The arrangement of lemmata in early glossaries might be local, thematic, or alphabetical. Some medieval compilers followed strictly alphabetical arrangement; some combined alphabetical order with other systems. Hugutio of Pisa and Osbern of Gloucester combined the derivational principle with first-letter alphabetical order. John Balbi and Firmin Le Ver combined full alphabetization with the derivational system. The anonymous compiler of the Latin-German Vocabularius brevilogus arranged his material in three categories - Nomina, Verba, Indeclinabilia, and arranged each of them alphabetically.
126.96.36.199.1. Local arrangement
"The explanation of glosses, obsolete or dialectical and so 'hard' words, at first frequently followed the order in which the words occurred in the texts in which they were found" (Daly 1967: 28). Most early Greek glossaries were arranged according to the occurrence of the source words in a text (local order), for example, the Homeric glosses (Hessels 1910: 124) and Erotian's collection of technical terms in Hippocrates (Daly 1967: 34).
Late classical and early medieval glossaries developed from interlinear and marginal glosses on classical texts, the Bible and Christian authors, assembled in the order of their appearance in a text. Collison (1982: 40) notes, "The late ninth-century Leiden Glossary extracted glosses from a number of classical works, keeping each set distinct and in the order of the original text. This was useful when the text was available, but its use as a general dictionary was practically nil." Other glossographers collected glosses and arranged them in alphabetical order (often only to the first or second letter), or combined two or more glossaries into one. However, glosses removed from their context in this way frequently lost their meaning (id. 1982: 40; Hunt 1991: I, 4).
While large collections were becoming alphabetical, small glossaries based on local order arrangement continued to be compiled throughout the Middle Ages. Those relating to the Scriptures formed a particularly rich field. For example, the lemmata in one part of the Reichenau Glossary are arranged in order of their occurrence in the Bible. The text of the Mammotrectus super bibliam of Johannes Marchesinus, a manual for the clergy, follows the progression of books in the Bible, and also the Church year, that is, the regular progress of scriptural readings in the Church calendar.
The scant evidence remaining to support the use of local order arrangement in Latin-French glossaries in the late Middle Ages is discussed in 2.3.
188.8.131.52.2. Thematic arrangement
Some early glossaries were also arranged systematically by subject, "since this was a more logical method than alphabetical arrangement which brings topics together only by the accident of initial letter order" (Collison 1982: 40). Subject order appears to have been followed by Aristophanes (third century B.C.) in the compilation of his Lexeis (id. 1982: 26), and by Phrynicus Arabius (second century A.D.) (id. 1982: 40). Glossaries might also follow an order which the author deemed to be logical, as is the case with the Onomasticon of Julius Pollux (Daly 1967: 28).
As we have seen in 2.2., works such as Nequam's De utensilium nominibus and John of Garland's Dictionarius, intended for teaching Latin, were written from the twelfth century onwards. Vernacular glosses were either integrated into the text or added between the lines and in the margins. From the thirteenth century, vernacular glosses on a variety of subjects were compiled independently (see 2.4.) and grouped under headings such as parts of the body, household objects, weapons, domestic and wild animals, etc. Doubtless these glossaries had a didactic purpose.
Collections of botanical terms, often collocated in manuscripts with medical receipts, were intended for pharmaceutical or medical use. A few of these glossaries are arranged alphabetically; in the main they are in random order, although it is possible that the glosses occur in the order in which they were extracted from some botanical text(s).
184.108.40.206.3. Simple alphabetical arrangement
Daly (1967: 11f.) tells us,
The only examples of nearly perfect alphabetical order in Greek lexicography are Galen's collection of Hippocratic Glosses (second century), the Ethnica of Stephanus (fifth century), and Suda (early eleventh century) (Daly 1967: 95).
The extent to which Varro (first century B.C.) used alphabetical order in his lost works is uncertain, but two lists of Greek writers on agriculture contained in Book I of De re rustica are in first-letter alphabetic order (Daly 1967: 52). Judging from the ninth-century extract of Festus's second-century epitome, Verrius Flaccus (fl. 10 B.C.) followed at least first-letter, and possibly second- or third-letter order in his vast Libri de significatu verborum (id. 1967: 58).The elder Pliny (first century A.D.) also observed first-letter order in numerous wordlists contained in his Historia naturalis (id. 1967: 38). However, since Pliny copied his material from mainly Greek sources, it is possible that the entries were already in this order in the source documents. Another instance of first-letter order is found in the lexicographical parts of the De compendiosa doctrina of Nonius Marcellus (fourth century A.D.) (id. 1967: 52).
The principles of alphabetical order, known in Antiquity, were later forgotten and were rediscovered in the Middle Ages. Miethaner-Vent (1986: 90) explains that the first step was grouping together all the words beginning with the same letter. Next came organization to the second letter, which was not always the second letter of the word but rather the vowel of the first syllable. Words whose initial syllable was pronounced in the same way were grouped together. It is this arrangement which is found in the Affatim Glossary (ninth century) and described by Daly (1967: 70): "[the order] is determined by the first letter and by the first vowel following the first letter of each word, regardless of what other letters may intervene. It results, for example, that words beginning in fla- and fra- may precede words beginning in fa-, and that 144 glosses intervene between affatim and affecta."
For the letter 'U', continues Miethaner-Vent (1986: 90), glossaries distinguished words beginning with a vowel from those beginning with a consonant; 'H' was often not taken into consideration; 'Ph' might be confused with 'F', and 'K' with 'C'. The determining factor in ordering entries was the syllable, not the letter.
The base manuscripts of the Placidus Glossary edited in CGL V date from the eighth or ninth centuries and full alphabetical order is generally observed. However, it is not possible to know whether this arrangement originated with Placidus. It is very likely the result of later rearrangement since the glossary exhibits the usual medieval practice of ignoring double consonants and medial 'h': Achates Acecurate Accentus Acceptorem Acera Aceris Accipienser, etc. (CGL V: 43f.).
The Abavus Glossary, edited in CGL IV: 301ff. from four ninth-century manuscripts, is in virtually absolute alphabetical order, as is the Liber glossarum or Glossarum Ansileubi  (end of the eighth century).
Papias (fl.1053) states that his Elementarium doctrinae erudimentum (eleventh century) is composed according to the alphabet, not only in the first letters but in the second, third and sometimes even further.  He carries out this intention fairly accurately, although he ignores double consonants and medial 'h' in determining the order, for example, acetum Achialon aciare accida.
When writing his Summa, William Brito (mid-thirteenth century) "followed the precedent of alphabetic arrangement but applied it more thoroughly [than Papias and Hugutio]. The general scheme is that of absolute alphabetization." (Daly/Daly 1975: xxxv). However, Brito occasionally departed from absolute order; for example, derivatives sometimes follow their primary word: onix onichinis, vermis vermiculatus; double consonants are often ignored in determining order: Melchon melodia melota Mello; as is internal 'h': cantus chaos capparis. There are also lapses which are not attributable to any apparent reason: magnatus magispluris magi magister malagma.
Most bilingual glossaries follow a simple alphabetical arrangement. Lemmata in the Latin-French Abavus series (thirteenth-fourteenth centuries), which is based largely on Papias, follow alphabetical order for the root, then tend to follow derivational order for the suffixes. Lemmata in the Aalma series (fourteenth-fifteenth centuries), which is based on the Catholicon, follow full alphabetical order, with only occasional variations. Exhibit 27 compares alphabetical order in the two glossaries.
|Exhibit 27: Alphabetical order in Abavus and Aalma|
Ms. Paris, BN lat. 7692 (Ab4)
Ms. Paris, BN lat. 13032 (A1)
|Roques 1936b: 334, 433|
The compiler of MP generally applied the principles of alphabetization; however, there are lapses. Some are due to disregard for medial 'h': Abolla Abhominor Abhorigenes, and double 'c': Aceries, Accessio, Acestes. Other errors are due to misreadings of the word in the source text, either the Elementarium or the Catholicon.
Alphabetical order in the Glossarium gallico-latinum is essentially first-letter only, as a result of the method which the author of ms. Paris, BN lat. 7684 used to compile his text. He obviously wanted to take some of the French forms from an unknown Latin-French dictionary and use them as lemmata in a French-Latin glossary. He first searched the letter 'A' in his source and extracted French glosses beginning with 'a', listing them in the order in which they occurred. Then he searched the letter 'B' for French words beginning with 'a' and added them on to the end of his first list. Next he searched the letter 'C' for French words beginning with 'a', adding them to the list, and so on. When he had completed his list of French words beginning with 'a', he applied the same technique to those beginning with 'b', and continued in the same way through the alphabet. 
As a consequence, French lemmata overall in GGL are only in first-letter alphabetical order, but within each letter-section where derivatives of one word-family succeed one another, groups of French forms follow second- or third-letter order. Latin equivalents frequently follow derivational order, and then are roughly in alphabetical order.  This phenonemon may be observed in the following sample from GGL.
|Exhibit 28: Alphabetical order in the Glossarium gallico-latinum|
Avorté abortus .ta .tum
Acourcer abbrevio .as .atum
Aurone herba, abrotanum .ni
Absent absens .tis
Absenté absentatus .ta .tum
Aluyne absinthium .thii herbe
Abstinent abstinens .entis
Abunder abundo .das
Abundaument abundanter - adverbium
Abusion abusio .onis
|Ms. Paris, BN lat. 7684|
Another difficulty in ordering the macrostructure of GGL arises from the syntagms and paraphrases which replace synthetic Latin forms. Analytic forms are usually arranged according to the first letter of the paradigm:
Du tout trencher
De an en an
amputo .tas .tatum|
annuatim - adverbium
Occasionally, they are placed under the principal lexical unit:
| Under 'F'|
Celuy qui fait faux
220.127.116.11.4. Alphabetical-derivational arrangement
Compilers of lexica based entirely on etymological and derivational principles applied alphabetization only to a limited extent. The alphabetical arrangement in Book X (De vocabulis) of Isidore of Seville's Etymologiae (beginning of the seventh century) is to the first letter only. Lemmata under letter 'B', for example, begin: Beatus Bonus Benignus Beneficus Benivolus, etc.
The Liber derivationum or Panormia of Osbern of Gloucester (twelfth century) consists of Derivationes in which articles compiled according to the derivational method are grouped in first-letter alphabetical order, and Repetitiones (glossaries) which repeat some lemmata from the derivational sections. The glossaries, too, are in first-letter alphabetical order, because they follow the original order of the lemmata in the derivations.
Daly/Daly (1975: xxxv) note that in arranging the lemmata in his Magnae derivationes, Hugutio of Pisa (end of twelfth century) "adopted an alphabetic order which took account of only the first two letters of each word, but in words beginning with a consonant [he] placed combinations of the consonant with the vowels before combinations of the consonant with other consonants".  Because the articles in Hugutio's work are based on etymological relationships, as are those in the Panormia, they also contain sub-lemmata which cannot be accessed alphabetically. It was not long before alphabetical tables were added to manuscripts of Magnae derivationes to improve its consultability (see 18.104.22.168.2.).
John Balbi created the first Latin dictionary in Part V of his grammatical treatise, Summa grammaticalis quae vocatur Catholicon, combining alphabetical and derivational arrangement. Balbi carefully describes absolute alphabetical order in the lengthy preface to his work.  Daly/Daly (1964: 237) remark that it "may be a long-winded explanation of absolute alphabetization but it is at least clear, and it was put into practice." Balbi follows alphabetical arrangement with very few lapses. He often includes derivatives as lemmata in the alphabetically-ordered macrotext, rather than as sub-lemmata in the article headed by the base form. These derivatives may be treated as lemmata, or they may contain a reference to the base word, e.g. Abigeatus ab abiges gis dicitur. vide in abigeus; Abnormis. in norma exponitur.
Firmin Le Ver applies absolute alphabetical order in arranging the lemmata in his Dictionarius (mid-fifteenth century). Each lemma begins at the left-hand column margin. In most articles, there are sub-lemmata (derivatives), each of which also begins at the left column margin on a separate line. Le Ver also adds derivatives which logically complete the macrostructure (Merrilees 1994a: xiv). Exhibit 29 compares alphabetical-derivational arrangement in the Catholicon and DLV.
|Exhibit 29: Comparison of alphabetical-derivational arrangement in Catholicon and DLV|
Marcidulus. la. lum. aliquantulum marcidus. a marcidus dicitur|
Marcidus. a marceo. ces. dicitur marcidus. da. dum. putridus, arescens. languens. Et comparatur. dior. simus. Unde hec marciditas. tatis. et marcido. das. .i. marcidum facere. et est activum et corripitur ci
Mainz 1460; reprint Westmead: Gregg 1971
MARCIDUS - a marceo, marces dicitur|
Marcidus .da .dum - .i. putridus, arescens, languens .i. flamis, come flourettes, pales, languissans, matis, secs ou pourris - comparatur
Marcide - adverbium - matement, palement comparatur Marciditas .tatis - flamissure, come de flourettes, paleur, mateur ou langueur
Marcitudo .tudinis - idem - Marcor idem
Marcido .das .datum - marcidum facere .i. faire flamir, sequier, palir, matir, languir act
Marcidatus .a .um - .i. marcidus factus
Marcidulus .la .lum - diminutivum - aliquantulum marcidus aucunement flamis, matis, come flourettes
Merrilees 1994a: 3, 293
Lemmata are ordered alphabetically in DLV, while sub-lemmata are arranged hierarchically according to an order which is determined by derivation. At the first level of derivation, substantives tend to generate adjectives, and verbs to generate participles, e.g. MARCIDUS - Marcidus. At lower levels, adjectives and participles often generate adverbs and derived nouns, e.g. Marcidus - Marcide - Marciditas - Marcitudo (Merrilees 1992: 79).
Derived forms also appear at times in the macrostructure in DLV, usually in cases where the orthography of the derivative is much different from that of the root, e.g. ABITIO .tionis - In Abeo, abis dicitur; LANUTUS .ta .tum - In Lana dicitur.
[Next] -- [Table of Contents]