising if Caxton (as Vinaver thought [Works, p. 1666])

or any of his predecessors had added to it, and done so alliteratively. It is much more likely that Malory himself altered Crete to Cairo, and that the scribe of W or a predecessor lost the name by homoeoteleuton, presumably as part of a phrase. The Christian island of Crete might well have seemed to Malory less topical and less in keeping with the theme of his story than Cairo, the capital of Moslem Egypt, which in his own lifetime had sent three fleets out to conquer the Christian outpost of Rhodes. Malory indeed had personal reasons to be interested in Egyptian operations against Rhodes. In his youth, his kinsman Sir Robert Malory, Prior of the Hospital of St John of Jerusalem in England, had taken a force out to Rhodes to resist another anticipated invasion by the Sultan of Egypt.[52] That invasion never materialised, but a decade later another Malory, who may also have been a close kinsman of Sir Thomas's, was in Rhodes during the Egyptian seige of 1444.[53] Before suggesting a wording, however, we need to consider the next name.

No. 18. It is a more economical postulate that the middle vowel of Capados was altered by the scribe of W or a predecessor than that it was altered by Malory and altered back again by Caxton's compositor or a predecessor. The latter would not only double the number of alterations, but require that whoever made the second alteration should either have hit upon T's vowel by chance, or have recognized and known how to spell Cappadocia. It seems best, therefore, on the pattern of the next sentence in W, to emend W's kynge of Capydos to read kynge of Cayer, and of Capados.

No. 19. From the variants, we must assume Malory altered Tartary to Tars. All the reference works assume Tars means Tarsus,[54] St Paul's city in Asia Minor, which in ancient times, before it silted up, was a well-known port — it was the scene of Cleopatra's famous meeting with Antony. If Malory made the change, one might speculate, particularly with the change from Crete to Cairo in mind, that he did so thinking of or even alluding to something about the Levant, perhaps something bearing on the fortunes of the Knights Hospitallers.

It is not, however, apparent why Tars should mean Tarsus. The verse-line requires a kingdom to balance Turkey, whereas in the fourteenth and fifteenth centuries Tarsus was a mere possession of the kingdom of Lesser Armenia. In Middle English, however, cloth of tars was a well-known type of rich cloth, and it seems more likely that the Tars intended here is the place that cloth of tars came from. The difficulty is that nobody then or now seems to have been very clear about where that was. Among the places of origin suggested have been Tartary, and Tarshish as Mandeville seems to understand it, as a part of Tartary, both of which have the advantage of

being kingdoms,[55] but it is quite possible that whoever was responsible for the place-name knew no more than that Tars was remote, exotic, and probably oriental.

Nos 25 — 26. These names at first sight suggest that Malory wanted to alter the name in his source, as he apparently did with Crete and Tartary; but that here, as with Tartary, W did not lose its reading. Perce seems a very plausible name for Morte Arthure. It is very frequent in Mandeville: of the names listed above, only India and Arabia occur more frequently in the standard text.[56] Textual criticism, however, must attempt to explain both the plausible and the implausible readings in any group; and in that context the very plausibility of Perce makes it suspect. There is nothing in the Morte Darthur to explain why Malory should have substituted Pontus for Persia. Pontus occur nowhere else in the Morte Darthur, but it is one of the names in Acts ii.9 — and Pamphilia, which follows it in the next verse of Acts, follows it in the same verse in Morte Arthure. It looks therefore as if Pounce was the reading of O and that Thornton or a predecessor substituted Perse. Pounce looks in fact like a classic case of difficilior lectio: an obscure word displaced by a better known word of similar meaning and form. The similarity is so close that the substitution could have been unconscious. T's Perse, therefore, should be emended to Pounce.

Nos 30 — 31. These names also set T and W in opposition, but in this case T's obvious rightness is genuine, whereas W's Nero is not even a place-name. There can be no question of Nero being a variant or a gloss: it looks like a substitution by a reader in whose copy of Morte Arthure the word Nylus was illegible, or who perhaps repudiated it as incomprehensible. Since the rest of the line alliterated on n-, he supplied from limited classical knowledge a name that he believed to be suitable. The word has disappeared from C, perhaps as part of a process of selective weeding of the less likely names, a process we may reasonably attribute to Caxton himself.

Unfortunately, there are insufficient grounds for emending W to Nylus. We must assume that O read Nylus, and that Nero was introduced by the scribe of E, or by Malory himself, or by the scribe of X. If the first, Malory reproduced efficiently the information provided in his source, if the second, he deliberately chose to introduce the word, and we have no right to improve what he wrote. If the third, we should emend, but the odds are two to one against emendation. Since the probabilities are that Nero, though a poor reading, is Malory's own, even a Malory edition based on Caxton should (unfortunately) insert it, though an edition of Caxton's prose would be entitled to leave it out.

Nos 32 — 35. The least unsatisfactory archetypal readings proposed so far for no. 33 are Gerasa and Gaddrys (Hamel, line 592n). Both are places in the Holy Land that alliterate on the right letter. W's Garese might derive

from by Gerasa by vowel metathesis: but vowel metathesis is not common in W or generally, and Gerasa has not been cited in any source known to be available to the poet of Morte Arthure. Garese might derive instead from a variant of Gaddrys, Gadres, by the loss of medial d and the addition of final -e. The first of those two changes may seem arbitrary, but the second could easily have been a reflex — a great majority of names in the passage probably originally ended in -e — and Gadres would have been available in a source from which the poet derived a substantial later incident and from which he may have taken the second name in this passage.[57] Gadres might therefore seem a strong contender, were it not that it, like Gerasa and every other suggested form, does nothing to explain T's Garyere. It is difficult to have confidence in an hypothesis that does not explain the readings of all the texts.

C's Gallacye might, as Dr Hamel tentatively suggested, be a conflation of Garese and Galily. Someone who was prepared to correct the spelling of Alexandria and Damietta would presumably have been prepared to correct a third name in the same passage. It is notable that this is the only part of the passage in which C fails to reproduce names that appear in W. We might suspect an illegible or damaged copy, but that is unlikely. In the part of the passage where the omissions occur, whoever was responsible for C was able to read the name of Syria, and was also able to read enough of Nazareth or Galilee to be prompted to a replacement name that not only alliterated with the nearest two names that disappeared, but was also from the right part of the world and from the same part of the Bible. It looks as if somebody who has been reproducing the list with some care has decided to purge it on three criteria: he did not want names that were textually corrupt or manifestly legendary or from the Holy Land. His single replacement, Galatia, though named several times in the New Testament,[58] is not in the Holy Land. The impulse suggested would also explain the disappearance of the outer isles (no. 11) and Prester John's land (no. 28).

This can only be a guess, but if it is accurate, the leading suspect must be Caxton himself, both because he shortened Malory's Roman War story, and because some other passages in C suggest he may have been prepared to make good what he felt was Malory's casualness on sacred subjects.[59] An editor of Caxton's prose would keep gallacye, but an editor of a Malory edition based on C ought therefore to substitute Nazareth, Garese, and Galely (even though one of them may be nonsense), and restore the oute yles and Preter Johanes londe because they are more likely to be what Malory wrote.

Nos 36 — 37 and 39. The agreement in the order of names between T and W suggests that C has transposed Cyprus and Greece, and therefore that a Malory

edition based on C should reverse that order, although an edition of Caxton's prose should keep it, since it may represent Caxton's preference.

Nos 46 — 47. In W Calabria and Catalonia, the predecessors of these two names, appear in a passable line of alliterative verse. Dr Hamel introduced that line into the text of T because she thought neither Malory nor the scribe of E was likely to have invented names that were so unfamiliar, and or to have put them into alliterative verse. She rejected Portugal and Spain, however, although they seem to preserve the shadow of an alliterative line, because they were more obvious, and suggested that Malory invented them to replace the pagan Prussians and Lithuanians, who in his time were no longer topical. On these grounds she placed the line about Calabria and Catalonia before T's lines naming Apulia, Prussia, and Lithuania.

There is force in this, but it does not allow for three things that have come to light in this essay. Most important to the point immediately at issue is that all the four chronicle sources include Spain: since Spain is in a source and a derived version of Morte Arthure, without a more compelling reason for Malory to have invented it than has yet been offered, the most reasonable assumption is that Spain was also in the intermediate text, the Morte Arthure itself.

Second, our examination reveals great differences between somewhat superficially similar treatment of their sources by Malory and Caxton. Caxton apparently omitted seven names supplied by Malory, most and perhaps all probably because he thought them suspect in one way and another. Those omissions and the names he improved and supplied all look like the product of a more critical temper and a better geographical sense than Malory possessed, based on real commercial knowledge of the world. Malory, in contrast, apparently saw his source as an "authorised" repository of Arthurian story, and his primary urge seems to have been to reproduce what it said. The most plausible explanation for the absence of eight names from his book is that they were absent from his source as well. In the remaining more than thirty cases he apparently reproduced his source's names, some of them very obscure and far from topical, as accurately as he could. Three small attempts at improvement that we may attribute to him, replacing Crete and Tartary and Nylus by Cairo and Tars and "Nero," show him less skilful and less ruthless than Caxton, and make one wonder if his changes were so limited not only because of his respect for his source but because he knew he did not know enough about distant places to improve on it.[60] Either motive would make it unlikely that he would add new names, nor is it particularly probable that any he added would have taken even approximately the form of lines of alliterative verse.

Finally, it may be suggested that Spain and Portugal bring the poet's splendid list of Roman vassals and allies to a particularly appropriate close. Three dozen lines ending in Lithuania would have been an anti-climax: late mediaeval England knew of more exotic places further away than Lithuania. Anti-climax is difficult to reconcile with the rhetorical artistry that Dr Hamel herself has shown the poet to have displayed in this very passage. The poet who began his sweep evocatively with the Orient itself (in line 571) and worked westward across three continents had a sense of form that demanded he should work to some proportionate conclusion. A sequence ending with Spain and Portugal shows the Roman Empire and its spheres of influence stretching from the extreme east to — give or take an Atlantic island or two — the extreme west of the known world;[61] and even without the support of the chronicle sources, that alone, it seems to me, would be a sufficient basis for believing that this section of the Morte Arthure should conclude after line 605 of the Thornton manuscript with two supernumerary lines containing Malory's phrases about Calabria, Catalonia, Portugal, and Spain.

The Texts

The texts below embody only emendations mentioned or argued for in this essay. Others are possible, particularly in editing the passage from C as Malory's work, but any attempt to take that process much further would only show that, at least for his Roman War story, an edition of Malory must be based on W.

• Thane Sir Lucius lordlyche   lettres he sendys 570
• Onone into the Oryente   with austeryn knyghtez:
• Till Ambaygne and Arcage   and Alysaundyre eke,
• To Inde and to Ermonye   as Ewfrates rynnys,
• To Asye and to Affrike   and Ewrope the large,
• To Irtanye and Elamet   and all thase owte ilez, 575
• To Arrabye and Egipt,   till erles and other
• That any erthe ocupyes   in thase este marches;
• Of Damaske and Damyat   the dukes and erles,
• For drede of his daungere   they dresside them sone;
• Of Crete and of Capados   the honourable kyngys 580
• Come at his commaundmente   clenly at ones;
• To Tartary and Turkye   when tythynges es comen,
• They turne in by Thebay,   terauntez full hugge;
• The flour of the faire folke   of Amazones landes,
• All thate faillez on the felde   be forfette fore evere; 585
• Of Babyloyn and Baldake   the burlyche knyghtes;
• Barons with theire baronage   bydes no langere
• Of Pounce and Pamphile   and Preter Johne landes,
• Iche prynce with his powere   appertlyche graythede.
• The Sowdane of Surrye   assemblez his knyghtes, 590

---

127




• Fra Nylus to Nazarethe   nommers full huge:
• To Garyere and to Galelé   they gedyre all at ones,
• The sowdanes that were sekyre   sowdeours to Rome.
• They gadyrede overe the Grekkes See   with grevous wapyns
• In theire grete galays   with gleterande scheldez. 595
• The Kynge of Cyprys on the see   the Sowdane habydes
• With all the realls of Roodes   arayede with hym one;
• They sailede with a syde wynde   oure the salte strandez.
• Sodanly the Sarezenes,   as them selfe lykede,
• Craftyly at Cornett   the kynges are aryfede, 600
• Fra the ceté of Rome   sexti myle large.
• Be that the Grekes were graythede,   a full grete nombyre,
• The myghtyeste of Macedonye   with men of tha marches,
• Pulle and Pruyslande   presses with other,
• The legemen of Lettow   with legyons ynewe, 605
• Of Calabre and of Catelonde   bothe kynges and deukes, W1
• And the Kynge of Portyngale   with many thousande Spaynardis. W2
• 572 Ambaygne] T Ambyganye Arcage] T Orcage 575 Irtanye] T Irrttayne 576 Arrabye] T Arraby 578 the] T and 582 Turkye] T Turky tythynges] T tythynnges 584 Amazones] T Amazonnes 587 Barons] T Bayous 588 Pounce] T Perce 603 Macedonye] T Macedone W1 — 2] om. T

Malory's Le Morte Darthur, from the Winchester Manuscript.

• Than the Emperoure sente furth his messyngers of wyse olde
• knyghtes unto a contrey callyd Ambaygne, and Arcage, and unto
• Alysaundir, to Ynde, to Ermonye that the rever of Eufrates
• rennys by, and to Asye, Aufryke, and Europe the large, and to
• Ertayne, and Elamye, to the oute yles, to Arrabye, to Egypte, 5
• to Damaske, and to Damyat, to noble deukis and erlys; also
• the Kynge of Cayer, and of Capados, and the Kyng of Tars, and
• of Turkye, and of Pounce, and of Pampoyle. And oute of Preter
• Johanes londe, also the Sowdon of Surrye, and frome Nero
• unto Nazareth, and frome Garese to Galely, there come Sarysyns 10
• and becom sudgettis unto Rome. So they come glydyng in
• galyes. Also there come the Kynge of Cypres, and the Grekis
• were gadirde and goodly arayed with the Kynge of Macidonye,
• and of Calabre and of Catelonde bothe kynges and deukes, and
• the Kynge of Portyngale with many thousande Spaynardis. 15
• 2 Ambaygne] W Ambage Arcage] W Arrage 3 Alysaundir] W Alysundir Ermonye] W Ermony Eufrates] W Eufrate 4 Asye] W Assy 5 Arrabye] W Arrabe 6 Damyat] W Damyake to noble] and to noble W 7 Cayer, and of Capados] W Capydos 8 Turkye] W Turke 9 Surrye] W Surre 13 Macidonye] W Macidony 14 Calabre W Calabe

Le Morte Darthur as Malory's work, from Caxton's edition.

• I shall sende for them all that ben subgettys and alyed to
• thempyre of Rome to come to myn ayde, and forthwith sente old
• wyse knyghtes unto these countrayes folowynge: fyrste to
• Ambaygne and Arcage, to Alysaundir, to Ynde, to Ermonye,
• whereas the ryver of Eufrates renneth, into Asye, to Auffryke, 5
• and Europe the large, to Ertayne and Elamye, to the oute yles,
• to Arabye, Egypte, and to Damaske, to Damyat and Cayer, to
• Capadoce, to Tarce, Turkye, Pounce, and Pampoylle, to Preter

---

128




• Johanes londe, to Surrye, to Nero, Nazareth, Garese, and
• Galely. And alle these were subgette to Rome, and many moo, 10
• as Cypres, Grece, Macydonye, Calabre, Cateland, Portyngale,
• with many thousandes of Spaynardys.
• 4 Ambaygne] C ambage Arcage] C arrage Alysaundir] C Alysaundrye Ermonye] C hermonye 6 to the oute yles om. C 7 Damyat] C Damyete 8 — 9 to Preter Johanes londe om. C 9 — 10 to Nero, Nazareth, Garese, and Galely] C and gallacye 11 Cypres, Grece, Macydonye] C Grece / Cypres / Macydone

Le Morte Darthur as Caxton's work, from his edition.

• I shall sende for them all that ben subgettys and alyed to
• thempyre of Rome to come to myn ayde, and forthwith sente old
• wyse knyghtes unto these countrayes folowynge: fyrste to Ambage
• and Arrage, to Alysaundrye, to Ynde, to Hermonye, whereas the
• ryver of Eufrates renneth, into Asye, to Auffryke, and Europe 5
• the large, to Ertayne and Elamye, to Arabye, Egypte, and to
• Damaske, to Damyete and Cayer, to Capadoce, to Tarce, Turkye,
• Pounce, and Pampoylle, to Surrye and Gallacye. And alle these
• were subgette to Rome, and many moo, as Grece, Cypres,
• Macydone, Calabre, Cateland, Portyngale, with many thousandes 10
• of Spaynardys.

For this purpose, C needs no emendation.

Notes