The selections above are from the MED. But why not extend the wealth? I mention just one example of a misconceived and misrepresented entry. In the Dictionary of Medieval Latin from British Sources there is an apparent hapax legomenon. Its single quotation is from the Trinity College, Cambridge, Latin-English glossary: "Circumpres anglice a prayer of a worde." Yet the interpretation of this quotation by the editor is revealed in the entry word and gloss "circumprex, ?prayer." Several Medulla manuscripts provide justification for the MED entry "preier(e n.(1): One who offers prayers." The MED, supporting circumpres by analogy with interpres, furnished the correct insight; the s indicates agency. In DMLBS (Latham and Howlett 1975-), the choice of circumprex reveals the nuance behind prayer: prex = prayer (the means of reverent communication, not the instrument).

What little reference there is to the Medulla in other literature of the period is for the most part to be seriously questioned because of the lack of palaeographical experience evidenced, and a want of philological training in Greek, Latin, and Middle English. A superficial knowledge will not do; rather, it seriously retards progress. Once a reading insinuates itself into a text, it is very difficult to remove it.

The benefit or shortcoming of a glossary is that it provides little or no context most of the time. It invites speculation in the form of additions to its apparently simple structure, the detailed study of which is well under way in the capable hands of our colleague, Brian Merrilees (1991), in his seminal work on lexicographical metalinguistics. Most of the manuscripts of the Medulla enunciate the following format: the entry or lemma, followed by the oblique form, that is, the genitive of the noun, the feminine and neuter endings, if an adjective, and the second person singular active or deponent, if a verb. This is then followed by the abbreviation for id est, and finally the interpretation or gloss. In the case of a few manuscripts, there are further additional glosses given in Latin or English in an attempt, it seems, to provide even greater clarity of meaning (and sometimes it is as helpful as not being distracting). One Medulla manuscript, Harley 2257, regarded by A. S. Way, the editor of the Promptorium Parvulorum, an influential glossary of the fifteenth century, as "perhaps the most valuable ms. of its kind in the British Museum" (p. lii), exceeds even these generous offerings by furnishing in many cases elaborate etymological notes as well as all known compounds of the verbal entries, and cognate forms of adjective, noun, or participle. As expected, these varying formats create endless difficulties in organizing and presenting a critical apparatus.

The tradition of the Medulla Grammatice provides remarkable variety regarding spelling of even the most common words. The English language in this period, i.e., 1100 to 1500, and in the case of this glossary, the l400's (early to late), was quite volatile. The word virtue, for example, is spelled differently within the same two or three lines, e.g., vertue and vertu, with i replacing first e frequently. In fact, during one of many forays into the Medulla, I came upon the word Armelausa, which I had earlier read in a Greek papyrus I had edited, a receipt for garments: (McCarren 1980: 48, and note line 11, p. 50). In the Medulla it is spelled three ways, arme-, arma-, and armi-, and it is normalized in Isidore as armilausa (Lindsay 1911). It is defined as that which is divided before and behind and is opened; closed only across the shoulders, as if armiclausa, but with the letter c removed. My text provided the first appearance of the word in papyri: meaning 'a military cloak'. It seems to have been a phonetic variation upon Indeed, it created as much difficulty in both languages, producing in Medieval Latin at least as many as 14 variant spellings. It might be said that the Greeks had considerable foresight in glossing by 'tongue', thereby emphasizing the remarkable fertility of the medieval mouth.

To grasp the full importance of a gloss is to understand thoroughly the significance of what we call the definition of a word. To appreciate it in its pristine context is to realize a different method of alphabetization and a unique grasp of grammatical and etymological principles, some of the material of which has not reached our handbooks and grammars of Latin and English.

A glossary is an amalgam of undistilled marginalia and supralineal insertions arranged somewhat alphabetically and otherwise in verbal families, and ultimately based upon a system of phonetics more or less known only to the scribe, which certainly upset normal alphabetical expectations. Attempts to alphabetize seem to have met with resistance at every stage. What, for example, can be said with any confidence about the alphabetization of a work which on the one hand exhibits a patch of twenty words perfectly alphabetized to the last letter, and on the other, no series of five words that can be sustained alphabetically even within initial letter order? Consider the Pepys entry "gera ge sanctus le", which doesn't belong under g, except (according to our scribe) phonetically. The Greek word is which is transcribed hieros. The letter n has its share of vocal turbulence: "nea ge nouem le" belongs under ennea ('nine'). We are not privileged with a legitimate shortened form as found in Stonyhurst. Nor will "noma ge" work for its gloss "nomen le". The correct form is onoma and obviously it doesn't belong under n where Pepys places it. A bit less foreign but no less to the point is the entry "lauda, a larke", apparently innocuously placed between laudo, 'to preyse', and its own diminutive laudula, 'a litel larke', in the Stonyhurst manuscript. There is just one hitch: no evidence anywhere shows that the word lauda can mean 'larke' or even that it in fact exists. The correct word here is alauda, which has no business being placed under l.

In other cases, also, the Medulla is quite disordered. What, for example, conditioned the Stonyhurst scribe to put an Ad- segment within Ac-? or more striking, why did the scribe of Harley 2181 insert 60-70 entries from Amamen to Amen between Accedior and Aciecula? Finally, what about the confused artistry in the Add. 33534 manuscript? The scribe develops an interesting alphabetical pattern: from Pabulum, the first word of P, to plaxillus, all is reasonably well except for the inevitable inconsistencies. At this point he resumes with Peani through pec-, pel-, pem-, pen-, to persuadeo and next doubles back to the pl- section he abandoned and picks up plebesco and then continues through to the end of P. The damage is that five and one-half columns or 229 entries are out of alphabetical order. The above examples represent the broader view of the problem, for which there might be, and in some case are, reasonable explanations. If, however, one looks more closely, one is continually jolted. Consider the word proditus in the Stonyhust manuscript. Glossed as "nobilis propositus, a waryson [garrison] or a tenement", I wondered about it. Everyone's gloss of proditus is 'betrayed'. Propositus is a fairly colourless word, but nobilis is not; it is quite the opposite of proditus. Also, "a waryson or a tenement" is out of order here surely since proditus, being an adjective, has nothing to do with a gloss revealing two nouns, especially the meanings they convey. One hopes that something will click. Logical analysis is probably not the answer. It's the sudden bark of the dog or erratic breathing in the pool that brings this sort of solution into view. I'm lucky. We have a dog and I swim regularly. Suspecting some spelling disorder, I found some two folios later the proditus I hoped for, glossed as 'bytrayed'. With confidence I lurched backward to the entry and gloss in question and realized that the misspelled word should be preditus, which has about it that "noble" aura; yet, preditus is an adjective and the two nouns in the gloss don't fit. The rest is simply textual. Two manuscripts, Harley 1738 and St. John's (Cambridge), point up that "waryson or a tenement" belong to predium, which is the following word in Stonyhurst and is a noun. The other manuscripts are not inconsonant with Harley 1738 and St. John's (Cambridge) and suggest that Stonyhurst might have been conflating two independent glosses, 'hauynge of toune or of fylde' and 'or a waryson or a tenement'. Perhaps this might be the explanation for the disjuncture. The Stonyhurst scribe wrote the one gloss, and then wanting to insert the other afterward, placed it with the wrong entry.

It becomes evident that the position of a word is sometimes a clue to its intended spelling. In Add. 33534, Eruro is found between Eructuo and Erudio. No alphabetical sense can be given until one realizes that there is no such word as eruro, but rather it is a mistake for erudero and so is again correctly placed but just miscopied.

[Continue]