Microsoft Word - studcat.doc 1 Pre-print of article published in Journal of Librarianship and Information Science, 38(3) (Sept. 2006), 173-85 Why Appoint Professionals? A Student Cataloguing Project K.E. Attar Abstract Students have provided cheap successful labour for routine retrospective cataloguing projects. The current article examines a library project which went further, using university students with minimum training to catalogue its undergraduate stock from the book in hand to AACR2, level 2, allegedly to professional standard. The article discusses the faults made in MARC coding, descriptive cataloguing and subject cataloguing, noting the nature of the errors and their results. The investigation concludes that intelligence alone does not guarantee library ability, and that cataloguing beyond the creation of minimum records is not an intuitive task to be picked up without training. Hidden expenses are attached. A derisory attitude towards library skills is unjustified, and a place remains for qualified librarians to do qualified library work. Keywords: cataloguing; project work; student labour, retrospective conversion. Is cataloguing a professional activity? For the cataloguing of manuscripts and early printed books from the book in hand it is generally accepted to be so, as demonstrated by the qualifications required for such cataloguers and the concomitant level of their employment. For special materials, such as maps and music, cataloguing may be regarded as a professional activity. But for modern monographs? For their own prestige and salaries it is in cataloguers' interests to insist that it is so, and the perceived devaluation of cataloguing since automation, within both many library school curricula and libraries, is a current concern on both sides of the Atlantic. On the other hand, financial pressures understandably render deprofessionalisation managerially attractive. Students and other non-professionals have successfully assisted in retrospective catalogue conversion projects involving the transfer of bibliographical data from cards to electronic format. The current study examines a project which went further. Assuming cataloguing to be an unskilled task, it used students working from the book in hand to catalogue an undergraduate College library. The students were expected to apply notes, added entries and Library of Congress Subject Headings, with the intention of cataloguing books to a professional standard. The article's purpose is to investigate whether in the light of this exercise professional work can be expected from non-professionals. The lament The devaluation of cataloguing is a widespread concern in its own right, and also part of a broader issue of the deprofessionalisation of librarianship. In a special issue of the American-based Cataloging & Classification Quarterly Janet Swan Hill points to the perception that computerisation has reduced the importance of cataloguing and the fallacy of this view, for example because computers cannot practise authority control (Hill, 2002: 11). Michael Gorman, in a piece headed: 'Why teach cataloguing and classification?' observes: 'Before the great gas bubble of digitization came along, 2 the answer would have been so obvious that only a ninny would have even posed the question', before spelling out the importance of the catalogue as a retrieval tool (Gorman, 2002: 2), and Heidi Lee Hoerman states: We start to think cataloging is something than can be done by anyone. Maybe it can. To be honest, at this stage, it is in many cases being done by anyone, and that anyone has very little training. We then take the cataloging done by this untrained person and 'share' it, unexamined, into our catalogs. (Hoerman, 2002: 36-7). In Great Britain, John Bowman echoes such opinions in the introduction to his 2003 textbook Essential Cataloguing: Cataloguing has long been unpopular and nowadays is little taught. Ever since computers began to be used in cataloguing ... there has been a school of thought that in some mysterious way computers would be able to do all the cataloguing that was necessary, and that it would no longer be necessary to employ human cataloguers. ... Cataloguing is important. It is the principal means whereby library users can find the contents of the collection. Now that most cataloguing is computerized, it is even more important to avoid errors, because a simple mistake can make a catalogue record - and therefore an item - irretrievable (Bowman, 2003: 2-3). Jane Read cites Bowman and states succinctly: 'Many library administrators ... think that automation of cataloguing has made it easy enough for a trained monkey to do' (Read, 2003: 12-13). In 2004, letters and articles in the CILIP Update and Gazette have called for a reinstatement of the value of cataloguing (e.g. Trickey 2004; Ward 2004), while in Catalogue & Index Rodney M. Brunt has argued the centrality of cataloguing for librarianship and library users, and urged the continuation of training: 'Library school is the place to make bad cataloguing decisions and assign unwise subject indexing which might emerge from an incomplete understanding of principles' (Brunt, 2004: 3). The background That cataloguing is expensive has been stated frequently. Most recently, Read discusses the expense and the consequent temptation to downgrade cataloguers and cataloguing to economise (Read, 2003: 14-16). Earlier writers to have made the same point include Hoare (1986: 97), Law (1988: 81) and Quedens (1991: 15), with the reminder that a library's aim should be to gain value for money; to be cost-effective, which is not synonymous with cheap (Peters, 1984: 162). In a university context, to consider student labour is a reflex action. Students performed retrospective conversion at the University of Freiburg in Germany, the University of Basel in Switzerland, and, in America, Oregon State University, Indiana University and the University of California at Riverside. These projects had several features in common: (1) they involved retrospective conversion from cards rather than recataloguing; (2) they were simple and well controlled by professional librarians; (3) they did not expect a professional level of work from the students. The most comprehensively described project was at Freiburg, which was extremely satisfied with the results. The students employed had passed their intermediate examination, taken after two or three years, but had at least 18 months to go before their final examinations. Catalogue records were defined as 'I-Niveau' ('interim level'), which meant that any library contributing to the union catalogue which 3 catalogued with the book in hand could upgrade the record; imposing subject headings was not part of the exercise; qualified librarians supervised the work and catalogued the 2% of titles deemed difficult (Maurer, 1991). At Basel, unlike Freiburg, students keyed in records from the cards instead of importing them where possible from external databases. But here, too, students worked under the supervision of two qualified librarians who undertook the difficult cases (Wessendorf, 1991). The assessment of the results was: 'quite acceptable, although it is clearly below the level of our current cataloguing' (Wessendorf, 1991: 55). In America, students at Oregon State matched cards against records on the OCLC database (Watkins, 1985). Nancy Douglas, who stresses the financial benefit of using students at the University of California at Riverside (Douglas, 1985: 13) emphasises the elementary nature of what was expected: 'The project requires very little expertise on the part of a library's staff' (Douglas, 1985: 11). This was another retrospective conversion project, cataloguing books from cards. Records were to contain apart from the Library of Congress classmark and card number only basic descriptive elements, namely author, title, imprint, physical description and tracings. Notes and ISBN were routinely excluded, and the title field sometimes excluded sub- titles and usually the statement of responsibility. Records were then sent away to be matched. The Indiana project (Mayer, 2002) differed in having students catalogue sound recordings from the items in hand. Mayer records that the students 'strove for an aesthetically pleasing yet meaningful display of the contents of an individual sound recording', without stating the degree of success, and notes the limitations: There is no attempt at authority control, nor is there an attempt to standardize structural metadata across different representations of the same work, although we recognize the desirability of such functionality and see it as an important area for future work (Mayer, 2002: 154). Projects which did not rely on students but which did use paraprofessionals have been described for the Universities of Hull and Botswana. The University of Hull employed typists to perform its retrospective conversion from cards (Dyson, 1984; descriptive cataloguing only), while the University of Botswana used O-level school-leavers (Kgosiemang, 1999) to catalogue books in hand. Both institutions reported unsatisfactory results, with the cost of errors at Hull becoming apparent only after completion of the project. In Hull, 73.65% of records had errors; the average number of faults per record was 2.16, and the highest number of mistakes found in one record was 16. Errors included typing or spelling mistakes and incorrect spacing. Records for books in foreign languages contained a particularly high level of inaccuracies. Errors were ascribed to four causes: carelessness; poor or illegible handwriting on the cards; failure to interpret instructions correctly; unfamiliarity with library and/or cataloguing routines (Dyson, 1984: 257). The Botswana experience led to the conclusion: 'To transcribe bibliographical information accurately in accordance with the necessary standards and codes requires general knowledge of cataloguing practices ...' (Kgosiemang, 1999: 93). In summary, the survey of literature indicates that university students are capable of undertaking undemanding retrospective conversion under controlled conditions and that they can do so more competently than other non-professionals. The Cambridge background and project 4 The current article describes the result of a project at King's College Cambridge which employed students to catalogue an undergraduate library from the books in hand, allegedly to a professional standard. Cambridge University Library had been computerised since 1978 and operated a union catalogue for the College and Departmental libraries of the University. However, while Colleges could use the cataloguing module of the University's library management system, the acquisitions and circulation modules were closed to them. Several Colleges therefore purchased their own library management systems so that they could operate automated loans. This meant cataloguing in the union catalogue in the first instance, then downloading records into their own library catalogues. Both catalogues had then to be maintained with all editing after the initial download being done twice, once on each catalogue. The project at King's College was motivated by its adoption of SIRSI Unicorn. It began in September 1994 and ended as a major project in December 1995. Between these dates a total of 37 students, a mixture of undergraduates and graduates, worked in the University holidays on the catalogue. As far as possible students worked on sections of the library aligned with their own subjects of study, to enable them to understand the works they were cataloguing and impose relevant subject headings. As the work remained unfinished in December 1995, two new graduates who had been among the undergraduate participants continued full-time until the end of 1996. The library contained approximately 65,000 books and periodicals covering almost all subjects of the undergraduate curriculum in accordance with its function to support taught courses in the University; only veterinary medicine, Oriental Studies and Anglo-Saxon, Norse and Celtic, as small subjects which were not supported at the College, were not represented. Music was a particular strength. The library further included works of local interest, books in the Modern Archives, a bibliography section, a collection of fiction written by members of the College, and a special collection of works by and about Jane Austen. All were included in the project. Cataloguing was done in UKMARC, following AACR2 and Library of Congress Subject Headings. It was done from the book in hand, with the cataloguers physically adding barcodes as they went along for circulation purposes. Students downloaded records from the section of College and Departmental libraries on the University's union catalogue where possible. For books which did not have records on the union catalogue, they downloaded records from the cataloguing database of the Consortium of University Research Libraries (CURL), or where necessary created new records (extra-MARC material, or EMMA). Statistics of the proportions were not kept. Calculations of cost were done on the basis of each student cataloguing seven books per hour. The College made a deliberate decision to have full catalogue records, for example transcribing complete title and statement of responsibility and including subject headings and notes. This meant that many records taken from the union catalogue, the first source for matching records, would require considerable upgrading, since a number of Colleges and departments had contributed short records in order to record quickly as many books as possible. The short records comprised surname and initial of author or editor; title; date and place of publication, library barcode and classmark (i.e. less than the elements required for AACR2's first level of description (AACR2 1.0D1). The librarian managing the project was unaware that the University Library's full catalogue records, available for books acquired after 1978, were available via the University's union catalogue. The students received two days of training (compared with approximately 20 hours of training per student at Freiburg and Riverside). Their work during the bulk of the project was checked by a graduate student, who was employed as Senior 5 Library Assistant and spent each morning reading the previous day's records. Having no more experience or qualifications than the rest, he could do little more than to pick up spelling mistakes. The work done in 1996 was not monitored. The notion of using student labour, as readily available and cheap, at Cambridge was not new. In at least one other College computerising its library stock, the Bursar suggested students for the task; the professional librarian successfully resisted the proposal. The desire for feasibility is clear from costings: at the equivalent pro rata of an annual salary of £7,645 and no overheads paid to the students at the commencement of the project, it was calculated that the cost per book at seven books per hour would be £0.55 (the actual cost was later revised to £0.60 per book). Junior professionals would typically be paid at the time an annual salary of £11,894 - £13,780, plus overheads. The novelty both in the Cambridge context and in comparison with the projects noted above was to expect a professional standard of work from students, including the imposition of Library of Congress Subject Headings, after two days of training, and not to build trained supervision into the project. It was argued that students were the most suitable people to catalogue the library because they were the main user group of the library. This was supposed to give them a vested interest in working well. A further assumption behind appointing students was that students were intelligent. For two years after the project ended the only quality control performed on the catalogue was unofficial, as librarians from other Colleges cataloguing their own overlapping stock viewed and compared the standards of records emanating from the respective Colleges. In 1998 the employment of an enthusiastic cataloguer at King's College led to the closer examination of the student cataloguing. It began piecemeal, as bibliographic records were examined during standard procedures, such as adding item records when purchasing second copies of popular or lost items. The systematic improvement of records escalated in autumn 1998, when the donor of the Jane Austen collection complained to the Fellow Librarian about the low standard of catalogue records for the books he had donated. Instant recataloguing was necessary to secure the receipt of remaining books in the donation. Later the music library and the German section of the main library were targeted for upgrading. Approximately 95% of the catalogue records contained errors, ranging from the insignificant (e.g. not including an illustration statement in books containing music; not including an optional note field to point out the presence of bibliography or index) to spelling mistakes which rendered books irretrievable. In the worst cases, such as the Jane Austen records, there was a mistake in every line, and up to 16 errors per record. While the project is now old, in terms of the swift progress of librarianship, the points raised remain relevant and have indeed perhaps become still more pressing in terms of the continuing debate about the importance of cataloguing. For reader-friendliness, examples in the following analysis are in display format unless MARC format makes a point more cogently. Several of the errors discussed have been corrected before the time of writing. The noting of which students used which barcodes renders catalogue records easy to trace, and examples have been taken from the work of a variety of students. General analysis showed that while one student (the initial Jane Austen cataloguer) was responsible for almost all the worst records on the system, the difference between this student and the others lay in the high quantity of errors per record across her records. The nature of errors was uniform across the students. The results: MARC coding 6 Three common errors emerged in the application of MARC: 1. In the 008 field, students often coded multi-volume works published over more than one year as 'm' (multiple date) and provided the inclusive dates of publication (US MARC practice), instead of using the code 's' with the date of the first volume published (UK MARC). The error arose from the discrepancy in practice between UK and US MARC, with records being imported which would have been created in the latter. The mistake at no time hindered retrieval. With the transition to MARC 21, the error has become correct. 2. Students exhibited lack of understanding when cataloguing multi-volume works. UK MARC allowed for several sub-level titles as in the following example from the UK MARC manual (British Library Bibliographic Services Division, 1980): 245 10 $aBritish Standard methods of analysis of fats and fatty oils 248 10 $gPart 1$hPhysical methods 248 20 $gSection 1.12$hDetermination of the dilation of fats Students sometimes instead used the various levels to list the various volumes of a multi-volume work as follows, rather than creating one catalogue record per volume or, in a single catalogue record, listing the titles of the individual volumes in a contents field: 245 10 $aTchaikovsky$ba biographical and critical study 248 10 $gVol. 1$hThe early years (1840-1874) 248 20 $gVol. 2$hThe crisis years (1874-1878) 248 30 $gVol. 3$hThe years of wandering (1878-1885) 248 40 $gVol. 4$hThe final years (1885-1893) 260 00 $aLondon$bVictor Gollancz$c1978-1991 300 00 $f4$nv$c24cm The results were confusing cataloguing records which had to be untangled when discovered. 3. There was widespread failure to comprehend the importance of non-filing characters, both in the title field and for series. Even after considerable correction, typing in 'The' in the browse section of the library still calls up 1,024 titles; 'A' calls up 181 titles, after the exclusion of correct titles beginning with French 'À'; 'An' calls up 39 titles, and 'Les', 35. Thus titles are misfiled, and some manifestations of a work are separated from others. For example, Henry James's The awkward age cannot be found by browsing the title, because it is filed under 'The', not 'Awkward'; of The collected works of John Maynard Keynes, 21 volumes can be found by browsing the title, while 12 are under 'The'. Occasionally a preposition was mistaken for an article and the MARC tag for filing characters adjusted accordingly, also resulting in the inability to retrieve records by browsing, e.g. 240 13 De aeternitate mundi contra Proclum which appears in lists as the grammatically non-sensical ‘Aeternitate mundi contra Proclum’. On the whole, however, the students coped admirably with UK MARC. Whereas they might have committed punctuation errors in MARC21, UK MARC rendered the insertion of punctuation unnecessary. Occasionally students reversed the order of the place of publication and the name of the publisher in the imprint field (e.g. 260 00 $aOxford University Press$bOxford for $aOxford$bOxford University Press), but the ability to place information in the correct fields and subfields was 7 generally excellent and compared favourably with the 95 % accuracy rates offered by retrospective conversion companies (Bridge, 2003: 2). This indicates an ease in dealing with the structure of databases. The focus of the training may have been on MARC coding, and the students would certainly not be alone in regarding the MARC manual as an easier reference tool than AACR2. The results: descriptive cataloguing While the mistakes in descriptive cataloguing resulted in unprofessional catalogue records, few affected the ability to retrieve items. They are as follows: When transcribing title pages, students frequently omitted statements of responsibility. Another common error was to omit the major statement of responsibility, while retaining subsequent ones, e.g. The monadology and other philosophical writings / Translated with introduction and notes by Robert Latta for: The monadology and other philosophical writings / Leibniz ; translated with introduction and notes by Robert Latta Often a statement of responsibility was preceded incorrectly by the word [by] in square brackets. This was a derived error, arising from the fact that MARC predates AACR2, which prescribes the transcription of a statement of responsibility exactly as it appears following a slash (1.1F1). In AACR1, a comma introduced a simple statement of responsibility and the insertion of [by] was essential for the sense, e.g. Written for children : an outline of English-language children's literature, [by] John Rowe Townsend (see AACR1, 136). Titles and statements of responsibility often contained errors of capitalisation, chiefly employing a capital letter for the verb or preposition beginning the statement of responsibility, and using a capital letter for the second word of an English title. Some spacing errors occurred. In transcribing foreign languages, students frequently ignored diacritics. Ignorance of German frequently led to the substitution of lower case letters for upper case in nouns, e.g. Leopold Mozart, 1719-1787 : portrat einer personlichkeit for: Leopold Mozart, 1719-1787 : Porträt einer Persönlichkeit. In edition statements, 'edition' was sometimes abbreviated as 'edn' (an abbreviation with which students may have been familiar from style sheets) rather than 'ed.'. Edition statements were frequently omitted for German literature, following ignorance of German publishing patterns and vocabulary, whereby the word 'Auflage' can mean either 'edition' or 'printing' and should be taken as an edition statement. 8 The worst error concerning the edition statement, and the most serious error in descriptive cataloguing, concerned the conflation of editions. An extreme example occurred for five editions of Frederick Bussby's Jane Austen in Winchester. The cataloguer created a catalogue record for the first edition of the work, including a note field: 'Copies include 2nd, 3rd, 4th and 5th editions'. These later editions, all published in different years and with different pagination, were effectively lost in a search. Typical errors in the imprint field included the inclusion of a second place of publication where irrelevant (e.g. Cambridge ; New York : Cambridge University Press), sometimes with irrelevant additions: e.g. Cambridge [Eng.] (unjustified by AACR2 abbreviations) or Cambridge [England]. This arose from using records ultimately derived from America, where the inclusion of the non-English place of publication was correct (see AACR2 1.4C5) and catalogue users might think of Cambridge as Cambridge, Mass. in the first instance. Names of publishers were sometimes given in full rather than being abbreviated to the briefest internationally recognisable form (AACR2 1.4D2), as in 'Victor Gollancz' for 'Gollancz' in the Tchaikovsky example above. The absence of publisher or place of publication was sometimes denoted by 'No pub.' and 'n.p.' respectively: abbreviations acceptable in style sheets such as the MHRA style book (Modern Humanities Research Association, 1991: 35; rule 10.2.1) and also in AACR1 (138C), but not AACR2. The possible ways to record an unknown date (AACR2 1.4F7) were ignored. Sometimes dates were recorded as 'n.d.', known from style sheets and acceptable according to AACR1 (142K), technically wrong, but clear and factually correct. Worse were wild and non-sensical estimates of publication dates, contradicting evidence elsewhere in the catalogue record, e.g.: Title: Statutes of Trinity College : translated from the original Latin statutes which were published ... 8 June, 1818 Imprint: Cambridge : n.p., 1800? Author: Bryant, Jacob, 1715-1805 Imprint: [S.l.] : [s.n.], [1700]? A common error in the statement of extent (MARC 300 field) was the omission of all pagination statements except the main one: i.e. of the frequent introductory sequences paginated in roman numerals; of leaves of plates. Note fields failed to record the presence of bibliography and index. An extension of the conflation of editions noted above was the failure to recognise discrete items bound together. These unwittingly received what were effectively misleading collection level descriptions. As not even a note denoted the contents, the discrete items were effectively lost. For example, one student record read: Author: Bentley, Richard, 1662-1742 Title: Tracts Imprint: London : Privately pub., 1710 Physical description: 1 v. ; 20 cm. (unpaged) Note: Spine reads 'Bentley Tracts' - no title page Subject: Trinity College (Cambridge) -- Source material Subject: Universities and colleges -- Source material -- Cambridge 9 Setting aside such errors as the incorrect form of the subject headings and the physical description field, the volume comprised five discrete items, all of which were paginated and had imprint statements (e.g. Printed for A. Baldwin ...; Printed for J. Morphew ...), and some of which were about rather than by Bentley; a search by author or title would fail to retrieve the items. Errors in descriptive cataloguing were exacerbated in catalogue records for the few early printed items catalogued. The mistakes were similar, including some omissions of statements of attribution; the effect was worse because the artefactual interest of books from the hand-press period renders accurate and full descriptions more important. For such materials, many libraries follow the detail presented in DCRB (Descriptive Cataloging of Rare Books), providing format in addition to size of items, accounting for unnumbered as well as numbered pages; presenting edition statements and the imprint in the exact words in which they are written; noting the presence of such features as errata and publisher's advertisements. By wanting any statement of physical description the student records fell below the minimum CURL standard even for modern monographs, e.g.: Author: Bryant, Jacob, 1715-1804. Title: Observations upon a treatise entitled 'A description of the Plain of Troy' by M. le Chevalier / Jacob Bryant. Imprint: Eton : s.n., 1795. Subject: Troy (Extinct city) -- History. for: Author: Bryant, Jacob, 1715-1804. Title: Observations upon a treatise entitled 'A description of the plain of Troy' by M. le Chevalier / by Jacob Bryant. Imprint: Eton : Printed by M. Pote ; sold by Messrs. Cadell and Davies, 1795. Physical description: [2], vi, 49, [1] p. ; 26 cm. (4to). Subject: Lechevalier, Jean-Baptiste, 1752-1836. Description of the plain of Troy. Subject: Troy (Extinct city) -- History -- Early works to 1800. In summary, students found descriptive cataloguing more difficult than MARC coding. Errors ranged from the cosmetic through the misleading to the inability to retrieve books. Subject cataloguing and access points The provision of access points, including subject headings, may on the whole be considered more difficult than descriptive cataloguing. In Germany, for example, subject headings are imposed by staff with superior qualifications and on a higher grade than those who do descriptive cataloguing ('Bibliothekare im höheren Dienst' vs 'Bibliothekare im gehobenen Dienst' respectively). In other words, this element is the more professional element of cataloguing, and therefore that in which one would expect non-professionals to have most problems. The students were expected to ensure that name headings conformed to the University Library's authority forms of names. The exception was for College members, for whom the College had developed its own style providing the fullest possible form of names, expanding all initials, providing dates of birth and, where relevant, death; and concluding with initials denoting Collegiate affiliation. 10 In fact, students did not check authorised forms, and the catalogue (no doubt like many others) abounded with multiple forms of names. The fault was worst for authors who were College members (the authors whom the College most wished to have right), as the additional local detail required entailed more scope for error, e.g.: Williams, Bernard Williams, Bernard, 1929- Williams, Bernard, 1929-, K.C.C. Williams, Bernard Arthur Owen Williams, Bernard Arthur Owen, 1929, K.C.C. Williams, Bernard Arthur Owen, K.C.C. Williams, Bernard Arthur Owen, 1929-, K.C.C. To have four forms of name for a single author was common. The maximum number of forms found was 16 for Sir Charles Bruce Locker Tennyson. This contravened a fundamental function of a library catalogue, to bring together all the works of a single author (Cutter, 1904: 12) Students frequently confused the roles of author and editor, giving the name of an editor main entry status for works without a single author, instead of using title main entry, with the name of the editor as an added entry, e.g. (for a collection of plays by four playwrights): 100 10 $aMarowitz$hCharles 245 10 $aNew American drama$ewith an introduction by Charles Marowitz for 245 30 $aNew American drama$ewith an introduction by Charles Marowitz 700 11 $aMarowitz$hCharles In this instance the student had introduced the error into a correct CURL record, as is evident from the fact that the student had imported the record twice, and a copy of the CURL record as imported, with the College's marker at the end, was still present in the interim database. The rule of three, whereby only the first of more than three authors in a single statement of responsibility receives an entry, and main entry is by title, was not always followed. The students coped well with uniform titles for single works, although they did not always apply them where relevant. They had problems with collective titles, which they devised idiosyncratically: e.g. 'Selected drama. German' for 'Plays. Selections' and 'Collected works. German' for 'Works' (in both examples the language is irrelevant, as the collective titles applied to original German works, not to translations). The collective titles thus lost their collocating functions. The library laid great emphasis on subject headings. Where possible, students catalogued books in their own subjects, in order to understand the contents and apply sensible subject headings. The theory proved well-founded, with notable errors pertaining to form rather than content. Examples of incorrect headings included corporate subject headings analogous with the lack of authority control over personal authors' names, such as: King's College, Cambridge. Chapel. King's College Chapel -- Cambridge. 11 King's College Chapel (Cambridge, England) alongside the correct form: King's College (University of Cambridge). Chapel. Sometimes inconsistencies resulted from ignorance of LCSH forms for types of works, e.g. for bilingual dictionaries: Language dictionary -- German -- English. German language -- English translations. English language -- German translations. instead of: German language -- Dictionaries -- English. English language -- Dictionaries -- German. Thus not all works on one subject were grouped under one heading, and a search by subject would not retrieve all relevant works. Students misapplied particular formulae, e.g. 'History and criticism' versus 'Criticism and interpretation' for literature. Sometimes they wrongly subordinated topics to places (e.g France -- Music). Sometimes they misunderstood the collocative purpose of subject headings to bring together works in different languages, such that for Leibniz's Fünf Schriften zur Logik und Metaphysik, a student imposed the subject headings 'Logik' and 'Metaphysik' besides, correctly, 'Logic' and 'Metaphysics'. The most prevalent errors concerned literature, which was catalogued by all the students because there was so much of it in the library and it was regarded as requiring no specialist knowledge. Works of literary criticism for which subject headings were appropriate frequently received such forms as: German literature -- Poetry -- 19th century for: German literature -- 19th century -- History and criticism. An overwhelming error was to impose subject headings upon single literary works, typically along the lines of the language and century (e.g. 'English literature -- 20th century'). This mistake arose from ignorance of the use of subject headings by the librarian training the students. Some headings were nonsensical: e.g. author: Storm, Theodore, 1817-1888. LCSH: German literature -- Prose -- 20th century. and, for a Dutch translation of one of Jane Austen's novels, Austen, Jane, 1775-1817 -- Translations into French. Some headings were correct in form, but frequent incorrect application rendered it difficult to find the works to which they applied among those to which 12 they did not: e.g. 'English literature -- 20th century' called up more novels than works of criticism. Many headings were not only incorrectly applied, but erred by addition of genre terms: e.g. English literature -- Prose -- 20th century (48 headings) English literature -- Prose fiction -- 19th century (133 headings) English literature -- 19th century -- Prose (54 headings) The above pattern was followed for other nationalities and genres of literature and was intensified when specimens of more than one genre appeared in a work, e.g. for Frank Wedekind's Prosa, Dramen, Verse: German literature -- Prose -- Drama -- Poetry -- 20th century. Errors in access points are more noticeable to users than many flaws in descriptive cataloguing and obscure the basic Cutter principle of grouping all works by a specific author and on a specific subject. Errors spanning descriptive and subject cataloguing The chief errors to span both descriptive and subject cataloguing were spelling and typographical ones. A keyword search for 'Correspondance' (conducted after several instances had been changed) resulted in 69 hits, of which 59 were spelling mistakes, a few in subject headings, the majority in titles (changed from 'Correspondence' in derived records). 'English literature' appeared in subject headings as 'English literatyre', 'English litertature' and 'English litrature'; 'Philosophy' as 'Philosopy', 'Philososphy' and 'Philosphy'. Below are two examples of catalogue records to contain multiple errors, besides the corrected form: Author: Mattingly, Garrett, 1900-1962. Title: Renaissance diplomacy. Imprint: London : Cape, 1955 (1963) Physical description: 322p[1 plate] ; 23cm. Series: The Bedfoord Historical Series Subject: Diplomacy -- History. Subject: Diplomacy -- Italy. Subject: Dimplomacy -- Sixteenth century. for: Author: Mattingly, Garrett, 1900-1962. Title: Renaissance diplomacy / by Garrett Mattingly. Imprint: London : Cape, 1962 (1963 printing) Physical description: 323p, [1] leaf of plates ; 23cm. Series: The Bedford historical series ; 18 Note: Originally published: 1955. Note: Bibliographical references: p. 299-300. - Includes index. Subject: Diplomacy -- History. 13 The two spelling errors (in a subject heading and in the series), the confusion about editions, the incorrect statement of pagination, wrong capitalisation in the series statement, and the lack of a statement of attribution from the title page are typical. The additional subject headings, although wrong in form, are not thoughtless (and are currently present on several records for the book on the relevant Union catalogue): a large proportion of the book in question concentrates on Italy and on the sixteenth century. They reflect ignorance of matters that can be known only if one is taught and if one looks things up: that 'Diplomacy' may not be subdivided geographically, and that a time division must follow either the subdivision 'History' or a subdivision considered to convey an historical concept (Chan 1995: 371). Author: Deutsch, Otto Erich. Title: Leopold Mozarts : briefe an seine Tochter / im Auftrag Gemeinde in Salzburg herausgegeben von Otto Erich Deutsch und Bernhard Paumgartner. Imprint: Salzburg-Leipzig : Verlag Anton Pustet, 1936. Physical description: 592p ; 22cm. Subject: Mozart, Leopold, 1719-1787. Subject: Composers -- Austria -- Biography. for: Author: Mozart, Leopold, 1719-1787. Title: Leopold Mozarts Briefe an seine Tochter / im Auftrag der Mozartgemeinde in Salzburg herausgegeben von Otto Erich Deutsch und Bernhard Paumgartner ; mit 32 Bildtafeln. Imprint: Salzburg : A. Pustet, 1936. Physical description: xvi, 592p, [32] leaves of plates : ill., coat of arms, facsims., ports. ; 21cm. Note: Leopold Mozart's daughter = Maria Anna Berchtold zu Sonnenburg. Note: Includes indexes. Subject: Mozart, Leopold, 1719-1787 -- Correspondence. Added author: Berchtold zu Sonnenburg, Maria Anna Mozart, Reichsfreiin von, 1751-1829. Added author: Deutsch, Otto Erich, 1883-1867. Added author: Paumgartner, Bernhard, 1887-1971. Added author: Mozartgemeinde in Salzburg. In this record the country of publication was coded as West Germany. Obvious errors are: failure to identify the author (whose name would not be found under an author search); positioning the first editor in the main entry field as the author; failure to note the second editor (which could impede retrieval); failure to check the valid form of the first editor's name; misinterpretation of the title page (even allowing for lack of knowledge of German which makes grammatical nonsense of the title, nothing in the layout of the title page justifies the interpretation Leopold Mozarts : briefe); lack of capitalisation; incorrect copying of the statement of responsibility; incorrect reproduction of the imprint (which does look like Salzburg=Leipzig, but elementary geographical knowledge should preclude such faults); incorrect statement of pagination, no reference to illustrations; an inappropriate subject heading. Most of the errors indicate carelessness. The omission of Maria Anna Mozart and of the Mozartgemeinde in Salzburg as access points require more advanced knowledge, and indeed have been omitted from catalogue 14 records of major libraries; the Mozartgemeinde in Salzburg does not appear in standard authority lists. Explanation of errors The root of problems was insufficient training and supervision. Training emphasised MARC (in which it succeeded) rather than AACR2. Moreover, students were unfamiliar with automated catalogues. They therefore were not in a position to make a connection between their work and the results for the user. The College was automating concurrently with other parts of the University. At the time, the post-1978 holdings of the University Library were catalogued electronically, but the University Library was a research library, little frequented by the undergraduates. The librarian in charge of the project considered the graduate students to be better cataloguers than the undergraduates; this could arise at least partly from the graduates' intuitive understanding through their dependence on the University Library with its partly automated catalogue. To the extent that students used catalogues, automated or otherwise, they would scarcely have analysed catalogue records, as they were usually searching known items by author and title to establish the classmark. The students did not understand the purpose of their work, nor the importance of such matters as correct spelling and non-filing characters for retrieval. They forgot instructions concerning the latter because they regarded such matters as irrelevant and petty fussing. One student cataloguer, who subsequently did a postgraduate library course, recalled that the meaning of the work she had been doing became clear only during her cataloguing option at library school. Shoddiness and indifference caused some mistakes, such as the inaccurate transcription of title pages and the ignoring of diacritics, and inadequate checking left them undetected. Frivolity inconceivable in a professional context caused others. A book published in Reinbek by Rowohlt about the Austrian poet Georg Trakl was later found to have on the catalogue record as its imprint: The Chocolate Factory : Willy Wonka; an error discovered only years later. Similar love for chocolate emerged in a subject heading for a literary work: 650 00 $aChocolate$xMars bars$xCrunchies$xBounty$xSnickers$xTwix$xKitkat$xMilky way. Well after the completion of the project, an entry for a stuffed toy was found on the catalogue. The inevitable unpredictability and sometimes inconsistency of Library of Congress Subject Headings (LCSH), reflecting their evolution over an extended period of time (e.g. Chan 1995: 51-53), exacerbated lack of training concerning LCSH. Rules are complex and are not intuitive. Furthermore, over time subject headings change, and the old, plausible form remains on many records. Wilfulness contributed to error: one student insisted on applying the subdivision 'Autobiography' rather than 'Biography', understanding neither that the former was a sub-class of the latter, nor the basic grouping function of subject headings for authority control. Significantly, the students were hardly aware of subject headings. Undergraduate reading lists supplied by lecturers eliminated the need to compile bibliographies, when subject headings become important. Catalogues throughout the University were author-title ones. To browse books by subject students went directly to the shelves, possibly asking staff for orientation first. Shelf browsing was efficacious in the College library owing to its refined classification system, the 15 second edition of the Bliss Classification (BC2). The students therefore did not rely on the intellectual collocation provided by subject headings, and had no occasion to note and unconsciously absorb them. Thus there was no familiarity with Library of Congress Subject Headings to substitute for the sound training which the students lacked. With motivation, the students could have read a manual on the topic in their own time, but they were expected to do holiday reading for academic purposes, besides which a vacation job cannot demand the commitment of a permanent post. As mentioned above, one of the most frequent errors, the application of generic headings as subject headings for single literary works, arose from the trainer's incomplete grasp of subject headings. Junior professionals might have corrected the trainer on the basis of their catalogue training at library school; students with no background in librarianship could not. An underlying flaw beneath the assumption that students were intelligent was the equation of intelligence with aptitude. Even the least intelligent of the students (whose academic capacity varied) earned respectable degrees; but a desire for pocket money through a holiday job and a leaning towards the attention to detail which marks library work are not the same. A correlation between academic subject studied and cataloguing ability reflects this. The worst of the student cataloguers read Archaeology and Anthropology, another noticeably poor cataloguer Social and Political Sciences, subjects which require broad reading, whereas the best read Classics, which, like cataloguing, demands precision. It must be emphasised that the students did not introduce all the errors found in the catalogue records. Errors in one context were not always errors in another: different libraries followed different authority forms (e.g. British Library versus Library of Congress name headings, before they merged); libraries adapted rules for the benefits of their own users; the rule about recording a second or subsequent place of publication if it is the country of the cataloguing agency renders records correct in an American context which are wrong in a British one, and vice versa; and some derived records had evidently been catalogued following AACR1 and become inaccurate with the application of AACR2 (cf Hoare, 1984: 101, 108). Furthermore, not all the records taken would have been produced by qualified cataloguers; and even the most expert professional can have an occasional memory lapse or a bad day. Some striking errors of subject headings (e.g. for dictionaries) were present in at least one other library catalogue on the system. In such cases students could not recognise and amend the mistakes as fully trained staff would have done. Consequences and analysis The relatively small size of the library, a good classification system and the modest requirements of catalogue users mitigated the results of cataloguing errors. The catalogue was used almost exclusively to perform known item searches for books on reading lists. Therefore in practice weird and wonderful subject headings caused little inconvenience, and other inconsistent or erroneous access points, as well as descriptive faults, also faded into insignificance. Subsequent discarding of stock and concomitant deletion of catalogue records further mitigated the result of the student cataloguing. The library's policy was to withdraw old editions of works as soon as a new edition appeared; thus in medicine, the natural sciences and to a lesser extent economics, the results of the students' work were soon deleted. Even without the replacement by new editions, some works were superseded and removed from the stock. Weeding of a secondary section of stock resulted in the rejection of a large number of poorly catalogued travel books. 16 However, owing to spelling mistakes in titles and/or authors' names some books were effectively lost, and duplicate copies inadvertently purchased. A higher level of staff time and expertise was necessary than should have been the case to establish whether the library possessed a book, for example when comparing possible donations against current stock. Staff needed to know that particular books should be in the library, and to have the flexibility and patience to continue searching when a correct search had failed: to look by author or title instead of both, or to try by ISBN, to circumvent the result of spelling mistakes. Poor cataloguing became a drawback when little-used or peripheral books were kept in the library but moved from open shelves so that physical browsing would no longer discover books. The library management system enabled virtual browsing, but users could not be expected to know relevant classmarks, which necessitated staff expertise to help. The removal of a section of theology books from display entailed a rapid recataloguing of the relevant books to ensure retrievability. As has been touched on, the most serious consequence of shoddy cataloguing occurred with the special collection of books by and about Jane Austen, catalogued to a standard well below that expected of junior library assistants. Faults included numerous spelling mistakes (including 'Persausion' for 'Persuasion' in 25 out of 27 titles), consistent failure to transcribe the title pages correctly, incorrect pagination, confusion between editions, and inappropriate subject headings (including the author's name for editions of her novels, thus introducing considerable 'noise' into the subject headings), description of Catalan translations of the novels as Spanish, and incorrect presentation of the imprint among other lesser errors. The donor had requested to see a copy of the catalogue records. He replied with a letter stating: 'I find it incredible to believe that a single person could make so many mistakes', and spent five pages, typed singly-spaced, listing them. Yet more disturbing were the books given which appeared to be missing (such as four of the five editions of the Bussby pamphlet noted above), implying Collegiate negligence; in fact, all books were present, but poor cataloguing, with the conflation of editions, had excluded them from the list. The donor's most valuable books, including the earliest editions, had not yet come to the College, and the student's work jeopardised the chances of their doing so. As a matter of high priority, all the student's records were deleted, and the books recatalogued, good money following bad. A second donation of Jane Austen books had been given to the College. The two collections were meant to be complementary, with approval having been given for the sale of duplicates. Reprints from different years were judged not to be duplicates for the purpose. The inaccuracy of the catalogue rendered it impossible to establish duplication from the catalogue; instead, every book had to be examined. From a financial viewpoint, the project therefore had hidden costs. The students were paid the minimum wage, at the bottom of the University's clerical scale (CS1; £7,645). Another College paid newly qualified librarians on the CS4 clerical scale (£13,504), a standard rate, for which it gained high-quality catalogue records. As long as records at King's College required no alteration, King's College gained financially. However, as soon as upgrading was required the costs mounted. The College paid a student a CS2 wage for a year manually to correct variant name forms of Collegiate authors. In large-scale correction across approximately 16 bays of the music collection, nearly every book required amendment. Recataloguing was quicker than the first cataloguing effort had been, because in the interim more records had been loaded onto both the University's union catalogue and the CURL database. The administration of two databases, however, the University's union catalogue and SIRSI Unicorn, slowed procedures, as upgrading records meant either upgrading the College 17 record on the University system or (quicker and more efficient for the poorer records) deleting them and copying another record. From there, the records on SIRSI Unicorn had either to be deleted and re-imported, or else overwritten. The money paid to the student became an additional expense to the CS4 rate paid for a record which was acceptable in the first place. Where upgrading was done by a permanent employee towards the top of the CS4 salary scale, the time required to correct student labours was time not available for other tasks. On the positive side, the recataloguing of the Jane Austen collection enabled the new cataloguer to exploit the collection more fully than would have been the case if the first attempt to catalogue the books had been satisfactory. Recataloguing led to a small exhibition of books from the collection and an article in a major bibliographical journal increasing awareness of it.1 Other Colleges in Cambridge benefited from the experience at King's College. Because College libraries were all purchasing the same books they continually borrowed each other's records and were in a position to compare the cataloguing of different libraries, readily identifiable from the record identification number imposed by the union catalogue which began with a unique code for each library. The student records functioned as a cautionary tale and rendered those funding other libraries more likely to pay to have the work done professionally. Conclusion Can students catalogue? Is cataloguing an intuitive, basic task that can be picked up with minimal training, as the project assumed? Students' general competence when imposing UK MARC codes indicates speed of learning and proficiency with computers. They can perform basic clerical tasks in libraries. At the time, several libraries were loading short records onto the Union catalogue in order to record their holdings in a basic form, to be upgraded as time permitted. Had King's College attempted this, the result would have been more accurate, with minimal scope for error, and, as records could have been created considerably more quickly, cheaper. One other library within the University used students to catalogue books in this way; the Librarian checked each record for accuracy, and found the result acceptable. The partial understanding of collective titles shows a degree of thought. Students cannot master two detailed, sometimes arcane sets of rules - AACR and LCSH - and provide library work of professional standard without concomitant training and adequate ongoing supervision. At worst their work will be inferior to that of (supervised) junior library assistants. The expense for the library, if it desires professional results, will be greater than to appoint professionals in the first instance. In this students are no worse than other non-professionals. The results at King's College resembled the experiences at Hull and at the University of Botswana described above (except for Hull's problem of poor handwriting on cards). Moreover, the variable quality of catalogue records to be found on the CURL database demonstrates that wherever cataloguing is uncontrolled and not highly regarded, quality will suffer. Results at King's College Cambridge make clear that computerisation, far from compensating for lack of accuracy and training, instead highlights it. The increasing quantity of records on CURL and other shared databases, as re-cataloguing and retrospective conversion continue and consortia grow larger, does not alter this: there is less call for original cataloguing, but at least an equal need to know rules for editing purposes and to prevent the proliferation of 1Attar, K.E. (2002) Jane Austen at King's College, Cambridge. Book Collector, 51, 197-221. 18 error. As the argument about professionalism generally and cataloguing in particular continues into the 21st century, the project results show that a derisory attitude towards library skills is unjustified. A place remains for qualified librarians to do qualified library work. Bibliography American Library Association et al. (1967) Anglo-American cataloguing rules: British text. London: Library Association. Bowman, J.H. (2003) Essential cataloguing. London: Facet. Bridge, Jeremy (2000) Retrospective conversion at the Tate Gallery library. Catalogue & Index, 136, 1-3. British Library Bibliographic Services Division (1980) UK MARC manual. 2nd edn. London: British Library Brunt, Rodney M. (2004) The education of cataloguers. Catalogue & Index, 153, 1-7. Chan, Lois Mai (1995) Library of Congress subject headings: principles and application. 3rd edn. Englewood, Col.: Libraries Unlimited. Consortium of University and Research Libraries, CURL minimum standards. Online resource at: http://www.curl.ac.uk/database/bibstandards.html. Cutter, Charles A. (1904) Rules for a dictionary catalogue. 4th edn. Washington: Government Printing Office Douglas, Nancy E. (1985) REMARC retrospective conversion: what, why and how. Technical Services Quarterly, 2 (3/4), 11-16. Dyson, Brian (1984) Data input standards and computerization at the University of Hull. Journal of Librarianship, 16, 246-61. Gorman, Michael (2002) Why teach cataloguing and classification? Cataloging & Classification Quarterly, 34(1-2), 1-13. Hill, Janet Swan (2002) Pitfalls and the pendulum: reconsidering education for cataloging and the organization of information: preface. Cataloging & Classification Quarterly, 34(1-2), xix-xxiii. Hoare, Peter A. (1986) Retrospective catalogue conversion in British university libraries: a survey and a discussion of problems. British Journal of Academic Librarianship, 1, 95-131. Hoerman, Heidi Lee (2002) Why does everybody hate cataloging? Cataloging & Classification Quarterly, 34(1-2), 31-41. Joint Steering Committee for Revision of AACR (1988). Anglo-American cataloguing rules, 2nd edn, 1988 revision. Ottawa: Canadian Library Association. 19 Kgosiemang, Rose Tiny (1999) Retrospective conversion: the experience at the University of Botswana Library. Cataloging & Classification Quarterly, 28(3), 67- 94. Law, Derek (1988) The state of retroconversion in the United Kingdom: a review. Journal of Librarianship, 20, 81-93. Maurer, Hansjürgen (1991) Retrospektive Katalogkonversion in einem Verbundsystem. Zeitschrift für Bibliotehkswesen und Bibliographie, 38, 109-28. Mayer, Constance (2002) Variations: creating a digitial music library at Indiana University. In: Cataloging the web: metadata, AACR and MARC 21, ed. by Wayne Jones, Judith R. Apronheim and Josephine Crawford. 149-154. Lanham, Md. and London: Scarecrow Press. Modern Humanities Research Assocation (1991) MHRA style book. 4th edn. London: Modern Humanities Research Assocation Peters, Stephen H. and Douglas J. Butler (1984) A cost model for retrospective conversion alternatives. Library Resources & Technical Services, 28, 149-62. Quedens, Jenny (1991) Retrospektive Konversion in den USA. Bibliothek: Forschung und Praxis, 15, 62-111. Read, Jane M. (2003) Cataloguing without tears: managing knowledge in the information society. Oxford: Chandos. Trickey, Keith. (2004) Revive the lost art - or we've only ourselves to blame. Library + Information Gazette, 26 Mar. 2004, 1-2. Ward, Ray (2004) Appalling downgrading of cat and class. Update, May 2004, 20. Watkins, Diane (1985) Record conversion at Oregon State. Wilson Library Bulletin, 60(4), 31-3. Wessendorf, Berthold (1991) Catalogue conversion in Switzerland. European Research Libraries Cooperation, 1, 55-60.