The pluralistic approach –
The first scholarly edition of
William of Auxerre’s treatise on liturgy [1]


What should a scholarly edition look like? A methodology of scholarly editing has, on the one hand, to be developed under conditions of ongoing change in text theory – as has taken place in a Material Philology – and concomitant changes of perspective on the extant documents. On the other hand, the shift to digital media is augmenting the options for representing and publishing texts and documents, and thus is a decisive factor in methodological enhancement. Stimulated by the fertile interaction of both theoretical and technical options, the digital edition of William of Auxerre’s treatise on liturgy is intended to be a contribution, drawn from practical experience, to the discussion of objectives and methodological standards in scholarly editing.

The present article (1) gives some basic information on the life and work of William of Auxerre and (2) presents the features of the digital edition of his treatise on liturgy. This edition then (3) is placed in a theoretical framework; finally (4) some technical issues and problems that appeared during encoding are pointed out.


1. William of Auxerre’s life and work


William of Auxerre [2] was born in about 1160. After education at Auxerre and Paris he soon became one of the leading theologians of his time. Among his students he was considered »le maitre le plus en vogue« [3] – the most popular teacher. William was very much involved in the foundation of the University of Paris. When a long-growing conflict between the citizens of Paris and students came to a head in autumn 1229 William was sent as a legate of the French court to Rome, where pope Gregory IX was expected to work out a directive. In the papal bull Parens scientiarum of 13 April 1231 Gregory assigns to the University the right to constitute a self-governing corporation, and thus guarantees its far-reaching independence from the local authorities as well as from the French court. William is mentioned as procurator, that is, as responsible for drafting the very document that thenceforth is considered the founding statute of the leading institution of intellectual life in medieval Europe, the University of Paris. Shortly after that, in the autumn of 1231, William died in Rome.


William of Auxerre is known as the author of three works that were probably written during his Parisian tenure:


1. a commentary on the Anticlaudianus, an encyclopaedic poem by Alan of Lille; this commentary has never been printed and is preserved in only one extant manuscript [4];


2. a very influential theological scholastic Summa, which was highly regarded among his contemporaries and successors and for this reason was called Summa aurea (the Golden Sum) [5];


3. the text here in question, namely the Summa de officiis ecclesiasticis, his treatise on liturgy, which has never appeared in print and which now for the first time is accessible in a scholarly digital edition.


Within his treatise on liturgy, William intends to describe and interpret aspects of all parts of Christian liturgy: of the mass, of the liturgy of the hours, of feasts and holidays, et cetera Apart from the apparent function of praising God and creation, all manifestations of liturgy, in the middle ages, are supposed to bear a symbolic truth. This truth is considered to be threefold: moral, allegorical and eschatological. These three spiritual senses of liturgy – as well as the three spiritual senses of scripture – have to be revealed by the theological masters. Accordingly, William describes order of ceremony, texts and prayers, gestures, objects and music, and exposes their moral impact on the believer’s existence. He reveals their allegorical meaning, which points to the history of salvation (as exposed in the holy scripture). And, finally, he demonstrates that every aspect of Christian liturgy, all of the history of salvation and every human being’s existence, is related and directed to the end to come; this eschatological sense or function of liturgy is supposed to remind the believer to prepare himself for the Judgment Day at the end of all times.


By doing this, William’s work takes its place in a rich literary tradition of interpretations of the Mass that began in the Carolingian era, and pushes it forward to a peak that was then followed, with the development of protestantism, by a drastic decline due to increasing criticism of scholasticism and of allegorical methods of interpretation. [6]

William’s liturgical treatise itself is attested by 15 extant manuscript witnesses, average length 80 manuscript pages. Dissemination was mostly limited to the kingdom of France, especially the Île de France and the North, in the 13th and 14th century. [7] Its influence, on the other hand, became extremely widespread thanks to the compilations of two 13th-century authors who closely followed William: William Durandus of Mende’s treatise on liturgy and Jacobus de Voragine’s famous compilation of the legends of saints, the so-called Legenda aurea. Both are among the most widely disseminated and popular works of later mediaeval times, real best-sellers. The former compilation, the Rationale divinorum officiorum by William Durandus of Mende, [8] is the most extensive and comprehensive treatise on liturgy from the middle ages, and William of Auxerre’s treatise is nearly completely and almost literally absorbed into it; compiled between 1286 and 1291 (circa 80 years after William’s own treatise) its dissemination in the 14th and 15th century is wider than that of any other book – except for the bible. [9] The author of the second compilation, Jacobus de Voragine, also makes extensive use of William’s treatise [10]; compiled between 1261 and 1267 the Legenda aurea [11] is considered the most popular collection of lives of the saints in the middle ages; it survived in more than a thousand manuscript copies and after the invention of the printing press was quickly and widely diffused in print, both in Latin and in vernacular translations. Accordingly, the indirect diffusion and influence of William of Auxerre’s Summa de officiis ecclesiasticis through the works of William Durandus of Mende and Jacobus de Voragine can hardly be overestimated.


Table I: Screenshot of the prologue in the digital critical text version with access to various other text versions such as the transcription of the principal manuscript witness (P1), the Cambrai version, print versions and the facsimiles.


2. The digital edition of the Summa de officiis ecclesiasticis

The technical realisation of the digital edition of William’s Summa de officiis ecclesiasticis has been accomplished in close collaboration with the Institute for Documentology and Scholarly Editing, a network of researchers working at various institutions on the application of digital methods to historical documents [12]. The edition is intended to cover all aspects of textuality and provides different representational levels of the Summa:


1. a critical text (Table I), with apparatus notes for the different readings in the manuscripts (marked by numbers), notes for sources, liturgical and biblical texts (marked by lower case letters), and notes for the passages used in compilations by later authors (marked by upper case letters);


2. a very close transcription (Table II) of the principal manuscript witness, [13] with the recording of codicological and palaeographic phenomena, such as column and line breaks, marginal text and rubrics, as well as original orthography, abbreviations and punctuation et cetera;


3. a transcription of a contemporaneously revised text version preserved in a manuscript from Cambrai in Nothern France; [14]


4. facsimiles of all extant manuscript witnesses located within a virtual archive of the entire manuscript tradition.


The reader navigates through the various texts via dropdown menus (Table III) or tables of content, or simply by browsing chapter by chapter and manuscript-page by manuscript-page. All the texts are closely interlinked by chapters and manuscript pages, enabling the reader to pass from each text version to each page of each manuscript witness.



Table II: Screenshot of the transcription of the principal manuscript witness (Paris, Bibliothèque Nationale de France, lat. 14145, fol. 41ra-b).



Table III: Screenshot of the dropdown menu providing immediate access to the facsimile of the Besançon manuscript, Bibliothèque Municipale, 41 (P).



Table IV: Screenshot of the extended description of the Besançon manuscript, Bibliothèque Municipale, 41 (P).

The edition also includes:


1. a description of each manuscript, both in a short version and in an extended version (Table IV), divided into description of layers, content, origin, physical material, texts and citations and closely interlinked with the reproductions of the whole manuscript (thus every statement or observation regarding the document can easily be verified by the reader);


2. indexes of words, of biblical quotations, of the sources and the reception, all closely interlinked with the relevant passage of the critical text;


3. a print version of the philological introduction; and


4. a print version of the critical text (Table V), with lemmatised apparatus notes for readings, sources and reception.


Table V: Screenshot of the print version of the critical text
with a threefold apparatus


3. Theoretical framework – The pluralistic approach towards text


What is the theoretical background for the diversity of the texts that are presented in this edition? The traditional scholarly edition according to the Lachmannian paradigm is only one possible approach or attitude towards the phenomenon of textual tradition – and it is quite dependent on the restrictions of the printed book. According to the pluralistic text theory of Patrick Sahle [15], textuality can be described by means of different attitudes towards the text, depending on the individual perspective on the text and its perception: text is what you look at!



Table VI: Text wheel according to Patrick Sahle


1. Text – or textual identity within the (re)production of a scholarly critical edition – can be understood in terms of its content, of its idea and its intention: text is what it means. From this perspective a five-minute performance of Shakespeare’s Macbeth in a London Underground station is still Shakespeare’s Macbeth.


2. But of course this performance is quite different from the text commonly understood as a structured work with a certain stable narrative and a bibliographic macrostructure, be it in English, German or French.


3. On the other hand translations that are considered to be the »same« text (from the perspective of being a certain work) are different from the perspective of being a linguistic code (a certain series of words).


4. The »same« text might exist in different text versions (in several print or online publications, for example), which would be considered as quite different texts if being used in a seminar. One version might be considered »better« or »more authentic« than others. A small change in wording, spelling or punctuation might lead to »another text«.


5. Text, furthermore, can be understood as a particular document that differs from every other document, even from a copy: a facsimile of an autograph is indeed not the autograph.


6. In between text understood as one particular document and the idea of a text there may be text as a complex visual sign, a document that points, by its material existence and by means of visual aspects, towards a certain idea or truth: medieval diplomata, the bible to swear on, or Shakespeare’s leather-bound collected works on the philistine’s shelf.


Table VII: Positioning of the various components
of the digital edition on Sahle’s text wheel


Which aspects of textuality are covered by the digital edition of William’s treatise on liturgy? There is the diplomatic transcription, attached very closely to one particular document, recording as far as possible the palaeographical and material aspects. But the boundaries between the different aspects are constantly in a state of flux and thus the diplomatic transcription also represents one particular version of the text. The scans of the manuscript pages (the facsimiles) provide the best possible impression of all visual aspects of the documents, while manuscript descriptions give all relevant information on the manuscripts both as physical witnesses of the text and as a material object of history, intentionally produced by human craft to fulfil a certain function. The critical text traditionally covers various aspects of the text. But, primarily, it establishes one authoritative linguistic code that, based on the manuscript witnesses, is intended to reconstruct a text version representing the assumed literal intention of the author. It is modernised, normalised and closely interlinked with the different text levels represented, and it establishes an abstract work structure that can be referenced. There is the Cambrai text which represents a variant text version, preserved by one single document, transcribed, slightly regularised and adapted to the work structure. The introduction, finally, gives information on all textual aspects that are necessary to understand »the text itself«, its meaning and the context of the work.


4. Problems concerning the text encoding: The pluralistic approach and the TEI


Most of the different texts and representational levels are generated from one large XML document that itself has been created by several transformations from one extensive and multi-layered Word document, and has been marked up as far as possible according to the TEI guidelines. [16] The domain of the TEI by nature, or by birth, is the field of linguistics and literary studies. Therefore, problems with the encoding naturally arise when textual phenomena should be encoded that are not in the realm of the work, of the linguistic code or of a certain version, but rather are a material or physical expression of text in a concrete document, thus belonging to the area of text as document. However, ideally a scholarly edition is the basis for research and interpretation; hence the editor has a great responsibility for a thorough and accurate documentation of the sources: the documents are fundamental for the (re)construction of every philological text. For this reason the interest of scholars and TEI enthusiasts is concentrated in Special Interest Groups (SIGs) such as the SIG for manuscripts and the SIG for texts and graphics (designed to elaborate sustainable standards to deal with specific text document issues).


But when the encoding of some rather common phenomena of mediaeval manuscript texts began there was – and there still is – a lack of appropriate TEI elements and attributes; or at least existing elements seemed to be too generic. Thus, for the needs of the diplomatic transcription, auxiliary elements and attributes had to be created, unless appropriate elements were supplied with the TEI guidelines.



Table VIII: Facsimile (detail) of manuscript Paris, BnF lat. 14145, fol. 41ra


Just a few examples from two chunks of two ordinary manuscript pages of the principal manuscript witness will be given here to illustrate several encoding problems. The first chunk (Table VIII) represents a small chapter on the four liturgical hours of the day in general (Prime, Terce, Sext and None), and the beginning of the chapter that deals with Prime (the early morning prayer at 6 a.m.) in particular. The second chunk (Table IX) is taken from the explanation of liturgical matters in the week before Palm Sunday.


1. Rubrics: In both cases, chapters are titled by rubrics (that is, red-coloured paratext): »De quatuor horis diei« (»On the four hours of the day«: Table VIII, line 1); and: »De prima hora« (»On the first hour«: Table VIII, line 14 = 3rd line from the bottom). A <rubric>-element has been introduced with the P5 guidelines released in November 2007. Unfortunately, this <rubric>-tag is only for use in the header – specifically in msItemStruct and model.msItemPart; it is not a paragraph-level element. However, to have this element also available for use on the transcription level would make it easier to deal with these text passages as a whole and individually as regards the different representation levels (for example, to have the option of suppressing them or converting them into headings). [17]


2. Initials: There should be ornamented initials at the beginning of each chapter, but they were never executed by an illuminator: »(F)inito« (Table VIII, line 2) and »(D)eus« (Table VIII, line 14). For the upper one, the scribe has noted by a representative letter in the margin the character to be executed. There are two further types of initials: »Su(n)t« (Table VIII, line 2) and »Q(ueritur)« (Table VIII, line 4). For the mark-up of all different types of initials that appear in the manuscript, generally the character-element <c> has been used with a non-standardised <type>-attribute for initials – being aware that the definition of the <c>-element is from the linguistics module and does not cover initials fully. The respective form of appearance has been defined by the rendition-attribute <rend>.


3. Margins: There is text in the margins. Margins may have similar functions to rubrics and a lot more: They can also be a correction, an addition, a headword, a commentary, a hint, a mark or, indeed, a representative letter for an initial. But in the guidelines there is no appropriate element for simply saying (without interpretation of function): »There is text in margin«. For this reason, for the diplomatic transcription a <margin>-element has been created and placed within the linear transcription in front and at the end respectively of the line where the margin is actually placed in the manuscript. The two margins in red ink on the first chunk (Table VIII) mark the structure of scholastic argumentation: »q(uesti)o« – »Sol(uti)o«, and they refer to »Queritur quare hoc fit in quatuor horis« (line 4) and to »Ad quod dicendum est quod corpus animale est et corrumpitur« (line 5).


4. Punctuation: Traditionally, original punctuation has been regarded as inconsistent and arbitrary and, for this reason, as an obstacle to understanding by today’s reader. Consequently, original punctuation in scholarly editions has always been disregarded in favour of modern punctuation by the editor to support the reader’s comprehension. However, medieval punctuation is significant. [18] In the present case the scribe of the principal manuscript witness generally uses two punctuation marks: a dot on the base line, in visual terms comparable to the modern full stop but not marking principally a grammatical break or a linguistic break or a break as regards content, but rather any sort of spoken or notional caesura (passim); and a kind of inversed semicolon (Table IX, line 1), marking a weaker type of caesura and the end of questions. These punctuation marks have been thoroughly encoded with the glyph-element <g> and defined by the <type>-attribute so as also to cover other special signs such as hyphens (passim), red-coloured paragraphs (Table VIII, line 3) and »ornamental« fillings of incomplete lines (Table IX, line 4). A modern punctuation, in contrast, has been supplied in order to meet the demands of a reader-friendly text version. Another problem for the encoding results from the fact that several functions might be attributed to a particular original punctuation mark: Dots, for example, are used both for various abbreviations such as ».i.« (for »id est«: Table VIII, line 12 & Table IX, line 5) or ».s.« (for »scilicet«: Table IX, line 3) and for enclosing Roman numerals (for example, Table VIII, line 2, 4 & 6). These different functions can be carried out by one single original punctuation mark, even at the same time, for example if an abbreviation or number is situated next to a caesura (such as an abbreviation at the end of a sentence of a modern text).


5. Non-linearity: In any case, when encoding the text, there is always a conflict between the complex order of relationships of the text segments in the document and the linearity of the text versions that are to be generated. For example, the place of the second rubric in the text given here is intended to be before the line where the initial is missing. But a more appropriate place – converted into a heading of a critical text version – would be four lines above (Table VIII, line 13) after the caesura marked by a scribal punctuation mark. Another example of non-linearity that makes diligent encoding of quite a simple feature become a challenging task is the line-break at the end of the penultimate line of the first chunk (Table VIII): The word »di-cens« is hyphenated, while the second part of the word is placed in the last line after the rubric of the following chapter, that itself is placed after the beginning words of the same chapter. In each of the cases, the »correct« order of the text depends on the understanding of text respectively as a document (TextD), as a certain version (TextV), or as a standardised linguistic code (TextL).


Table IX: Facsimile (detail) of manuscript Paris, BnF lat. 14145, fol. 59vb


The pluralistic notion of text is not fully supported by the TEI yet. The notions of text covered by the TEI standard is – despite its universal claim – still strongly related to text theories that, in principle, understand text as a literary work, as linguistic code or as realization of text in a particular version. For several textual aspects related to other notions of text (as expanded on the text wheel), individual modifications and creations of auxiliary elements and attributes are still necessary. But, in fact, most of the document-related phenomena listed above – rubrics, margins, initials, punctuation and non-linearity – are common for many kinds of text documents, not only for medieval manuscripts. Thus it is merely a matter of time until appropriate elements, attributes and methods are established within the TEI guidelines, so as to facilitate a standardised encoding of these phenomena. [19]


Another, more severe problem concerning encoding, according to the pluralistic approach that has been undertaken for the digital edition of the Summa de officiis ecclesiasticis, results from the complexity of the different representational levels of the text: There is no »base text« in the sense that this is understood in the TEI guidelines chapter on apparatus criticus. The transcription of the principal manuscript witness is fundamental; textual criticism is encoded within notes referring to the code of the transcription. The textual criticism is the basis for generating a critical text version and must be thoroughly expounded in annotations and apparatus notes. But these annotations and apparatus notes must refer to the critical text version, which of course does not yet exist, and which has to be generated by textual criticism via collation and corrections – which textual criticism is encoded in the notes referring to the transcription of the principal manuscript witness. The transformation of references and referenced text is thus a complex and confusing task liable to lead both to technical and to editorial failures.

I would like to thank the following people for their support in preparing this paper: Anthony Harvey, Barbara Levergood, Eoghan Ó Raghallaigh, Dot Porter and Patrick Sahle. A slide show of this paper is available under [1].
For William of Auxerre see: Arnold (1995: 1–15); cf. Ribailler (1967: 1192–1195), Ernst (2004: 48–51).
Chenu (1964: 34, n. 1).
Ms. Paris, Bibliothèque nationale de France, lat. 8299, fol. 13–87; see Hauréau (1890 : 351–356), Gauthier (1982: 321–374, esp. 340–344).
William of Auxerre (1980–1987).
A history of treatises on liturgy is still needed. For an overview see Häussling (1980: 1083–1090) and Reynolds (1982: 624–633); for a more recent discussion on interpretation of liturgy throughout history see Angenendt (2001).
French provenance can be proved for the manuscripts from Besançon, Brussels, Cambrai, Carpentras, Douai, Tours, Uppsala, the Vatican Library and both of the Parisian mansucripts; for stemmatological and geographical reasons the manuscript from Treves might be added here as well as three lost manuscripts from Douai, Tournai and Paris. From Italy originate the manuscripts from Graz, Milan and Subiaco near Rome; a manuscript from Montecassino is lost. Only the abbreviated version of the manuscript from Klosterneuburg can be assigned to neither of these manuscript traditions. For further information see the philological introduction with the manuscript descriptions.
William Durandus of Mende (1995–2000).
It was being translated into French in 1372 (see Brucker [2003, 13–33]) and into German in 1450 (see Buijssen [1966–1983]). Between 1459 and 1500 it accounts for 44 incunable prints, 111 prints altogether.
A thorough examination of Voragine’s dependency on William of Auxerre is still a desideratum. For evidence see the philological introduction.
Jacobus de Voragine (1999).
Institut für Dokumentologie und Editorik; website: Essential support to the creation of the digital edition of William’s treatise has been given by its members, among them Patrick Sahle (Cologne) and Bernhard Assmann (Cologne). A revised and enlarged version of the edition is intended to be published in spring 2009. For a preliminary online version see Franz Fischer [1].
Ms. Paris, Bibliothèque nationale de France, lat. 14145, deriving from the monastery library of Saint-Germain-des-Prés, Paris, and perhaps originating from Corbie.
Ms. Cambrai, Bibliothèque Municipale, A 259 (249). This highly elaborated codex was presumably produced in the 2nd quarter of the 13th century in an excellent scriptorium in the Île de France, perhaps in the Cistercian monastery of Ourscamp.
Sahle (forthcoming).
Texts generated from the XML file are (1) the transcription of the principal manuscript testimony, both (2) the print and (3) the digital version of the critical text, including apparatus references to the different manuscripts readings, to liturgical and biblical texts, to literary sources and receptive texts; including also (4) the references to each manuscript testimony by pages and by columns, (5) references to the other text versions and (6) indexes, tables of content and menu navigation following the text divisions of books, chapters and paragraphs.
Cf. the recent thread on the TEI listserv that discusses this general topic [4].
See generally Parkes (1992).
A proposal to introduce a punctuation element with scribal and editorial attribution has recently been made at TEI members’ meeting 2008 within the SIG for manuscript description by Alexei Lavrentiev [3].