Ignorant scribe and learned editor: Patterns of textual error in editions of Anglo-French texts¹

   In 1943 the Anglo-Norman Text Society published an edition of two manuscripts of La Seinte Resureccion under the names of four distinguished professors, with no fewer than six other prominent scholars being cited as having helped in the work.² On three occasions in the edition the place of origin of Joseph of Arimathaea is printed as Arunachie from the Paris manuscript, whilst the Canterbury one carries the expected Arimathie. Conscious of the difficulty here, Miss Pope, who played the principal role in completing the edition, confirms the Paris reading in a Note.³ In terms of theoretical medieval orthography she is no doubt correct, but in terms both of overall orthographical practice as seen in scores of medieval texts written in Anglo-French and also as viewed from the standpoint of medieval civilization in general, this is clearly a sacrifice of common sense on the altar of philological theory. To be chosen to copy precious documents concerning the faith must have been a considerable privilege for any religious, so to assume that a scribe entrusted by his community with copying a text dealing with such a fundamental article of Christian belief as the Resurrection would have been ignorant of the spelling of a key name as important as 'Joseph of Arimathaea' is beyond belief. Nor is it remotely credible that readers of the text, members of a small literate elite amongst a predominantly unlettered population, would have registered the incomprehensible Arunachie as anything other than Arimathie.

   The reading Arunachie hangs on two highly dubious tenets of medieval French palaeography: firstly, that i is always distinguished from other minims (e.g. those that make up m/n/u/v) by having a dot above it and, secondly, that c and t are always distinguishable by the top of the t being visible above the cross-bar that so often joins it to a following letter. Would that it were so! Indeed, Miss Pope would appear to contradict herself when, after stating that 'symbols such as c and t, n and u, m and iu, ui [are] clearly differentiated' (p.xvi), she goes on to qualify this entirely positive assessment by a somewhat naïve statement to the contrary: 'The reading is only uncertain when the letter is not made with the scribe's customary precision [...]' (ibid.). It could hardly be otherwise. She maintains that such 'scribal inadvertencies' are, however, 'very rare' (p.xvi), but then introduces a purely hypothetical explanation to account for other, hitherto unmentioned, 'errors' in the text, claiming that: 'More frequent are the instances in which the copyist appears to have been misled by a lack of differentiation of letter or symbol in the text he was copying' (p.xvii). Amongst such failures of differentiation she includes Arunachie. So the text is not, in fact, as orthographically unblemished from the standpoint of the modern philologist as she made it out to be in her first statement. Unfortunately, the original that the scribe was copying and which is claimed to be responsible for his setting down Arunachie on three occasions is not available for examination, so the reader has no means of  ^p2 determining the degree of care or otherwise that the original author brought to his work in general, or, in particular, whether he led the later copyist astray in the matter of Arunachie/Arimathie.

   As so often, hapless 'ignorant scribes', author or copyist long deceased, are saddled with the blame for any resulting difficulty experienced by modern scholars approaching the text from their own perspective rather than that of the medieval audience for whom it was written. In effect, the reader of the edition is being asked to accept on trust that the original author's failure to separate beyond any possible confusion every case of c and t in combination with a following letter, a failure that cannot possibly be substantiated on account of a total lack of surviving evidence, led the later Paris scribe to write down an outlandish form for a very well-known religious personage. Furthermore, Miss Pope asserts that this negligent scribe's version of the text shows a 'definite trace of continental Norman influence' (p. xxxix), and another member of the editorial team writes: 'The scribe of P [...] imparts a distinctly continental colouring to his language. [...] Turning to the language of C we are at once aware of a greater remoteness from strict continental standards' (p.xxxix, Note 2). This would strongly suggest that the Paris scribe with his 'distinctly continental influence' and his leaning towards 'strict continental standards' would be the more likely of the two scribes to get 'Arimathie' right, but, perversely, it is the insular Canterbury scribe who sets down the correct form on each occasion. That the Paris scribe fails to do so can only mean either that he must have been slavishly wedded to his supposed carelessly drafted original, to the point of transcribing what both he and his readers knew to be nonsense, or else, more probably, that he was simply not concerned to follow to the letter the strict orthographical rules drawn up for him centuries after his death by modern philologists to bolster their view of the evolution of French. Miss Pope inadvertently confirms this latter interpretation on pp.xli-xlii, where she lists characteristics of the Paris scribe's spelling in other texts, revealing that his 'orthographical system' in general left something to be desired as regards conformity to received twentieth-century wisdom in this department. A similar unruly attitude displayed by another admittedly well-qualified scribe, this time on the north-east border of France, will be commented upon later. On the other hand, for the Canterbury scribe, who, it is affirmed, was not just copying the Paris version, and who 'was no great expert in Old French [...]' (p.xvii), to get the form right on all three occasions of its use would presumably imply that he corrected the supposedly faulty original as he copied it, despite his alleged handicap of a limited knowledge of Old French, a truly remarkable, but highly improbable feat. Alternatively, as far as the tangible evidence is concerned, the original may not have been corrupt at all, so that the Canterbury scribe may just have copied it tel quel 'with meticulous accuracy' (p.xl) as, so we are told, was the case with the other texts attributed to him.

   This apparently trifling question has been dealt with at length because it encapsulates an attitude of mind that has dominated the study of medieval French for many decades. Rather than accept textual reality as their touchstone, basing any conclusions regarding either continental or insular French of the period directly on the evidence that the sum of such textual reality provides, without any pre-judgements, philologists have judged the written legacy of their subject by reference to a model of what they thought it ought to have been in order to conform to phonological rules drawn up in the nineteenth and early twentieth centuries, consequently regarding as reprehensible blunders anything that did not fit into their construct. The fact that no less than ten specialists in Old French on both sides of the Atlantic were associated with this edition and must be considered to have been in agreement with the kind of gratuitous  ^p3 contortion illustrated above is an indication of the authority behind this approach to the editing of Anglo-French texts.

   All such attempts to explain away what is no more than a commonplace scribal 'failure' to differentiate unequivocally between c/t and the minims in im/un/iv/vi are the product of a modern anachronistic preoccupation with form as against substance that goes back to the efforts of nineteenth-century philologists to trace the development of the individual Romance languages. Since first-hand evidence regarding pronunciation in the Middle Ages is in very short supply, in contrast to that provided by the unimpeachable tape-recordings now at the disposal of the researcher into present-day languages and their dialects, the philologists were thrown back on what evidence could be gleaned from a study of the writings of the period, establishing connections between the written forms they found in medieval verse texts and the sounds these forms were thought to have represented at the time the texts were written. This meant constructing a picture of phonetic changes occurring over centuries that had to be based ultimately on small marks made on parchment by generations of mostly unidentified scribes about whose place of origin, possible movements from one area to another in the course of their careers, exposure to external linguistic influences, dates of activity and methods of training often little, if anything, is known. Interpreting this evidence from written forms in terms of sound, the philologists built up over decades a complicated apparatus of sound-values and sound-change over different periods and from different areas, the apparatus that lies behind the title of Miss Pope's major work: 'From Latin to Modern French: Phonology and Morphology'. The second part of this title is highly significant in that it makes no mention of lexis or semantics, without which essential ingredients no language can fulfil its primary purpose of communication.

   Far from covering the whole spectrum of the surviving textual material in medieval French, phonological findings are based on evidence provided in very large measure by texts in verse, which are numerically far outweighed by those in prose. This imbalance may be readily appreciated by comparing the size of the List of Texts given in Miss Pope's book (pp.xxi-xxvi, ranging from the earliest times up to the end of the seventeenth century, and pp. 483-5) with the vastly greater one of Tobler-Lommatzsch which does not go beyond the middle of fourteenth century, or even AND¹. What is more, when the current work being carried out by M. Jean-Loup Ringenbach with the aim of constructing Godefroy's bibliography from the thousands of references in his vast dictionary is complete, this comparison will be even more striking.⁴ A further limitation on the scope of phonological enquiry is that, within those texts in verse, it concentrates overwhelmingly on the final syllables of words at the rhyme. This means that the proportion of the total written material surviving from the medieval period which actually contributes to the phonological history of French as found in the manuals is minute: a very small tail is wagging a very large dog.

    More important than this quantitative factor, however, is the qualitative attitude that it engenders. For the purposes of traditional phonology words are regarded as combinations of individual characters whose phonetic values vary according to their position in the arrangement of letters, together with the time and place of composition of the work in which they occur. The indispensable requirement for the success of such an attempt to deduce sounds from written characters is that the form of these letters made on parchment should be consistent in the work of each individual scribe and consistent also from  ^p4 one scribe to another over time, so that, by extension, the words composed from them should similarly be spelled in a consistent manner. However, although the characters used in medieval French manuscripts may correspond to those in modern French books, this does not necessarily mean that the writers' or their readers' attitude to them was the same as that of the modern writer or reader. The revolution brought about by the printing press has endowed the modern individual printed character with a uniformity and an independence which cannot be taken for granted in medieval writings. The standardized orthography developed in modern times as a result of this revolution posits an equally standardized form of instruction applied over the whole area in which the language is used, a state of affairs found in France and England only from the post-medieval age in which the early philologists themselves lived. Such a process of standardization may be seen at work today in the way in which children are gradually schooled through national systems of education to write in a standard script at the level of both individual letter and word, although their own untutored handwriting and spelling may be perfectly comprehensible, if somewhat idiosyncratic. This reflects the medieval position when the ability to read and write was not widespread and when there was no acknowledged central authority issuing prescriptive directives in respect of spelling and pronunciation.

   The prime requisite for any centralized control over a language is a general availability of dictionaries which alone make possible the dissemination of both a standard spelling and, concomitantly, a generally accepted system of semantic values. In the absence of such dictionaries the wide variety of equivalents found in the profusion of glosses surviving from the medieval period, especially in Britain where two foreign languages were in use alongside the native vernacular,⁵ demonstrates the lack of uniformity in both spelling and meaning current amongst the many scribes who compiled them in different areas at different times, without, however, any necessary implication of backwardness or ignorance on their part. Consequently, any philological approach to medieval French in general and Anglo-French in particular needs to be along the lines used by the writers of the period themselves, concentrating on the complete word rather than on the individual letters which go to make it up, treating the word not in isolation, but situating it in its context. Only by consideration of the full context surrounding a particular term can it be interpreted correctly. Medieval semantic fields differ from those current today no less than the variety of medieval spellings differs from the standard modern spelling. For example, the meaning of the Anglo-French chambre, in all its twelve different spellings so far attested, extends from its generic sense of 'compartment, enclosed space' to the different ways in which that generic sense was realized in the medieval world - 'bed-chamber', 'privy', 'private quarters', 'judge's quarters', 'private domain (of royalty or nobility)', 'private retinue', 'treasury (of king, noble or community)', 'treasure or fortune', 'hangings of a room'.⁶ The precise sense to be attached to the word is not determined by its spelling, but by the context in which it is found. Nor can this linguistic feature be conveniently regarded as a characteristic applicable only to a 'half-known' semi-incoherent language, because the DMLBS entry for camera is along very comparable lines, proving that the same attitude to semantics was current even in Latin, the language of prestige in the Middle Ages. This is hardly to be wondered at, since the same scribal hands and brains were guiding the quill in both  ^p5 languages. A similar state of affairs may be seen in the medieval French astele, whose senses range from 'stick (for fire-wood)' through 'hame', 'sheath of the scabbard', 'splint', 'shaft (of an arrow)', to 'cart-pole' and 'wooden pillar'.⁷ Here again, the DMLBS is broadly in agreement. This need to allocate meaning by reference to context rather than spelling holds good for the lexis of medieval French in general, continental as well as insular. On occasion, it is virtually impossible to apportion correctly the role of scribe and editor respectively in a medley of forms found in a printed edition, as when the variants for seyns (=in this court) are given as follows: seins, feuz, feinz, seinz, leinz and seuz ( BRITT i 362). If correctly interpreted orthographically by the editor, something difficult to check, these variants would show an apparent confusion of s/f, s/l and n/u, but the only certainty is that, however interpreted, they must all have been correctly understood by students of the law for whom this text was fundamental, as witnessed by its twenty-six surviving manuscripts. In sum, the phonologist's preoccupation with individual letters serves only to divert attention from the real purpose of medieval texts - intelligible communication - and so is inimical to the promotion of a correct understanding of the language he is studying, at the level of both word and text.

   The detrimental effect of focusing on the individual character as against the complete word is illustrated by the consequential wholesale dismissal of a mass of non-literary works written in later Anglo-French which fail to meet the phonologists' cardinal requirement of consistency in spelling and thus prevent the connections between letter and sound from being established. When allied to a similar lack of consistency obtaining in matters of grammatical form this lack of orthographical consistency prompted Miss Pope's reference in her book to the 'debasement' of the language after the middle of the thirteenth century, describing it as 'characterised by a more and more indiscriminate use of words, sounds and forms, but half-known' (p.424). In contradiction of these strictures, however, incontrovertible evidence shows that Anglo-French lived on as a major language widely used in Britain well into the fifteenth century,⁸ a survival difficult to explain if such sweeping condemnation were well founded. Leaving aside the thorny question of how anything can be known for certain about the 'sounds' of this later Anglo-French in the acknowledged absence of confirmatory rhymes, if, as alleged, neither its writers nor its readers had a native or even an adequate grasp of the 'half-known' language in which they were trying to communicate, a language claimed to lack coherence also as regards its lexis and its syntactical markers, later Anglo-French could not have served as the instrument of any meaningful exchange of information, public or private, or been used as the linguistic tool of government at national and local level and the language of the law which relies implicitly on a stable system of semantics.⁹ Yet abundant documentary evidence survives to show that it did, indeed, fulfil all these roles. Nor would it have been possible to compile, hundreds of years later, a dictionary of this insular French, illustrating coherent and often complex semantic patterns of words, together with a mass of locutions that bear witness to a semantic continuity that lasted for generations  ^p6 over a wide area of Britain in specialized registers as well as in general vocabulary.¹⁰ In reality, the essence of Anglo-French lies not in sound-values deduced selectively from rhymes and considered in isolation, independently of meaning, but in complete words having widely accepted semantic content embedded in a recognizable syntactical framework so as to constitute a viable means of communication. Sound and form in Anglo-French may be variable, its syntactical system may not conform to what philologists in the past would have us believe to have been a uniform form of French current throughout medieval France, a credo now increasingly under threat, as will be shown, yet its consistent semantic structure nevertheless guaranteed its intelligibility for those who used it. The concern of those who compiled the voluminous body of documents of all kinds in later Anglo-French that are an anathema to the traditional philologist was not to display their command of the minutiae of an imagined rigid orthographical conformity, but to communicate, and their work would be valued by their own and subsequent generations for its content, not its spelling. The modern philologist, driven by the requirements of his speciality, seeks the unattainable and blames the medieval writers and scribes for not providing it.

   A few years after the appearance of the Seinte Resureccion the preoccupation with 'authentic' form and its concomitant disregard of meaning was taken to far greater lengths when the Early English Text Society published in 1958 an edition by W.H. Trethewey of a prose version of the Ancrene Riwle preserved in three insular manuscripts, all dating from around the end of the thirteenth century, roughly a hundred years and some four generations after the Seinte Resureccion. This later work is highly germane to the study of Anglo-French, not because it is unique in its treatment of the text, but rather because, being of a literary nature yet set down in prose at the time when extensive administrative and legal works were being produced, likewise in prose, it straddles the two forms of writing, literary and non-literary, and would fall under the strictures of Miss Pope as mentioned above. In general terms, it may be said that the editor approaches his task from the usual standpoint of traditional philology as found in most editions of literary texts in verse, but without being able to draw verifiable phonological conclusions from this approach owing to the absence of rhymes. He examines the manuscripts in close detail, finding their spelling defective in many places and assuming, as did the editors of La Seinte Resureccion, that his own readings are in every case correct and that the responsibility for the many perceived 'deficiences' these readings reveal lies with the scribes who failed to copy correctly the now lost original. Again, as was the case with the earlier text, he gives no consideration to the purpose for which the manuscripts were written. Significantly, his edition has no glossary or Notes to assist comprehension, although the number of readers capable of using it profitably in its present form must be very small. Like the earlier text, however, the primary purpose of the Ancrene Riwle was, in fact, as an aid to salvation, so that intelligibility must have been a sine qua non. The didactic purpose behind the compilation is expressly stated: A ses duz chers freres e suers en deu hommes e femmes de religion e a tuz icels e celes [ki] cest escrit lirront ou de autre lire le orront, saluz [...] (p.159); later on the same page the scribe writes: E ke vus puissez tost trover en cest escrit ceo ke vus querrez [...]. The text was indisputably meant to be read in private by individuals and also to be read aloud to  ^p7 groups of hearers (E atret le lisez ou deuant vus lire facez p.160.26-7). Moreover, not only is there repeated and correct reference made to Biblical authority and that of the Fathers for the prescriptions being laid down, but the frequent quotations in Latin taken from such authorities are set down without any of the 'errors' found in the Anglo-French. This disparity between the scribes' Latin and their French could be interpreted as meaning that they were better equipped to handle Latin than French, or, perhaps more plausibly, that the French of England at the end of the thirteenth century was not considered by those who used it to demand the same level of strict conformity to the niceties of an accepted system of orthography and grammar as was required for the hierarchical Latin. Like the scholars before him who produced the Seinte Resureccion, the editor is using an anachronistic criterion by which to judge his text. This judgement is supported by an Anglo-French legal text of the period in the very area of language where rigid orthographical precision might be most expected -- the law: 'qe nulle procés des pleez soit discontinué en nule place (i.e. court of law) pur defaute de une lettre ou de une sillable' ( Harriss 520).¹¹ The primacy of meaning over orthography and of substance over form in the minds of the small group of highly educated Englishmen who used Anglo-French to run the country's legal system in the Middle Ages is clear from this statement and tallies with the non-standard language found in many legal and broadly adminstrative texts produced in the later thirteenth and fourteenth centuries.

   During the four generations or so separating the Seinte Resureccion from the Ancrene Riwle the number of people in England who still preserved linguistic family ties with France must have been constantly diminishing, whilst the amount of non-literary writing in Anglo-French increased, so that it is perhaps not surprising to find a far more wayward form of French in the later Ancrene Riwle. This means that the editor of such a text might expect to experience even greater difficulty in squaring medieval textual reality with modern philological theory than his predecessors did when dealing with the twelfth-century Seinte Resureccion. Trethewey writes in his Introduction that:

The text here printed is that of Trinity College, Cambridge, MS.R.14.7, reproduced without emendation according to the plan being followed in the editing of the English manuscripts of the Ancrene Riwle. Consequently obvious errors and doubtful readings remain in the text and must be controlled by the variants given at the foot of the page. The orthography, word-division, and punctuation are strictly those of the base manuscript (p. xxvii: my emphasis).

   In the first place, he admits that the abbreviations 'have been silently expanded', but does not specify what form they take even in the base manuscript, not to mention the two others, where they occur or how he has resolved them, thus leaving the reader with a text that definitely cannot be claimed to be 'without emendation', a text in which emendations are admittedly made, but are not signalled to the reader. He does, however, explain one category of his expansions, saying: 'The abbreviation for re or er stands frequently for e final [...] and has been so read in numerous instances when intrusive between u and r': leuereres 93/23, couerir 190/29, ouerez 105/25, &c.', but only these three out of the 'numerous instances' of this editorial intervention are given, so any others, correctly identified and resolved or not, must join the rest of the scribe's abbreviations that have been silently expanded and thus remain hidden from the reader. Moreover, the editor's admission that er, re and final e can all be represented by the same abbreviation in the manuscript not only means that his intervention in such cases cannot be other than idiosyncratic, but raises the wider issue of the treatment of abbreviations in general.

   This question has recently been explored in some depth by Laura Wright. In Sources of London English she uses a special font to replicate exactly the whole range of abbreviation and suspension signs occurring in her manuscripts. She writes of 'the adaptability of the abbreviation and suspension signs', showing how one small sign <⁹> 

could signify a Latin morpheme when appended to the end of a word (port⁹, "portus"), another morpheme when appended to the beginning of a word (⁹gruo, "congruo"), and the letter graphs "us" when appended within a morpheme (Latin vni⁹, "unius", English brigho⁹, "bridgehouse"). Further, signs could signify whole words [...], or just indicate general suspensions [...]. They could also be otiose. Scribes used the same abbreviation and suspension marks for all three languages [...] (p.9).

   The 'adaptability' of the medieval system of abbreviations referred to by Laura Wright may be appreciated, in particular with reference to Trethewey's treatment of them as referred to above, and as regards Anglo-French in general, by examining the variety of meanings attaching to a modest selection of them in no more than a few pages of late Anglo-French taken from the authoritative Statutes of the Realm which have been printed with the abbreviation signs still in place (vol. 2, pp.278-287). At this point it must be stressed that the language in which these records are couched was not restricted to internal communications within the British Isles, but, as will be shown, is found in a wide range of communications dealing with personalities and authorities abroad, for whom it can have presented no problems of intelligibility, so the Statutes may be regarded as broadly representative of the system of abbreviations used by later scribes in Anglo-French for compiling documents of national and international importance.
1. The tilde over a written character can carry any of the following senses:
(a) over the final m in Westm it probably indicates the full form Westm[oster] (p. 278, 281, etc.);
(b) over n it makes tentz into ten[emen]tz (pp.279, 281, etc.);
(c) over o it makes coe into co[mun]e, coes into co[mun]es or co[mmun]es (p.278, etc.);
(d) over p (i) it makes espuelx into esp[irit]uelx or esp[erit]uelx, espale into esp[eci] ale (p.278); (ii) on the other hand, over the p in drap (twice on p.284) and in draps (seven times on p.284 and again repeatedly on p.285) it is otiose. The form draps is given without tilde on pp.284 and 285, whilst drape occurs three times on p.285;
(e) over q it indicates an expansion to q[e] or q[ue] , either by itself or used in the body of a word such as queconq[e] or queconq[ue], adonq[es] or adonq[ue]s, and turns solonq into solonq[e] or solonq[ue] ;
(f) over r (i) it indicates a missing ot or ost, making nre into n[ost]re or n[ot]re (p. 279, 280, etc.) both forms being given in full elsewhere; (ii) in br, brs or bris it denotes br[ef]]or br[ief ], br[efs], br[eifs] or br[ief]s, but the same result is obtained by the scribe writing brs with a stroke through the initial b, both forms being found on consecutive lines of p.286 (but the stroke through a b can also represent er, as in robbies, to be read as robberies, and herbgiez, to be read as herbergiez, both on p.278, or it can stand for ie as in sibn, to be read probably as sibien, also on p.278); (iii) in Londrs it indicates a missing e –; Londr[e]s (p.283); (iv) over the final r of Surr it probably indicates an expansion to Surré or Surrey (twice on p.278).  ^p10 
2. The hook frequently found over a letter in a word usually indicates either er or re, e.g. for tsdissolute govnance with the hook over the first t, read t[re]sdissolute and gov[er]nance; similarly, read t[er]res for tres, aut[re]s for auts (all on p.278), etc. Sometimes both meanings of the abbreviation are found in a single word, e.g. in pms on p.281 with a hook between p and m and again between m and s, in the first instance it is to be read as re and in the second case as er, hence p[re]m[er]s. In this role it is equivalent to the stroke through b referred to above. Over the initial letter of visemblabement, however, the same hook must indicate ra, i.e. v[ra]isemblablement (p.286).
3. The superscript a can represent ra, as in g^aundes for g[ra]undes (p.278, etc.), but it is also used in cases where it is either otiose or represents au, e.g. tesmoign^ance may be read as tesmoignance or tesmoignaunce, dev^ant as devant or devaunt, enav^ant as enavant or enavaunt, etc. (p.278, etc.).

   This sketchy excursion into the various meanings represented by no more than a few of the abbreviations used in a very small piece of authoritative Anglo-French text demonstrates that they are capable of carrying more than one sense and that a particular sense may be rendered by more than one abbreviation. In many cases, the intended spelling behind the abbreviation cannot be determined with any certainty, so successive readers over the years must have interpreted it as they thought fit. Indeed, the wide range of spellings to be found not only all through the standard dictionaries of medieval French but also in the body of the texts themselves from one line to the next is proof that the medieval scribe was most unlikely to have had one particular spelling in mind. His motivation in writing was semantic, not etymological or phonological. In other words, as Laura Wright says, the system of abbreviations plays a semantic rather than a morphological role, allowing the scribe to move quickly over the page to convey his message without necessarily attending to every little detail of orthography. The other side of this coin is that the medieval reader must have been accustomed to carry out the process of abbreviation in reverse, automatically interpreting the symbols in a manner consonant with the sense of the text. This attitude would naturally carry over from the abbreviation into the other individual characters in the word, with the reader registering its overall sense rather than picking his way through it letter by letter and fretting over the presence or absence of a dot over an i or a bar across a t. This latter approach has been brought in by the modern phonologist, dependent on such minutiae to validate his claims regarding the sounds represented by individual characters set down long ago as he attempts to attach precise values to each one in order to establish a necessary connection between the spellings used by a scribe and his pronunciation. Although such a connection can only be even contemplated in relation to texts in verse, the assumption of a 'correct' orthography which it engenders has been unwarrantably carried over into medieval texts as a whole. To base overarching conclusions regarding the link between spelling and pronunciation on such flimsy evidence is to build on sand.

   In this regard a pertinent, although unusual, example of the relationship between abbreviations and rhymes is provided by the opening verses of a thirteenth-century Anglo-Norman sermon on shrift,¹⁴ a text probably pre-dating the period of 'degeneracy' and so not to be automatically dismissed as not counting for the purposes of phonology:  ^p11 

   

Or escutez lais e clers Ceste romaunce qe vint apers Coment vous devietz estre confés [...]

   In the absence of any editorial explanatory comment on the pairing clers/apers, the reader is presented here with a printed couplet that rhymes but makes no sense. Although the editor does not explain his treatment of abbreviations or mark the places where they occur, simply referring the reader to his previous edition of a similar sermon, it is reasonable to assume that his manuscript would have used the same abbreviation at this point for er and re in what are transcribed as clers and apers. The form clers is undoubtedly correct, the stock phrase clers e lais being widely used, but apers would mean 'open' or 'obvious' and destroy any sense in this context. If, however, the apparent preterit vint is read as the present tense vent 'comes' (see AND sub venir) and apers read as aprés , the sense becomes: 'Now listen lay and clerk (=cleric) to this tale which follows (literally, "comes after"), how you must go to confession [...] '. If this hypothesis is accepted, good sense is obtained, but the rhyme is lost. This would mean that the writer was creating a rhyme for the eye, not for the ear, a rhyme based on the reader's visual, not aural, interpretation of an abbreviation sign. This is far removed from the modern belief in a consistent, 'correct' medieval orthography capable of reflecting details of pronunciation.

   Just as the reader was expected to interpret the abbreviations in his text in accordance with the sense required, he was expected likewise to take in his stride variations in the spelling of words set out in full, because the principle of variability seen in abbreviations extended also to complete words. For example, endings such as -ance/-aunce could be rendered by -antz/-auntz without any loss of intelligibility. In fact, this is a fairly common feature, especially in later administrative texts. It is perhaps difficult for modern readers to accept that a variable orthography carries with it no necessary stigma or slur on those who use it, yet if it was good enough for Chaucer and Shakespeare it can hardly be branded as a mark of ignorance or incompetence. As has been said above, for the medieval scribe meaning, not form, was paramount. This crucial fact seems to have escaped the modern phonologists, whose speciality has nothing to do with meaning or communication and contributes but little, if anything, to the understanding of medieval language and so of the medieval civilization it reflected.

   This question of variability arises yet again in respect of Trethewey's declared intention to produce a text 'without emendation' when he explains his treatment of the variant readings in the Ancrene Riwle offered by the two manuscripts that he does not print. On page xxxii of his Introduction he admits that the variants given in his edition are 'selective'. Although he details the precise criteria which he has followed in deciding whether or not a variant is to be quoted (pp.xxxii-xxxiii), his list of the categories to be ignored under this system contains grammatical ones, and any variant that fails to meet one or other of his tests for inclusion is simply omitted without any reference,¹⁵ so the reader remains in ignorance of their very existence and cannot possibly judge for himself where they occur, how numerous they are or whether they would have had any relevance to the form printed in the edition. They may well add  ^p12 nothing to the comprehension of the basic text, but, as in the case of the abbreviations, the editor is once again coming between the reader and the text of his manuscript: negative interference it may be, but interference nonetheless. In view of his stated aim to produce an 'authentic' text, the two manuscript versions not printed ought at least to be available to the reader through the medium of a fully comprehensive system of variants if the edition is to be regarded as an accurate reflection of the transmission of the Anglo-French Ancrene Riwle in prose. The base manuscript belongs to Trinity College, Cambridge, with the others coming from the Bibliothèque Nationale and Bodley. Although the Bodley text, being incomplete, could not be considered for the role of base manuscript, Trethewey admits that 'the readings of BN (i.e. the Bibliothèque Nationale manuscript) have equal authority against those of TrBd' (i.e. the Trinity and Bodley mss.) and that this manuscript is 'not markedly inferior and its readings can usually be used to correct the rather numerous lapses in the the text of Tr[inity]' (p.xvii). Furthermore, he states that 'both were carefully checked and corrected at the time of copying' (ibid.). These editorial observations are significant, because they not only argue in favour of printing at least the BN ms. in full alongside the Trinity text (as was the case with the two mss. of La Seinte Resureccion) but they remove any suggestion of careless errors of copying, although the editor's reference to the 'rather numerous lapses in the text of Tr[inity]' is not easy to reconcile with this postulated careful checking. These 'lapses' are precisely the non-standard spellings which are the subject of the present study and examined in some detail below. Be that as it may, the scribes must have been satisfied that the finished texts would be intelligible to their readers or hearers. The logical inference from this in the present context is that any remaining lack of intelligibility in the printed edition may be the responsibility of the editor, not the scribe.

   Regarding the c/t relationship discussed above in La Seinte Resureccion, Trethewey writes: 'In principle, t and c are distinct in form, but the letter in the manuscript is often ambiguous. [...] with cel or tel, for example, the choice has proved embarrassing [...]. The situation is similar for n and u. When the form in the manuscript is clearly one or the other, I so transcribe, even though it is incorrect, cf. tront for trout 190/30 and Plesante for Plesance 255/14' (p.xxvii-xxviii). This is a candid affirmation of editorial superiority, but if the 'letter in the manuscript is often ambiguous', his alleged 'principle' must be regarded with some scepticism. In fact, his self-confidence turns out to be misplaced. The reading tront, reinforced by 'sic' at the foot of the page, makes no sense as it stands. The BN manuscript gives the very common tro, so the sense is not in any doubt: it means 'hole', in this context the hole in the privy. The manuscript form read by Trethewey as tront and corrected by him to trout makes sense only when read as trouc, with or without a cross-bar on the final letter. This form is listed in the AND as a variant of tro, and also in the entry traucum 'loch' in the FEW (13 ii 228b), amongst other forms ending in c (traouc, trauc, trawk), so the editor prints a form with two errors, corrects one of them but leaves the other unnoticed. If he had known the form trouc, it would have been better to give the scribe the benefit of the doubt and print it. After all, as a perceptive scholar once said of the medieval scribes: 'leur ancien français était meilleur que le nôtre', however unpalatable that may be to some scholars. The truth of this remark is, in fact, demonstrated here, because earlier in his text (p.2.10) the scribe had compared the human nose and the mouth to deus trous de une chambre foreine 'two holes of a privy', a clear indication that he was no stranger to the correct form, so why he should later be judged to have a sudden lapse of memory and write nonsense calls for an explanation on the part of the editor. As regards the Plesante/Plesance pair, it is safe  ^p13 to assume that Trethewey must have noticed that, despite using what he transcribes as plesante in the role of a noun on p.255.9 and desplesante also as a noun at p.255.9 and 10, the Trinity text has the correct substantival forms pleisance and desplesance on p.268.11 & 14. So, if the editor's transcriptions are to be believed, the scribe must have known and used the correct form on p.268 and nevertheless deliberately or carelessly written nonsense on p.255, in contradiction of the claim referred to above that the manuscripts were carefully checked.

   Alternatively, in the light of Laura Wright's work and as suggested above in respect of the Seinte Resureccion, perhaps the scribe was not ignorant at all, but simply not concerned to set down every jot and tittle as demanded by twentieth-century philologists guided by the 'principles' crucial to their phonological deductions, leaving his text 'ambiguous' in places, to use Trethewey's term, confident that his readers would have enough intelligence to read the words correctly. We perform exactly the same operation today whenever we read a hand-written message in which the calligraphy is less than perfect: the mind registers the sense of the message without necessarily having regard to the shape of the letters. Only the gratuitous assumption by modern editors obsessed with phonology that c and t are ipso facto always distinguishable if the scribe is competent creates the 'embarrassing' problem. In Plesante foreine si est sancte de cors (sancte must obviously be read as sancté, 'health') manger e boiuere (p.255.14), the sense is clearly 'exterior pleasure is bodily health, eating and drinking', with Plesante understandable only as the noun Plesance. Were it not so, with plesante interpreted as an adjective, foreine would have to be regarded as the noun to which that adjective refers, thus giving the meaning 'a pleasant privy ...', hardly what the scribe had in mind. In similar vein, to print pesance terre (p.230.14), followed on the very same line by pesante char (p.230.17), is to exalt this 'ambiguous' orthographical feature of c/t into an immutable dogma, overriding both semantics and elementary common sense, as Miss Pope did by maintaining her meaningless Arunachie. If correct, these discrepant readings on the very same line would make the scribe appear an utter fool, an unwarranted assumption to make. This is seen yet again when Trethewey prints: une poigne de uerges tant come eles sunt liez en semble forces sunt adespescer. Only when the scribe is regarded as competent and his statement treated as coherent and intelligible, rather than as a succession of disparate written characters signifying nothing except, perhaps, individual sounds, does the perfectly sensible statement emerge: une poigné de verges tant come eles sunt liez ensemble fortes sunt a despescer ('a handful of rods as long as they are tied together are hard to break') (p.26.17-18).

   Another similar example in the Ancrene Riwle is that of enclinante for enclinance (Godefroy 3.106a) in the phrase [...] sanz boce de tote conscience e de enclinante a pecché (p.161.24), i.e. 'inclination to sin'. In this case there is no variant given, so that, according to the method of edition as laid down in the Introduction, all three manuscripts must be assumed, rightly or wrongly, to carry exactly the same faulty enclinante at this point, thereby making all three scribes equally illiterate and unintelligent, unable to tell an adjective from a noun, a conclusion based on nothing more than, possibly, one tiny mark on a sheet of parchment. An alternative, although less than flattering, explanation might be that the apparatus of variants as provided in the edition is perhaps not totally infallible, but, as was pointed out above, the reader has no means of knowing. In similar vein, the presence of a correct form with t in essartrer (p.192.9) does not deter the editor from printing on the very next line the noun derived from it as essarceure, despite the variant essartreure, thereby adjudging the Trinity scribe to have known the correct form of the verb, but not that of the noun, although this verb and noun together represented an integral agricultural procedure of daily medieval life. On p.168.31 the editor prints checun aucel com autre [...], and so  ^p14 confident is he of the supremacy of form over substance that he puts in the footnotes: 'aucel: sic: autiel' (i.e. the variant autiel is given in the other two manuscripts, according to the rules set down on p.xxxiii). So again the Trinity scribe, apparently failing to cross his t, is deemed to be an ignorant fool who set down nonsense, thus baffling potential readers and hearers, whilst the other scribes presumably inserted the little cross-bar to make c into t and so provided an intelligible text for the modern editor. Elsewhere, again according to the editor, the scribe of this Trinity text 'written throughout in a clear, even, miniscule book hand' (p. xii) apparently did not even know the French for 'mouth', setting down on p.42.19 the plural form louthes: les jugelours au dyable ke ne sevunt servir de nule autre chose fors de [...] bestorner lour louthes et de (l. dé) oiz roillier ('the Devil's jugglers who are useful only for twisting their mouths and rolling their eyes'), although the variant bouche is common to the other two texts. This is reminiscent of the case of trou referred to above. That a trained, responsible scribe, capable of understanding the Latin of the Bible and the Fathers, one who has at the beginning of the work written bouche quite correctly (p.2.9), should be considered to have committed later in the same work not one but two gross errors of spelling in such an everyday French word in a semantically transparent sentence calls into question either the competence of the medieval scribal class in general or the judgement of the modern editor. What is not in question is the fact that the readers and hearers of the day would not have shared the editor's interpretation of the word. (Other cases of error in the reading of b in other editions will be seen again below).

   Without labouring the point further, a final example taken from the numerous similar c/t difficulties to be found in the Ancrene Riwle leads from Anglo-French into continental Old French and the modern Romance languages. With reference to the well-known Biblical quotation from Matthew ch.23, v.24: 'Ye blind guides, which strain at a gnat and swallow a camel' (King James version), the Trinity scribe writes, at least according to the editor: Mes meint homme [...] eschieue la tincerele e transglout la musche (167.28-29). The BN text, again according to the editor, inverts the c and t, reading cinterele. Lest it be thought that this is just an illustration of the woeful lack of understanding of medieval French endemic in later medieval Britain, two Anglo-French manuscripts of the Commentarius of John of Garland, roughly contemporary with the Ancrene Riwle, give cincerele as a gloss for the Latin zinzala ( TLL i 227), and the correctness of the c form as against the t in both positions in the word may be traced without difficulty from Classical Latin into its medieval form and not only into medieval French but also into modern Spanish and Italian. The DMLBS has cincenella and Godefroy (2.136b) gives cincele, cincelle, cinciele, chinchielle, chincielle and sincelle as forms of the root word, with the diminutives cincenelle, chinchenelle (2.136c) and cincerele, cincerelle (2.137a), as used in a variety of texts from different areas and different periods. The Tobler-Lommatzsch (2.436) adds scincenele to Godefroy's list, quoting additional sources. Modern Spanish has cénzalo and zénzalo, whilst Italian has zanzara. In all this variety of attestations there is no trace anywhere of a form beginning with t, or of any with a t in the middle of the word. All the texts from continental Europe are ultimately based on spoken languages, and even the insular DMLBS (cincenella) and TLL (cincerele (BC) zinzala, i 227) are in line with them: the Ancrene Riwle edition alone is out of step, its forms representing merely the personal interpretation of its editor, either in ignorance or defiance of the dictionaries, and supported by nothing more than his conviction that he can distinguish infallibly between two written characters whose forms are  ^p15 separated by only a tiny stroke made by a quill and this, as he himself admits, by no means clearly in many cases.

   A similar dogmatic approach is to be seen in the treatment of minims in the Ancrene Riwle. At p.214.20 the text as printed says that for a nun to take her hands out of her habit unnecessarily est daunerie a avoir hontage e aceuerie de potein. The difficulty here is the noun aceuerie. This form is not found in the dictionaries and, since no variant is given, it must be assumed to be present in all the manuscripts, otherwise the apparatus of variants must be deemed to be faulty. In fact, the word is a noun derived from acener 'to beckon', so the correct form is acenerie, not attested in the dictionaries, which have only acenement. The meaning is that the showing of the nun's hands is effrontery to be ashamed of and equivalent to the beckoning invitation of the prostitute. On p.45.24 the editor prints nul doz en ceus, where the separation between en and ceus (the difficulty of determining in which of the numerous similar cases such gaps are intentional is referred to on p.xxviii of the Introduction) needs to be closed and ceus read as cens in order to arrive at the correct encens given in the variants as the BN reading, i.e. nul doz encens ('no sweet incense'); yet again, the editor is so convinced of the correctness of his reading that he puts 'sic' after it. He repeats this just a few lines later: si vus ueez ore pour e lessez uostre errour (p.45.32-46.1), where yet again the BN reading gives the correct form - si vus ore ne eez pour [...] ('if you are not now afraid [...] )'. In both these cases, on the strength of his belief in a theoretical distinction between the two, and in flat contradiction of his statement in the Introduction (p.xxvii) that: 'The situation is similar for n and u' (i.e. similar to that for c and t, with its 'ambiguities') he is saying, in effect, that the Trinity scribe was writing nonsense. If this were the case, either his medieval readers would make the necessary changes to the text in their heads or would not understand it, thus negating the whole purpose of his toil and casting doubt on his fitness to be entrusted with his important task. Elsewhere, the editorial reading: La contineure de justice ceo est de dreiturel-te si est silence. Cultus inquid iusticie: silencium. Car silence contiue dreiturelte (p.186.31-187.2) makes no sense as it stands, but it requires no more than a different reading of the minims to make it perfectly intelligible: contineure only needs to be read as coutiveure and contiue as coutive to produce the perfectly acceptable sense: 'the cultivation/furthering of justice, that is of rightfulness, is silence. For silence fosters rightfulness'. (See CONCEVEURE/COUTEVEURE below). The Latin confirms this reading unequivocally, yet the editor's fixation with the minutiae of what he believes is a 'correct' spelling is so great that it leads him not only to set down nonsense in Anglo-French but even to set aside the clear evidence of the Latin which is the basis for the French.

   The editor's perceived 'difficulty' in interpreting minims is increased when other characters are also involved. For instance, des entres peines (p.206.16) makes no sense, but the correct form des autres peines involves nothing more than the reading of a for e, the categorical separation of n and u being, as Trethewey himself acknowledges, highly dubious. The same entre for autre occurs again at pp.223.3, 252.2 and 257.30. On p.253.9 mention is made of la maladie de la goute chauie, the reading 'guaranteed' by the editor's 'sic' and with the BN manuscript said to give the variant chaine. Sadly, neither chauie nor chaine makes any kind of sense: we are dealing, of course, with the well-attested goute chaïve 'falling sickness', where the alternation n/u/v is accompanied by the presence or absence of the dot on the i. To imagine that the medieval readers or hearers of either the Trinity or BN manuscripts would have interpreted the texts as Trethewey does is perverse, but his failure to  ^p16 provide the correct form in a footnote suggests that he himself might not have shared their familiarity with the term. Even when the text as printed patently makes no sense and the editor knows the correct version, he still maintains his reading and attributes the error to the scribe or scribes: dealing with the Devil's huffing and puffing harder to blow godly women off the course of righteousness than evil ones, he prints: de tant sont les boffees au deable e les nerz de ses temptacions sur celes (sc. the godly women) plus forz [...] (p.254.8). The form nerz is patently nonsensical here, and the correct venz is given in the footnotes, but the reading nerz is not questioned.

   Similarly, on p.255.16-17 the printed text runs: Plesante denzeine si est si com aucune fause leesce ou beaute de avoir pris e loos e fame [...]'). Louise Stone, who read the manuscript for the AND, indicated that the original correct reading Plesance had been 'altered in later ink to t', but Trethewey not only keeps this grammatically incorrect Plesante (even though the t turns out to be no more than a later mis-correction), but also the semantically incorrect beaute (l. beauté), putting a footnote as follows: 'beaute sic, both MSS (error for beance?)'. The sense is: 'Interior pleasure is like some false joy or desire' (beance) 'for reputation, praise or fame [...]'. If Trethewey's reading is to stand, the inference can only be that not one, but both scribes must have independently read u for n and t for c, both of them thus producing identical nonsense, neither of them knowing the familiar term beance. There could not be a clearer affirmation of the superiority of dubious calligraphy over grammar and semantics, of learned modern editor over ignorant medieval scribe. Again, the editor twice prints the incorrect form soudiuant/soudiuanz for souduiant: ('the sacrament of the altar destroys completely all the wiles of the Devil') et ses soudiuanz turs ('and his treacherous tricks') p.34.10; Car nule guise de daunoer si soudiuant ne est si "coluerte" (l. colverte) [...] ('For no kind of lover is so deceitful nor devious') p.199.10-11, together with the associated adverb soudiuaument: ('the wiles, deceits and shams') dont il (sc. the Devil) meintes genz soudiuaument engigne et deceit (p.205.27). These words are derived from the verb suduire, sudoire, etc. and are correctly given as soduiant in Rom Chev ANTS 3250 and suzduiant in Proth ANTS 122 and 4050. The whole 'family' of words based on the verb souduire is set down in extenso in Godefroy 7.495b-496b. Elsewhere Trethewey prints eschinel (p.246.24), with a variant eschaniel, when the sense demands eschamel 'stool', where literally one mere dot on an i separates the readings. After printing baing ('bath') on p.146.31 and the plural baingz on the following line, he then prints banig only three words later and gives 'banig sic' in the footnotes, although the BN manuscript also has baing. Yet again, this form banig hangs merely on the positioning of one tiny dot. The absence of this dot leads elsewhere to the reading of a mythical demer (p.206.19), with the footnote: 'demer apparently, both MSS' (italics as printed in the edition), showing that the editor's determination to view a text as being no more than a long string of individual written characters completely blinds him to the semantically obvious. If he had allowed for the not unusual possibility of the minims in what he read as m in both manuscripts being in reality vi he would have arrived at the correct devier: (Christ) pur nos ordes enormitez forment deskes au devier penetrez, ('Christ for our foul heinous transgressions pierced right unto death'), devier 'to die' being used as a substantivated infinitive. Again, on p.180.5 the printed text reads: ensuiez nostre duce dame en tesant e ne mie la breante e-ue en ianglant. Leaving aside the ensuiez which could well be read as ensivez ('follow'), the reading e-ue clearly puzzles the editor, since he writes in the variants at the foot of the page 'eue in inner mar[gin] and also interl[inear] BN' (his italics, my expansions), thus showing that the BN manuscript confirms the reading of Trinity, but he offers no explanation for this strange unanimity in scribal obscurity. Eue (or eve) is, in fact, a form of ive 'mare/nag', hence  ^p17 'follow our sweet lady (i.e. the Blessed Virgin) in keeping quiet and not the chattering of the braying nag'. Under equa (3.233a-b) the FEW lists a form igue from the Franche-Comté region, glossed as the pejorative "vieille jument". So the 'ignorant scribe' is right yet again, his knowledge of medieval French being manifestly superior to that of the modern editor. Another clear case of this scribal superiority occurs in the following line (p.180.6): (when the hound of Hell comes rushing in) ne gisez mie en peis, ne ne seez nul dur pur ueer ke il ueut fere [...]. The variant for nul dur is given as 'nudur (or midur) ', so when ueer is correctly read as veer 'to see' and ueut as veut 'wants/intends' the sense becomes: 'do not remain inactive, nor sit around for a moment to see what he intends to do' (p.14.16). On p.xvii of the Introduction it is asserted that the variant form nudur of BN 'is correct beyond doubt', being based on the Middle English nouder, whilst 'the reading nul dur of both Trinity and Bodley is a corruption'. The reality, however, is quite different: dur as an independent noun indicating a tiny space or amount is well attested in Godefroy (2.748c-749b), Tobler-Lommatzsch (2.2028-9) and FEW (3.192b), all of these dictionaries being readily available for consultation in 1958, so the Trinity and Bodley manuscripts were right and the editor’s preferred Bibliothèque nationale version wrong. The term goes back into the middle of the twelfth century and is found also in Anglo-French from the end of the twelfth century: (The wood of the idols) nul dur ne sent ne veit ('does not feel or see anything at all') and tresor Ne truvai, nul dur d'argent ne d'or ('I found no treasure, not a scrap of silver or gold', Set Dorm ANTS vv. 350 and 1324 respectively); again, Queor e boche ad si adiré N'ad sur nul dorre poesté ('his heart and mouth were so deranged, that he had not a whit of control over either') Salemon v.10892. It also occurs later in a legal text from the early fourteenth century as nuldour YBB Ed I 21-22, p.529. All this would suggest that Middle English took the word from Anglo-French, as so often happened, rather than the other way round. The editor's reading of the variant midur is no more than an uninformed guess, based on nothing tangible, having the same number of minims as nudur, but with the editor presumably adding the 'all-important' dot on the misread i, thus creating a form without meaning.

   Trethewey applies this pursuit of orthographical regularity more widely than Miss Pope did in the Seinte Resurreccion, positing a separation 'in principle', one honoured, however, more perhaps in the breach than in the observance, not only between n and u, whether correctly read or not, but also between u and v, the former being said to have a vocalic role, the latter a consonantal one. The result in the edition as printed is somewhat odd: Vne ueue ke vus ueez (p.19.5); uaut in the base text, but vaut given in the variants at the foot of the page (p.89.14); la ueu lei in the base text, with vars. veuz, veille (92.23); une ueilie, vars. vielle, veille (p.95.30), Les uns var. Les vns (p.164.19); la temptation […] vus uout amener […] (p.17.29); ysbosech dormi e si mist vne femme a garder la porte ke uenteit forment. e vindrent les fiz […] e trouerent la femme ke cesseit de venter en dormant (35.27-29), with venter occurring again at p.36.5. It would be strange indeed if the scribe deliberately wrote uenteit and then venter just a few words later, with venter again just over the page, or if he wrote vus (p.84.2) followed by uus (p.84.4), repeating the same alternation at pp.181.2, 181.30 and 188.12-3. A similar fluctuation is found again: ieo ameroie mout mieuz ke ieo vus uoisse uus recluses e vus noneins […] pendre au gibet […] ke ieo ueisse une seule de vus doner un seul beiser a acun homme du secle uiuant ('I would much prefer to see you recluses and you nuns hanged than to see any one of you give a single kiss to any living layman' p.214.15). This meaning, however, can only be obtained by ignoring the editor's claimed distinction between u and v, and thus putting semantics before  ^p18 dogma. Again, his distinction impedes understanding in the following clause: Car mout a ennuiz de parte li un de le autre (p.147.21). As it stands, this makes no sense. However, in the context of the mutual love between body and soul, the variant reading depart i.e. present indicative third person singular of departir 'to separate, etc.' instead of the locution de parte, i.e. 'on the part of, in respect of', provides the first clue to the correct sense, and then, if the editor's a ennuiz, made up, apparently, of the noun ennui, and the third person singular present indicative of the verb aver, hence, 'has trouble', is read with the consonantal v to make the adverbial locution a enviz/ennviz ('unwillingly'), an intelligible interpretation of the clause emerges: 'For the one is reluctant to be separated from the other'. The exercise of elementary common sense obviates the necessity for the contortions called for by the editor's reading of the text . Elsewhere it is the insistence on reading i for t that gives rise to another difficult passage at p.28.22-23, where the editor prints this: Si li diable entre vus venie nule ire […], in which the transitive verb venie cannot be construed to make good sense, although one of the variants has vente, thus giving: 'If the Devil puffs/exhales amongst you any anger […]'. There is no doubt that this variant has the correct form, and the difference on parchment between an i and a t is not great enough to exclude the possibility of confusion, on the part of either scribe or editor. Either way, the concern of readers, then and now, must lie with the meaning of the text, not with the minutiae of its spelling. A similar rigid editorial interpretation of another pair of 'ambiguous' characters elsewhere leads once more to both grammar and semantics being overridden. When the Trinity scribe writes: E la recluse […] ne vout pas fermer ses fenestres de ses euz, ne ses euz en serrez en countre la greve mort de l'alme (p.177.15) no variant is given for either ses fenestres or en serrez (my emphasis) which would imply that the other scribe or scribes (the reader has no means of knowing whether both the other manuscripts contain this passage, or only the BN one), are also in error. That at least two, if not three, scribes should make two identical grammatical mistakes in this one sentence (for ses read les; for en serrez read enserrer) is unlikely, to say the least. Later in the text we read: hommes e femmes sages ke se font (my emphasis) en serrez contre le leon de enfern ('wise men and women who take up the life of seclusion against the lion of Hell' p.245.13). This time the BN text gives the correct reading se sont enserrez, but the editor prints the incorrect font in defiance of the grammar. The manuscript characters l/s and f/s are notoriously hard to distinguish (see below), but the guiding light of intelligibility based on semantics is not allowed by the editor to get in the way of a highly dubious orthographical certitude.

   Such 'problems' may not have worried the medieval reader, but the difficulties confronting the modern reader of this edition as a result of the editor's devotion to theoretical orthography are not confined to the interpretation of individual characters. As was mentioned above, in his Introduction he remarks on numerous cases of separations occurring within individual words and equally strange agglutinations, noting the frequent separation of the prefixes 'in, de, des, mes, for, and i (of icel, &c.)' from the main body of the word and going on to comment on the difficulty of deciding whether the space left between the prefix and the following word was sufficient to warrant separating them in the transcription, admitting that 'many instances were not easy to decide' (p.xxiii). On the other hand: 'Contrariwise, certain words, especially short forms, were frequently attached by the scribe to the following word, for instance, e (enoit 46/2, a (avendre) 39/22 […]. However, in a closely written hand many border-line cases inevitably occur and I do not claim to have been entirely consistent' (p.xxiii). Yet, despite these admissions, 'word-division' figures  ^p19 amongst his list of variants that are discounted (see Note 11), and he proceeds to act as though the evidence for mangling the words were irrefutable. In the text as printed the reader comes across le diable e nuai (var. enuai) deus souers de religion: here the difficulty caused by the word-division is compounded by the u/v problem examined above, so that to obtain good sense the printed e nuai must be read as one word, the variant enuai, and the vocalic u read as the consonantal v, giving envaï : 'the Devil attacked two nuns' (p.107.4). On p.114.28 le vus en ion ieo must be read as le vus enjon jeo: 'I enjoin you'; for li pot de uostre uentre est si pres a joint au de sa fete menbre (p.129.12) read […] ajoint au desafeté menbre: 'the pot of your belly is so closely linked to the unruly member' i.e. genitals; pers a gieus means pereçeus ('idle, lazy', p.213.24), etc. All these artificial difficulties placed in the path of the reader not only do nothing to enhance the appeal of what the editor makes into a rebarbative text, but, when allied to the many exceptions to his declared desideratum of a totally faithful reproduction of the manuscript, as referred to above, show that the printed text cannot be accepted uncritically as conveying the intended message of the scribe(s), but must be seen rather as an exercise in palaeography based on the tenets of a theoretical orthographical orthodoxy.

   This importance attached to the ideal of a uniform spelling which would facilitate the drawing of phonological conclusions from the scribe's orthography extended, as might be expected, well beyond the editors of the Seinte Resureccion and the Ancrene Riwle. It may be illustrated again, for example, by McMillan's La Chanson de Guillaume from the mid-thirteenth century (vol.1, p.xi and vol.2, p.116). Embarrassed by the lack of such a uniformity in his text, he writes: 'nous espérons arriver à pouvoir faire abstraction du rôle d'un scribe qui, à maintes reprises, laisse entrevoir son embarras, ses hésitations, et même ses erreurs' Ch Guill vol. 2, p.86. The problem for the editor lies in 'les nombreux cas où le scribe représente un seul son par plusieurs graphies', and he goes on to refer to 'le maintien, même sporadique, de beaucoup de graphies françaises en face de graphies anglo-normandes bien caractérisées […]' (ibid.). It would be difficult to find a clearer instance of the way in which, as a result of the traditional devotion to the ideal of a recognizable insular as opposed to an equally recognizable continental spelling, the unruly scribe is perceived as a nuisance whose orthographical vagaries come between the editor and what he thinks ought to be the true form of his text. Consequently, the editor's expressed aim is to neutralize this pernicious scribal input ('faire abstraction de' is how he puts it), and he invokes the work of Miss Pope as one of his models in his efforts to do so (p.87. Note 3). Like Trethewey, he chooses to lay what he perceives to be difficulties at the door of the scribe, rather than accept variability for what it was, an inescapable fact of medieval reality. However, in the much later Chevalerie Vivien, published posthumously in 1997 (Senefiance 39), McMillan's Introduction shows greater awareness of the unreliability of phonological and orthographical data.

   In all this concern about the failings of the scribes in their recording of Anglo-French it is tacitly assumed that the prestige language of the Middle Ages, Latin, was free from blemish. Yet the scribes who dealt with the one dealt also with the other and could, on occasion, commit errors in both. In a fourteenth-century Anglo-French translation of a Latin medical text giving dietary advice for the different seasons of the year¹⁶ reference is made to the desirability of eating in spring: un manere des  ^p20 oysels qe sunt dist conturnices. These are 'quails' (Latin: coturnix DMLBS; contornice Godefroy 2.267a). In summer, the same birds are again recommended, but this time they are given as: les oysels qe l'en appele cucurbites. Cucurbites, however, are 'cucumbers' (Latin: cucurbita DMLBS; cucurbite Godefroy 2.390c). It would, perhaps, be tempting to attribute this error to the 'ignorant scribe' who made the translation into Anglo-French, were it not for the fact that the editor states that: 'The Latin has simply "cucurbite"' (Note 23). The difficulties facing scribes when dealing with registers outside their normal range – medicine, ornithology, etc. - must have been very considerable in both Latin and the vernacular, as a perusal of the DMLBS shows.

   The customary and enduring condemnation of an orthographically wayward Anglo-French, its wide variations in spelling attributable to the fact that those who wrote it did not know the 'correct', etymological forms, postulates, whether explicitly or implicitly, the existence of its opposite against which it can be measured, a far more consistent continental French, its regular spelling related to its etymological origins. Yet the validity of this model form of French on which the neat and tidy separation of insular from continental French is based has been seriously questioned on the continent for some time now, although there is little sign of this being yet appreciated in Britain. It rests on the comfortable premiss that the elusive francien developed in a direct line 'From Latin to Modern French' and was exempt from the various kinds of 'error' that made later Anglo-French so unpredictable. However, over the last few decades a range of influential scholars such as Cerquiglini, Dees and Pfister have seriously challenged the long-held view of the prevalence in France as a whole from as early as the thirteenth century of a regular, widely accepted francien against which could be set the reprehensible insular variety. More recently, Dominique Lagorgette has launched a powerful attack on the continued acceptance of the traditional view of the nature of medieval French as taught in the university system in France itself.¹⁷ She writes: 'A lire les études de phonétique historique (mais aussi de morphosyntaxe, fondée sur un système d'édition des textes traquant et corrigeant les “fautes”), on repère que l'idéologie sous-jacente est plus proche de celle des néo-grammairiens que de celle des sociolinguistes et phonologues modernes' (p.2). Again, 'croyons-nous vraiment savoir comment les locuteurs de l'ancien français prononçaient et articulaient les sons?' (p.3). Her conclusion as regards this question is stark: 'Montrer qu'il existait un système phonétique national homogène au moyen âge alors que l'étude des manuscrits prouve que les différents traits dialectaux s'entremêlent constamment relève de la gageure, de l'alliance des mots ou de l'aveuglement' (p.3). She goes on to describe the hallowed notion of francien as 'un mythe' and proceeds to lambast the denial of the idea of variation in medieval French as a gommage (ibid) .

   So the issue is no longer a simple one of good continental French contrasted with a bad insular variety. Further evidence providing a pertinent parallel to the spelling situation in later Anglo-French dealt with above in the Ancrene Riwle has recently been provided for the border region lying to the north-east of France in an article by Günter Holtus and Anja Körner.¹⁸ Dealing with a legal document in French drafted in  ^p21 Luxembourg concerning the payment of homage for land, they examine the spelling of the original, dated 1275, therefore roughly contemporaneous with the Ancrene Riwle, and four copies made of it up to the middle of the fourteenth century. The dialectal features noted in the copies are not germane to the present enquiry, but the writers comment on peculiarities in word-division – Auffälligkeiten bei der Worttrennung (p.452) – which recall those found in the insular Ancrene Riwle, although they are not as numerous, and then turn to the u/v confusion and the n/u problem. The first of these involves merely the use of v by all the versions of the document in varont (=verront), whilst only the original and IV keep the u in Thio(n)uille, the others having Thionville. The n/u problem, however, is more interesting:

Einen komplizierten, wenn auch für die Entstehungsgeschichte der Abschriften bedeutsamen Fall stellen die Graphien <n> und <u> dar: So wird im Original deutlich zwischen die beiden Graphien unterschieden, während der Schreiber von IV in seiner recht flüchtigen Kursivschrift <n> und <u> zu einem Allograph zusammenfasst (ähnlich wie in einer modernen, schnell geschriebenen Handschrift). Die Zuordnung bleibt in dieser Abschrift letzten Endes für den verstehenden Leser jedoch eindeutig, und man gewinnt den Eindruck eines des Französischen mächtigen Kopisten, der nachvollziehen konnte, was er abschrieb, und der es nicht für nötig befand, eine sorgfältige Trennung der beiden Buchstaben für spätere "Abschreiber" vorzunehmen. (p.455)

   So the earliest copyist (the number 'IV' does not indicate posteriority) made a perfectly intelligible transcription of the original without observing the 'rules' for n and u , whilst the three who made later copies of his work ('I', 'II' and 'III') were unable to separate the two in his 'flüchtigen Kursivschrift' and consequently made errors. In a footnote (no.19, p.456) the authors emphasize that they are aware that the n/u question is not straightforward: 'Uns ist bewusst, dass die Transkribierung von <n> und<u>im Falle von mittelalterischen Urkundenkursiven stets ein heikles Unterfangen darstellt', but claim that they have carefully separated the two graphies in the faulty copies I, II and III. Significantly for the scribal practice as illustrated above in the Ancrene Riwle, they attribute the errors in these copies to the scribes' lack of familiarity with French: 'Die Fehlinterpretationen beruhen offensichtlich auf der Unkenntnis der französischen Sprache' (p.456), contrasting it with the mastery of French possessed by the scribe of IV who did not find it necessary to make the clear distinctions in his spellings. The authors go on to state that: 'der Schreiber von IV [...] [unterscheidet] ebenfalls nicht eindeutig zwischen <c>, <e> und <t>'(p.456). So this scribe, who is thought to come from a Romance background and knew French better than his fellows whose background was apparently germanic, treats the text he is copying in much the same way as the Trinity scribe treated the Ancrene Riwle, confident that his readers would not be disconcerted by any lack of precision in his calligraphy and would read the text as a meaningful communication, making any 'corrections' automatically and unconsciously as they read. As the authors of the article rightly say, this is precisely what we all do today when reading a handwritten message. Yet this does not address the question of the three scribes who set down erroneous forms of commonplace words such as tonz for touz, or the fact that two of them wrote chost for chose and one spelled the modern Trèves as Trienez (both cases on p.456) even though the city lay almost on his own doorstep. In fact, the authors go further, stating that in the case of one scribe the level of discrepancies in his text: 'verstärkt […] den Eindruck, dass der Schreiber die Vorlage weder beim ersten noch beim zweiten Abschreiben sprachlich verstand' (p.462). If such a profound 'Unkenntnis der französischen Sprache'(p.459) existed in the ranks of professional  ^p22 scribes working in a border zone which cannot have been monolingual at the higher administrative level in those days, it is legitimate to wonder why they were entrusted with a task outside their competence, resulting in faulty copies of important documents being preserved in the archives. The officials who read and used these documents must have interpreted them intelligibly, regardless of the niceties of the spelling, just as many thousands of pages of records of all kinds drafted in a similar Anglo-French were preserved and actively put to use over many generations in Britain.

   Along similar lines, in a separate article in the same volume, 'Kontinuität oder Variation? Die Sprache der Luxemburger Grafenurkunden des 13. Jahrhunderts in Original und Kartularabschrift' (pp.393-417), Anja Körner gives other examples of the kind of French that found its way into the archives, e.g. 'fais cognisait au tous ceaus qui ces letres v(er)ront …' (p.403), commenting on 's/f- Abweichungen' (p.405, Note 39) and referring to ces for ses, ci for si, ce for se and ceroit for seroit (pp.405-6). This is followed by a contribution by Marie-Guy Boutier, 'Études sur des chartes luxembourgeoises' (pp.419-447), the first of which (11 June 1237) begins as follows: 'Gie Maheus dus de Loheregne (et) Marchiz faiz conussant a touz ke …'(p.420). The lengthy piece by Martina Pitz, 'Volkssprachige Originalurkunden aus Metzer Archiven bis zum Jahr 1270' (pp.295-392) provides eleven documents (ten in French, one in Latin) with photographs, all of which French texts contain forms of language far removed from the norm. These scholars are continuing the earlier work of Jacques Monfrin in his Documents linguistiques de la France. To take just one example amongst many, in his volume dealing with the Haute-Marne (1972) not only are outlandish forms found on a regular basis in local documents (ataigtes as p.p. of ateindre p.2 ; eutoibre p.24, but otouvre p.75, possibly to be read as otonvre = octobre, etc.), but also grammatical 'errors' like those mentioned above, conesant a toz ceaus […] p.20; je metera, je porra p.31, etc.). Taken collectively, these studies provide a body of strong evidence pointing towards similarities between later insular French and that used in medieval French documents from France itself and its border regions not only from the orthographical, but also the grammatical point of view. In fact, much of this writing would qualify for Miss Pope's condemnation of later insular French as given above, but she could hardly apply her tag 'but half-known' to the language of similar official documents drafted by native French scribes.

   The foregoing fleeting digression into continental French has been made in the present Anglo-French context simply to demonstrate that the 'half-known' French of Britain was, in fact, by no means the pariah that it has traditionally been made out to be.¹⁹ All these different strands come together to show that Trethewey's difficulties with the many 'ambiguities' in his text were in all probability not a mark of scribal ignorance and, in any case, were not the sole prerogative of insular scribes. As was maintained earlier, the root of the problem on both sides of the Channel lies in the preoccupation of scholars with spellings as potential indicators of sounds, leading to a predilection for the study of verse texts from the earlier period rather than the later ones in prose, an attitude which, by extension, is detrimental to research in the areas of lexis and semantics. This affirmation is now supported by Gleβgen in his article in  ^p23 the Trier volume referred to above²⁰ with reference to one of the foremost scholars in the field of French language studies during the second half of the twentieth century: 'Für GOSSEN ist das Verhältnis von Graphem zu Phonem das "Zentralproblem der mittelalterlichen Spracherforschung" ohne weitere Betrachtung von Syntax oder Lexik' (p.259), an observation perfectly in line with the main thrust of the present paper. The inability of such a concentration on phonology to achieve the desired result emerges clearly from yet another paper by a scholar of repute in this important volume of conference papers. In 'Sind Schreibdialekte phonologisch interpretierbar?'²¹ Jakob Wüest writes of the recognition for some decades now that: 'die Schrift nicht einfach ein Abbild des gesprochenen Wortes sei, sondern dass die schriftliche und die mündliche Sprache als zwei Systeme betrachtet werden müssen' (p.38). He concludes as follows: 'Es gilt deshalb abschliessend festzustellen, dass es keine unfehlbare Methode gibt, welche uns erlauben würde, von der Graphie direkt auf die Phonie eines Textes zu schliessen' (p.49). The extension of this stark conclusion from the domain of continental dialect to that of Anglo-French would transform the study of the latter. If there is no guaranteed correspondence between written form and sound in regions where the writers are using their own native form of French, there can be little prospect of establishing such a relationship between the spellings set down by English scribes using an acquired French and the pronunciation of that French up and down a country lacking any unity in the spelling or pronunciation of its own language at that time. All these independent testimonies go to show that for many years in both France and England the concentration of scholarly attention on a small proportion of literary works in medieval French in search of phonological knowledge from rhymes to the neglect of the far greater body of non-literary material containing a far wider range of cultural evidence has distorted the perception of the respective value of the two and hindered enquiries into other areas of language.

   Evidence demonstrating the fragility of the link between pronunciation and spelling on which phonology is based is not, however, confined to new research into French dialects, but has been readily available in Godefroy for the past century, although the dictionary has been regarded solely as a mine of lexicological information, having no necessary connection with other aspects of linguistic study. To take just one example: Godefroy has two substantive entries, ERRANMENT (3.327a-b) and ERRAUMENT (3.328a-c), which are given identical glosses – 'promptement, en courant, avec impétuosité, aussitôt', so that these are unquestionably the same word. In the light of the frequent palaeographical hesitancy between n/u as seen above, it has to be accepted that some of the forms in Godefroy's lists may well have been separated into their present entries on the basis of the readings attributed to them by editors of the texts in which they are found rather than by the hand of the relevant scribes. Short of going back to study for himself each of the manuscripts concerned, the reader can only take on trust the accuracy of the distribution of numerous forms between the two headwords. Mindful of this caveat, he will find that under ERRANMENT Godefroy lists sixteen different forms, under ERRAUMENT eighteen, almost all of them of continental origin. He also has another separate entry ERREEMENT (3.330b-c), with a further two quotations and the glosses vivement, promptement, clearly another variant of the same word. So, at least according to those  ^p24 who transcribed the texts quoted in these entries, medieval authors or scribes wrote this single word in no less than thirty-six spellings. Twelve forms have also been found in Anglo-French, but this 'suspect' insular material with its notoriously wayward spellings will not be taken into consideration in the present case, attention being concentrated on the 'good' French of Godefroy's texts. If the 'rules' for the phonological development of words as laid down in the manuals of historical French grammar are to have any validity across the full sprectrum of surviving textual material in medieval French, as distinct from being a largely theoretical exercise based on a conveniently small sample, this abundance of diverse spellings can only be regarded as the product of 'ignorant scribes'. Yet the great majority of them must have been natives of mainland France, who were therefore no less remiss in their spelling than their insular counterparts, although without being able to have recourse to the excuse that they were dealing with a foreign language. The Altfranzösisches Wörterbuch of Tobler-Lommatzsch confirms the variant spellings for erranment/erraument, but, unlike Godefroy, groups all the 'word family' together under errer (3.775-779), going from the verb to its present participle errant and on to the adverb formed from that participle. Etymologically speaking, this means that all the eighteen forms under Godefroy's ERRAUMENTand the two under ERREEMENT must be erroneous in that they are based on an incorrect u or e spelling instead of the etymologically correct n. (See the item SACHAUMENT/SACHANMENT below). Whilst the n/u 'confusion' here might be construed as being the result of editorial errors of transcription, the forms with e in the body of the word cannot, neither can those in which the 'erroneous' u has been replaced by l (e.g. er(r)alment, er(r)alement, esralment, etc.). Furthermore, it is not just the -ant ending of the present participle that is subject to variability in these entries: Godefroy and Tobler-Lommatzsch give the following forms at the beginning of the word: her-, herr-, esr-, eir-, ierr-, air-, ar-, arr-, aur-, an-, en-. Yet, despite this welter of divergent forms, their semantic content does not vary in the examples given, therefore all the scribes knew the meaning of the term: they just were 'unable' to spell or, presumably, pronounce it 'correctly'. In fact, it is impossible to explain by etymology and phonology the many forms that in practice represent erranment in written continental medieval French. In these circumstances it is difficult to avoid the conclusion that there was not one consistent pronunciation any more than there was one consistent spelling. The same question is raised by the dictionary entries in Godefroy under DONT (2.747b-c) and Tobler-Lommatzsch under DONC (2.2005-10). Both entries cover the senses of the modern French donc only (not dont) and both have a mixture of forms ending in both c and t, yet intelligibility does not appear to have suffered in the slightest. (See DOUNC/DOUNT below).

   These examples have links with the problem in Luxembourg referred to above and are matched by a similar difficulty found in the various manuscripts that provided Anglo-French glosses to the Exoticon of Alexander of Hales around the same time as the Ancrene Riwle and which use a variety of spellings to gloss the Latin calopodium, itself spelt in different forms. The 'standard' vernacular French form that provides the headword in both Godefroy (3.586b-587a) and Tobler-Lommatzsch (3.1330-1332) is estache, and, although Godefroy gives sixteen different spellings for the word and Tobler-Lommatzsch about half this number, all the forms in their entries have t. In contrast, the Anglo-French glosses to the Exoticon contain, along with the 'correct' kalopynda: gallice estache ( TLL i 320), forms with c/ch/sc/sch and even asch: calopinda: gallice eschache, calobinda: eschace gallice, calopinda: echace ( TLL i 320), calopodium: aschaxe and ka[l]opinda : ka[l]opinda: scoche ( TLL i 321). The escache form is also found in two collections of  ^p25 Anglo-French administrative documents, the Red Book of the Exchequer: les esaches des dites tailles Red Bk iii 964 'the sticks of the said tallies', and the Novae Narrationes, where variant readings give estages, lostages and stakes Nov Narr (p.206), thus retaining the correct t, but altering the ending –ac(h)e to -age. As in the case of erraument etc. these spellings could point towards a confusion or a variability in sound as well as in written form. The Gothic root word stakka, supported by the German noun Stecken and the English 'stick' indicate that, etymologically speaking, the t forms are the correct ones. The use of the incorrect s sound as represented by c/ch/sc/sch in the insular texts might well be regarded as just another case demonstrating the ignorance of the insular scribes, were it not for the awkward fact that in Godefroy 1.391b under aressier the form eschaces occurs in an erotic context in the continental Roman de Renart referring to the erection of the penis. The same quotation, but from a different manuscript, is found again in Godefroy 4.630a, under jamberesce. So the absence from the main entries in both Godefroy and Tobler-Lommatzsch of this spelling suggestive of a pronunciation with an s sound cannot be taken as conclusive proof that it did not exist in the 'correct' French of the continent, with consequential implications for the notion of 'correctness' now challenged by Dominique Lagorgette. The ch for t here does not fall into the category of the simple c/t error examined above, and it be would be difficult to claim that eschache in these two manuscripts is nothing more than a scribal or editorial error. The case of estache/eschace is hardly likely to be an isolated one, but, until it becomes possible to search the big dictionaries electronically, there is no sure means of finding others. We are left with the possibility that future research will uncover increasing evidence of such similarities between the insular and continental forms of French in their infractions of philological wisdom.

   The foregoing examination of ERRANMENT and ESTACHE serves to highlight the abiding dichotomy in real terms between the tiny area of the French language suitable to be put under the the phonologist's microscope and the much wider sweep of medieval textual reality as represented by Godefroy and Tobler-Lommatzsch, which is excluded from his studies. Neither of the terms in question can be used to provide any phonological information in the traditional manner, even if situated at the rhyme, where only the endings -ment and -ache would count for rhyming purposes, yet the abundance of their forms cannot legitimately be dismissed as irrelevant to questions of the relationship between form and sound. Any progress towards a better understanding of the civilization of France and England in the Middle Ages to be derived from medieval French will have to come through study of the whole of its textual legacy, not from a concentration on individual written characters or syllables in no more than a handful of works that lend themselves to exercises in phonology.

   Returning to the circumscribed area of later Anglo-French, its disparagement by the experts over many past decades has meant that it is not generally appreciated to what extent this faux français d'Angleterre, with its manifold imperfections and regrettable similarities to the non-literary French of mainland France, was used both at home and abroad. Edward I wrote to his falconer in Anglo-French²² and his son, Edward II, has left a volume of letters in French;²³ mattters concerning Cornish tin-mines are set  ^p26 down in French;²⁴ royal scribes used French not only to record the proceedings of Parliament, but also to communicate with foreign diplomats and also with dignitaries and institutions, both English and foreign;²⁵ the Mayor of London wrote reams of letters in Anglo-French in support of his fellow-citizens in trade disputes, cases of piracy, etc. involving not only francophone authorities along the Channel coast but also those in Dordrecht in 1364 and even Danzig in 1367,²⁶ as well as English nationals and municipalities; a large part of the records of medieval York kept in their York Memorandum Book are in French;²⁷ the port authorities at Southampton and London recorded the movement of shipping and goods originating in Europe or the Middle East in Anglo-French for well over a century;²⁸ the mercantile companies used it for their records;²⁹ the legal fraternity compiled numerous authoritative treatises in it; countless court cases³⁰ and the wills of nobles³¹ and commoners alike were recorded in this 'half-known' language; the sheer volume of medical writing in Anglo-French points to its widespread use alongside Latin at all levels of the profession;³² the religious of both sexes and differing ranks up and down the country used it to draft their rules and record the business of their houses as well as for their active correspondence, both official and private; even cookery recipes appeared early in Anglo-French.³³ The phonological value of this mass of writing is at best minimal, but semantically and lexicographically it is crucial to an understanding of the development of many aspects of the English language and hence to a correct understanding of English history: scripta manent.

   The problems of intelligibility associated with a rigid interpretation of spelling forms which have been discussed above arise particularly in Anglo-French because editors of insular texts often fail to recognize the basic differences that separate the three languages in use in later medieval England. Of the three, only Middle English can be said to lend itself to the kind of orthographical/phonological exercise undertaken for the Ancrene Riwle, because Middle English alone is grounded in a living vernacular. This is not the case with the insular form of French from the later medieval period, which had become over many years increasingly a language for the eye rather than for the ear, being transmitted from one generation of native English scribes to another  ^p27 largely on parchment. Like Middle English and continental French, as has been shown above, Anglo-French had an abundance of different spellings, but these latter could be far more outlandish than the former, because the scribes who wrote them did not necessarily have any connection with France itself, their French being entirely derivative. On the other hand, Medieval Latin shows little appetite for variant forms. Like Anglo-French, it did not rest on a living vernacular, but, having been for centuries the chosen linguistic vessel of the Church, hence the primary repository of the wisdom and knowledge of a supranational elite, its forms were shielded from serious haphazard modification at national level. These differences are clearly reflected in the relevant dictionaries. Whilst the Middle English Dictionary and the Anglo-Norman Dictionary have extensive variant spellings for their entries, the Dictionary of Medieval Latin from British Sources has very few. A recent edition of trilingual medical texts written in England a little later than the Ancrene Riwle illustrates this point. In the Three Receptaria from Medieval England,³⁴ the spelling in the Middle English sections is certainly not 'regular', but the links with one specific regional dialect are so close that Michael Benskin is able to establish the precise area of its composition; no such localization, however, is possible for the Anglo-French element in the texts, although the range of spelling variants is broadly of the same order as in the Middle English sections, being based, however, not on geographical differences, but on the training or social position of those who used it. In contrast, the Medieval Latin is far more 'regular' than the other two and presents few problems of interpretation, but, again, cannot be attributed to any particular location.

   Another factor which makes the difficulties of comprehension in editions of Anglo-French texts more widespread than in continental ones is that they are produced by different categories of scholar. Whilst the problems in the Seinte Resureccion, the Ancrene Riwle and other similar literary texts result in large measure from the editors' unswerving confidence in the correctness of their readings based on their phonological training, in contrast, many of those found in editions of the non-literary works produced in Britain from the time of the Ancrene Riwle onwards stem in the main from their editors' inadequate mastery of French per se. Unlike those who edit literary texts, these editors are not usually specialists in French, medieval or modern, and so have often approached their task lacking the linguistic expertise necessary to deal correctly with the difficulties presented by their manuscripts.³⁵ Additionally, whilst the French editor of a later medieval prose text in continental French has to deal with only one foreign language, the ubiquitous Medieval Latin, the other language of his text being no more than an early version of his mother tongue and hence not usually totally hermetic, the British editor of a similar text in insular French is faced with two foreign languages, Medieval Latin and Anglo-French, in a country where monolingualism is the norm. Any specific help towards understanding the French found in British medieval documents not furnished with class-room aids such as glossaries or translations has largely been directed towards the solution of scribal abbreviations, as in C.T. Martin, The Record Interpreter (2nd edition, 1910) and the more recent Introduction to J.H. Baker's Manual of Law French (2nd edition,1990), which contains a useful table of common abbreviations and contractions (pp.20-23).

   However, although a missed or misinterpreted superscript abbreviation mark or a final  ^p28 contraction incorrectly expanded may hamper intelligibility, many of the textual difficulties that remain for the reader in the eventual edition arise less from the editor's inability to deal correctly with such curtailed forms than from the presence of words set down in full but not correctly identified orthographically or syntactically and which remain, therefore, semantically opaque. As far as the insular variety of medieval French is concerned, the grammars have in the past tended to offer, at best, patterns of phonological and morphological development that did not take into consideration the variability dealt with above and which are, therefore, largely theoretical and of little use to a prospective editor whose speciality is history or the law when faced with a text in Anglo-French.³⁶ Nor can such an editor expect much help from those engaged in the teaching of Medieval French, because the few who edit insular works usually choose verse texts from the earlier period and subscribe to the traditional view of the development of the language. The result of this fundamental division between the editors of the two kinds of text is that each group goes its own way, with only minimal contact, if any, between them. Consequently, the editions they produce have little in common. The introductory matter provided for an edition of a literary text will in most cases describe the manuscript(s), deal with the date(s) of composition, with the phonology, morphology or syntax of the text, and mark any glossary as 'selective', an adjective that usefully covers a multitude of sins, primarily of omission: on the other hand, the introduction to a non-literary work will often take the language of the text completely for granted and focus all attention on its overall place in the history of Britain, despite the fact that, especially in legal documents, the number of surviving manuscripts often greatly exceeds those in literary texts, the inevitable variant readings they produce thus providing a fertile ground for lexicological and lexicographical study. For example, Nicols' edition of the key legal text Britton lists no less than twenty-six manuscripts (Introduction pp. xlviii-liii), with the comment that this is 'far from being a complete catalogue of existing manuscripts of this work'(p. xlviii). Collas used fifteen manuscripts in both of his editions of Year Books (Selden Society vols. 70 & 81). Such numbers are by no means unusual across the legal field and especially in the many volumes of the Selden Society that have been appearing on a regular basis for well over a century. When this often neglected factor is taken into consideration, the amount of non-literary Anglo-French will be seen to be seriously out of proportion to the attention, or rather lack of attention, it receives from specialists in medieval French. In neither literary nor non-literary editions is it usual for the lexis to be afforded much consideration, semantics being the weakest part of most editors' armoury. It invariably occupies the bottom rung on the ladder of importance, so that a considerable number of avoidable errors find their way into print. Despite the above differences, however, many of the errors found in editions of both literary and non-literary Anglo-French texts fall into the limited number of identifiable categories illustrated above in respect of the Seinte Resureccion and the Ancrene Riwle, so that an understanding of their nature may help editors to avoid them and readers to deal with them if the editor does not. The principal letters that cause difficulty, often when in combination with another letter or letters, are c/t, the groups of minims n/v/u and m/ui/vi and f/l, f/s, l/s; the confusion r/n and ei with ie and e with r is also found, but less commonly.

   Naturally, mistakes of a non-recurrent kind also occur in printed editions of both literary and non-literary texts, but their correction is usually possible by reference to context, provided that the reader is conversant with the whole range of Anglo-French, not merely with the well-trodden but narrow path of the standard literary register, and that he is content for meaning to be his guide rather than mere form. For example, the Oxford Psalter reads: Kar tes sajetes entichedes sunt en mei Oxf Ps¹ 37.2, but the Latin: sagitae tuae infixae sunt mihi shows that entichedes is an error for enfichedes and the Cambridge Psalter at this point has the correct enfichees. In one of Bozon's Sermons the printed text reads: Doun est dounk saunz nul fail De regarder le ray du solail BOZ Serm p.29.81-2, where Doun must be read as Boun to make sense ('it is doubtless good to look at the ray of the sun'), or again, Hounte faut e chet en cendre, Pecché crest […] (ibid. p.53.33-4), where, in the context of what the editor calls 'the ominous triumph of evil', it must be Bounté ('goodness') that 'lies in the dust', not Hounte ('shame'). This latter misreading of b and h is mirrored in another religious text from the thirteenth century: ceste desestance Ki entre mes suers est hastie in Robert Grosseteste's Chasteau d'Amour Chast v.384. The editor's hastie makes no sense in this context, the correct form being bastie: 'this trouble that has been caused amongst my sisters'. In the Petit Bruit the waters are further muddied by a second error of transcription in the same sentence as hastoit: parmit (l. parunt) il hastoit (l. bastoit) cel enpoysonement ('by which he set up/plotted that poisoning' Petit Bruit p.7.12. The b/h error turns up again in an administrative text, The Affairs of Ireland,³⁷ where the printed text reads: appela le seint corps Dieu diablerie e hurdys Affairs of Ireland p.133, where hurdys must be read as burdys: 'he called the body of Christ devilry and a fraud'. In the edition of the earlier part of the Anonimalle Chronicle (1991) the error is the other way round: for the jousting in London's Cheapside: furent faites barures d'une part et d'autre (my word divisions) de bone meryn et fu graunt burdiz (l. hurdys) fait en haut en travers le rue ('barriers of stout wood were set up on both sides and a great palisade built high across the street' Anon Chr² p.146. 'B', 'D' and 'H' are not always clearly differentiated on the page, especially when used as initial capitals, but recourse to semantics instead of reliance on a dubious spelling seldom fails to sort them out.

   The same applies to words in the body of a sentence. In one of the earliest pillars of English jurisprudence the editor, a lawyer, transcribes: chiminage, murage, cariage ou reles autres custumes ( Mir Just p.17). The first three words are well-known as taxes for the use of roads, building of walls or cartage, but reles makes no sense in this context. It obviously ought to read teles, 'such'. Likewise, in the Rotuli Parliamentorum the reading: prient pavage & murage a dorer pur vii aunz ( Rot Parl¹ i 423) the infinitive should be understood as a form of durer 'to last/endure', rather than 'to gild'. In the edition of the early part of the Anonimalle Chronicle referred to above it is recorded that John Crabbe equipped ten boats from Flanders: et mist leinz tut l'estor qe li estoit bosoigne pur guerre, et des mellours in venceux de Berewyk Anon Chr² p.150. The final part of this sentence is gibberish as it stands and totally defeats the editors, but all that is needed to produce good sense is for the three minims in to be read as iu (hence ju) a