As of 22 April 2009 this website is 'frozen' in time — see the current IFLA websites

This old website and all of its content will stay on as archive – http://archive.ifla.org

IFLANET home - International Federation of Library Associations and InstitutionsAnnual ConferenceSearchContacts

62nd IFLA General Conference - Conference Proceedings - August 25-31, 1996

International standards in authority data control: costs and benefits

Alan Danskin
Head of Authority Control
The British Library, Boston Spa


National and international co-operation and exchange of bibliographic data are well developed. Outside the United States, the exchange of authority data is in its infancy. There are benefits to be derived from such exchange, but to achieve these through strict adherence to principles of Universal Biblographic Control is not economic. The long term solutions may be a new model of access control record, but potential intermediate solutions are indicated by the development of an Anglo-American Authority File and Project AUTHOR.



The theme of this paper is "economic benefits through resource sharing" which by coincidence, is the theme chosen by the Section on Cataloguing. Great strides have already been made in this direction through the exchange of bibliographic records and it is this which has sparked interest in the exchange of authority data. Despite some success for national co-operatives, of which NACO is t he pre-eminent example, little authority data is exchanged internationally. In contrast to the success IFLA has enjoyed in encouraging the creation of national bibliographies and the exchange of bibliographic data, the effort it has devoted to encouraging the extension of these principles to authorities has borne little fruit. I want to review the benefits which may be derived from the exchang e of authority data and then consider to what extent these may be realised.

Economic benefits of co-operation

Sharing authority data can benefit individual libraries in three ways:

Despite interest in the transnational exchange of authority data, little progress has been made. It is much easier to identify the economic benefits of resource sharing than to enjoy them.

Universal Bibliographic Control and Authority data

IFLA has helped to develop standards for the formulation and identification of authority data through GARE (1), the UNIMARC(A) (2) format and the publication of Bourdon’s International Co-operation in the field of Authority data(3) . The tendency has been to apply the principles of Universal Bibliographic Control for bibliographic data to authority data. I believe that these principles are ill suited to this context and their implementation will prove prohibitively expensive.

These principles are:

A great deal has already been written about the difficulties that arise from prescribing a particular form of heading for use by all libraries and I do not intend to repeat it here, it is sufficient to note that Tillett (3) and others have advocated a new kind of access control record which acts as a node for the collocation of all variant forms of a particular name. In this model the ISADN acts as the unique identifier, therefore it is not necessary for the headings itself to fulfil this role.

What is to be the fate of all the authority records which already exist? National Bibliographic Agencies have invested millions of pounds over the years to develop national authority files (see attached table). Is each national authority file to be thrown into the melting pot and assigned a unique ISADN on the basis of nationality?

It is fundamental to UBC that the nationality of authors (and therefore the NBA with responsibility for establishing the name) can be readily determined. However, a brief survey of national authority files shows that most NBAs do not make this distinction, (see table). For example, neither the British Library nor the Library of Congress code authority data for language or nationality; the Bibliot eca Nacional de Madrid and the Istituto da Biblioteca Nacional et do Livro (Portugal) code language but not nationality; the University of Helsinki and the Bibliotheque Nationale de France are exceptional in that they code both language and nationality. Even if the coding issue could be resolved retrospectively, the thorny issue of nationality would still remain, as Samuel Johnson noted, The ch ief glory of every people arises from its authors”. Individual NBAs would in any case be unable to make their authority files available and it is unreasonable to expect other agencies to defer cataloguing until a name has been established.

The establishment of a single preferred form, even if feasible would not be sufficient in itself: provision also must be made for the maintenance of authorities over time. Volatility is one of characteristics distinguishing authority data from bibliographic records. In general, once created a bibliographic record changes very little; further, any changes which do occur will normally be carried o ut locally and limited to a single record in a single collection. This self-containment means bibliographic records are easily exchangable. Much as we may wish that authority data exhibited such stability, we have to accept that they are subject to change whenever references or source data have to be added or when authors change their names or when a conflict has to be resolved by addition of q ualifying data. Changes to authority records are dynamic: they are not limited to a single bibliographic record; nor even to a single catalogue; they may require changes to other authority records or the creation of new authority records. If the transnational exchange of bibliographic data is to be facilitated by encouraging consistency in authority data it is essential that users have access t o the latest data.

What format is to be used to exchange authority data? IFLA has developed UNIMARC (A), but most of the world’s authority data is stored in USMARC (A) or related formats (See table). UNIMARC(A) was developed primarily as a switching format but, as is well known, data is corrupted or lost during conversion and therefore the integrity of data transferred between systems using different formats cann ot be guaranteed. This has not been a significant obstacle to the exchange of bibliographic data, which is normally in one direction only and for which a certain level of degradation can be tolerated. Degradation of authority data is not tolerable because spurious access points may be generated and because the mutability of authority data means that it may be subject to erosion over successive t ransfers.

These obstacles to implementing the UBC objectives are profound and the resources required to overcome them are immense. The benefit to be derived from such an unfocused investment is negligible, indeed the result would probably be to degrade the quality of access in any given catalogue.

There are positive gains to be made from transnational exchange of authority data, but they can best be attained by focusing resources where the greatest benefit will be derived. There are already projects which provide models and perhaps the foundations for a better cooperative structure. Two complementary approaches can be envisaged: firstly to develop closer co-operation between libraries work ing within the same cataloguing culture, by which I mean language of cataloguing, cataloguing standards and publishing industry. These clusters can strive towards the very significant savings attainable by developing a shared authority file and thereby enhancing the efficiency with which bibliographic data are exchanged. The second route is to make authority data as accessible as possible throug h remote access. This will reduce the duplication of research to identify or link headings and should enable NBAs which belong to different cataloguing cultures to match the form of their headings as closely as possible within the constraints imposed by national standards.

Case Study 1: Anglo-American Authority File

The Anglo-American Authority File is an ambitious undertaking by the Library of Congress and the British Library to develop a single shared authority file in place of the existing British Library Name Authority List (BLNAL) and the United States National Authority File (USNAF). The project is being implemented in three phases.

The starting point for the AAAF was a consideration of the differences in cataloguing practice and record creation between the British Library and the Library of Congress. The differences uncovered were quite extensive, but in detailed negotiations during the period 1993-1995 they have either been resolved by or, in a few cases, excluded from AAAF; - I regret to report that we could not agree on how to romanise Chinese characters. The British Library has used Pinyin since 1958 (4), but the Library of Congress still retains Wade-Giles. Romanised headings have been excluded pending a decision by Library of Congress to adopt Wade-Giles. The other main category of exclusions are uniform titles for which no appropriate conversion between MARC formats could be derived. The most difficult ta sk facing the AAAF project so far has been to develop a methodology by which the idiosyncrasies of the respective MARC formats can be accommodated, without corrupting the data to be exchanged. The announcement in January this year that USMARC, CANMARC and UKMARC are to be aligned in a single format by 1999 is an indication of the limitations of format conversion.

The first Phase of the implementation, completed earlier this year, was to load a copy of the USNAF on the British Library’s cataloguing system. To guarantee the integrity of the data USNAF has been loaded in USMARC. The currency of the British Library copy is maintained by overnight file transfer of new and amended records from the Library of Congress’ master copy. USNAF is searched in conjuncti on with the BLNAL and cataloguers at the British Library are already using the USNAF as a resource file. When a heading is required for which no British Library headings exists, records can be copied from USNAF into BLNAL instead of creating a new authority record. This ensures that the British Library does not duplicate work already carried out by the Library of Congress. The Library of Congre ss benefits from the transfer of new authority data created by the British Library, which has been in progress since 1994. However this exchange of data is of limited long term value, as neither library can track changes to the data it has contributed. In the course of time, it is inevitable that unless updates can be automatically applied, the forms of headings will drift apart.

In Phase 2, the implementation of sophisticated MARC conversion software will enable the British Library to flip authority records between USMARC and UKMARC. Prior to the implementation of MARC Harmonisation only those records which can be converted without changing the data content of the record will be authorised for inclusion in AAAF. It will be possible to create an authority record at the Br itish Library in UKMARC, convert it to USMARC and export it directly to the USNAF Master File at Library of Congress. A cataloguer working at the Library of Congress, or in one of over 200 other NACO libraries, will be able to use it, add to it or if necessary amend it. If the record is changed in any way it will be redistributed to the British Library as well as all the other NACO libraries within 24 hours. Work on Phase 2 is in progress and it is scheduled for completion later this year.

The third Phase of AAAF will be the retrospective convergence of the British Library’s authority file (BLNAL) with USNAF. This will eliminate duplicate forms of headings from Anglo-American cataloguing, which will be a major undertaking. The BLNAL comprises some 700,000 authorised name authorities. It is a reasonable assumption that approximately half of these will be held on USNAF and of these at least half will be held in a variant form. Therefore at the very least 175,000 name authority headings will have to be amended. The general principle of preferring the USNAF form has been agreed. Phase 3 will not be implemented until 1999, by which time, the British Library will have an authority file linked to its catalogues

I have reviewed the AAAF programme in some detail because it illustrates the commitment of time and resources necessary to integrate authority data created by separate NBAs. Such investement has to be justified. The British Library creates 60,000 name authority records per annum, research suggests that as many as 44% (5) may duplicate records created by Library of Congress. Note, that duplicate d records are not the same as duplicated headings. The British Library derives approximately 50% of its English language cataloguing from Library of Congress records but, since it does not use the same authority file as Library of Congress, each bibliographic record has to be checked against BLNAL and, if necessary, amended to match BLNAL. Therefore the full potential of derived cataloguing is not realised. Estimates based on research conducted by Ed Jones (6) indicate that this may affect up to 12,000 bibliographic records per annum. The efficiency gains to be achieved through the use of AAAF justify the investment and are directly attributable to the shared cataloguing culture and the very extensive overlap between publishing in the UK and USA.

Case study 2: AUTHOR

AUTHOR is a good example of how efficiencies can be achieved by making authority data public, as advocated by IFLA. Each of the AUTHOR partners will have access through the internet to the authority files of the other partners. The British Library cannot directly incorporate authority records from these sources into its own authority file; there are technical difficulties which would be very e xpensive to resolve, but more importantly neither the form of name nor the language of cataloguing would necessarily be acceptable in the British Library catalogues (7) . However the British Library would find information which helped to establish the identity of an author or clarified the relationships between corporate bodies very useful.


More information is needed about the productivity gains libraries would make by sharing authority data. This is particularly true for exchanging data across linguistic boundaries because authority co-operation comes with different price tags. A high investment will not necessarily result in a high return. There are pre-conditions which have to be satisfied before really large benefits can be a chieved. IFLA has done some valuable work through recommending standards for authority data entries. However it is not the case that a single prescribed form will meet all needs. Standardisation of cataloguing codes and formats remains a prerequisite to further integration, however it is likely that standardisation of language -however much monoglot Anglo-Saxons may secretly desire it -is beyo nd the power of even IFLA. To facilitate the work of colleagues in other libraries NBAs should follow the example set by Library of Congress in making their authority data available over the internet (8) . In the long term there may be a role for the ISADN in providing each bibliographic identity with a unique identifier independent of language and cataloguing code. This is most likely to be achieved through cooperation between NBAs which share a common language. By developing a single preferred form, these agencies will create clusters of headings from which an access control record could be developed if economically viable.


Not available, please contact Author.


1. Guidelines for authority and reference entries, recommended by the Working Group on an International Authority System (London: IFLA International Programme for UBC, 1984)

2. UNIMARC/Authorities: Universal Format for Authorities, recommended by the IFLA Steering Group on a UNIMARC Format for Authorities (München: K.G. Saur; 1991) (UBCIM Publications - new series, 2)

3. Bourdon, Françoise International cooperation in the field of authority data (München: K.G. Saur; 1993) (UBCIM Publications - new series, 11)

4. Tillett, Barbara 21st century authority control: what it is and how do we get there? In: The future is now: reconciling change and continuity in authority control. Proceedings of the OCLC Symposium, ALA Annual Conference, June 23rd, 1995. - Dublin, Ohio: OCLC, 1995, p. 17-21.

5. Lu, Suping A study in the Chinese romanization standard in Libraries” Cataloguing & Indexing Quarterly, Vol. 21(1) 1995

6. Agar, Sharon Survey of hit rate using an Anglo-American Authority File. British Library. Unpublished internal report, 1994.

7. JONES, E. A Consistency in choice and form of main entry, 1982 and 1989: a comparison of Library of Congress monograph cataloging with that of the British Library and of the national libraries of Australia and Canada. (Ph.D. dissertation, University of Illinois at Urbana-Champaign, 1994).

8. Jacobowitz, Neil A. “A comparison of AACR2R and French cataloguing rules”, Cataloguing and Classification Quarterly, Vol. 20(1) 1995

9. There is a Z39.50 interface for USNAF on LC MARVEL: URL: http://lcweb.loc.gov/z3950/mums.html