Le Médiéviste et l’ordinateur
Le Médiéviste et l’ordinateurHistoire médiévale, informatique et nouvelles technologies
n° 42 (Printemps 2003) : La diplomatique


Nicholas Karn & Richard Sharpe
Université d’Oxford

Anglo-Norman royal acta are among the most fundamental sources used by historians of the territories of the Norman kings of England in the later eleventh and earlier twelfth centuries. They are widely used to illustrate and clarify issues in political, legal, social and religious history. Despite this popularity, grave problems remain in using them, and a database has been planned as a tool for dealing with these. Richard Sharpe began some years ago to collect the printed texts from many sources, now filed by beneficiary, as part of his research in support of the teaching of diplomatic in the Modern History Faculty in Oxford. During the year 2001–2002, however, funding has been made available by the Aurelius Trust to convert these paper files into a database, and Nicholas Karn has worked full-time on this task. The database contains the texts of around 350 acts of King William I, 220 of William II, 1650 of Henry I, 1060 of Stephen, among them small numbers of royal acts in the names of William I’s queen Matilda, Henry I’s first wife, also Matilda, and their daughter Matilda, known as the Empress, and Stephen’s queen, another Matilda, and also of some royal children. The texts amount to somewhat over 500,000 words.

The most basic problem facing users of Anglo-Norman royal acta is difficulty of access. The material survives in some abundance, but is dispersed among a few hundred institutional and familial archives ; much has been published, but generally in relation to the context in which it has been preserved, through editions of cartularies and comparable materials. For two reigns only are there modern editions, of the acta of Kings William I and Stephen [1]. No collected editions have ever been produced of the charters of Kings William II and Henry I, only calendarings of the texts, now outdated [2], which suffer both from incompleteness (well over 100 addenda for Henry I are already known from printed sources alone, and we have not yet begun any systematic searching of manuscript sources) and from the limitations of the calendar form. The texts of the acta are not themselves presented, so that the necessary study of words and formulae is all but impossible ; and the descriptions of each text are commonly so elliptical that they do not form an adequate guide to the acta. Not infrequently the calendaring shows a misunderstanding of the document in question which only becomes apparent if the user has recourse to the text itself. Some of the terms used to describe the acta, such as mandate and precept, disguise the contemporary category, in these cases, the writ ; and the manner in which the calendars were the work of several hands has produced a certain imprecision and inconsistency in description, sufficient to confuse readers and, it seems, the original editors themselves. It has sometimes happened that a single text has been calendared more than once in greatly differing terms, presumably reflecting the practices of different editors [3]. The usefulness of the Regesta calendars is further compromised by the inadequacy of the indexing.

The problem of access is a practical one, not a challenge to understanding the texts, but it has had important adverse consequences for that understanding. Serious diplomatic study depends on ready access to a multitude of texts. The calendars provide only a starting-point, and it is a great labour, requiring recourse to hundreds of volumes, not all of them easy to find, even to see the printed texts. Studies of the acta of William II and Henry I have thus been very few [4], while even for the reigns of William I and Stephen the difficulty of locating the texts of their reigns in the context of preceding or succeeding diplomatic developments has retarded the study of the documents. It cannot be said that the development of the writ in the time of Henry I, and the origins and development of the general address, are fully understood. Even relatively simple diplomatic questions, such as the date at which Henry I took the title of ‘duke of Normandy’, and the manner in which he used it, are not understood, and indeed have scarcely been discussed.

The database has thus been conceived as a means of achieving easy access to the acta of the Anglo-Norman kings. Apart from its practical usefulness, it will be a powerful tool for a fuller understanding of the texts. It is a step in a larger project on Anglo-Norman royal acta that envisages three principal outcomes : editions of the acta of William II and Henry I, a public form of the database, perhaps made available through the web, and a coherent account of Anglo-Norman royal diplomatic in monograph form. It is necessary that the database should be usable for all kinds of searches based upon different attributes of the acta, and this need for flexibility has determined much of the design of the database.

The most fundamental kind of searching is for words or phrases in the acta themselves. Flat-file text-searching of this kind is possible simply using modern word-processing programs and does not require the elaboration of a database. It is, however, a crude form of searching in that the results are limited to the search-term itself. It would, for example, be impossible to search for occurrences of a word or phrase in authentic acta alone, because word-processing software has no means of distinguishing between authentic and forged acta if both are in the same file. The database in MS-Access has been designed to allow this kind of searching. Each act constitutes a single file, and grouped with the text are fields recording attributes of the document in a standardized form. The formulaic nature of diplomatic documents means that the number of categories for many of these attributes is small, and can be reduced to a menu of standard options ; thus, for example, the acta can be categorized by document type as writs, writ-charters, charters, letters, diplomas or pancartes. These standard classifications can be combined with text searches, so that, for example, writs containing the phrase ‘portuum maris’ can be found. This pattern has been repeated for many of the other attributes where fixed categories can be drawn up in the form of a standard menu. In order to remove unwanted references in such searches, a language filter has been introduced. As the language of most of the acta, Latin is the default setting, but the option exists for Old English to be entered where relevant, and it is intended that options be added to accommodate the handful of bilingual Latin and Old English texts and the small number of acta surviving only in the form of early-modern English or French translations [5].

In selecting attributes, those in most general use in studying and describing English royal acta of this period have been used. The categories do not express anything new in the analysis at this point, but might of course be extended and altered to take account of developments in understanding of the texts. The categories are simple. The attribution of each act to one or other of the Anglo-Norman monarchs is relatively straightforward, and has been done by a simple list of names, comparable to the method by which the category of document was recorded [6]. The forms of address, which provide the most immediate route to a categorization of the documents by type, have been entered in a similar manner. The standard addresses to a shire court or to several shires, and in the second half of the period the general address, account for a large proportion. Writs of various types increase the number of forms, and the identification for the first time of more than twenty relatively usual forms of address has lengthened the list.

All Anglo-Norman royal acta have a beneficiary, but categorization of the beneficiary has necessarily been managed in a different manner from the criteria described above. In the other cases, a simple menu was practical because the number of options is never great, while the number of beneficiaries is very large, comprising some few hundred persons and institutions. The descriptions of persons and institutions in the documents are rather inconsistent, and the creation of a single authority-file for each person or institution has allowed a uniform and consistent nomenclature to be imposed, reducing scope for possible confusion of similarly-titled persons or institutions. Linking all occurrences of a person or institutions to a single file also has the advantage of conveying some sense of unity of activity. Moreover, at least in the case of the personal beneficiaries, these same people appear elsewhere in other acta, especially as witnesses to the charters. It is therefore possible to find all occurrences of a person or institution by one search, regardless of how many different roles are represented for each individual, and regardless of changing titles or offices.

The attribution of dates to the acta has also been handled in a different manner, largely because of the difficulties in dating texts that generally do not themselves include any explicit date. Dated texts account for only a small minority, but most acts can be dated within defined limits, for example, after the accession of Henry I and before the death of Robert Bloet, bishop of Lincoln, that is, 3 August 1100 x 10 January 1123. At first it was thought adequate to provide six boxes [year, month, date x year, month, date], with the option that any boxes could be left blank, so that a specifically-dated text could use the first three boxes only. This would also allow for imprecisely-dated events to be used as dating criteria, so that, for example, the death of Urse d’Abetot, sheriff of Worcester, at some unknown point in 1108 could be used to date a Henry I charter 3 August 1100 x 1108. Experience in assigning dates to the texts and in using the database, however, demonstrated that this was not a fully satisfactory means of including acta that were datable only by imprecisely-known events. For example, the election of William Warelwast to the bishopric of Exeter is datable only to some point between the death of his predecessor Osbern fitz Osbern in 1103 and his dated consecration on 11 August 1107. Bishops tended to be given their new titles after election rather than consecration, and so any occurrence of this important figure with his episcopal title must therefore be datable after [1103 x 11 August 1107]. The addition of an extra box to cope with this and other cases of imprecisely-dated events has been necessary for the accuracy of the entries in the database, but has resulted in some formidable date-complexes needing further explanation. Thus a field for reasons for datings given is placed below those for recording the dates themselves.

Most of the criteria used to describe the acta are more simple, and are managed in the way used for the type of document described above, by selecting an option from a limited list. It has been necessary to include the option of simply leaving some field or fields blank.

This is obviously necessary for those criteria that apply only to documents surviving as originals, which comprise perhaps ten per cent of Anglo-Norman royal acta. In these cases attributes relating to the external appearance of a document can be seen that are lost in copies. Thus the hand of the royal scribe responsible for writing the act can often be identified, and this information is included where possible, using the identifications of hands made by T. A. M. Bishop [7]. Given that it often proved difficult to match named scribes known from other sources with the hands seen in the acta, Bishop used numbers for his identifications, and his numeration has been taken over without modification. However, a substantial minority even of the authentic originals are written in hands other than those of the professional royal scribes, most commonly where a beneficiary used a private scribe to prepare a text which was then presented to the chancery for sealing alone, and thus this box need not be filled.

Description of the sealing of originals has been treated in a similar manner, because again acta surviving as copies do not have seals, and because even many authentic originals have lost seals that were formerly attached. Separate fields cover the manner of sealing (on tongue, on tag, on cords, etc.), the identification of the seal itself (Henry I’s forged ‘first seal’, Henry I’s second seal, etc.), and the colour of the wax of the seal. Similarly relevant to the originals alone has been a simple text field in which any endorsements can be noted.

The field that gives a guide to authenticity has been a difficult matter to settle. It is rare for judgements about authenticity to be expressed in categorical terms ; more commonly, elaborate phrases expressing qualifications and doubt are used in commenting on the acta. For the functioning of the database it has been necessary to reduce the many possible assessments of authenticity to a simple numerical scale, which can then be combined with other terms to search, for example, only indubitably authentic acta. A five-point scale has been used, in which the highest authenticity rating (5) has been given only to surviving originals whose texts and external features suggest no forgery. The next rating (4) is given to texts that survive as copies but which bear no detected marks of inauthenticity. This rating also allows for the capacity for minor error in even the most conscientious of copying [8]. The next rating (3) denotes texts that seem mostly in order but which have small oddities or traces of minor rewriting ; whereas the next rating (2) indicates heavily interpolated texts which nevertheless seem to have substantial basis in a presumed authentic act. The division between these two is necessarily based on a somewhat impressionistic judgement. The final rating (1) is used for texts forged without reference to an authentic source, whether surviving as originals or as copies.

The use of five ratings maintains both ready intelligibility and the capacity for allowing fairly sophisticated analysis. The number of categories could have been far greater, for the number of possible permutations of authenticity and inauthenticity is considerable. The gradations of interpolation could have been expressed more finely, and ratings could have been generated for unusual survivals such as pseudo-originals preserving the texts of authentic acta for which the genuine originals have perished, or for erased originals which have had new, inauthentic, texts imposed upon authentically-sealed parchment. Such cases do not fit well into the five categories, but the fewness of such eccentricities has limited the consequences of not providing for them.

The linking of occurrences of a person in a text to a personal identification file was described above in relation to the beneficiary field. Persons mentioned in the text are identified in a similar manner. There are, however, some complications. Persons can be mentioned in acta in one or more of four roles : as the addressee of an act, as a simple mention in the body of an act, as witness to an act, or as per, an obscure category that occasionally follows on from the witnesses. When the link is made between person- and text-files, at each occurrence the person’s role is selected from a menu, and so it is possible, for example, to search only for acta addressed to Roger, bishop of Salisbury, and exclude any witnessed by him.

Place-names in the acta present, if anything, greater variation in form and especially spelling, and so a similar structure of separate files is used ; they also present the same problems of quantity, and well over two thousand have been entered so far. Place-names generally occur in the body of the text without any roles comparable to those of persons ; but one special category, for place-dates, must be recognized. The place at which an act was given is frequently important for dating purposes, as well as for constructing the itineraries of participants, and so this link is made to a different field to preserve the distinctness and usefulness of this information.

In addition to this there are fields for noting the archival survival of the texts, and also there are fields in which the publication history of the individual acta can be noted, which will obviously be of relevance to the creation of the fully worked-out editions that are intended to follow.

The database of Anglo-Norman royal acta is still in progress, but was substantially complete in September 2002. It will still depend mostly on printed texts of varying quality. A later stage in the project should involve the substitution of texts based on more accurate transcripts of the originals and copies. In its initial state, however, it will be a usable and, hopefully, useful version. If the project as a whole prospers, it is hoped that the database can be substantially improved through the addition of further texts discovered in the course of work in archive repositories [9], through refinement of the dating limits of individual acta as the introduction of diplomatic features is more accurately dated and the careers of important attestors and great officials is understood more fully, and perhaps through the inclusion of the acta of the early Plantagenet kings, Henry II and Richard I, in collaboration with the Angevin acta project of Professor Sir James Holt and Professor Nicholas Vincent, based in Cambridge. The completion of these further stages is dependent upon our finding substantial further funding. The development of the database has made rapid progress from printed texts.

[1]Regesta Regum Anglo-Normannorum : The Acta of William I (1066–1087), ed. D. R. Bates Oxford, 1998 ; Regesta Regum Anglo-Normannorum, 1066–1154, III, Regesta Regis Stephani ac Mathildis Imperatricis ac Gaufridi et Henrici Ducum Normannorum, 1135–1154, ed. H. A. Cronne and R. H. C. Davis, in continuation of the work of the late H. W. C. Davis, Oxford, 1968.

[2]Regesta Regum Anglo-Normannorum, 1066–1154, I, Regesta Willelmi Conquestoris et Willelmi Rufi, 1066–1100, ed. H. W. C. Davis and R. J. Whitwell, Oxford, 1913 ; Regesta Regum Anglo-Normannorum, 1066–1154, II, Regesta Henrici Primi, 1100–1135, ed. C. Johnson and H. A. Cronne, from the collections of H. W. C. Davis, Oxford, 1956.

[3]. See, for example, Regesta II, nos. 1436 = 1991, 841 = 1582, 1522 = 1615, and many others. The criteria for distinguishing and attributing texts is often inadequate ; one text of Queen Matilda was attributed to Matilda of Flanders, wife of William I, in Regesta I, no. 189 ; and to Edith-Matilda, wife of Henry I, in Regesta II (no. 1190). However, it must be an act of Matilda of Boulogne, wife of King Stephen, in that it is addressed to Ansfrid dapifer as sheriff of Kent (for another occurrence of him see Regesta III, no. 142). The act thus substantiates the unproven claim that Matilda acted as regent during at least part of her husband’s absence in Normandy in 1137.

[4]. The most recent major studies are : Facsimilies of English Royal Writs to A. D. 1100, Presented to Vivian Hunter Galbraith, ed. T. A. M. Bishop and P. Chaplais, Oxford, 1957 ; P. Chaplais, ‘The Seals and Original Charters of Henry I’, English Historical Review 75, 1960, 260–75 ; reprinted in his Essays in Medieval Diplomacy and Administration (London, 1981), no. XVII ; T. A. M. Bishop, Scriptores Regis : Facsimiles to Identify and Illustrate the Hands of Royal Scribes in Original Charters of Henry I, Stephen, and Henry II, Oxford, 1961 ; yet few have ventured to build on these works.

[5]. It is not intended to provide fully lemmatized searching. In a language as highly inflected as Latin, lemmatization is a considerable challenge. This lack is not likely to be a major problem, however, for so much of the acta is common form, in which phrases appear in consistent wording, and because intelligent use of wild-card characters in place of inflexions should enable most occurrences to be isolated.

[6]. However, attribution of texts to issuers has occasionally been a problem. The acta of Williams I and II are sometimes difficult to distinguish (see Regesta Regum Anglo-Normannorum : The Acta of William I (1066–1087), ed. D. R. Bates, Oxford, 1998, nos. 6, 7, 44, etc.), and there are some examples that seem to merge characteristics of the acta of Henries I and II (see Regesta II, nos. 1209, 1222, etc.). The occurrence of three Queens Matilda has also caused some confusion in the past; see note 3 above.

[7]. T. A. M. Bishop, Scriptores Regis, Oxford, 1961.

[8]. For example, original charters written by royal scribes have rex Angl’ for the royal style, representing rex Anglorum (as seen on seals and coins); from 1199 onwards, however, the royal style used rex Angliae, and many cartulary copies have this anachronistic expansion.

[9]. A substantial collection of addenda to the editions and calendars has been made already, but the work of collecting acta from still unexplored sources goes on. If any Anglo-Norman royal acta not in any of the editions (for William I and Stephen) nor the calendars (for William II and Henry I) are known to readers of this report, it would be much appreciated if these could be notified to Professor R. Sharpe, Modern History Faculty, Broad Street, Oxford OX1 3BD, or to richard.sharpe@history.ox.ac.uk.

© CNRS - IRHT 2004 - Contacts - Infos légales - Rédaction