EDMORIAL BOARD

Editor and Chair of the Editorial Board JENNIFER YoUNGER

E d i t o r i a l A s s i s t a n t . . . . . . . . D A V I D H . T H O M A S

Assistant Editors:

Cunlsrrer* BorssoNNAs for Acquisitions Section

C,rnlril Ruscnorn for Cataloging and Classification Sectiotr

JrtNr Tnreowrll for Collection Management and Development Section

L E E D I R K S , . . . for Preservation and Reformatting Section

SyLVtA O ManrIx for Serials Section

PRUDENCE \ \ ' . DALRYVPLE . . Special Editor

Book Revierv EditorMencRnpt Ronov

Er-Officio Members:

Er-erNr YoNrz, Delegate ol the Chair, Council of Regional Groups

Kanrr" Mullrn, Executive Director, ALCTS

Dals Swrxsrr'-, Editor, ALCTS Netosletter

Library Resources b Technical Serxices (ISSN 0024-2527) is published quarterly by the

American LibraryAssociation,50 E. Huron St., Chicago, IL 60611. It is the official publication of

the Association for Library Collections & Technical Services, a division of the American LibraryAssociation. Subscription Price: to members of the Association for Library Collections & Techni-

Librarg Resources Cr Technical Serxices, S & S Computer Services, Inc , 434 \\i Dou'ner, Aurora,IL 60506

Library Resources Cr Technical Serxices is indexedin Library Literature, Library d: InformationScienci Abstracts, Current Indet to lournals in Education, Science Citation lndet, and Informa-

tion Science Abstracts. Contents "r'"

li*t"d in CALL (Current American-Library Literature).Its reviervs are included in Book Rexieu Digest, Book Rexieu lndex, and Rexieu of Rexieu;s

Instructions for autl-rors appear on p 72-73 of the January 1998 issue and on the LR?S \\Ieb page at

*tm'.ala.orglalctsnrts. Copies of books lbr revies'should be addressed to Margaret Rohdy, Book Revies'Editor, Libranl Resources d: Technical Sencices, Van Pelt-Dietrich Ubrary Center, University of Penn-

sylvania Libraries, 3420 \\zalnut St., Philadelphia, PA 19104-6206 (e-mail: rohdy@pobox.upenn.edu).

@ American Library Association 1998

All materials in this journal subject to copy'right by the American Library Association may be

photocopied for the noncommercial purpoie ofscientific or educational advancement granted by

Sectionil0T and 108 ofthe Copyriglit Revision Act oft976. For otlrer reprinting, photocopying,

o r t r ans la t i ng ,add ress reques t s to theALAOf f i ceo fR igh t sandPe rm iss i ons ,50E Hu ronS t . ,

Chicago, IL 6061 I

The paper used in this publication meets the minimum requirements of American National Stan-

dard for Information Scigngg5-pslrnxnence of Paper for Printed Library Materials, ANSI

Z,39 48-1992 *

Publication in Library Resources U Technical Sen;ices does not imply official endorsement by

the Association for Library Collections & Technical Services nor by ALA, and the assumption of

editorial responsibilityis not to be construed as endorsement ofthe opinions expressed by the

editor or individual contributors

/257

Referees | 998These referees are those to whom manu-scripts received in lgg8 were sent KimFisher was missed on the 1997 list and sois included here even though the manu-script was sent to her in December 1997This list includes referees for papers re-ceived by October 1.-Ed.

Ahronheim, JudyAvery ChristineBaumgarten, RichardBeacon, MatthewBogan, RuthBoissonnas, ChristianBudd, JohnCalhoun, KarenChrzastowski, TinaCooke, EleanorCox, BonnieDean, BeckyDiedrichs, CarolDirks, LeeDuke, JohnEl-sherbini, MagdaErcelawn, AnnGrenci, MaryHayman, LynneHemmasi, HarrietteHixson, CarolHolley, RobertHolt, CarolineHorning, KathleenHughes, Janet

Jacobs, El in

Jedlicka, Beth

]ohnson, Nancy Becker

Johnston, GeorgeKelley, SherryLarnbrecht, JayLeazer, CregLeBlanc, JimLee-Smeltzer, ]anetLeighton, LeeMacEwan, BonnieMartin, SylviaMcCombs, GillianMcGarry DorothyMorris, DilysMusser, LindaPerry Emma BradfordRichter, CaroleRiemer, JohnRuscholl, CarlenSaylor, JohnSchmidt, KarenStine, DianeSun, DajinTreadwell, JaneTuohy, |udyVanPulis, Noelle\l/alters, Carolyn\\/eiss, Paul!\'iberley, Stephen, Jr.\\/ilson, Flo\\blven, Bob\\'ool, Gregory

258/

Rising to the Top: Evoluoting theUse of the HTML META Tog lolmprove Retrievol of World WideWeb Documents throughInfernef Seorch Engines

Thomqs P. Turner ond Lise Brockbill

imp rooe re s ourc e dis c ot: ery.

rn|. he problern ot'finding materials on the

\\brld \\'ide \\zeb has been discussed inlibrary and in{brmation science journals,computer literature, and the popular me-dia Internet search engines have beendeveloped to aid in finding rnaterials; how-ever, their per{brmances vary considerably.Numerous re.searchers have evaluatedthese tools and have detailedtheir strenethsand rveaknesses. Melee's lndexing Cover-age Analysis (MICA) report. issued rveekly,

details the number of pages indexed byvar-ious Intemet search enflnes; in addition,the speed ol the systerns is evaluated(Mele; 1998). Other authors have analyzedparticular aspects of Internet search en-gines, such as their retrieval precision(Leighton and Srivastava 1997), their us-abillty (Pollock and Hockley 1997), andtheir indexins methods (Srinivasan, Ruiz,and Larn 1906). Some researchers ha't'eofl'ered advice to the authors of Hypertext

THoMAs P Tunr-rn (tpt2@corr-rell edu) is lvletadata Librarian, Albert R Mann Library, Cornell

University, Ithaca, Nerv York Ltst BnacxsILL (liseb@nethost multnomal.r lib,or us) is Technol-

ogv Trarner, Multnomal.r County Library Portland, Oregon Manuscript received February 4,

1998; accepted lbr publication April 17, 1998

LRTS . 12(4) . Rising to the Top /259

Markup Language (HTML) documentsabout improving retrieval of their materialsThe current research rvas designed to de-termine how uselul one rnithod, theHTML META tag. is in irnproving accessi-bility via Interneisearch engnes; here lvefocus on indexins rather than on search en-gine per{ormance.

METADATA AND THE

HTML META TeC

Much has been rvritten about the impor-tance of metadata for understanding andusing electronic resources This l i ter i turesheds light on the types of issues that theHTML META tag (see {igure 1) is

<HTML><HEAD><T ITLE>Pou l t r y , p roduc t i on and va fue< /T ITLE><META NAME="keywords " CONTENT="USDA, Mann L i b ra r y , pou l t r y p roduc -L i o n a n d v a l u e , a g r i c u l L u r e , I i v e s t o c k , d a i r y , p o u l t r y , a g r i c u l -t u r a l e c o n o m i c s , b u s i n e s s , t r a d e , c o m m o d i t i e s , s t a t i s t i c s " >< M E T A N A M E = " D e s c r i p t i o n " C O N T E N T = " T h i s f u l l - t e x t f i l e p r e s e n t s t h eannua l es t ima tes o f p roduc t i on and va lue f o r commerc ia l b ro i l e r s ,e g g s , t u r k e y s r a i s e d , a n d c h i c k e n s s o l d b y s t a l e s a n d U . S . T h i srepo r t i s a supp lemen t t o B ro i l e r ha t che ry , Ch i ckens and eggs , andTu rkey ha t che ry - ">< / HEAD><BODY>< A H R E F = " h L L p : / / w w w . m a n n - I i o . c o r n e l I . e d u / g a t e w a y . h c m 1 " > M a n n L i b r a r yH o m e P a g e < , / A > :<A HREF= " h t t p : / /www. mann f i b . co rne l l . edu / ca ta f og / ca ta f og . h tm l ' , >ca te -way</A><H1>< img s r c= " h t t p : / /ww . r nann l i b . co rne l I . edu / i cons /wo r l d . g i f " ALT- " I W o r ] d l " >Pou lL r y , p roduc t i on and va lue .< /HL><HR><P>< f o rm ac t i on= " h t t p : / / tm . mann l i b . co rne f I . edu / cg t i - b i n , / connec t . cg i "me thod= "pos t " >< inpu t t ype= " subm i t " name= , ' Connec t " VALUE= " Connec t , ' >< f npu t NAME= ' t he lD ' TYPE= " h i dden ' , VALUE- " 7 28 ' , >< / f o rm><b r c l ea r=a l l ><HR><H3 >Desc r i p t i on< /H3 >Th i s f u l 1 - t ex t f i l e p resen t s t he annua f es t ima t . es o f p roduc t i on andva lue f o r commerc ia l b ro i l e r s , eggs , Lu rkeys ra i sed , and ch i ckenss o l d b y s t a t e s a n d U . S . T h i s r e p o r t i s a s u p p l e m e n t t o B r o i l e rha t che ry , Ch i ckens and eggs , and Tu rkey ha t che ry .<p>Resou rce t ype : Fu l l t ex t<P>Update Frequency: Annual fy<P>Summary Ho ld i ng rs : 1995 -<P>P u b l i s h e r : W a s h i n g t o n , D C : N a t i o n a l A g r i c u l t u r a l S t a t i s t i c s S e r -v r c e , < P ><H3>Access No tes< , /H3 >N o a c c e s s r e s t r i c t i o n s a p p l y . < p ><HR>< B > C r o s s r o a d s < / B > . . . f r o m h e r e y o u c a n : < p >< D L > < D D > c o t o o t h e r t i t l e s o f s i m i l a r s u b i e c t :<DD><UL><L I><A HREF= "h t t p : / /www.mann l i b . co r -n e l l . e d u / c g i - b i n / s u b j . c g i ? a g - e c o n " > A g r i c u l t u r e - A g r i c u l t u r a l E c o -nom ics< , /A><L I><A HREF= "h t t p ; / /www.mann l i b . co r -n e 1 1 . e d u / c g r i - b i n / s u b j . c g i ? a g - l i v e ' , > A g r i c u l t u r e - L i v e s t o c k , D a i r yand Pou l t r y< /A><L I><A HREF= "h t t p : / /www.mann l i b . co r -ne l l . edu / cg i - b i n / sub j . cg i ?bus - t r a , ' >Bus iness and Econom ics - T radeand Commod i t i es< /A>< / u L > < / D L > < P >< /BODY>< /HTML>

Figure l. Example HTML Document s'ith Embedded META Tags

260/ LRTS . 42(4) . Tu.rnerandBrackbil l

intended to address. Metadata is com-rnonly defined as data about data A morecornolete definition notes that rnetadatapro*'ides "a user (human or rnachine)with a means to discover that the re-source exists and how it miEht be ob-tained or accessed. I t can covir rnany as-pects, such as subject content, creators,publishers, quality, structure, history,access riqhts and restrictions, relation-ship to oiher works or appropriate audi-ence" (Efthimiadis and Carlyle 1997, 5)Metadata is important fbr what it enables;its strength is not description but the sup-port it provides for resource discovery anddata use (Lynch I998). Metadata also pre-vents ambiguity about data (Lide 1995).\\'eibel (1995) describes metadata as thecenterpiece ol inlbrmation gathering. Heargues that new types of metadata need tobe developed to facilitate docurnent dis-covery and suggests the Dublin Core ele-ment description set as a solution Ibrmetadata problems.

HTML permits document authors tocontrol not only how text, graphics, andmultirnedia materials are displayed, butalso the inforrnation available about thedocument itself throuph the use of theMETA tag. Several authors have sug-gested that the HTML META tag can beused to enhance infbrrnation retrieval, es-pecially through Internet search engine.sAltaVista Search Network (lgg7) docu-mentation suggests that authors use thekeyrvords and description attributes o1'the META tag to improve retrieval andcontrol the descriotion ol the documentthat appears on i search results pageBremser (1997) offers more detailed ad-vice to \\/eb authors about using di{Ierentaspects of the META tag

The META tag has also been seen as arvay of providing additional types o{'metadataabout documents. Mil ler (1995)discusses the potential use of the METAtag to contain formatted in{brmation de-fined by the Dublin Core element setThe Dublin Core orovides a means of cre-atine basic metadita about a resource in asirnple rnanner and is not formally con-nected to the HTML META tag. Hou'-ever, the META tag is the best section ol'the HTML specification in which this

data can be placed (\\/eibel 1997). Manyo{'these authors envision resources thatare "self-declaring" because the itemsprovide importanl information aboutthemselves to hurnan catalogers and auto-rnated indexers.

The HTML META tas resides withinthe header and can have the attributesCONTENT, HTTP-EQUIY or NAME.It is intended to provide "a place to putrneta-information that is not delined bythe other HEAD elements. This allows anauthor to more richly describe the docu-ment content for indexing and catalogingpurposes" (Graham 1995, f47). In this re-search. we are most concerned with trvoattributes: CONTENT and NAME. TheNAME attribute requires that a CON-TENT attribute also be present. Al-though the NAME attribute can take thevalues of author, document type, distribu-tion, keylvords, and descripiion

"*otlgother values, most of the Intemet searchengines that currently support use of theMETA tag recognize only those NAMEattributes defined as keyl,vords or descrip-tion. The keylvords attribute provides im-portant terms associated u'ith a docu-ment, rvhile the description attributebriefly details it and is often used as asummary on the results page generatedby Internet search engine queries. Thisexample of a META tag from the headerof the USDA report "Agriculture andtrade: Eurooe" illustrates the use ofboththe keywords and description attributes:

<META NAME="Kelt 'ords" CON-TENT="USDA, Mann Library agricul-ture, Europe, agricultural economics,international agriculture, business, eco-nomics, trade, commodities, statistics">

<META NAME="Descript ion" CON-TENT="Database contains macroeco-nomic data on \\'estern Europe, budgetand price data, and time-series data on sup-ply and utilization of agricultural commod-ities for the EC- 12 and the European FreeTrade Association ">

Several authors have voiced some con-cerns about the potential misuse and fail-ure of the META tag. Kuhn (1996) notes

that although the META tag can be usedfor certain infonnation, there is notenough agreement about the qpes ofin-formation that can be irnolemented. He isespecially concerned a6out in{brmationrelated to authors of documents, ab-stracts, and document content beyondkepvords assigned by authors. One con-cern about the use of the META tas in-volves the various opinions about the no-menclature for the NAME attribute.Currently some search engines recognizeNAME designated as keywords and de-scription, but other options, such as his-tory, access restrictions, and audience,are ignored. \\/ithout consensus about thenornenclature amonq the HTML stan-dard develope.s, aniho.s using HTML,and Internet indexinq services, theMETA tag will ne'uer 6e rvidely irnple-mented (P1'a{Ienberger f995). Currentcourt cases rvill also set orecedents for theuse of the META tag. Using names in aMETA tag that have nothing to do withthe content ol'a site has been called intolegal question by companies whose namesappear in documents with which theyhave no connection (Kaplan 1997).

PRoBLEM Srerrlroxr

The Albert R. Mann Library at CornellUniversitv works in coniunction with theEconomic Research Service, the NationalAgricultural Statistics Service, and the\A/orld Asricultural Outlook Board o{ theUnited States Department of Agricultureto produce the USDA Economics andStatistics System (http://usdamannlibcornell edu/usda/). Thii system providesaccess to over 300 statistical reports anddata sets in various agricultural iommod-ity and business areas Like other largecontent providers, producers rvant theiematerials to be discovered by lnternet us-ers who might not have been previouslyaware of the service. Users would {indthese materials relevant whether theywere searching fbr agricultural econom-ics materials in general or specific com-modity ligures, such as watermelon pro-duction statistics.

The META tag might help publishersensure that their materials are found

LR?S . 42(4) . Risingto the Top /26I

when appropriate searches are executed.Although the META tag is being put touse by many !\/orld \\/ide \tt/eb publish-ers, its effectiveness has not been evalu-ated. In this study, we examine the follow-ing questions related to the use of theHTML META tag:

l. Do pages that use the META tag havehigher retrieval ranks than pages thatdo not?

2. Is one method of META tasauthoring more effective than otheimethods?

3. Do pages that use both of the METAtaq attributes have better retrievalranks than pages that use only one at-tribute?

To answer these questions, it is necessaryto understand how search engines dealwith the META tag.

Mrrnoo

At the t ime oI ' this research. Mann Li-brary provided access to many USDA re-ports and data sets, as well as other net-worked electronic resources, throuqhthe Mann Library Gateway (http://www.library.cornell.edu/). All resources pre-viously available through the Mann Li-brary Gateway are norv available throughthe Cornell University Library Gateway(http://www. librarycornell.edu). Thegateway is a searchable database of elec-tronic resources that allol'r's users to con-nect via a hv-oerlink to resources thatmatch their queries Searches yield dy-namically generated HTML pages withlists of appropriate records. A gatewayrecord lists the title of the work, a de-scription, the publisher, the publicationdate, update frequency, type of material,summary holdlngs, access information,and general subject categories Userscan connect to the resource by clickingon a hyperlink fiorn the record. For thisexperiment, static gateway-like HTMLdocuments were created to test how ac-cess to this type of metadata record couldbe imoroved.

Twenty HTML docurnents were cre-ated in five subject areas: agricultural

262/ LRTS . a2Q) . Turner and

TABLE 1LIST OF USDA ECoNoMIC AND

Srerrsrrcs SYSTEM DocuvrNrs Usup

Agricultural trade behveen Asia./Near Eastcountries and the United States

Agricultural trade o{ former Soviet RepublicsAgricultural trade policies "Redbook"

Agriculture and trade: EuropeCotton and rvool outlookCotton and rvool yearbookCotton ginningsCotton/citrus productionFarm business balance sheetFarm operatirrg and {inancial characteristicsFarm production expendituresFarm sector balance sheetPoultrv outlookPoultry slaughterPoultry yearbookPoultry production and valueVegetable yearbookVegetables ar.rd specialtiesVegetables annual summaryVegetables

trade, {'arnr business statistics, poultrv sta-tistics, vegetable statistics, and cottonstatistics Static pages were used lbr trvoreasons: dynamically generated pagesiue created frorn a script that u'ould notallorv {br different HTML markuo and.more importantly, many Internet iearchengines do not index dynarnically gener-ated pages Four pages rvere created ineach subiect area: one with no METAtags, one rn'ith a META tag using thekeyrvords attr ibute, onewith a METAtagusing the description attribute, and oneu'ith M ETA tags using both the keyrvordsand description attributes The reportand data set titles used are listed in tablel. These pages contained al l infbrrnationpresent {br the titles in the gatervay re-cord, including a \\'orking hyperlink tothe resource These documenrs wereplaced on a separate server and rvere notlinked to the gateway. As far as anInternet user f inding these pages is con-cerned, they are functioning gateway re-cords. However, we were able to alter theHTML markup for research control pur-poses, andthe unique location allowed us

Brackbill

to determine quickly *'here these partic-ular documents resided on the ranked listof search results

Keylvord and description attributeterms were chosen {iorn descrintive in-lbrrnation avai lable throush the docu-nlentation for that reoort r-,r data set asu,ell as from cataloging records created atMann Library. The nurnber of kepvordterms chosen for the keywords attributesranged {iom 10 lbr "Poultry Slaughter," to40 fbr "Vegetables." The average numberof keylvord terrns assigned using thekey'rvords attribute was 18.4 and themode rvas 12. Keyrvords ranged flom spe-ci{ic terms, such as "watermelon," to ab-stract or general terms, such as "busi-ness." Descriotions matched surnmariesroutinely provlded in cataloging recordsibr each itern and contained keylvsld5 a5-sociated rvith the title

All 20 pages were subrnitted to thethree Internet search engines that supportthe use of the META taq-AltaVista,HotBot, and Inlbseek-in late Decemberi996 Of these three services, onlyAltaVista and Infbseek indexed the pagesIn{bseek indexed the pages within t$'odays of submission. AltaVista indexed thedocuments after a rnontht delay and a{ierseveral submission attempts over asix-rveek neriod. HotBot failed to indexthe pagesiller several requests. Frorn theperspective of a content provider, \ve sug-gest that the process fbr subrnitting pagesandthe speed ofindexing sites by Internetsearch engines rnight be irnproved

Once the pages were indexed, searches$,ere perlbnned in Altavista and Infoseekto {ind tenns colnmon to all pages as rvellas for terms paired with each keyword anddescription term contained in the METAtags. The only search conducted fbr all 20pages was the search fbr "Mann and agri-culture." Searches follorved a set {brrnat inu'hich "Mann," "statistics," or "USDA"

rvere combined rvith the kelvrord tennscontained on the page For exarnple, thesearches "Mann and poultry" "statistics

and poultry" and "USDA and poultry"u'ere each perforrned "Mann and [key-rvord]" u'as searched to test a specilic terrnand name u'ith a varietv of soecific andgeneral keyllords "staiistics^ and Ikey-

LR?S . 42(4) . Risingto the Top /263

TABLE 2Salrpr-n Quunre s Usro ro Snencn FoR USDA EcoNoturc

AND STATISTICS Svsrnu Docutr.tnNrs

Mau and IKelvordl USDAarrd IKervord l S ta t i s l i csar rd fKervor , l ]

Mann USDA

Mann statistics

Mann poultry production

Mann value

Mann agriculture

Mar.rn livestock

Mann dairy

Mann poultry

Mann agr icul tural economits

Mann business

Mann trade

Mann commodities

USDA Mann Library

USDA statistics

USDA poultry production

USDA value

USDA agriculture

USDA l ivestock

USDA dairy

USDA poul tw

USDA agricultural economics

USDA business

USDA trade

USDA commodities

Statistics Manl Library

Statistics USDA

Statistics poultryproductiol

Statistics value

Statistics agriculture

Statistics livestock

Statistics dai4,

Statistics poultrv

Statistics agriculturaleconomrcs

Statistics business

Statistics trade

Statistics commodities

rvord]" was searched to test a general terrnsearch with a range ol'specific' and generalkepvords 'USDA and [kepvord]" wassearched to test a commonly expectedgeneral name u,ith a number rr1'speci{icand general kelwords. Table 2 lists a sam-ple ofqueries. In total, 579 search combi-nation results were recorded.

All Inlbseek searches were completedin January and early February 1997. AllAltaVista searches rvere comDleted dur-inglate March andApri l I997^The lengthof time required to complete the searchesrvas due partly to delays ir-r indexing andpartly to the time required to complete allsearches More ellicient automatedmeans of checking the ranks of pages,such as software like \\/ebposition Ana-lyzer, rvere not available at the tin-re thesearches were conducted These delaysare not expected to have had a significantimpact on the results obserwed becauseseirches repeated at inten als during thisprocess did not reveal any changes inrankinq Once search results \\'ereachieved, the first 200 results were exam-ined to determine rvhich pages l-ell rvithinthose retrieved-iten'r lists. If a page rvaslbund, the rank ofthe page rvithin that listwas recorded.

To evaluate the effectiveness of using

the HTML META tag to irnprove re-trieval of HTML documents searched onInternet search enqines, the ranks ofpaqes retr ieved by both AltaVista andit-tl,.".k rvere recorded. For each set o{'rnarkup comparisons, searches were ex-amined in q'hich the terms appeared onboth of the pages analyzed either in thetext or in the description or kepvords at-tributes of the META tag To be consid-ered for analysis in a given cotnparison,the search must either have retrievedboth pages being compared or have beenexpected to retrieve both pages.

By basing queries on terrns known tobe present in these documents, severalconcerns rnust be noted. In his analysis ofthe ASLIB Cranlield Research Project,Swanson (1965) argues that it should notbe assurned that nonsource docurnentsbehave in the salne rnanner as source doc-uments Srvanson also notes that the on-portunity t<-r tind unexpected resuiismieht be lessened if the lbcus remains ona relati rely small set of documents andsearch terms. In addition, terms searchedmight not reflect the rvide range of tennsused in actual searches {rorn a diverseuser populat ion (Furnas et al. 1987).Searches per{brmed in this research u'eredesigned to determine horv search engines

264/ LRTS . 42(4) . Turner and Brackbill

TABLE 3NUMBER oF RANK Sconns Coupennp rN MANN-\ /HITNEY U TESTS

BY META TAG CoMPARISONS

Attributes ComparedNumber of Rank Scores Compared

(sanple size)

AltaVista and Infoseek Ranks Combined

None versus ke;rl'ords

None versus description

None versus both keryords and description

Keyvords versus both keyrvords and description

Description versus keptords

Description versus both kenvords and description

AltaVista Ranks Only

None versus ke;ru'ords

None versus description

None versus both ke;r'ords and description

Kepvords versus both kelu'ords and description

Description versus ke;r'ords

Description versus both kepvords and description

Infoseek Ranks Only

None versus ke;r'ords

None versus description

None versus both kepvords and description

Kepvords versus both keyrvords and description

Description versus kepvords

Description versus both keyu'ords and description

index page text in relation to META tagtext rather than to sirnulate user behavior.The results generated re{lect an ideal sce-nario. Page ranks might be lower withmore diverse search terms and search en-gine retrieval might be less e{Tective un-der those conditions.

The first 200 documents retrieved wereexarnined and the ranks of source pageswithin those {irst 200 were recorded Incases in which more than 200 sites wereretrieved. the first 200 were examined asan arbitrary cutoffpoint supported by allthe search engines used Harman (1993,371) reported using 200 as a retrievalthreshold, although she concluded thatfor the purposes d;tailed at the first TextRetrieval Conference, this point was too

198

96

194

506

166

174

96

44

96

: D J

84

88

102

J Z

98

254

82

86

low. I fa search could have resulted in re-trieving a page but it was not in the top200 sites, we gave that page the rank of201 lor that search. As a result, rankingsranged {rom I (highest) to 201 (not re-trieved) and the number ofsearches ana-lyzed in each comparison varied.

The ranks ofthe di{Ibrently coded pageswere compared using the Mann-\4/hitney Utest The U statistic measures "the numberoftimes that the rank ofa score in one groupprecedes the rank ol'a score in the othergroup" (Kiess 1989, 468). This provides for acomparison of hvo sets of ranked scores todetennine whether or not the sets can be ex-Dected to fall $'ithin the same distribution. Iflhe ranks are nart of the same statistical dis-tribution of ranks, then the addition of

Notes :No{erver than22searcheswererunforeachseto fcomoar isons Eachsearchqer rera tedranks foreachofthepagesbe ingcompared Asaresu l t , thenumbero fsearches lsha l f tha to l ther r r rml re ro f rankscompared Forexample, 42 searches resulted in 84 ranks being generated, 42 r anks for pages with the description attribute and42 ranks for pages with the ke;words attribute

LRTS . 42(4) . Risingto the Top /265

TABLE 4MANN-\ ZHITNEy U TEST RESULTS pon META Tac Colrrenrsols

Attributes Compared Siqtrif i cance Level (p=)

AltaVista and Infoseek Ranks Combined

None versus kepvords

None versus description

None versus both kep'ords and description

Kepvords versus both kepvords and description

Descriptior.r versus kelnvords

Description versus both keyvords and description

AltaVista Ranks Only

None versus ke;'u'ords

None versus description

None versus both keyvords and description

Keyrvords versus both keyu'ords and description

Description versus ke;ru'ords

Description versus both keyrvords and descriptior.r

In{bseek Ranks Only

None versus kepl'ords

None versus description

None versus both kepvords and description

Kepvords versus both ke;nvords and description

Description versus keryords

Description versus both kepl,ords and descriptior.t

'IDdicates significance at the 0l level

0000.

3900

0000.

3913

0000.

.0000"

0000.

.3806

0000"

2898

0000"

0000.

0000"

3605

0000'

.4439

0000"

0000"

rather than relevance percentages be-cause not all search engines provided rele-vance Percentage infbrrnation

The Mann-\\/hitney U test was run tocompaJe several sets of search result rank-ings {br pages with: no META tag andkepvords attribute META tags; no METAtag and description attribute META tags; noMETA tag and both keyrvords and descrip-tion attribute META tags; keywords attrib-ute META tags and both ke1'rvords and de-scription attributes META tags; descriptionattribute META tags and kepvords attributeMETA tags; and description attributeMETA tags and both keywords and descrip-tion attribute META tass In each test thenumber of times that the rarrk of one set ofscores exceeded the rank of another set ofscores was tallied. The results are summa-rized in tables 3 and 4.

Table 3 lists the number of searchesu'hose rankings were cornpared in each pair

META tags has probablynot affected the re-trieval rank Ifthe U test shows thatthe ranksare most likely not from the same distribu-tion, then the $pe of META tag markupused has probably affected the rank.

The Mann-Whitney U test was chosenbecause it tests rankings oI'at least ordi-nal-level data. The rankings received fromthe search engines were Jonsidered ordi-nal-level dati because AltaVista andInlbseek have di{I'erent alporithms {brranking materials. In addition, the degreeof relevance attributed to a site by a searchengine is not directlycorrelated to its rank.For instance, one search miqht yield a re-sult in rvhich a site ranked first is consid-ered l00vo relevant to the search queryrvhile in another search, a site given thefirst ranked position is consid ered,7SEo rel-evant to the query. As a result, the distancebetween .rikr is not consistent for allsearches analyzed. Ranks rvere recorded

266/ LRTS . 42(4) . TurneranclBrackbil l

TABLE 5MEorar. MoDE. ,rNp RaNcr FoR META Tac CovpeRIsoNs

(AlreVrsra exo INposrrx CovsrNnp)

Attributes Compared Paee TrDe Mediari Mode Ranqe

None versus keym'ords

None versus description

None versus both keynvords and description

Keryords versus both keyvords and description

Description versus keyt'ords

Description versus both kepvords and descriptiol

'Multiple modes exist Smallest mode is shown

No f'erver than 22 searches (yielding 44rankinqs) were observed for each set ofcomDaiisons Table 4lists the results of theU teit for each comparison by recordingu,hether a statistically significant di{I'er-ence was noted at the 0l level. Sets o{ranks are considered statistically signili-cant in their differences if the srnaller Uscore observed is less than or eoual to thecritical U value {br that samnle iize

The U tests \\'ere run tbr-all searches inAltaVista and Infoseek combined as rvell as{br AltaVista and Infoseek results sena-rately. Results are repoftecl lor AltaVistaand In{bseek ranks cornbined to determinehoq' rvell META tags rvork regardless o{ thesearch enqine used AltaVista and In{bseeks,ere also-considered separately to deter-mine rvhether one search engine'.s per{or-mance skewed the combined ranhngs.These results show horv Internet search en-gines deal u'ith META tag data rather thanhovv rvell Intemet search engines rvork toretrieve known iterns.

In addition to the Mann-\\/hitney Utest, the rrredians, rnodes, and range.s ,, [ 'ranks were generated to test the practicalsignificance of the findings {rorn theMann-\\/hitney U test If the U test sug-gests a statistically signilicant diff'erencebut the rnedian, mode, and range are sirni-

lar, then the U test difTerence may be saidto have less practical significance. Hovr'-ever, if the U test suggests a statisticallysisnilicant diflerence and the median,mode, and ranqe are different, then the Utest can be said to be reflect a nractical asrvell as statistically signiticant difl'erence.The median, rnode, and range values aresunrmarized in tables 5, 6, and 7.

DATA ANALYSTS : CotuPARrsoNSoF PRESENCE AND ABSENCE OF

META Tlc

The use of the HTML META tag vyas ex_pected to irnprove the ranking ,, f a docu-ment u'hen searched using an Internetsearch engine. \\'e tested this assertion bymeasuring the rankings of three sets o1'comparisons: pages rvith no META tag tothose rvith keyr'vords attribute METAtags, pages rvith no META tag to thoservith description attribute META tags,and pages u'ith no M ETA tag to those u'ithboth keyrvords and description attributesMETA tags. The inclusion of tliekernvords altribute, with or u'ithout thede.scription attribute also present, consis-tently improved the accessibility ofHTML documents. Unexpectedly, horv-ever, the inclusion ofonlythe description

None

Kelvords

None

Description

None

Both

Ke;nvords

Both

Description

Keprords

Description

Both

59 7 199

8 3 1 8 1

3 9 5 3 1 9 7

43

79 15 198

t4 2 ' 193

I4

20

O J

t3

a l

t2

2 t97

I 196

14 198

3 r09

4 198

2 r92

LRTS . 42(4) . Risingto the Top /267

TABLE 6Mrpr,q,N. MoDE. a,No ReNcr ron META Tac Coupentsor.ls

(ArreVtsre ONry)

Attributes Compar ed Page Ttpe Median Mode

None versus keyvords

None versus description

None versus botl.r kenvords and description

Kelnvords versus both kepvords and description

Descriptior.r versus kepvords

Description versus both ketruords and descnption

t97 5 r98 195

t7 5 5" 199

196 5 198 199

3 4 5 4 1 9 9

Surprisingly, the cornparison of' theranks of pages lvith no META tag and ofpages with only the description attributeMETA tag revealed no signilicant di{Ier-ence The ranks of96 searches were com-Dared. $'ith a U test result that failed toieach the .01 level (table 4) In addition,median, mode, and range were very sirni-lar for both sets of ranks (table 5). Thissuggests that the e{I'ect ofthe descriptionattribute was negligible

The U test and median, mode, andrange were generated f<rr AltaVista andInfoseek scores separately to verify thatthe similarities noted in all scores corn-bined were not the result of the bias ofone particular search engine The U testfbr AltaVista and {br In{bseek (table 4) didnot reveal a statistically significant differ-ence betrveen these sets of ranks. More-over, the rnedian, rnode, and range forboth AltaVista (table 6) and Infbseek (ta-ble 7) validated the U score results Al-though it is possible that the AltaVistascores, which re{lected a higher medianlbr ranks of pages rvithout i tr,letR tagthan for pages rvith the description attrib-ute alone. mav have in{luenced the me-dian for the icores cornbined, it is notlikely that the results lrom eitherAltaVista or Infoseek biased the corn-bined totals

None

Ke;r,ords

None

Description

None

Both

Ke;nvords

Both

Description

Ke;nvords

Description

Both

153 5 156 197

l0 5 r81

r27 148 196

r34 65" 197

153 156 r98

255 12 199

235

J 4 D

4 181

3 193

attribute did not improve ranking of apage over a page rvith no META tag.

Pages containing the kepvords attrib-ute META tag rvere consistently ranked byInternet search engines as more relevantthan pages lacking a META tag Pagerankings were lbund to have statisticallysignificant differences at the .0I level. TheUlest results of pages employing only thekeyu'614r attribute versus pages lacking aMETA tag reflect a comparison of 198search ranks. The results detailed in table5 suggest that pages that use the keyr,vordsattribute META tag are consistentlyranked as more relevant than those with-out the META tag In addition, the dispar-ity an-rong the median, mode, and rangefor these tu'o groups (table 5) is consistentwith the findings of the Mann-\\/hitneytest. Similar results u'ere obtained whencomparing the ranks of pages rvith noMETA tags and ol pages with both the de-scription and kewvords attributes METAtags-. This comparison o{ 194 search re-sults (tables 3 and 4) revealed that havinqboth attr ibutes in the META tac resultei iin higher rankings than having no METAtag present As rvas the case with thekep,r'ords attribute only results, the rre-dian, rnode, and ranqe (table 5) rvere suffi-ciently difl'erent fbr-each set to reinlbrcethe results of the Mann-\\/hitney test

268/ LRTS . 42(4) . Tu.rner and Brackbill

TABLE 7MrpraN, MoDE, aNp RRNce ron META T.rc Coup.rnrsoNs

(Iunosrox ONI-v)

Attributes Compared Paqe Tlpe Median Mode Range

None versus kepvords

None versus description

None versus both ke;nvords and description

None 26

Kepvords 5

None 18

Description l9

None 37

Both I

Ke;r'ords versus both kelnvords ar.rd description Ke;rvords 12

Both I I

Descriotion versus kelryords Description 27

Ke;nvords 8

Description versus both kepvords and description Description

Both

22

4

'Multiple modes exist Smallest mode is shown

The differences noted here rnicht bedue to the ways in which Internet iearchengines index and weigh the text from apage in comparison rvith the text in a de-scription attribute. According to in{brma-tion that we received from AltaVista Sup-port, if the description attribute isindexed, the page text is not (Alta VistaSupport 1998). This rnight be the casewith other search engines as well andwould naturally {avoi pages with noMETA tag over those with only the de-scription attribute META tag becausemultiple occurrences of the sarne termsmight be more likely in the text of a pagethan in the description attribute

Dere Auarvsrs: CoMPARrsoNs oFDtrrBnnNr META Tec Artnrsurrs

The Mann-\\/hitney U test was also per-forrned on the ranks ol pages using vari-ous attributes within the META tae. Uscores were generated to cornpare theserankings' pages with onlythe keywords at-tribute META taq to those with bothkewvords and dlscription attributesMETe. tags; pages rvithbnly the descrip-tion attribute META tag to those withboth kepvords and description attributesMETA tags; and pages with only the

keywords attribute META tags to thosewith only the description attribute METAtag. It was assumed that using both thekewvords and descriotion attributeswould improve retrievil in relation tothose pages with only one b?e of METAtaq. In addition, it was assumed that thekeywords attribute META tag would re-sult in better retrieval than the descrip-tion attribute META tas.

No statistically signif-icant di{I'erenceswere found between the ranks of pageswith both the kepvords and descriptionattributes and those of pages containing

keyrvords attribute. Although the U testdid not uncover a statistically significantdifTerence between the sets of' scores(table 4), the median rank for thekeywords attribute only pages was some-what better than the median rank forpages with both the keyrvords and de-scription attributes (table 5).

The ranks ofpages using only the de-scription attribute META tag q'ere com-pared to ranks of pages using only thekey"words attribute and to ranks of pages

2' r99I I I 9

3' 79l . R .

3 ' 198

2 168

4 r97

2 196

19 198

I 103

r9 197

2 4 0

using both the keywords and descriptior.rattributes. In both cases, the U test re-corded a statistically significant difler-ence (table 4). Additionally, the rnedianranks observed in both casei validated theU score results (table 5) The evidencesuggests that using the kepvords attrib-ute, eitherwith orwithoutthe descriptionattribute, improves retrieval rank over us-ing only the description attribute. Thismight be the result of the weight given to

Cot{ct usrou.lNn Furunr Wonx

Enabling users to lind materials on the\\'orldVttide \\/eb is an important problemfaced by librarians, search engine design-ers, and Internet publishers and contentproviders. As new standards in metadataemerge, such as the Dublin Core (\\,eibel

one current method, the HTML METAtag, is in improving accessibility viaInternet search engines. Because onlyAltaVista, Infoseek] and HotBot cui-rently recognize and use the META tag,\4'e suggest that search engine designeisenable their services to accept META tagdata because it does benefit retrievairank. Concern over improper uses of ' theMETA tag do not. iust i ly fai lure to indexthern when appropriately used.

This research serves as a snanshot of'how current forms of embedded metadataare processed by Internet search engines.Newer technolosies and methods tbi ern-bedding and indexing \\brld \\'ide \\'ebdocurnents are evolvins that will alter theview presented here T[e searches used totest indexinq methods reflect an idealizedsituation because terrns searched wereknown to be present in the documentssought. Retrieval rates based on more re-alisiic search scenarios might reveal lorr,er

LRTS . 42(4) . Risingto the Top /269

rankings. It was fbund that using thekepvords attribute of the HTML METAtag, u'ith or without the description attrib-ute, consistently irnproved the retrievalrank of a \\/eb docuilent. Mann-\\/hitneyU test comparisons of rankings of pageswith the keyrvords attribute versus thosewith no META tags revealed a statisticallysisnificant difference at the 01 level.Horvever, using only the description at-tribute in the META tag did not appear toimprove retrieval over not using a METAtag The Mann-\\zhitney U test did not re-veal a statistically significant improve-rnent in retrieval rank between pages us-ing only the description attribute andpages rvith no META tag. Furthermore,the U test shows that pages with only thedescription attribute were given consis-tently less-relevant ranks compared topages ernploying either the kelvords at-tribute alone or pages containing bothkewvords and description attributes.'

This discrepancy-may refl ect diflerentrvays in which Internet search engines in-dex and weigh the text from a page andthe text in a description attribute and the{ailure to index page text when the de-scription attribute is present. This pro-cess favors pages with no META tag overthose with only the description attributeMETA tag because there can be multipleuses of the same terms in the text of a pagervhile the description attribute is likely touse a given term f'ewer times. It is impor-tant to bear in mind that the descriptionattribute is designed to provide a displaysummary of the resource rather than toimprove retrieval of a document. \\/e sug-gest that search engine designers con-sider indexing the full text of a page re-gardless of the presence of a descriptionattribute or improve the relevance assess-rnent of'text in the descriotion attribute.The Mann-\\zhitney U te]st dld not un-cover a statistically significant differencebetween the ranks of pages with only thekepvords attribute and pages rn'ith boththe keyrvords and description attributes\\te suggest that \\'orld M'ide \\'eb authorsuse at least kepvords attribute METAtaes in their documents.

More research needs to be done to de-terrnine the types of kepvords that are

270/ LR?S . 12(4) . TurnerandBrackbtll

most ef{'ectively used in HTML METAtags. In this research, rve relied upon ide-alized search situations because all theterms searched were known to be presentin the documents indexed. Further re-search is needed to determine whetherMETA tag data assists users in findingknorvn documents when a more diversegroup of search terms is used. Thernethod used here rnight have resulted inbetter retrieval rates than rnore realisticsearches would generate Not sulpris-ingly, this research also suggests that be-cause searches fbr abstract or generalterms retr ieve larqe result sets, the inclu-sion of a META tag rnight not have practi-cal significance in improving.the retrievalrank of a page. The inclusion o{'many ab-stract terms in a META tag might notraise retrieval rank.

It is clear. however. that the inclusionof specific terms in combination withmore abstract ones rvill have an imoact onretrieval rank when less general tei,r-tr u."searched. The question

-of the use{ulness

of abstract and specific kepvord terms re-quires more study. Finding new ways ofembedding significant metadata into doc-uments will enhance the exneriences ofboth content providers and users ofInternet search engines

Wonxs Crrrr

AltaVista Search Netrvork f 997, Tire METAtag: Controlling horv vour \\'eb page is in-dexed by AltaVista, Online Available:http ://urrrr altavista, digital com,/avlcontenVaddurl meta l.rtm

AltaVista Support 1998 E-mail message tothe autbors. 4 Jur-re

Babbie, Earl 1992 The practice of social re-search 6thed Belmont, Calif :\\/adsu'orthPublishing Co.

Bremser, \\'ayne 1997 Gain fame u'ith METAtags. Internet uorld 8,no 10: 94-96

Eftl'rimiadis, E{ihimis N , and Allyson Carlyle1997 Introduction to special section: Or-ganizing Internet resources: Metadata andtlre \\'eb Bulletin of the American Societyfor Information Science 24, no I: 4-S

Fly,nn, Peter 1998 Frequentlyasked questionsabout the Extensible Markup LanguageVersion I 3 Online. Available: htto://utrru t c i e / x m V 1 J u r r e

Furnas .G \ \1 ,T .K Landauer ,L M Comez,

and S T Dumais 1987 The vocabularyoroblem in human-svstem communication.bommunications o|ihe aCM 30: 964-71

Gibbons, Jean Dickinson. 1993 Nonarametricstatistics: An introcluction Quantitativeapplications in the social sciences, no 90Neu,bury Park: Sage Publications

Graham, Ian S 1995 The HTML sourcebook:Acomplete guide to HTML3 0 2d ed NervYork: John \\riley and Sons

Harman, D.K., ed 1993 The first Text Re-triexal Conference (TREC-I) Gaithers-burg, Md: U S Department of Com-merce, National Itrstitute ofStandards andTechnologr.

Kaplan, Carl S 1997. Cvber larv journal' Legalroadblocks starting to deter lvlETA-tag hi-jachng. Neu' York Times 16 Oct OnlineAvailable: htqr://search riJtimes com/library/cyberlauy'l0 l69Tlarvhtml

Kiess, Harold O 1989 Statistical concepts forthe behaxioral sciences Boston: Allyn andBacon

Kuhn, Heinricl.r C 1996 A proposal for struc-tured iudexing rvith tl.re meta-tag OnlineAvailable ; http://rnnv glvdg.de/-hkuhn1/vrnvcaV mtprop0l html

Leighton, H Vemon, and faideep Srivastava1997 Precision amons \\rorld \\ride \\'ebsearch services (search engines): AltaVista,Excite, Hotbot, Inlbseek, Lycos 16 June.Online Available: http://*ut'.u'inonamsus edu/l ibrary/rvehind* l ebindf .htm

Lide, David R f 995 Metadata: A descriptionLibraru hi tech 13. nos l/2: 33-34

r-ynch, cliflbrd 1998 The Dublin Core De-scriptive Metadata Program: Strategic im-olications for libraries and netrvorked in-formation access. ARL: A bimonthlynervsletter of research library issues ar.rdactions no 196: 5-10 Available: http://urrrv arl org/nes'sltr/196/dublin html

Melee s Indexing Coverage Ar.ralysis 1998. TheMICA report. 9 Aug. Online. Available:htto ://sqrr'. m elee com/mica./ir.tdex. html.

Millea-Erlc J 1995 Nehvork centric computrng:Issues of document descriptior.r in HTMLO nli r re. Available : ht q>'//sinr'.oclc org/oclcy'researclr/publicationVlevieu05/partl,4rtml. htm

Pfaffenberger, Brian 1995 Web search strate-gias Nerr'York: IvIIS

Plrillips, Jolrn L,, Jr 1992. HotD to think aboutstatistics. Revised ed Nerv York: \\'.HFreeman and Co

Pollock, Annabel, and Andreu'Hockleyr 1997\\'hat's u'rong s'ith Internet searching.D-Lib magazine. Marcli. Online. Avail-able, http://rrtt' dlib orgldlib/m arch97/bV03pollock html

Srinivasan, Padmini, Miguel E Ruiz, and \\hiLam. 1996 An investigation of indexing onthe \\1\1\l ln Global iomplexity : Inforiu-tion, chaos, and control: Proceedings ofthe59th ASIS annual meet ing, Bal t imire, l l , l .October 19-21, 1996, ed. Steve Hardir r :79-83. \\hite Plains, N Y: Knou'ledse In-dustry Publ icat ions.

Slanson, Don R. 1965. The evidence ulderlr.irrg the Cranfield results. Library quartcily35: I-20.

LRTS . 42(4) . Risingto the Top /271,

\\'eibel, Stuart 1995 Metadata: The Ibunda-tion of resource description D-Lib naga-;ine luly Online. Available: http:/Arsrrdlib org/dlib{uly95/0Trveibel html

- 1997 The Dublin Core: A simple con-tent descr ipt ion model for e lectr in i t . re-sources. Billettn of the American Society

for Information Science 24, no 1: 9-11

PreservationProtective EnclosuresPamphlet BindersMusic BindersBound Four Flap EnclosuresTan Archival BoardGreyMhite Archival BoardDrop Spine Archival BoxesAcademy FolderManuscript Folder3" Ring Binder Album

PO Box 1413. Des Moires. Iona 50305-1413pil. 80t,-526-5040

Gafl tor ,^eroi-nrqa COmpfete E-Mln. archiaal@tx netcom com

CatalOg wr,B snE http://nuu.archixal com

272/

Specificity, Syndetic Slrucfure,ond Subiect Access to Worksobout Individuol CorporoteBodies

Mory Dobney Wilson

The eaohttion of subject access to uorks about indiaidual corporate bodies inAnglo-American,subj ect cataloging 7made to the similar problems of uorkcomparable lecels of ,subiect access fo,Automation may hare been part of tmain.tenance routines, it offers the bemendations are macle that uou.ld restru.sing syncletic referen.ce stntcture anentru Ltpon ohich most subject accessare based

W*u, about individual corporate 1986, the pendulurn was reversed, leav-

bodies are not unlike biographies o? indi- ing works of individual biographywith th_evidual people. ln lhct, miny are just that: fuller subject access. In this paper, I will

historles treatinq the birth, iif'e, and death revierv the evolution of'subject catalogingof a particular institution or enterprise. practiceforworksaboutindividualcorpo-\\tith^regard to subject access, a catalog iate_ bodies and suggest changes that

.ts". wo,ild probably expect to find a simi-- would return par,ity to subject access for

lar Ievel of lreatrnent 16r works about in- the two classes of rvorksdividual corporate bodies as that providedfor biographies

,of individual persons. TuB ..yt l low pec's,, DTLEN'MA

Ho\{'ever. ln catalogs lollo\\'rng Practlcesestablished by the-Library of -ong."rs \\/hen assigning subject headings for. a(LC) that has not been the case in ilore work abouian individual corporate body,

than l0 years. For years the pendulum of without hesitation a cataloger will assign

subject access favored corpbrate bodies the name of the corporate body in the

ou"i ir-,clilrid,.,"ls. Though uiing quite dif- fbrm in which it is established fbr author

{erent approaches, a sort of e-quilibrium entry i.e., the {brm in the name authority^

began ti^emerge in the late 1970s. By record. For example, the cataloger of

Manv Dag^'uv \\rllsctr- (mdu'@tamu.edu) is Director of Cataloglng, Sterling C. Evans Library,

Texas A&M University, College Station. Manuscript received Maich 16, 1998; accepted for publi-

cation fune 18, 1998

LRTS . 42(1) . Specificity, Syndetic Stntctu,re, and Subject Access /273

Case Western Reserce: A History of theUn.iuersity, 1826-1976, will first assignthe heading Case Western ReserveUniversity-History. Thereafter, manycatalogers will instinctively think in termsof providing access to the work for thoseusers who want information about col-leges or universities in Ohio, but mightnot happen to know the narnes olthe in-di-vidual institutions about which a workmight exist in the particular database orcatalog. The cataloger reasons that suchusers would approach the catalog througha category or generic heading such as

ofsuch categories as banks, credit unions,dry cleaners, hospitals, libraries, schools,rnuseurns, etc. within the local dialine do-main. Thus i t might be reasonable to as-sume that a user would expect to find list-ings of books, etc. about individualcorporate bodies under a category head-ing in a librarys catalog. A user can findindividual biographies"of United Statespresidents under the category headingPresidents-United Stites, et en

section on subject headings for serials:

15 2 5 Assigning heaclings then a corpo-rate body is the ntb.ject Material about acorporate body often receives a lieading forthat corporate body alone

lfO 2 \\/.R Grace & Co.245 00 Annual report/$c\\iR Crace &Co.61020 \\iR Grace &Co. gxPeriodicals

Note, hos'ever, for annual reports ofcorpo-rate bodies that are responsible for certainactivities, it may be appropriate to assign

headings for both the individual corporate

body and the corresponding activity orIield For example, for the biennial report

of the Minnesota State Board of MedicalExaminers, the following headings couldbe assigned:

610 20 Minnesota State Board of Medi-cal Examiners $x Periodicals.

6500 Medicalpersonnel$xLicenses $zMinnesota $x Statistics $x Periodicals

Hou,ever, do not assign a general headingto reflect the type of corporate body s'hen

only one body is discussed. For example,

650 0 Health occupations licencingboard $zMinnesotat $x Periodicals

is not an appropriate heading for the bien-nial report of the Minnesota Board

Similarly, regarding subject headingsfor rvorks about individual coroorate bod-ies, Chan ( Igg4, 254) states: '

The name of the corporate body, as estab-lished accordin g to Anglo-American C ata-loguingRules is assignedas the subjectheading for a s'ork about an individual cor-porate body, even ifthe subject entry du-plicates the main entry or an added entryGeneric headings representing tlpes ofcorporate bodies are not assigned

These instructions are unequivocal; thecataloger does not assign a category or ge-neric subject heading in these situations.This principle dates back at least as f'ar ascutter (1904, 66-67) Until 1986, in thecase of u,orks about individual corporatebodies. the swrdetic structure o{'the cata-log supplied'&rectional assistance to theuser through a subject-to-name "See Also"re{erence fiom the broader category head-ing to the lndividually named instance.Since 1986, LC has ceased to make suchreferences and in fact prescribes the elirni-nation ofthem from older authority recordswhere some still reside. Therein lies thedilemma: the cataloger is prevented frommaking the category heading in the biblio-graphic record because ofthe principle ofspecific entry and is also prevented frornmaking the reference by current nationallevel instructions However. evidence

274/ LRTS . 42(4) . Wilson

abounds in records in the national utilitiesthat rnany catalogers elect to provide thebroader subject access by adding categorysubject headings, either in ignorance of, orperhaps in spite of, the prohibition againstthem.

Inconsistent, and perhaps confusingto catalogers, is the manner in which pre-ciselv the same oroblern is addressed ifthe work is about an individual Dersonrather than an individual cornorati bodv.I n LC subject cataloging praciice, sublectheading for a class of persons is assignedin the bibliographic record, a practiceclearly in opposition to the principle ofsoecific entrv. Horv did this variation inapproach come about?

HrsronY

Cutter (1904) recognized that therewould be a tendency to want to assignheadings at two levels o{'hierarchy. In hisdiscussion of rule 161 lbr specific entryhe noted that di{ficulty rvould arise whenthe public, accustomed to using a classedcatalog in his tirne, approached a dictio-nary catalog thinking of certain subjectsin connection with their including classes.He stated (67):

there is a temptatioD to enter certairbooks doubly, once under the specificheading to satisfy the rule, and once underthe class to satisfy the public The dictio-nary principle does not {brbid this If roomcan be spared, the cataloger mav put theless comprehensive rvorks also under theirrespective specific headings The objec-tion to this is that, if all the specifics arethus entered, tl-re bulk of the catalog isenormously increased; ar.rd that, if a selec-tion is made, it must depend ertirely upontl-re " judgement," i e, the prepossessionsar.rd accidental associations, of the cata-ioger, and tl.rere rvill be an end to all unifor-mity, and probably the public u'ill not bebetter satisfied, not understanding rvhythey do not find class-entry in all cases

For Cutter, guiding users {rom the cat-egory to the instance r'r,as the role ofs1'ndetic ref'erence structure in the cata-Iog. In rule 187 Cutter (1904, 79) stated:

Make references from general subjects to

their various subordinate subjects and also

to coordinate and illustrative subjects

Cross-references should be made by Full

from classes of persotts (Merchants, La'rv-

yers, Artists, Quakers, etc ) to individuals

belonging to those classes

Note that when Cutter used the term"by Full," he was rnerely indicating the

size of the catalog He stated (1904, f I ) :"to avoid the constant repetition of such

phrases as 'the

full catalog o{'a large Ii-

brary' and 'a

concise {inding-list', I shall

use the three words Short, Medium, and

Full as proper names. . . "

Haykin (1951, 16) echoed the s;mdetic

approach for individual biography, stating:

Such references to names of individuals

are indicated from headings designating

occupations, for example'

Architects, Britishsee also

\\/ren, Sir Christopher, 1632-17 23

Economists, Americausee also

Veblen, Thorstein, 1857-I929

\\ralker, Francis Amasa, 1840-1897

Painters, Frenchsee also

Bonheur, Rosa, 1822-I899Mat isse, Henr i , 1869-

Meissonier , fean Louis Ernest , 1815-f891

Haykin ( f951, f 7) was just as expl ic i t

regarding the recotnmended technique

foi *'.,.Ls about individual corporite

bodies '

A re{erence iiom the subject heading for a

particular kind of society or institution

should be made to the names of indlvidual

societies of that kind as a guide to such of

their publications as describe their pur-

poses, activities, history, and proceedings

This obviates the necessity of using the sub-ject heading designating the kind of society

or institution lbr every entry of this charac-

ter. It is exactly paraliel u'ith tlie use ofthe

reference from the name o{ the occupation

to the names of ir-rdividuals u'ho lbllorv that

LRTS . 42(4) . Specificity, Syndetic Stntcture, and Subjea Access /275

occupation Ibr the purpose of guiding thereader to autobiographical material

Medicine-Societiessee also

Academy of Medicine of CincinnatiColorado State Medical SocietyMedical \\bmen's Intemational Association . . .

For works about individual corporate bod-ies, the "See Also" re{'erence techniquewas employed fairly consistently by cata-logers at LC over a period ofyears. Therewere instructions (H 390, canceled in1985) in Subject CatalogingMnnual: Sttb-

ject Headings (SCM:SH i98+) that de-tailed LC procedures for cataloging a rvorkabout_ an individual corporate-body thatrvould result in the addltion of' sub-ject-to-name re{'erences to the existingname authority record for the body. InCataloging Sercice (CS 1975, 2&29), thepractice of mahng references lvheneverthe name of a corpbrate body was first as-signed as a subleci heading is described:

Such references are ofvalue since they rep-resent to the catalog user the key link be-ts'een the discipline of interest ar.rd thespecific organizations represented in tl.recatalog active in that discipline \\/itl.routthem pertinent information may be over-looked LC provides for sucli references fbrall corporate bodies except commercialfirms and geographicjurisdictions, e g ,

Societies International Kart Foundationxx Karting-Societies, etc

Otago Home Economics Associationrx Home ss6nsrni65-\srv Zealand-So-

cieties, etc

Agencies Manitoba \\/ater Seruices Boardxx \l/ater-s upply-Manitoba

Institutions Art Museum of South Texasxx Art-Corpus Christi, Tex -Galleries

and museumsCorpus Christi, Tex -Museums

Museums-Texas

Dom. Fulda. Gerxx Cathedrals-Germany, \\'estFulda, Ger -Churches

Frorn the examples given above, it isclear that only some of the rel'erenceswere from category headings. Many rveretopical subjects rnade into category head-ings with the addition of the subdivision"societ ies, etc.," and some represented afield ol'activity. It will also be noted thatthe {brrn of many would have been af-fected by the "city flip" tn which manysubjects {brmerly constructed as [geo-graphic name]-[topical subdivisionlwere f l ipped to I topical headingJ-lge,>-graphic subdivision] {brms. Examples ofclass of corporate body headings abovervould be Museums-Texas and Cathe-drals-Germany, \\/est

INolvtouel BrocRApHy

In 1976, LC announced changes in prac-tice related to biography that included as-signing an a:rray of appropriate topicalsubiert headings to a wbrk ol individualbiography, one of which was {br the classof persons (CS 1976). Before 1976, i t isapparent that sorne, if not most, of'Haykins class oI'persons See Also ref'er-ences were not made in LCk manual {ile.El-Hoshy (1998) indicated that only onthe manual authority card lbr Christo-pher \\7ren was "xx Architects-GreatBritain" pencilled in. The other Haykinreferences did not appear on LCt manualauthority cards fbr the other narnes heused as examples, and so it is clear thatHaykin's reco;nmendations were not fie-quently lbllowed In 1976, LC rational-ized its new practice by saing that up tothat tirne. tonical information in individ-ual biographies had been ignored andthat by changing the practice, i t was en-abhng a userlo ietrieve by form Ibiogra-phy] "in his particular field oI'interest.""Field or discipline" would be manifestedby a heading of the type: [class of per-sonsl-[place]-Biography. Later, LCprovided extensive nerv instructions tolhe basic provisions above to includeheadings f6r events or \r,ars and some-times elhnic and gender afftliations (CSB

276/ LRTS . 42(4) . Wilson

1983). \\/ith regard to the class ofpersonsheadins, SCM:SH (1996-, H1330, 2) in-cludes instructions that now state:

Note: This headlng is assigned to individ-ual biographies primarily for the berefit ofpublic library users n,ho are seeking biog-raphies ofa particular type ofperson ratherthan a particular individual The headingshould be selected u'ith that in mind. If thebiographee belongs to no discemable classof persons of the type judged likely to besought by the typical public library user, itmay be omitted

It is not immediately clear rvhat it isabout the technique ofassigning a subjectheading for the class ofperson to each bi-ography ol Abraham Lincoln that is inher-ently superior to the practice of making asinqle See Also reference from Presi-dents-United States to Lincoln.Abraham, 1809-1865 on Lincolntnarne authority record. The answer, how-ever, lies in catalog maintenance issuesfrom the card catalog era to today's auto-mated library systems The general-to-specific or subject-to-name referencetechnique only works when the refer-ences are made and maintained in cata-logs. Pahner (f986, 7f) noted in a study oI'subject heading practices in card catalogsthat many libraries never employed SeeAlso references: "not even the larsest li-braries were able to provide the'Sei Also're{'erences upon which LC assignment ofsubject headings is based." Thus a majorcomponent of the subject access systemwas ornitted in actual practice, throwingthe eflicacy of the entire systern intoquestion.

I{ libraries were unable to supply SeeAlso references explicitly listed in the Li-brary of Congress Su.bject Headings(LCSH), then theywere even less likely toprovide the subject-to-name reI'erences,because only institutions purchasing orreceiving depository LC authority cardsrvould even have had knowledge of theirexistence Furtherrnore, the need forsuch ref'erences is institution-soecific. de-pending solely on the local col l ict ions andthe occurrence therein ofworks about in-dividual persons or corporate bodies

Only the rnost rneticulous librarians everprovided cornplete arrays ofeven the ex-nlicitlv listed See Also references rvithinihei.

"^t"logr. They rvere much less likely

to make and maintain ones that theYwould have had to create and for whichonly the sketchiest guidelines were avail-able Instead, and as a substitute forsyndetic structure in the catalog itself, thevolumes of LCSH were placed strategi-cally near the catalog fbr the benefit oftheusers and the librarians who helped themnavigate subject terrninology. Until auto-mation. svndetic structure never hadnruch ol 'a'chance to perform.

AurouettoN: THE PRoBLEM

Factors contributing to the lower visibil-ity and utilization of subject-to-namerel'erences rvere and are related to auto-mation. The first records coded and dis-tributed in electronic form were biblio-graphic records, not authority records. A.r." l i - inatw edit ion of Authori l ies: AllenC Format rvas published in 1976,u'ith the first editio;appearing in 1981(USMARC Format for Au.thority Dataf993-). The first authority records be-gan to be distr ibuted through USMARCdistr ibution tapes in 1984 (CSB f983).\\'hile the ability to handle authority re-cords and ref'erence structures in sys-tems has improved, further refinementsare st i l l needed, especial ly in the area ofautornated catalos maintenance. Underpressure to update terminologY inLCSH, LC was f'aced with a situationwhere the l i rnitat ions of i ts own systeminten,ened rvith regard to sub-ject-to-name ref 'erences. Subject head-ings residing in narne authority recordscould not be rnaintained. In October1986, the Subject Cataloging Division atLC announced it would cease makingsubiect-to-narne references in natne au-thoiity records {br corporate bodies.Sorne tirne later LC instructed NameAuthorities Cooperative (NACO) partic-ipants to rerl love sub ject-to-name refer-ences u' lrenever the authority record wastouched {br any other reason This prac-tice is still in e{lect and is recorded in theNACO Participant's Manual (NPM 1996,

LRTS ' 42(4) ' Specificity, Syndetic Stnrcttt're, and Subject Access /277

0 0 1 n 8 5 - 4 2 0 90 1 0 n 8 5 0 0 4 2 0 90 3 5 ( O C o L C ) o c a 0 1 6 4 3 9 2 6040 DLC DLC AzTeS1 1 0 C o l l e q e o f I d a h o ( C a l d w e l l , I d a h o ) { A A C R 2 }4 1 0 C a l d w e l l , I d a h o . C o f l e g e o f I d a h o { o l d c a t a l o g h e a d i n g } { d o

no t make )5 1 0 A f b e r t s o n C o l l e g e o f I d a h o { l a t e r h e a d i n g i550 Un i ve rs i t i e s and co l l eges - I daho6 1 0 I t s A l u m n i d i r e c t o r y , 1 9 8 5 : t . p . ( C o l l e g e o f I d a h o ,

C a l d w e l l . I d a h o )6 7 0 L C d a t a b a s e , L 2 - 2 7 - 8 5 ( h d g . : C a l d w e l l , I d a h o . C o l l e q e o f

loano I67A Pe rsona l commun i ca t . i on f r om A lbe r t son Co l l ege l i b ra r y ,

3 /20 /96 (name changed f r om Co l l ege o f l daho t o A lbe r t sonC o l l e g e o f l d a h o i n 1 9 9 1 )

6 1 5 A l u m n i d i r e c t o r y , 1 - 9 9 5 : t . p . ( A l b e r r s o n C o l l e q e o f I d a h o )

Figure 1. National Level Authority Recoril Containing a Subject-to-Name Reference

81). The author did not find any solicita-tion {br public comment on this issue, norwas ther; any announcement of the cessa-tion of subiect-to-name references inCSB An example that currently remainsin the national authority file can be seenin figure 1. The rnaintenance problem oc-curs r.vhen a change is needed in subjectheading terminology (in this example, theheading "Universities and colleges") andthe system neither detects nor corrects amatching subject string when it resides inan authority record rather than a biblio-graphic record.

Gnunner "SEE ALSo" REFERENCESAND THEIR PNONI.NTUS

Using general See Also references is an-other sanctioned technique to guide theuser liom a category heading to u'orks inthe catalos about members of the cate-gory Haykin (1951, 15) contrasts the useof specific See Also references with gen-eral See Also references:

Frequently, horvever, u'hen the headings re-ferred to are obviously indiridud membersofa single class or category the reference ismade not to dre individual members but tothe class, and several members are added byu'ay o{ example. . . It may rvell be arguedthat general references defeat the sl,ndeticaspect of an alphabetic subject catalog inthat the specific subject headings to rvhichthe reference should lead are not allnamed, but are represented merely by an

example It is, hos,ever, unlikely that anybut the occasional reader u'ould seek allthe material in the library covered by allthe speci{ic headings comprehended bythe broad one from u'hich the general ref-erence is made Ifhe should, he u'ould bewiser to seek his material through bibliog-raphies, or to reler to systematic treatiseson tl.re broad subject for the topics fors'hich there u'ould be headings in the cata-log should the library possess separatervorks on them The purpose ofthe generalreference is primarily suggestive

In his logic for the efficacy ofgeneralSee Also references. Havkin doubts thatusers will want all occu;rences, when itseems much rnore likely that a usersearching a category heading will notknow specific headings to search and in-stead needs to see a listinE from which tomake a selection The weakness ol 'Haykin's logic can be illustrated by apply-ing it to an entry that might appear in yel-lo\ / page telephone directories. Most li-brarians would agree that an entry under"Banks" that advised users to "search

u'hite page listings under: First NationalBank, etc." would not supply a satisf'ac-tory level of guidance to users.

There is some evidence in SCM:SHthat, for topical subjects, general See Alsoreferences for generic class-to-classmember (e.g., "Tools See Also Axes, Fi lesand rasps, Harnmers . ") might be fad-ine. In H 37I (SCM: SH 1996-, H37r, 1)the followlng is stated:

278/ LR?S . 42(4) . Wilson

0 1 0 s h 9 3 0 0 2 5 1 8040 DLC Sc DLC0 0 5 1 9 9 3 0 5 2 1 L I A 1 2 6 . I151 0 Eu ro D i sney land (Ma rne - l a -Va l l ee , F rance )45L 0 Eu ro D i sney (Ma rne - I a -Va l l ee , F rance )45 I 0 Eu ro D i sney Reso r t (Ma rne -1a -Va l l ee , F rance )451 0 Eu rod i sney land (Ma rne - l a -Va l l ee , F rance )550 0 Amusemen t pa rks Sz F rance gw g

Figure 2. Authority Record for Corporate-Like Entity s'ith Rel'erence Iiom Broader Term

Tl.re practice of making neu're{'erences ofthis type has norv largely been abandonedin favor of making a specific referencefrom a broader heading u'henever a nervheading is established Existing generalSee Also references of this t;pe are beingretained in the subject authority file untilall individual headings that I'rad formerlybeen covered by the general reference areactually linked to the broader heading byBT/NT relerences

This irnplies a move toward exhaustivelistings for topical subject class-to-classmember leadlngs. However, H 371 goeson to say that there are other categories ofgeneral See Also references that have6een made and may still be made. Amongthese types ofgeneral See Also referencesare those to categories or tn)es of nameheadings, such as:

Church buildingsSA names ofindividual churches

In various SCM:SH instructions, thereare other specific provisions for generalSee Also references (indicated inUSMARC as 360 {ields in the authority re-cord for the category) that at best can onlysuggest a type of'heading that rnight beavailable to the user, but still require theuser to know individual narnes, such as:

Universities and collegesSA names of individual institutions

HospitalsSA names of irdividual hosnitals

Railroads

SA names of individual railroads

Concentration campsSA lames ofindividual concentration camos

MuseumsSA names of individual rnuseurns

PRoBLEMS AssocrATED wrrH THEDrvlsroN OF THE WORLD

In attempting to assign responsibility fbrestablishing categories of arnbiguous en-tities, LC has engaged in what is com-monly known as "the Division of the\\'orld." SCM:SH H405 provides two listsof entities: those established in the nameauthority file and governed by descriptivecataloging conventions, and those estab-lished in the subject authority ftle andgoverned by subject cataloging conven-tions. Some of these ambiguous entities,when established by subject catalogers.will have a broader term reference oro-vided fbr the category ol'entity to the ipe-cifically named entity (unless it falls un-der the provisions for assigninent of ageneral See Also as prescribed by LCabove) (see figure 2)

\4/hen an entity is moved from the sub-ject file to the names file, as has recentlybeen the case rvith concentration campsand arboreturns, an un{brtunate phenom-enon takes nlace. The broader term refer-ence is immediately expunged. Again, thisis because NACO suidelines fbr establish-ing names {brbid the use of subject-to-name ref'erences. Individual concen-tration camps provide a case in point.\\then concentration carnps were estab-lished by subject cataloger-s as geographicnames, a broader term reference fromWorld War. 1939-1945-Concentra-tion camps-[place] to the individualcamp was generally rnade. The result in acatalog that displayed See Also referenceswas a listing of individual carnps similar tothat found in fipure 3. As these entities mi-grated to the name authority file, the

LRTS ' 42(4) ' Specificity, Syndetic Stnrcture, and Su'bject Access /279

Wor ld War , 1939 -1945 -concen t ra t i on camps -GermanySea rch a l so

B e r g e n - B e l s e n ( G e r m a n y : C o n c e n t r a t i o n c a n p )Buchenwa ld (Ge rmany : ConcenL ra r i on camp)n r - h . ' , / C o r m r n r r . c ^ n - o n r r : f i ^ n ^ ^ m h \

Ravensbruck (Germany : Concentrat ion camp)

Figure 3. Typical Re{erence Array Before Subject Headings Became Name Headings.

broader term references were eliminated.A sirnilar phenornenon took place

when individual computer software pro-grams migrated from the subjects to thenarnes file as uniforrn titles. \\/hen estab-lished according to subject catalogingprinciples, general-to-specific referencesu'ere made (see figure 4). After the move{rorn the subiects file to the names file fbrindividual nimed entities, a user accus-torned to finding an exhaustive listing isrequired to know the names of specificconcentration camps or word processingso{'tware in order to find individual worksabout them

AUTOMATION: THE SOLUTTON

The onlyreason offered by LC for the ces-sation of subject-to-narne references wasthat they had not been rnaintained, andthe references continue to be removedlbr the same reason. One of the specificmaintenance issues at LC has been that itsautornated systern is unable to perfonnglobal changes Thus, when a headingneeds to be changed, as is frequently thecase to meet the demand for uDdated ter-minology or the recommen^dations ol1991 Subject Subdivisions Conference,records must be updated manually. Cou-ple this manual editing with the bifurca-tion of authorityfiles betrveen names andsubjects and the division ofpersonnel be-tween descriptive and subject cataloging,and it is easy to see why maintenance issuch an issue.

Maintenance in a rnanual environ-ment is a signilicant and costly proposi-

tion, but in an autornated environment it

should be much less of an issue, provided

that syster-ns include the necessary main-

tenance routines. Programming based on

existing USMARC formats and data ele-

ments could and should elirninate the

drudgery involved in rnanual mainte-

nancJ oi 'headings in both bib l iographic

and authority records.

Many integrated library systems al-

ready have in place some level o[ valida-

tion and error detection routines. Some

of the necessary features needed to re-

store subiect-to-name references as an el-

ement of spldetic structure of the catalog

and to eliminate almost all rnanual main-

tenance include the capability:

. To generate a reference from a name

authority record 550 to display in the

subject index (if separate) rvhen the

550 has a control subfield rv that is

coded as a broader term, valid fbr

subject reference structure only("gbnn") ;

. To detect that such a ref'erence is

valid if the name heading to s'hich it

points (autliority lxx) has been used

as a subject (bibliographic 6xx), u'ith

or u'itlrout further subdivision;o To set automatically tl-re values in the

control sublield u'of the authority550 for nondisplay-that is, broader

term, valid for subject reference

structure only, do not display("gbna") if the authority heading has

not been used as a bibliographic sub-ject heading (u'ith or s'ithout lurther

subdivisior):

! { o rd -p rocess i ng -Compu te r p rog r amsSea rch a f so

M a c w r i r e ( C o m p u t e r p r o g r a n )I ^ l ^ r d D 6 , f o - f / C ^ m n , , t s 6 r h r ^ ^ r ^ m l

Words ta r (Compu te r o roq ram)

Figure 4. Typical Reference Array Before Subject Headings Became Subject Uniform Titles

280/ LRTS . 42(4) . Wilson

o To set automatically the controlsubfield rv in the authority 550 tobroader term valid for subject refer-ence structure, display ("gbnn") lfthelxx in an authority record u'ith a sup-pressed 550 (control sublield s,"gbna") is used as a subject heading(witli oru'ithout further subdivision );

o To validate the use ofthe main head-ing as a subject in the local file and toset automatically the control subfieldrv to the appropriate value u'hen ar.rauthority record enters the local sys-tem for the first time. if a 550 is nres-ent as ment ioned above;

o To change globally a target subjectstring rvhen it occurs in authority re-cords 550s (specifically in name au-thority 550s) as s'ell as u'hen it occursin bibliographic records This routir-reis needed in general subject authorityrvork to support the updating of topi-cal subject reference structures Cau-tion u'ould be needed not to initiate aglobal change v'hen an old subjectterm is replaced by trvo or more newterms

. To detect and report (not change)strings in name authority Sxx re{er-ences that match on 4xx strings in sub-ject authorities. Tlris routine urculdtake care of the problem of updatingterminology across Iiles u,l.ren subjectheadings are changed Previousterms u'ould be recorded as 4xx refer-ences, but because the change is {ie-quently for a heading split, the termin the name authority 5xx u'ill need tobe reported {br human manipulatior.r

The routines described above wouldenable the subiect-to-name referencetechnique to be applied to works of individ-ual biography as well.

It should be recosnized that even ifmaintenance of subiect-to-name re{'er-ences can be automated, there rvill still besorne cost associated with the intellectualeflbrt ofdevising the references. Horvever,there is a similar level of intellectual rvork todetermine the class of persons subjecthea&ng for a work of individual biography.Using class of'corporate body or class ofpersons references would offer a savings in

e{Ibrt over in&vidually assigned suQectheadings, the current practice with individ-ual biography. It would be minimally morecostly to provide a comparable level of ac-cess {br works about individual comoratebo&es, and more importantly, not provid-ing any level of subject access apart fiomthe name itself is patently inconsistent

CoNcLUsroN AND RECOMMENDATTONS

The history ofsubject access in relation toworks about individual corporate bodiesand also to works ofindividual biographyhas been outlined. The principle of spe-cific entry upon which subject access indictionary catalogs is based requires asubject heading for the named entity andnot one for the generic category to whichthat entity belongs. Vl/ith other types ofsubjects, syndetic structure provides thedirectional cues needed to move fromcategory to instance. In the case of indi-vidual biography, the principle ofspeci{icentry is violated by assigning subjectheadings for the class of persons, but atleast a form of category access is beingprovided. In the case ofworks about indi-vidual corporate bodies, those cues havebeen eliminated, leaving a void in subjectaccess. The only sanctioned remedy, theuse of qeneral See Also references, hasbeen shown to be weak, being merely sug-gestive of possible terms for searching

To restore comparable levels of ac-cess, there are threi choices:

l. Follow the path used {br individualbiography and assign a class ofcorpo-rate body subject heading to worksabout individual corporate bodies.

2. Leave individual biography treat-ment as i t is, but restore sub-iect-to-name references on nalne au-thority records when rvorks aboutindividual corporate bodies occur inthe catalog.

3. Use subject-to-name references forboth and cease assigning class oI'per-sons subject headings {br individualbiography.

The path most consistent u'ith princi-ples would be the third option. At the least,

LRTS . 42(4) . Specificity, Syndetic Stnrctu.re, and Su,bjoct Access /281

it is recornrnended that the previous prac-tice of generating subject-to-narne refer-ences when a corporate body is used as asubject be reinstituted The specific guide-lines in the canceled SCM:SH H390 arernore developed than those described inCataloging Sen ice (1975) and in Chan(1986, 107-108), and those H390 suide-lines could be resurrected. The only-provi-sions that should be reconsidered ar"ihor"that call for addition of subiect-to-namereferences lbr headines other than class ofcorporate body Specifically, those H390guidelines that call fbr subject-to-namereferences that indicate the field ofactivityfor {irms, banks, corporations, govem-ment agencies, etc should be reconsid-ered according to the guidelines enunci-ated in Conser Calnloging Manrral. Thoseguidelines state that field oI activity, if ap-propriate to the work, should be assignedas a subject hea&ng to the individual re-cord rather than as a sublect-to-name re{'-erence. The issue of the specificity of geo-gaphic subdivision (to the first ordergeographic name ortothe local level) shouldbe opened for discussion.

For NACO contribution and editinq,the subiect-to-name ref 'erence should biallowed rvhen a participant identifies aneed lbr the relerence in a local file. Inediting older records that have sub-ject-to--name re{'erences, the NACO par-ticipant should update the heading to cur-rent subject cataloging terrninologz andconstruction. This practice would be sim-ilar to that required u'hen other 4xx and5xx rel'erences need to be evaluated.

LC is on the brink of selecting a new in-tegrated library system. making this an aus-picious time to restore these rel'erencesand to consider their applicability to indi-vidual biography. An autornated or an auto-mated-assisted solution to the loss of sub-ject-to-name relerence structure over theIast hvelve years should also be explored.

Woms CrrEo

Cataloging Sen:ice (CS) 1975 Bulletin 115(Fall): 2&29

Cataloging Seraice (CS) 1976. Bulletin 119(Fall): 19-22

C ataloging Seraice Bulletin. 1983, no 22: 5947Chan, Lois Mai 1986 Library of Congress

subject headings: Principles and applica-tion.2ded Littleton, Colo : Libraries Un-Iimited, Inc

Chan, Lois Mai 1995 Library of Congresssubject headings: Principles and applica-tion 3d ed. Engleu'ood, Colo : LibrariesUnlimited, Inc

Conser catalogingmanual 1993-, ed. Jean LHirons. \\'ashington, D C: Serial RecordDivision, Library of Congress, distributedby Cataloging Distribution Service

Cutter, Charles Ammi. 1904. Rules for a clic-tionaru catalog \\/ashington, D C: U SGovernment Printing Office

El-Hoshy, Lynn 1998 Personal correspon-dence, LC, Cataloging Policy and SupportOffice. Feb. 6

Haykin, David fudson I95I Subject head-ings, a practical guide \\lashington, D.C :U.S. Government Printing Office

NACO participants' manual 1996 2d ed.\\/ashington, D C: Library of Congress,Cataloging Distribution Seruice

Subject cataloging manu.al: Subject heaclings1984 Preliminary ed Subject CatalogingDivision, Processing Services \\hshing-ton, D C: L ibraryofCongress

Subject cataloging manu.al: Subject headings1996- Cataloging Policy and Support Of-fice, Library of Cotrgress \\'ashington,D C : Cataloging Distribution Service.

Palmer, Joseph \\i 1986. Subject authoritycontrol and syndetic structure-Myth andrealities C ataloging Cr classification qu.ar-terlu 16:71-95

USMARC fomnt for authority tlata 1993-Prepared by Nehvork Development andMARC Standards O{Iice. 1993 ed. \\rash-ington, D.C.: Cataloging Distribution Ser-vice, Library of Congress

282/

The lmpoct of Subiect HeodingAssignment on Circulotion ofDisserlolions ol Virginio Tech

Richord E. Sopon-While ond Mory Honsbrough

Subject headings for bibltographic records for dissertations are no lon.ger as-signetl at some academiclibrnries,but the impactthis mighthaae on disserta-tion u.se has not been eualuated In this study, btbliographlc and circulntionrecords for a sample of 248 academic dissertations u;ere examined to deter-mine the ffict of controlled subject headings in the records on. circulation ofthe items. Titles uith LC subject headings Loere compared to those uithou.tChi-square analysis shouecl significant differences in circu.lationforthe totalsample (p=.04),but not for inditidu.al areas of study. Discttssion of sou.rces ofbias and su,ggestions for future research are included

I-lissertations represent a unique andimoortant {brm of research literature. OI-ten, they are the first peer-reviewed rvrit-ings of new researchers and represent theinitial appearance of fresh ideas and dis-coveries in many disciplines. In theUnited States, over 41,000 dissertationswere written during the 1995-1996 aca-demic year, representing degreesawarded at over 750 institutions (UMI1997). Finding dissertat ions on.a giventopic can be challenging because usuallythey are held only by the library of the in-stitution where the dissertation was writ-ten. One way to locate this type ol re-search is through subject searching.

Subiect access to theses and disserta-tions g"enerally takes three forms: pub-lished lists, various products published by

UMI, and bibliographic records in librarycatalogs and bibliographic utilities. Listsof notable dissertations in a particularfield are sometimes published by prof'es-sional societies or universitv denartments.These lists are sornetirnes indexed by stan-dard indexing and abstracting services.

UMI actively microfilms theses anddissertations for subsequent publishingon demand. It publishes its database in avariety ofproducts, such as Dissertation.sFirst, Theses Abstracts, and DissertationAbstracts, so that researchers can locateand purchase micro{ilm or print copies ofthe papers. Subject access to the databaseis limited as titles are only characterizedbv subiect in the broadest terms InUMIt electronic products, it is possibleto search by keyword.

Rlcueno E. Sepor--\\rHtre (Richard Sapon-\\'hite@orst.edu) is Catalog Librarian at the ValleyLibrary, Oregor.r State University, Coruallis, Oregon Meny Har*snnoucH (maryhans@vt edu) isScience & Technology Cataloger at Virginia Tech University Libraries, Blacksburg, Virginia Theauthors u'ould like to thank CliffPereira for advice on statistical analysis, and Paul Metz, Janet\\/ebster, and Debbie Hacklemal for comments on an early draft of the article Manuscript re-ceived February 12, 1998; accepted for publication fune 9, 1998

LRTS . 42(4) . The Im.pact

Librarians in most academic librariescatalog their local dissertations and con-trlbute the bibliographic records to an on-line bibliographic utilrty, such as theOCLC Online Computer Library Center,Inc. Online Union Catalog or the Re-search Libraries Group Research Li-braries Inlbrmation Network (RLIN)The degree to which subject access is pro-vided to this form ofliterature varies fiomlibrary to library. Some assign subjectheadings and classify by subject whileothers assign no subject headings andplace all theses and dissertations in a sin-gle class number for the institution.

In recent vears. sorne academic librar-ies (e g , University of Florida and UCLA)have chosen to cease assigning subjectheadings to some theses and disserta-tions, often as a means of controlling cata-loging costs. Another lactor in thJdeci-sion might have been that subject accesswas thought to be sufficiently providedvia kelnvord searching in the local catalogor by the use of Disierroiion Abstracti.Howe.,er, keyword searching is qualita-tively different from subject searching us-ing a controlled vocabulary. In addition,the UMI databases can only benefit thosesearching specilically [or theses or disser-tations; they do nothing for online catalogsearchers seeking subject-speci{ic infor-mation owned by that particular library.

The impact -on

utl" of dissertationscaused by tle decision to suspend subjectheading assignment remains largelyunevaluated. ln this paper, we examinethe impact on use as evidenced by circula-tion counts following the implementationof this decision at Virginia Tech.

Llrnnetunr RBvrBw

There is a dearth ofliterature on the cata-loging of theses and dissertations. Muchof what does exist is out-of'-date and inneed of'being replicated in an environ-ment of online bibliographic databasesand library catalogs. In many of thesestudies, authors describe how acadernicIibraries catalog theses and dissertations,including how Jubject analysis is handled(Patterson, \\/hite, and \\'hittaker 1977;Harris and Huffman 1985; Ryans 1991;

of Subject Heading Assignrnent /283

Khurshid 1995). Others investigate howresearchers locate dissertat ions (Reppand Glaviano 1987; Lee-Smeltzer andHackleman 1995) andthe ease of search-ing for dissertations in bibliographicdatabases (Perry and Salisbury 1995), orevaluate how kepvord searching com-pares with controlled vocabulary search-ing (Keller 1992). Only in two studies doauthors evaluate the impact of subjectheading assignment on the circulation ofdissertations (Sullivan et al. 1992;Sapon-$/hite 1997).

Patterson, \\/hite, and \\/hittaker(1977) surveyed librarians at 90 univer-sity libraries about their local practicesfor binding, cataloging, classifying, andstoring theses and dissertations. Theyfbund that 76Vo of respondents assignedsubject headings to all dissertations,whrl,e lTVa did not assign any subjectheadings The remaining respondents as-signed only broad subject categories,used locally devised headings, or assignedsubject headings on a selective basis. Theauthors also recorded the statement frornone respondent that "subject analysis istoo costly for the cataloging departmentand a rvaste of tirne in relation to theamount of use." This unsubstantiatedstatement was subsequently cited by laterauthors (Harris and Huffrnan 1985;Khurshid lg95). It contrasts with the hiehpercentage of'libraries reported to be as-signing subject headings to dissertationsin these and later studies.

Harris and Huffman (1985) reportedon a survey of practices at 84 academiclibraries concerning thesis and disserta-tion cataloging practices They found that897o assigned subject headings, whilel07o did not Two institutions surveyed,the University of Kansas and the Univer-sity of New Orleans, had previously per-formed subiect analvsis for their disserta-tions but hid discontinued the practice.Harris and HufTman sneculated that as li-braries switched from card cataloss to on-line catalogs with keyrvord searching ca-pabilities, the long descriptive titles ofdissertations would obviate the need forsubject analysis

Ryans (1991) surveyed Association ofResearch Libraries (AitL) libraries in an

284/ LRTS . 42(1) . Sapon-White and Hansbrough

attempt to replicate the work of Harrisand Huf{inan ( 1985) She found that 927oassigned Library of'Congress (LC) sub-ject headings, u'ith some respondents alsousing in-house subject headings.

Khurshid ( 1995) noted that at his insti-tution, the decision rvas made to assignonly broad LC subject headings followedby a local fonn subdivision "Theses anddissertations." At his universitv. the Ibrmsubject heading "Dissertations, Aca-demic" subdivided by subject was also as-signed. He stated that kepvord access tothe titles and abstracts oftheses and dis-sertations cornpensated lbr the lack ofmore specific subject headings.

Repp and Glaviano (1987) studiedhorv and rvhy extramural researchers(those outside the institution where a dis-sertation is produced) access locally pro-duced dissertations. They concluded thatthese scholars usually seirch for disserta-tions by subject, though their searchterms were o{len the broad subject cate-gories used by Dissertation Abstracts andrelated products

Lee-Smeltzer and Hackleman (1995)surveyed library users, asking those whohad checked out a thesis or dissertationabout the strategy used to search the on-Iine catalog The'authors concluded thatlibrary useis pre{'erred keyr,vord search-ing in subject fields over other catalogsearching methods, although kepvordsearching in author fields and exact au-thor searches were also freouentlv used.

In an interesting study by Perry andSalisbury (1995), OCLC's \\brldCat data-base and RLIN's Dissertation AbstractsInternational (DAI) database were com-pared for ease ofretrieval ofbibliographicrecords for dissertations Though they fo-cused primari ly on ease of searchingandthe availability of abstracts, they also com-pared the databases for coverage of dis-sertations. They pointed out that thereu'ere trvice as many records for theses anddissertations in\\zorldCat as there were inDAI (2 6 rnillion versus 1.3 million) andthe \\'/orldCat records appeared muchsooner than those in DAI. Moreover,while coverage in the trvo databases didoverlap, coveiage rvas by no means identi-cal True subject searching in DAI, via a

controlled vocabulary, r'vas lirnited by thebroad subject terms used, rvhereas manyofthe records in \\/orldCat were assignedLC subject headings

Keller (1992) looked at how rvell thewords in subject headings matched wordsin the titles of theses. Her purpose was tosee whether keFvord searching could re-place subject searching u'ith a controlledvocabularyin an online environment. Shefound that matchingwas not high enoughto warrant abandoning subject analysis.

Sullivan and others (1992) were one ofthe few groups of researchers looking atthe ef-fect of subject headings on the cir-culation of dissertations They looked atcirculation records fbr the first two yearsof shelf li{'e for dissertations, notinqrvhether or not the titles had circulated.They fbund that those dissertations thathad been assigned subject headings weremore likely to have circulated than thosethat did not have subject headings. Intheir study, it was impossible to tell howmany times an item had circulated

Sirnilarly, Sapon-\\'hite (1997) com-oared circulation counts for individual ti-lles of 1993 theses at Oregon State Uni-versity belbre and alier the assignrnent ofsubject headings. Circulation counts ofti-tles increased 30Eo following subjectheading assignment. Much of the in-crease appeared to be due to a small num-ber o{'titles that circulated many tirnes af'-ter the assignment ol subject headings.

BACKGROUND

Virginia Tech in Blacksburg is Virginia'sland-grant university and its leading re-search institution. Its research expendi-tures place i t in the top 50 inst i tut i ,ons inthe nation. Virsinia Tech has over 200 de-gree programsind nearly 25,000 studentson campus, including approxirnately4,200 graduate students.

The university libraries are a memberof ARL and a selective depository for fed-eral documents. The libraries hold over1.9 million printed volumes, 17,000 serialsubscriptions, 5 million rnicroforms,130,000 audiovisual and machine read-able pieces, and 120,000 rnaps. In addi-tion to the rnain library the university

LRTS . 42(4) . The lmpact of Subject Heading Assignment /285

libraries include fbur branch libraries: Artand Architecture, Geosciences, Veteri-nary Medicine, and the Northern VirginiaGraduate Center

The university libraries have openstacks and use VTLS as the online catalogsystern VTLS is searchable by keyrvord inalmost all text fields, including author, ti-tle. series, and all sub.lect hea-ding flelds,as well as several note fields (lncludingthesis and surnmary notes) by perltrrmin[a "rv/' (word) or "b/" (Boolean) search.Keyr'vord searching can also be restrictedto the author, title, or subject headingfields if desired. A subject search, "s/,"will retrieve only records rvith LC subjectheadings in the MARC 650 {ield

METHoD

\\'hile Sullivan et al (1992) shorved thattitles with subject headings were morelikely to circulate than those without sub-ject headings, they did not take into con-sideration the number of tin-res a title cir-culated To take an extreme example, it isconceivable that sorne titles u'ithout sub-ject headings circulated many t imes,while titles with subject headings circu-lated infrequently In order to get a com-plete picture of hor.v subject headings a{-fect circulation, one must consider thenumber of times a given title is checkedout of the library T-he present study wasdesisned with this in mind

T-he university libraries receive twocooies of each dissertation The first ishoused in the Special Collections Depart-ment and does not circulate The secondis placed in the general stacks and doescirculate. In some cases, a third, circulat-ing copy is deposited in one ofthe branchlibraries. Dissertations are classed to-gether in a single call number and areorganized by year, then alphabetically byauthor's last name. Basic descriotive cata-loging is per{brrned by a student assistantusins an OCLC constant data fbrrn. In thepast, a catalog librarian then assigned LCsubject headings to each dissertation. A{-ter the records are revierved and edlted bya catalog librarian, they are updated inOCLC and exnorted to VTLS.

Prior to July 26, 1995, LC subject head-

ings were assigned to bibliographic re-cords for all Virsinia Tech dissertationsDissertations cataloged a{ter July 25,1995, received no LC subject headings,but sorne were assigned uncontrolled sub-ject terms, suggested by the author ofthedissertation, in MARC {ield 653. Becausethe WLS system indexes both title rvordsand field 653 words fbr general keywordsearching, author-supplied terms thatmatched title words rvere considered re-dundant and lvere ornitted from the re-cord during cataloging

This study rvas designed to determinethe effect of this change in su\ect head-ing treatment on dissertation circulationA program was run against the library da-tabase on October 21, 1996, to identi{y allcopies oI'dissertations that entered intothe svstem between Mav I and October3I, 1095, thus collecting 12 to 18 monthsworth of circulation data. Entry date u'asbased on the day that the itern's barcodewas entered into VTLS. This correspondsclosely to the date the itern becomes avail-able to the public. After being barcoded,the item is labeled and sent to the shelv-ing unit to be placed in the stacks, a pro-cess that usually takes only a day or two.The noncirculating Special Collectionscopies were excluded Iiorn the study.

The output of the program listed au-thor, title, call nurnber, barcode number,and coov number fbr each dissertationthat Iit'ihe study criteria, as u,ell as thenumber of tirnes each item had circu-lated The circulation count included thenumber of times the item circulated fromthe tirne it entered the library's databaseuntil October 21. 1996. the date on rvhichthe program was run. Because VTLS up-dates the circulation count rvhen itemsare returned, this figure does not includea count {br iterns tliat rvere checked outwhen the program was run or that hadbeen renewed by the borrorver a{ter theywere checked out. The circulation countdoes include items checked out to and re-turned from interlibrary loan

Followinq this iniual data collection,each bibliographic record vras searched inWLS and vieu'ed to find the date the copyrvas entered into the systern and the aca-dernic deoartment foru'hich the dissertation

286/ LR?S . 42(4) . Sapon-White and Hansbrough

TABLE TACADEMIC AREAS oF STUDY

Area o fs tud \ Academic DeDaf lme l ts

ArVArchitecture

B usi ness/Management

Education

Engineering

Science

Environmental Design and Planning

Housing, Interior Design, and Resource Management

B usiness

Business Administration and lvlanagement

Hospitality and Tourism Management

Hotel, Restaurant, and Institutional Management

Management

Management Science

Administration and Supenision of Special Education

Adult and Continuing Education

Com muni ty Col lege Educat ion

Counseling/School Psychology

Counseling and Student Personnel Sen'ices

Counselor Education

Curriculum and Instruction

Educational Administration

Educational Research and Evaluation

Instructional Technology and Desigr.r

Vocational and Technical Education

Aerospace and Ocearr Errgi r reer ing

Aerospace Engineering

Agricultural Engir.reerir.rg

Biological Systems Engineer ing

Chemical Engineerir.rg

Civil Engineering

Electrical Engineering

Er.rgineering Mechanics

Elvironmental Engineering

Industr ia l and Systcms Engineer i r rg

Materials Engineering Science

Mechanical Engineering

Mining and Minerals Engirrecr ing

Animal and Poultry Science

Animal Science

Biochemistry and Anaerobic Microbiology

Biology

Botany

Chemistrv

Computer Science

Crop ar.rd Soil Environmental Science

Fisireries and \\lldliie

Food Science and TechnologyForestryGenetics

(Continued on next page)

LR?S . 12(4) . The lmpact of Subject Heading Assignment /287

TABLE I (cont )ACADEMTC Anues op Sruov

Social Science

Horticulture

Human Nutrition ald Foods

Mathematical Physics

Mathematics

Physics

Statistics

Veterinary Medical Science

\\rood Science and Forest Products

Agricultural and Applied Economics

Clinical Psvchologr

Family and Child Development

Psychology

Public Administration/Public Af fairsScience and Teclinologr Studies

was rvritten Il uncontrolled subiect termsrvere encountered in a 653 MARi field. thevwere counted arnd recorded lorlateranalysii.

The circulation count lbr dissertationsrvith subject headings was then comparedto the circulation count ofthose catalogedwithout subject headings. Data were in-tered into an Excel spreadsheet for analy-sis Each entrywas assigned to one of sixbroad areas of'study (art and architecture,business and rnanagement, education,engineering, science, social science). Thein{brmation was sorted bv acadernic areaso that t l re data could be analvzed bv sub-

.ject categories as well as for the wholesample. The areas ofstudy and the depart-ments they included are listed in table l.

Anomalies in the data included 6 titlesrvith 2 copies that could be checked out.In 5 of ' these 6 cases, I ol ' the 2 copies didnot circulate at al l In al l 6 cases, the circu-lation counts fbr the 2 copies were addedtogether and entered as ifthere were onlyI copy. Also, 2 dissertations were encoun-tere-d'that cornprised 2 volumes each. Inone case, neither volurne circulated. Inthe second case, I volume circulated oncewhile the other did not circulate at all.Again, the circulation inlbnnation rvasadded toeether and entered as if it wereone volume.

Although titles rvith subject headingswere on the shelf for a longer period of

time than those without subject headings(potentially 15-18 months versus 12-15months), and therefbre had a greater like-lihood ofcirculating, this is probably not asignificant factor for two reasons First,the extra rnonths occurred in the sum-mer, rvhich is a time when the library re-ceives the lon,est arnount of foot trafficand circulation of dissertations in bothgroups rvould be expected to be ex-tremely low. Second, although the datepararneters {br selection of the sarnplecover a 6-month period (3 months beforeand after the chanqe in treatment), in ac-tual lact 92Vo o{ ihe titles with subjectheadings entered the online system be-trveen June 2I and l]UJy 25, 1995;58Vo oI'titles rvithout subject headings enteredthe systern betrveen luJy 26 and August15, 1995 Anybias as aresultof the dlf fer-ence in time on the shelf was thereforeconsidered to be very small

Data were analyzed using thechi-square test o{'independence (Sheskin1997). Because Sull ivan et al. (1992) hadshorvn that titles u'ith subject headings\\/ere more likely to circulate, this studyr.vas designed to evaluate whether titleswith subj-ect headings circulated more of-ten than those vvithout subject headings.The null hwothesis was, The circulationcount {br di.ssertations is independent o{'the subject headings assigned to their

288/ LRTS . 42(4) . Sapon-White and Hansbrough

TABLE 2DISTRIBUTTON By ACADEMTc AREA AND Sunyncr HEADTNG TnnarrratNr

Academic Area Witl i Subiect Headinss Withorrt Srrbiect Headinqs Total

Art/Architecture

Business/Management

Education

Engineering

ScienceSocial Science

I

b

40

51

3 l

l l

Z

A

2428341 n

3I

o4

79O D

28

Total 139

bibliographlc records A value of .05 waschosen for deterrnining significance ofthe results.

Rrsur,rs

A total of248 dissertations rvere catalogedbetween May 1, 1995. and October 31.1995. Of these, 139 were assigned LCsubject headings. Of the I09 witliout sub-ject headings, 36 had author-supplied un-controlled subiect terms added in theMARC 653 field of the record. Althouphthe program parameters were set to col-lect 6 rnonths worth of cataloging data, inactuality, all of the dissertations in thestudy were cataloged between June 21,1995, and October 3I, 1995 Therefbre,the data collected represents 16-17rnonths u,orth of circulation informationfor dissertations with LC subiect head-inqs and I2-I5 months worth of infbrma-tion for those without subject headings

The distribution of the records amongareas ofstudy is shown in table 2.

The mean number of circulations forthe total sample was 58Va higher for re-cords with subject headings r,vhen com-pared to those without subject headings.Engineering records shorved a mean 69Vohigher for records rvith subject headings,while Science records rvith subject head-ings had a mean l28%o higher than thosewithout subject headings. Education re-cords, however, shorn'ed nearly identicalmeans for the two groups.

The number of titles in ArVArchitec-ture, Business/Management, and SocialScience was too small to analyze using thechi-square test but are included in the

overall analysis. Results of the test areshorvn in tables 3-6 Ibr all dissertations,Education, Engineering, and Science, re-spectively. Statistically significant resultswere found for the aggregated sample(chi-square=9.99; df=4; p=.04), but notfor Education (chi-square=I.27; df=2;p=.53), Engineering (chi-square=3.74;df=z; p=.I5), or Science dissertations(chi-square=1 f4; d{=1; p= 28)

DIScUSSIoN

The results shown in table 3 indicate that,for all titles in this study, dissertationswith subject headings were more likely tocirculate and circulated a greater numberof t imes than t i t les without suhiect head-ings. The null hypothesis that circulationcount is independent of the presence ofsubject headings is therefore rejected.One explanation forthese results is that li-brary patrons used subject searching inthe online catalog and fbund their rvay todissertations, as rvell as other materials,on their subiect ofinterest. Conventionalsubject searching retrieved rnaterials thathad subject headings on their records,while records lacking subject headingswere not retrieved In the case ofkeyrvordsearching, the addition of the subjectheadings might have enriched the vocab-ulary present in the bibliographic record,thereby increasing the likelihood ofsub-ject retrieval by kepvord. These resultsindicate that library users are searchingby subject often enough to have an impacton the circulation of dissertations

One might reasonably expect that sub-ject terms supplied by the dissertation

LRTS . 42(4) . The lrnpact of Su,bject Heading Assignment /289

TABLE 3ALL DISSERTATIoNS

Number of Circulations

0 1 2 3 > 3 T o t a l M e a n

Observed

\\rithout subject headings 58

\\zith subject headings 53

Total l l l

Expected

\\rithout subject headings 48 79 31 21 17 14

\\rith subject headings 6221 39 79 21 86

Total lll 71 39

Chi-square=9 99; p=0 0{

7 0 1 0 9 0 7 31 3 7 r 3 9 1 1 5

20 7 248

8 7 9 3 0 8 1 0 9ll 21 3 92 13920 7 248

29 t542 2471. 39

authors (and recorded in the 653 MARCIield) would boost the circulation of titleswithout subject headings In the presentstudy, any such boost was not greatenough to be detected. This might be dueto the fbct that only 36 titles of the I09without subject headings (337o) had un-controlled subiect terms added to the bib-liographic record. In order to fully evalu-ate the effect on circulation of'addineuncontrol led subject terms, one wouldneed to compare a group ofrecords con-taining such terms with a control grouplacklng both author-supplied terrns andLC subject headings.

These results help to con{irrn the rvorkdone by Sullivan et al. (1992) In thatstudy, researchers could only tell whethera dissertation circulated or not; there wasno wayto account forthe nurnber of tirnesa title circulated. In the present investiga-

tion, by looking at the circulation counts{br each title, we show quantitativelyhigher circulation counts for dissertationsrvilh subject headings.

The results frorn the academic fieldsof study tables are less revealing as therewere no statistically significant di{Ier-ences between titles with and those with-out subject headings in any disciplinaryarea The chi-square test, horvever, is notpowerful enough to detect differences inthe distribution of circulation countsrvhen the sample size is Smdll. A more so-phisticated test, beyond the rneans avail-able for the p.et"tti study, rnight have thepowerto detect a directional shift in thesetables. That shifi is evident if one corn-pares the observed and expected valuesu'ithin the engineering or science tables;titles with subject headings circulatedrnore than expected while titles without

TABLE 4EDUCATION DrssentatroNs

Number of Circulations

0 I >1 Total

Observed

\\'ithout subject headings

\\'ith subject headings

Total

Expected

\\tithout subject headings

\\'ith subject headingsTotal

Chi-square=l 27; p=Q $ll

6I

I D

5 4 89 5 2

I J

6l 622

8,0313 9722

t 1I J

26

9 4 916 5126

2340O J

234063

t 2 6r .25

290/ LRTS . 12(1) . Sapon-White and Hansbrough

TABLE 5ENGINEERING DISSERTATIONS

Number of Circrrlations

I o ra l Mean

Observed

\\uithout subject headings

\\'ith subject headings

Total

Expected

\ltithout subject headings

\\'ith subject headingsTotal

l o

a l

10 9920 013 1

l318

6 3 8I I 6 218

82230

l0 6319 3730

0 8 9t D l

285179

28D I

7q

Chi-square=3 74; p=0 15

subject headings circulated less than ex-pected. The need to "collapse" the cells toperfbrm the chi-square analysis, partiallydue to the small sample size, is the likeli-est reason u,hy no statistically significantprobability is calculated {br the scienceand engineering tables. The a{bremen-tioned directional shift is clearly missingfrom the education table

Harris zurd Hu{Ihan (1985) contendedthat keyrvord searching might provide auseful substitute Ibr subject searchingwhen titles are sufliciently descriptive ofcontent In this study, all titles were highlydescriptive o{'their content. Though somedissertations in the humanities have non-descriptive, even rvhimsical, titles, nonesuch works were present in this study. Inthe case of whirnsical titles. retrieval bvkeyrvord in the absence of subject headingassignment rvould be highly unlikely.

These results have implications for un-derstanding user behavior. Clearly, dis-sertation users are not only conductingknown-author or knorvn-title searches.Subject access to disserlations is impor-tant, though one cannot tell from thisstudy whether the catalog searchers arespecifically looking for dissertations or ifthey are searching fbr material on theirsubject in any {brmat. As previously men-tioned, authors oftq'o studies have lookedat this issue and, in both instances, con-cluded that, lbr dissertation users,seaiching by subject is signi{icant (Reppand Glaviano 1987; Lee-Srneltzer andHackleman 1995)

V[/e have used circulation counts as anindication of use, but do not considerin-house use. A rnore complete picture ofthe efTect of subject heading assignmenton use q'ould have to include such use as

TABLE 6SctuNcn DISSERTATToNS

N r r m b e r o f C i r c r r l e t i o r r s

Observed

\\'ithout subject headings

\\'ith subject headings

Total

Expected

\\'ithout subject headings

\\'ith subject headingsTotal

Chi-square= I 14; p=Q l$

272I48

25 tL22.8948

710L7

8 8 98 1 1

t7

343 I65

343 165

0 2 10 4 8

LR?S . 42(4) . The lmpact of Su,bject Heading Assignment /29I

rvell. It does not seern likely that in-houseusers rvould behave differently frorn userswho check out materials, and so the re-sults here are not likely to be biased by thefocus on circulation counts.

Though beyond the scope ofthe pres-ent study, the question still remains as torvhether subject searching in DissertationAbstracts could comnensate {br the lacko{'subject headings issigned to disserta-tions in a local catalog. It would be inter-esting in a future study to compare thedescriptors used in DAIk subject search-ing with the controlled vocabulary oI LCsubject headings in a librarys catalog.Such a study would shed further light onthe merits of these two subject searchingsysterns and add to our understanding olhow users find the materials they need.

Further research, perhaps using alarger sample and a more powerful statis-tical test, is indicated to evaluate dif{'er-ences between academic areas of studyand to corroborate these results with anassessrnent ofuser behavior when search-ing Ibr dissertations.

CoNclustotrt

Cost-saving rneasures in academic librariesmust be carefully evaluated in terms of theireffect on service to library usen. One suchmeasure, the suspension ofassigning subjectheadings to dissertations, has a significant im-pact on dissertation use. The results of thisstudy indicate that dissertations with subiectheadings in their bibhogaphic records aremore likelv to circulate (and circulate moreoften) than those without subject headingsFailure to add subiect hea&ngs to biblio-graphic remrds for dissertations, while in-deed saving tirne in cataloging, noticeably re-duces the l-ikelihood that the items will beIbund and used by patrons Library adminis-trators choosinq to rnake such a decisionshould do so undentanding that there will bea resulting loss ofuse ofthis unique form of'research literafure.

Wonxs Crrno

Harris, Georqe, and Robert Hulfman 1985Cataloging oftheses, A survey. Cataloginglr classification E)arterhJ 5, no. 4: 1-I5.

Keller, Barbara 1992 Subject contentthrough title: A masters theses matchingstud1, 21 Indiana State University Cata-Ioging 1: classification quarterhT 15, no 3:69-80

Khurshid, Zahiruddin 1995. Improvisationsin cataloging of theses and dissertationsCataloging b classification quarterly 20,N O ] : D I _ D Y .

Lee-Smeltzer, Ianet, and Deb Hackleman1995 Accesi to OSU theses and disserta-tions in Kerr Library: Hou' they areused or are they? Cataloging b classi-

fication quarterlg 12, no. 4:25-43.Patterson, Kelly, Carol \\/hite, and Martha

\\/hittaker 1977 Thesis handlins in uni-versity libraries . Library resourcei dt tech-nical serxices 2I, 27 4-285

Perry Stephen, and Lutishoor Salisbury 1995.Access to infbrmation in both CitaDel andFirstSearch : A comparative study of disser-tation coverage lnformation technologyand libraries 14, no l: 17-29

Repp, Joan M , and Cli{TGlaviano. 1987. Dis-sertations: A study of the scholart ap-proach CoIIege and research libraries 48:148-59

Ryans, Cynthia C 1991. Catalogingtheses anddissertations: An update, Cataloging Ctclassi.frcation quarterlg 14, no l: 83-87

Sapon-\\'hite, Richard. 1997. Impact of sub-ject heading assignment on circulation o{'theses at Oregon State University. Postersession presented at the annual meeting ofthe American Library Association, SanFrancisco, Calif

Sheskin, David f . 1997 Handbook of paramet-ric ancl nonparametric statistical proce-dures Boca Raton, Fla.: CRC Press.

Sullivan, Michael V, and others. 1992 Ef-fects on circulation of dissertations ofomitting subject headings In Academic Ii-braries: Achieaingexcellence in higher ecl'ucation, ed Thomas Kirk 122-27 Chi-cago: Associat ion ofCol lege and ResearchLibraries

UIvII f997 Anterican doctoral dissertations,1995-1996 Ann Arbor, Mich : UMI

292/

Noles on Operolions

Evqluqtion of Three Record Types forComponent Works in Anolytic OnlineCotologs

Herbert H. Hoffmon

I-llbliographic databases that make uplibrary online catalogs contain biblio-graphic records that represent the books,discs, cassettes. and other i tems a l ibraryhas on its shelves. \\'hen such an item con-tains one sole rvork, the item and the workit contains are perceived to be one and thesame thing It is rarely difficult to retrievesuch works. But when one item containstwo or more rvorks, it nlay be difficult tolink the rvorks to the item. That may bepart of the reason rvhy online catalogs arestill hard to use, as Borgman recently reit-erated (Borqrnan 1996).

Much qroundwork has been done todlstlnguis[i works fiom items. Lubetzkyheld that the "work"-the "literaryunit"-uras the basis of bibhographic de-scription (Lubetzky 1963). Hoffman(1976) has attempted to shorv that all pub-

Hrngrnr H Horrvex (ho{Im@earthlink net)Ana, California Manuscript received January 1,

lications, regardless of rnediurn or {brmat,are first ofall "rvorks" (i.e., essays, poems,novels, plays, symphonies, etc.) con-tained in "books" (i.e , rnor-rographic pub-lications, collections, periodicals, cas-settes. etc.) that come in "sets" of one orseveral volumes, and that there are no ex-ceptions. The ERIC thesaurus of lg78, inits basic list o{'publication type catego-ries, supported this vieu' by defining"book" as "pure lbnn or 'empty con-tainer,"' contrasted u'ith "creative works"such as "poetry, literary rvorks, essays,novels, short stories ." (EducationalResources Infonnation Center 1978,f 78D). These definitions lend weight toLubetzky's proposition that it il thework-the intellectual creation-and notthe package or the container on rvhich thecatalog.should lbcus

is Catalog Librarian, Santa Ana College, Santa1998; accepted {br publication April 4, 1998

The bibliographic description of "ana-

lyzed parts" versus "containing items" rvastreated in detail by McCallum nearly hvodecades ago (McCallum 1980). Still, asHowarth (1997, 9) described the situa-tion, todayt online catalogs seldorn man-age to represent the "r,r,ork independentlyof the physical fbrrnat." As \\reintraub andShimoguchi phrased i t (1993, 178), a rel i-able analytic catalog rvould not only re-main a "vehicle lbr trar-rsmittins blblio-graphic information about u'hole books"but would "reveal infbnnation couchedwithin these books" as well. In such a cata-log, the bibliographic records of all of anauthort works, larqe or small, those thatstand alone and tliose that are "cornDo-nent parts," would be indexed so that inyofthem can be searched lbr and retrievedrvith one simple keystroke or click.

\\'hile a true analytic online databaseon a national scale does not vet exist. acase could be made {br it Ho{hnan andMagner (1985) found that {br every itemon the shelf that is listed in the cataloq,there might be live u'orks ernbedded incollections and anthologies, rvorks thatare not listed-not in the catalog nor inany other index, t inding l ist. oi bibl io-graphic aid. Thus, leaving the periodicalliterature aside, a library of 100,000 itemsmight contain half a rnillion works that arenof easily accessible to patrons searchingthe online catalog. Poulsen ( 1990), in stud-ies at six other lforaries, had similar lind-ings. Such data justi{y the conclusion thatour libraries contain enough uncatalogedor insu{Iiciently cataloged works to makeimproved analytic catalogs rvorthwhile.Hagler (f997, 13) seconds this objectivewhen he suggests that future editions ofAACR should require that "an agency pro-vide access to every u;ork . .. appearingwithin each cataloqued document."

The sheer nurn-bers ofworks. however.call for a coooerative e{lbrt Given the ad-vances being made in computer technol-ogy and telecornmunications today, it isconceivable that in addition to a MARCdatabase of books qu.a containers, a com-parable utility fbr component rvorks rvillbecome f'easible, a utilitv that rvould en-able readers to access the rvorks con-tained in library collections and antholo-

LRTS . 42(4) . Notes on Operations /293

gies, much as they now use INFOTRACand sirnilar indexes to access the r'vorkscontained in periodicals The organiza-tional and administrative asDects of suchan undertaking, however, aie not rvithinthe scope of this paper. \\/e restrict our-selves to a discussion ofhow di{Terent bib-liographic record str-uctures all'ect thesearch for, retrieval, and display of corn-ponent works in analytic or partially ana-lytic online catalogs

Let us consider what an analytic cata-log ought to achieve to f'acilitate the re-trieval of soecific u'orks. It seems thereare four miior goals'

1. Once a work has been identi{ied, thecataloe should retrieve all versions ofthat w"ork the library orvns, not justsome. At present, {-ew online catalogsprovide complete and reliable accessto works embedded with others incollections and antholoqies. Most on-line catalogs are at least partiallyblind to component u'orks becausethe contents of collections and an-tholoeies are not listed in the bibho-graphlc records that rnake up thedatabase. A sample of 44 anthologiesof drama retrieved {rom the catalogof 'one large American university re--vealed that only 15 ofthen were rep-resented by bibliographic recordsthat contained tables o{' contents.There was no clue to the contents oI'the other 29 antholosies. In that li-brary. a student looking lbr a givenrvork rvill find some manifestations ofthat work but rvill not find all mani-festations for that rvork that the li-brary owns And because most librar-ies today download their recordsfiorn the same pool of MARC re-cords, f'ew other online catalogs arelikely to give better ser-vice

2. The catalog should retrieve only onlyversions of that rvork and no others.There should be no irrelevant or unre-lated titles retrieved, that is, no falsedrops. In rnany online catalogs, a searchfor author "Beethoven" and title keyrvord'Octet" will retrieve rnany exarnples thatdo not contain any ofBeethoven's octets.For exarnple, this search will retrieve a

294/ LRTS . 42(1) . Hoffman

record that contains Beethoven's Septetop. 20 and Mendelssohn's Octet op. 20.Both search terms are contained in therecord for the item, but they do not de-scribe one u,ork; theyare noi linked spe-cilicallv to a oarticular raork.

3. It should be possible to achieve thedesired results in one pass. Readersshould not have to try several ap-proaches before they are confidentthat they have exhausted all possibili-ties. A search for author "Schnitzler"and title "Game of love" in many cata-logs will draw a blank. Libraries tlratorvn Corrigan's Masterpieces of theModemt Central European Theatre,however, do have a copy of the play.The MARC record has a searchablecontents field for the title. But there isno searchable author field {br"Schnitzler." To find Schnitzler's play,then, the reader must do a secondiitiesearch, which, paradoxically, will onlyq'ork if the author's name ii left out.'

4. The retrieved records should be col-located in an uncluttered, unambigu-ous screen display. In todayls onl inecatalogs, as Carlyle (1997) points out,the second obiective of the ParisPrinciples tends to get short shrift.The second obiective requires thatthe catalog collocate all of an author'sworks tha[ the librarv owns (Veronat97 l ) To d isp lay a work in the on l inecatalog *""ttr to shorv its title and, ifapplicible, its author or authors, in aprorninent position on the screen. Auser searching for Shawt play Armsandthe Man,for example, expects tosee a screen that displays both the au-thor name and the title. However, inmost systems, a briel'dlsplay will in-clude data from MARC field I00 (fbrauthor) and MARC lield 245 ({br ti-tle), u'ith results similar to those seen

in lrigrile I This display is not as clearas might be expected. From the infor-mation given, it is not clear that onlythe last two items contain the playsought. Moreover, neither "Shaw" nor"Arrns and the man" are disolaved onscreen. The reader rnust seaich forthat in{bnnation in the full display ofeach record on subsequent screens.

To construct analytic catalogs thatachieve all four goals, librarians must deter-rnine',vhich record fbrmat will produce thedesired results. In the Anglo-AmericanCatalogting ktles, 2d ed.. igSS revision(LACR2R),three record stmctures for ana-lytics are mentioned They are contentsnotes, added entries, and "In" analytics.

The first rnethod-contents notes-isdescribed in rule 13 4A. The rule states(p. 300) that it is "the simplest means ofanalytics" and that it is "usually limited toa citation of title or name and title."Blackwell's "Blackwell Table of Contents"project focuses on this method, making ti-tle-searchable contents lists for selectedcollections and antholosies available fordorvnloading.

The second method-added entries-is described in rule 13 2A, which reads inpart (p. 300): "this method is appropriatewhen direct access to the part is wantedrvithout creating an addltional biblio-graphic record for the part " It is clearfrom the wording that direct access tocomponent rvorks has ahvays been a de-sirable feature of hbrary catalogs It mustbe remernbered that the rules date backto a time rvhen the standard was a cardcatalog. An added entry then was an extracard \\'hen Iiled ln proper order, readershad direct access to a qiven title and foundall rnanifestations ofiineatly collocated inthe drawer. In the online world. however.an added entry is no longer a separately

AUTHORH I I T S , L . T i U S T

Grey, M CameronKe rnan , A l v i n B .R : r n o f C r r ' l r r : n

TITLE PUBDATL u s t , v i o l e n c e , s i n , m a g i c c L 9 9 3A n g e l s a n d a w a k e n i n g s 1 9 9 4C l a s s i c s o f t h e m o d e r n t h e a t e r [ ] - 9 6 5 1E r o h - o - e r f e o - c d i e . c L 9 5 8

Figure L Display of Titles Retrieved for Shauis Play Arms and the Man

LR?S . 12(1) . Notes on Operations /295

1 0 0

5 0 5

1 01 00

Sab l i l l i ams , Tennessee ,$aE iqh t p l aysS a T h e o l a q s m c n a o e - i e

and smoke -- The RoseOrpheus descend ing - -^ F F L ^ i ^ , , ^ - -

a 9 u a r r a .

s d 1 9 1 1 - 1 9 8 3 .

- - A s t r ee t ca r named des i r e - - Sunnne rta t t oo - - Ca t on a ho t t i n r oo f - -

Swee t b i r d o f you th - - The n i gh t

Figure 2. Use of Single Contents Note {br Collections.

{iled thine but a lield in a MARC record.Fields in MenC records do not automati-cally produce collocated displays. Nomatter what so{'tware a library may havebought, access to embedded rvorki is notas direct as it once was.

The third method-"In" analytics-isdescribed in great detail in rule 13.5A andB The rule siates that an "In" analytic en-try be rnade (p. 300) "if more biblio-graphic description is needed {br the partthan can be obtained by displaying it inthe note area "

CournNrs NorEs

Many librarians will say that a contentsnote in the bibliographic record fbr abook or other itern containing dill'erentworks will suffice. There are trvo rvays ol'entering data into the contents {ield of aMARC record (field 505). The basicmethod uses subfield a; similar results canbe achieved by placing the titles o{'conr-ponent rvorks into the subtitle, i e. MARCfreld ZqS subfield b. The second, en-hanced method uses subfields t and r inMARC 505.

Bestc 505 FIELD FoR CoLLrcrrous(Ar.r, \\/onrs BY THE Senn euruon)

Let us {irst look at the basic contents notemethod for collections Librarians raised

on AAC R2R, Akers' Simple Library C ata-Ioging, and similar classics might not be inthe habit o{ 'dist inguishing clearly be-trveen collections that contain works bythe same author and collections that con-tain works by different authors The hvotypes of iterns, however, are cataloged bydifferent rules. That is why u'e prefer adistinction and use the term "collection"

here to designate the fbrmer type of itenr.\\/e shall use the term "anthologi' lbr thelatter. In a collection, then, the author'sname would be in field I00 of the biblio-graphic record. The titles of componentworks can be added in {ield 505 sub{ield aor, occasionally, in245 sublield b It mightseem that this simple method of analyticsmakes good sense, but there is a problern.Because all strinss ln 505 subfield a and ir-r245 subfield b are in one subfield, only a ti-tle key+r'ord search is possible. There is norvav to instruct a comDuter to search forthe exact title phrase

An example rvould be the collectionEight Plags by Tennessee \\/illianrs (seel igure 2). The author's name appears inthe searchable 100 field. But a search lbrthe exact title "summer and Srnoke"would I'ail. Only a search fbr author's nameplus one or more title keywords rvill re-trieve this collection. \4/hen the retrieveditern is displayed on the screen, o{'course,it rvill show the title "Eight plays," not"summer and smoke," thus hidinq the col-

1 . W i l l i a m s , T e n n e s s e eC o l f e c t e d p l a y s . . . . . . . 1

2 . W i 1 l i a m s , T e n n e s s e eE i , g h t p l a y s . . . . . 1

3 . W i l l i a m s , T e n n e s s e eSummer and s rnoke . . . . . . 1

4 . I , ' l i I l i ams , TennesseeS w e e t b i r d o f y o u t h , a n d L w o o t h e . . . . . . . . . . I

Figure 3. Displav of Results for a Search for \\'i l i iams' "Summer and Smoke."

296/ LRTS . 42(4) . Hoffman

245 0A $aN ine teen th - cenLu ry B r i t i sh d rama5 0 5 0 $ a T h e C e n c i , b y P . B . S h e l l e y . - - B l a c k - e y , d S u s a n , b y D .

Je r ro l d . - - The l ady o f Lyon , by Lo rd Ly t t on . - - Londona s s u r a n c e , b y D . B o u c i c a u l t . - - C a s C e , b y T . W . R o b e r t s o n .- - T h e b e l f s , b y L . L e w i s . - - E a s t L y n n e , b y T . A . P a l m e r .- - H . M . S . P i n a f o r e , b y S i r W . S . G i l b e r t . - - T h e C o u n t e s sC a t h l e e n , b y W . B . Y e a t s . - - T h e s e c o n d M r s . T a n q u e r a y ,by S i r A . W. P ine ro . - - ?he impo r tance o f be ing Ea rnes t ,b y O . W i l d e , e t c .

Figure 4. Bibliogrraphic Record for an Anthology Using MARC Field 505 Sub{ield a

location f'eature. If the library owns severalversions <rf Srrm merand Smoks, the readerrnight see a summary display like that seenin ligure 3. Such a display confuses thesearcher because it is not clear which linesactually point to the play sought.

The basic table of contents method is un-able to respond to exact title searches Thisis a ploblem lbr the reader who begins withan exact title search, then finds out that theretrieved item is not available. The readerhas trvo choices: glve up or do a secondsearch, justto be sure that there really are noother copies of the play in the library. Andeven ifthe library does not own any othercollections or anthologies that contain theu,ork, this second pasJ is still necessary tobe sure all avenues were exhausted.

There is another problem. Keyr,vordsearches are vulnerable to f'alse drops.Suppose a reader searched the catalog iorauthor's narne "Shaw" and title keyrvord"Man" (as in Man and Sttpermon). I t i .easy to see that there might be a hit, butthat the play the reader retrieved wasArm.s an.d the Man instead. A keywordsearch lbr Palestrina's "Ave Maria" mishts'ell retrier,e Hyperion CDA 66850, anitern that does not contain the u'ork butdoes contain one titled "Aue verum cor-pus, natum de Marin virgine " Unless thelibrary's catalog takes u'ord order and prox-imity into account, a reader looking for thepoern Losf \Vorld might well retrieve thestory "\\brld \\/ell Lost" instead.

Besrc 505 Frnlp Ron ANTHoLocTES(\\7OnTS BY DIFFERENT AUTHoRS)

If an autlort u,orks are mntained in an anthol-ogz together u'ith other authors' works andfield 505 sublield a is the onlv analvtic field inthe host item recond, then author iearches orauthor and tide searches will not work.

An exarnple is the anthology Nine-teenth-Centunl British Drama (see {igure4). Although Oscar\\'ilde's ?he Importanceof BeingEam.esf is contained in this anthol-ogy, a perfectly logical author/exact title (oreven an author/title kepvord) approach willfail because authors'names, in this case, arenot contained in any author-indexed field.Pragmatical souls rvill say that all one hasto do to achieve the desired retrieval is totreat the authort narne as a title ksy'\a'e1dand include i t in the search argumentBut this is a stopgap approach, inconsis-tent rvith the concept ofauthorship, andtherefore not recommended as a perma-nent catalog design {eature. Library us-ers should not be required to do mentalrearrangements hke this in order to suc-ceed at the catalog

ENHANCED 505 FInr,p r.onCollncrroNs (ALL \\/ORKS BY TrrESeun AurHon)

The enhanced 505 tield has reoeatablesublields t lbr titles. An example would beThree Plarlsby August \\/ilson (see figure 5).

1 0 0

5 0 5

1 0l 0

SaWi l son , Augus t$aTh ree p l aysS r M a R a i n e v , s h l a e k ! 6 1 1 6 p - -

n n m a r n d n n n a

S t F e n c e s - - S t J o e T u r n e r ' s

Figure 5. Bibliographic Record for a Collectior-r Using IvIARC Field 505 Subfield t

LRTS . n2(1) . Notes on Operations 1297

L . B r o o k s , c a r l hN o f e n c e s . . . . . 1

2 . Symons, HarryF e n c e s . . . . . . . . 1

3 . W i f s o n , A u g u s tT h r e e p l a y s . . . . . . . . . 1

Figure 6. Display ol Results fbr a Search for \\/ilson's "Fences

Unlike the basic 505, the enhanced 505lvith subfield t can, if'the library's soft-ware allou's it. be fullv title indexed andcan then be searchld by exact titlephrase as r,r'ell as by kepvord The au-thor's narne is in field 100, the title in 505sub{ield t Thus, exact title, kepvord, andauthor/ title searches rvill succeed. Thepossibi l i ty of 'a false drop is remote. Butieaders would lind it hard to notice thecollocation feature For even il one hassearched for and fbund a rvork title in abrowsing display, the machine is likely togenerate a final item display that showsthe title contained in field 245, not thework title contained in {ield 505. An un-suspecting, maybe less sophisticated userlooking fbr the tltle Fences rnight, on seeingthe surnmary display in flgure 6, pick line 2and take honle the wrong book.

ENsexcep 505 Frnlo ponAurnolocrns (\[/oRKS BYDrnnunnNr Auruons)

In the case ofan anthology, one would addseparate sub{ields r for authors' names.An example .'orid be Zora NealeHtrrston, Eu.lalie Spence, Marita Bonner,andOthers:The Prize Plays (see figure 7)But sublield r, being a part of field 505,cannot be controlled the wav seDaratelields 100 and 700 are, becauie 505 is atranscribed {ield, while I00 and 700 useauthorized lbrms of names. Field 505subfield r. there{bre. cannot be used reli-

ably for author searches. And even in caseof a correct retrieval by title, a searcherlooklng for Aftermath by Burrell will beshown a screen that says "Zora NealeHurston, Eulal ie Spence, Marita Bonner,and other" instead of "Aftermath," a situ-ation that will probably confuse all but themore experiericed libiary users because itlooks as il the desired rvork was not fbund.The "table of contents" method rnay bethe sirnplest method of analytics, but it isnot the best Let us examine another al-ternative.

ADDED Eurnrns

Some believe that the best wav to provideanalytic access to component wor-ks is to,x. idd"d entries. ThiJ is the method ad-vocated in such nrles as 13.24 and21 30Ml of AACB2R. It requires the addi-tion ol'a suitable cornbination of DO( fieldsto the host itern's bibliographic record,either alone or together with a 505 field.

7XX Frnlps FoR CoLLECTIoNS (ALL\ \ 'onrs BY THE SAME Auruon)

For a collection it would be easy to add anumber of 740 fields to the item record,one for each comDonent work title. Be-cause in such a sit-uation the name of theauthor of all the works is in the f 00 field,an author/title search would be success-ful. An example is Three by Tennessee (see{igure 8). As in the situations described

2 4 5 A 0

s 0 5

$aZo ra Nea le Hu rsLon , Eu fa l i e Spence , Ma r i t a Bonne r , ando c h e r s : S b t h e p r j z e p J a y s .$ t t t a g a r a n d I s h m a e l , $ r c h a r l o t t e T e l l e r H i r s h . - - S t M i n eeyes have seen , S rA l i ce Moo re Dunba r -Ne fson . - - $ tA f t e rma th ,S rMary P . Bu r re l l . - - $ tThe pu rp l e f l owe r ,S rMar l t a Bonne r .- - S tThe hunch , $ rEu la I i e Spence ,

Figure 7. Bibliographic Record for an Anthology Using MARC Field 505 Subfields t and r

298/ LRTS . 12(4) . Hoffinan

1 0 0 1 0 $ a W i l l i a m s , T e n n e s s e e , S d 1 9 1 1 - 1 9 8 3 .245 L0 SaTh ree by Tennessee1 4 0 0 2 $ a S w e e t b i r d o f y o u t h7 4 0 4 2 $ a T h e r o s e t a t t o o140 42 $aThe n i gh t o f t he i guana

!!8ure S. Bibliographic Record for a Collection Using MARC Field 740 for Titles of Component\\/orks.

above, however. when the host item is dis-played on the screen, the title shown willbe that contained in the 245 field, thus hid-ing the identity ol'the retrieved work be-hind the item or docurnent title.

7XX Frnlps FoR ANTHoLoGTES(\A'onrs By DTFFERENT AurHoRS)

For an antholos/, one or more 700 fieldswould be indicated, rvhere subfields a and

t contain the authors and titles, respec-tively, of cornponent u'orks It rnight lookas if exact title, title ke;u'ord, as well as au-thor/title searches are provided for in thiscase. There is aproblern, though, when au-thor names in 700 subfield a and titles in700 subfield t are not linked. For example,a search for author "Palestrina" and title"Ave Maria" in most catalogs would re-trieve a record for an item ihat containsPalestrina's "Sicut Cer"rrs" and Robert

0 2 8 0 2 S a 7 7 0 9 - 2 - R G S b R C A V i c t o r c o l d S e a l0 3 3 r $ a 1 9 5 1 - - - - S a L 9 5 2 - - - -040 SaDLCScDLCu 4 t i a s y l a o r l a r o0 4 8 S b s a 0 l S a o a0 5 0 0 0 S a R C A V i c t o r G o l d S e a l 7 7 0 9 - 2 - R G1 0 0 1 0 $ a H e i f e t z , J a s c h a , $ d 1 9 0 1 - g 4 p r f2 4 5 I 0 S a S h o w p i e c e s S h I s o u n d r e c o r d i n g l .2 6 A A $ a N e w Y o r k : S b R C A V i c t o r c o l d S e a l , S c p 1 9 8 8 .3 0 0 S a 1 s o u n d d i s c : S b d i g i t a l ; i c 4 3 / 4 t n .3 0 6 S a 0 0 2 4 0 5 S a 0 0 0 9 0 7 S a 0 0 0 8 1 4 g a 0 0 1 3 1 6 S a 0 0 0 8 2 6500 SaThe 3 rd wo rk o r i g , i na l l y f o r v i o l i n and p i ano .505 0 $aSymphon ie espagno le : op .2L / LaLo (24 :05 ) - - Havana i se :

o p . B 3 / S a i n t - S a e n s ( 9 : 0 7 ) - - Z i g e u n e r w e i s e n : o p . 2 0 /S a r a s a t e ( 8 : 1 4 ) - - P o d m e : o p . 2 5 / C h a u s s o n ( 1 3 : 1 6 ) - -I n t r oduc t i on and rondo cap r i c c i oso : op .28 / Sa in t -Saens (8 :26 ) .

511 0 SaHe i f e t z , v i o l i n ; RcA V i c to r Symphony Orches t ra ;W i l l i a m S t e i n b e r g , c o n d u c t o r ( 1 s t - 3 r d a n d 5 t h w o r k s ) ;I z f e r S o l o m o n , c o n d u c t o r ( 4 L h w o r k ) .

S a R e c o r d e d i n 1 9 5 1 a n d 1 9 5 2 .Sacompac t d i sc .( : a n r l n a r o c n z n i -- - - - . - . - - s .

0 S a S y m p h o n i e s ( V i o I i n w i L h o r c h e s t r a )0 S a V i o l i n w i t h o r c h e s t r a .0 $aV io l i n w i t h o r ches t ra , A r rang red .

1 0 S a S t e i n b e r g , W i l l i a m , 9 d l 8 9 9 - 1 9 7 8 . S 4 c n d1 0 $ a S o l o n o n , I z L e r , $ d l 9 1 0 - S 4 c n dl 2 S a l , a l o , E d o u a r d , $ d 1 8 2 3 - f 8 9 2 . g t s y m p h o n i e e s p a g n o L e . S f 1 9 8 8 .2 2 S a S a i n t - S a e n s , C a m i l l e , S d 1 8 3 5 - f 9 2 1 . $ t H a v a n a i s e . 9 f 1 9 8 8 .1 2 S a S a r a s a t e , P a b l o d e , $ d 1 8 4 4 - 1 9 0 8 . $ t Z i g e u n e r w e i s e n ;

5 o a . r r - 5 r t v d u -1 2 $ a C h a u s s o n , E r n e s t , $ d 1 8 5 5 - 1 8 9 9 . S t P o d m e , S r n v i o l i n ,

o r c h e s t r a . S f 1 9 8 8 .2 2 $ a s a i n t - S a e n s , C a m i l l e , $ d l 8 3 5 - 1 9 2 1 . $ t l n t r o d u c L i o n e t r o n d o

c a p r i c c i o s o . $ f 1 9 8 8 .2 0 S a R C A V i c t o r S y m p h o n y O r c h e s t r a . s 4 p r f

5 1 85 0 05 0 06 5 06 5 06 5 07 0 07 0 07 0 07 0 07 0 0

7 0 0

7 0 0

7 1 0

Figure 9. Tu'o lvlARC 700 Fields Used in the Bibliographic Record for a Compact Disc to Specifythe Cor.rductors of Di{I'erent Pieces.

LRTS .

Parsons'"Ave Maria" because both terrnsoccur in that record. The rnachine findsthe author "Palestrina" and the title "Ave

Maria" and registers a {'alse hit.This will rernain a oroblem until newer

software such as Holizon, Voyager, andothers make it possible to limit the searchfor an author and a title to the sarne lield,excluding all other hits Because"Palestrina" and "Ave Maria" in the exam-ple given above are not in the same 700field, such an advanced systern wouldavoid the f'alse drop Even that, however,will not solve the problem of coauthors,editors, performers, conductors, and thelike. In bibliographic records that standfor single-work items, or "standaloneworks" is we could call thenr. such names.entered in separate fields, are ipso factolinked to the work The rnornent trvo orrnore rvorks are described in the samebibliographlc record those relationshipsbecome blurred. Consider the case of theRCA Victor Gold Seal recordinq7709-2-RG. The OCLC record has a 245subfield a "showpieces " There are sev-eral added entrv-fields. two of them forthe conductors'(see figure 9) This CDcontains five pieces, fbur of u'hich aredesignated by analytic 700 fields Even ifauthors' names and work titles in theseanalytic 700 fields rvere linked, there isnothing in any indexed field that would al-low a library user to focus on the works forwhich each conductor is responsible. Asearch for conductor (in rnost online cata-logs, conductors'naffles are included inthe author index) "Solomon" and title"Zigeunerweisen" will retrieve some-thlne, but it will not be that work con-ductid by Izler Solomon.

There are still other cornplications\Abrks often appear in tianslationCalderont Lile Is a Dreatn, when publishedas a separate book, is represented by a bib-Iiographic record that automatically linksthe English title rvith the original La \tidaEs Sueno in a separate {ield The same goesfor nicknames and altemate titles If asearch for "Eine kleine Nachtrrusik" brinqsup a standalone work titled "serenade inG," we can be sure that rve are dealing rvithone and the same u'ork.

But when the u,ork is one of many in an

42(4) . Notes on Operations /299

anthology and there is only one biblio-graphic record for the itern and its con-ienis, the links betr,r'een such separatefields and the works to which they belongare more difficult to establish. The cata-log retrieving Chekhov's "Vishnyorry sad"should eouate that rvork with "The

Cherry orc-hard." However, as long as alltitles contained in a multiwork item are inseparate fields, the catalog has no preciseway to establish and display this link.

One other detail needs to be consid-ered. \\'hen an added entry in the 7xxfields names a work that is contained inthe item named in {ield 245, the secondindicator is given the vahte of"2" to showthat the added entry is an analytic one.This indicator is not as helpful {br the ana-lytic retrieval ofernbedded rvorks as onewould hope. In the first place, there is, atthis point, no library softu'are availablethat would, on retrieving such an embed-ded work, create a display such as the onefound in fiqure 10

But even i{ there were such softlvare,it would work only if the 7XX fields con-tained the actual title ofthe rvork sought.In many situations this is not true. A cer-tainrecordinq (EMI CDC 7 49656 2)con-tains Beethovens Symphony no 5 in Cminor, op. 67. A title search lbr this workwould, in most libraries today, not re-trieve the EMI recording because the rel-evant analytic added entry uses the col-lective uniform title "Symphonies." Thesecond indicator, according to the rules,marks this 700 field as one that designatesan analytic work contained in the item inhand. But in this case subfield t desie-nates a class of 'works, not the specif icwork sought. Needless to say, a search forSymphony no. 5 finds no match.

It would appear that, {brworks containedin collections and anthologies, not all au-thor/title/subject searches can be handled byMARC 700 added entry {ields, even if ana-lytic indicators and "in-the-sarne-field" strat-agems are employed, unless the field inquestion is expanded to contain coauthors,editors, performers, translated titles, subj ectheadinqs, etc.-all the inforrnation that instandaloneworks appears in separate fields.

Conceivably, added entry analyticscould also be constructed using separate

300/ L R T S . 4 2 ( 4 ) . H o f J m a n

0 1 00 2 80 2 80280 2 80284280280 3 30 4 00 4 L0 4 50 4 70 4 80 4 80 5 01 0 02 4 0

2 6 03 0 04 4 05 0 0

5 0 05 0 05 0 05 1 1

S a 9 1 7 5 1 7 3 6 / R / r 9 1 200 SaCDs 7 49852 2$bEMI0 0 S a c D c 7 4 9 7 4 6 2 S b E M I0 0 S a c D C 7 4 7 6 9 8 2 $ b E M I00 SaCDC 7 491 -01 2SbEMI0 0 S a C D C 7 4 9 6 5 6 2 $ b E M r0 0 S a C D C 7 4 9 8 1 6 2 $ b E M I00 SaCDC 7 4922L 2ibEMI2 $ a 1 9 8 6 0 7 - - S a 1 9 8 8 0 8 - - $ b 5 7 5 4 S c L 6

SaDLCScDLCSdDLC( r i a a r ( c n a r a n n f r o ( h n a r ( n a n n f r a a a r e h a n n

s b d 1 7 9 9 S b d 1 8 2 4 0 1SasyS aovSaoa$bva0 1 $bvb0 1 $bvdO 1 $bvf 0 1 $aca$ aoaSaEMI CDS 1 49852 2S a B e e t h o v e n , L u d w i g v a n , S d l 7 7 0 - 1 8 2 7 .Sao rches t ra mus i c . SkSe lec t i onsSa9 S in fon ienSh Isound reco rd i ng ] =$bSymphon ies / $cBee thoven .S a H a y e s , M i d d l e s e x , e n q l a n d : $ b E M I , S c t 1 9 8 9 1S a 6 s o u n d d i s c s : S b d i g i r a 1 , s L e r e o . ; S c 4 3 / 4 t n .

0 SaRe f l exe$aEMI : CDS 1 49952 2 (CDC 7 49746 2 , C rx :7 47698 2 , CDC 7 491012 , cDC 7 49656 2 , CDC 7 498L6 2 , CDC t 49221 2 ) .SaThe 12 th wo rk sung i n Ge rman .S a T i t l e f r o m s l i p c a s e .$aTex t o f t he l as t movemen t o f t he l , 2 t h wo rk by Sch i l 1e r .

0 S a Y w o n n e K e n n v . q ^ n r . n ^ q ^ r ^ h { ^ 7 e l k F . m e ? z n - q 6 n r r - n p ^ f r i c k

Power , t eno r , Pe t t . e r i Sa lomaa , bass , Schu t z Cho i r o f London1 2 t h w o r k ) ; L o n d o n C l a s s i c a l P l a y e r s ; R o g e r N o r r i n g L o n ,conouc to r .

5 1 8 S a R e c o r d e d J u l y 1 9 8 6 - A u q . 1 9 8 8 , N o . 1 S t u d i o , A b b e y R o a d ,London.

5 0 0 $ a c o m p a c t d i s c s .500 SaProg ram no tes by Dav id Wyn Jones and pe r f o rmance no tes

by Roge r No r r i ng ton i n Eng l i sh w i t h F rench and Ge rmant r a n s l a t i o n s a n d t e x t o f t h e l a s t m o v e m e n t o f t h e l 2 t hw o r k i n G e r m a n w i t h E n g l i s h a n d F r e n c h t r a n s l a t i o n s ( 9 5p . : i 1 1 . ) i n c l u d e d .

5 0 5 0 S a S y m p h o n y n o . 1 i n C m a j o r , o p . 2 1 , - - S y m p h o n y n o . 6 i n Fm a j o r , o p . 6 8 : P a s t o r a l e - - S y m p h o n y n o . 2 i n D m a j o r , o p . 3 6- - Symphony no . 8 i n F ma jo r , op .93 - - P rome theus ove r t u re :o p . 4 3 - - S y m p h o n y n o . 3 i n E t l a t m a j o r , o p . 5 5 : E r o i c a - -S y m p h o n y n o . 4 i n B t i a l m a j o r , o p . 6 0 - - S y m p h o n y n o . 5 i nC m i n o r , o p . 5 7 - - O v e r t u r e t o C o l l i n ' s t r a g e d y C o r i o l a n :o p . 6 2 - - O v e r t u r e t o G o e c h e ' s C r a g e d y E g m o n t : o p . 8 4 - -Symphony no . 7 i n A ma jo r , op .92 - - Symphony no . 9 i n Dm i n o r , o p . 1 2 5 : C h o r a l .

6 5 0 0 $ a S y m p h o n i e s .6 5 0 0 S a o v e r t u r e s .6 0 0 1 0 S a S c h i l l e r , F r i e d r i c h , $ d 1 7 5 9 - 1 8 0 5 S x M u s i c a 1 s e t t i n g s .7 0 0 1 0 S a K e n n y , Y v o n n e . $ 4 p r f7 0 0 1 0 S a W a l k e r , S a r a h . $ 4 p r f7 0 0 1 S a P o w e r , P a t r i c k , S d 1 9 4 7 - $ 4 p r f7 0 0 1 0 $ a S a l o m a a , P e t t e r i . g 4 p r f7 0 0 1 0 $ a N o r r i n g t o n , R o g e r . $ 4 c n d7 0 0 I 2 S a B e e t h o v e n , L u d w i g v a n , $ d 1 7 7 0 - 1 8 2 7 . S L S y m p h o n i e s . $ f 1 9 8 9 .7 0 0 1 2 S a B e e t h o v e n , L u d w i g v a n , $ d 1 7 7 0 - 1 8 2 7 - $ t G e s c h o p f e d e s

Prome theus . $pOuve r tu re . S f 1 989 .7 0 0 I 2 S a B e e t h o v e n , L u d w i g v a n , S d 1 7 7 0 - 1 8 2 7 . S t c o r i o l a n . 5 f 1 9 8 9 .1 0 0 1 2 S a B e e t h o v e n , L u d w i g v a n , S d 1 7 7 0 - 1 8 2 7 . S t E g m o n t

S p o u v e r t u r e . S f 1 9 8 9 .7 I 0 2 0 S a H e i n r i c h S c h u t z C h o i r . S 4 p r f' 7 I 0

2 0 S a l o n d o n C l a s s i c a l P - l a y e r s . S 4 p r f

02

0 01 01 00 00

Figure 10. Bibliographic Record {br an "In" Analytic Structured in Accordance *'ith AACR2rule rJ D.

LR?S . 42(4) . Notes on Operations /30-

1 - 0 0 1 0 S a W i l l i a m s , T e n n e s s e e , S d l 9 1 f - 1 9 8 3 .245 L0 SaSummer and srnoke' 7 1 3

$ w l l i n k t o h o s t i t e m ]

Figure 11. Analytic Record Example

700 and 740 fields instead of 700 subfielda and subfield t. The problerns are thesame: the comDuter cannot tell rvhich au-thor's name go-es vvith what title, or whichsu\ect or performer might belong towhich 740 field, and so on.

The added entry rnethod using MARCfields 7XX has merit, especially if proxirn-i ty and "in-the-same-f ield" l imits, as wellai use of the second indicator, are incor-oorated into librarv software. But rnisses^and

false drops ale still possible eventhen. Nor can collocation or clear disolavofretrieved work titles be achieved in ailcases. AIso, as long as all componentworks are listed in one and the same bib-liographic record, the title ultimately dis-played to the reader is bound to be theone in field 245 If the table of contentsmethod for documenting analytics has itsshortcomings, the added entry method isnot completely satisfactory either.

. . IN,,ANALYTICS

The "In" analytic, described byrule 13.5Aof AACR2R, is a third possibility. In an on-line catalog, such analytics consist ofsepa-rate bibliographic records for all compo-nent works, whether contained incollections or anthologies, each with itsown MARC 245 and lxx field, as appropri-ate Such a record rvill also contain a "Host

itern entry" MARC 773 field, a field that isdescribedin MARC rnanuals as containingin{brmation concerning the host itern foithe constituent unit described in the re-cord (vertical relationship) As the scopestatement for 773 explains, this {ield is pro-vided in order to enable the user to locatethe physical piece that contains the com-ponent part being described

For the example shown above, EighfPlays by Tennessee \\/illiarns, eight sepa-rate "daushter" records would be created(see figuri 11). This stratagern producesreliable retrieval because the author and

title are unequivocably linked. In an au-thor/title search, there is no possibility oflinkinq an author and a title that do notbelong together, because there is onlyone title in the bibliographic record. Ifthis method is used consistentlv. the cata-log, searched by author and tiile, rvill re-trieve all rnanifestations o{'a'",r'ork in onepass. There can be no false drops. Andthehisplay will clearly collocate ail retrievedmanifestations of a work without inter-vening titles that are not those ofthe rvorksought (see figure 12).

The "In" analytic also solves the prob-lem of linking joint authors, performers,alternate titles, uniforrn titles, subjectheadings, etc , to the u,orks they pertainto because each work is represented by itsown bibliographic .ecord. All lieldi inthat record are. bv de{inition and withoutspecial programming, linked to the worknamed in field 245. A search for author"Beethoven" and title keyword "Octet"will retrieve all and only records that carrythis author's name and this title ke;nvordin one ofthe fields A false drop such asthe London 421 093-2 disc, mlntionedabove, is not possible. The reader lookingfor Euripides' Medea.willnot ever be con-fused again by being presented withAlcestis instead. If you are looklng fbrSchnitzlerk Game of Loaa, you will notonly find it, but find it on the first try

The subject approach to cornponentworks should not be lbrgotten either.Manv studies in librarv literature confirmthe importance ofthe subject approach,especially in acadernic libraries. Larson

l . W i 1 1 i a m s , T e n n e s s e eS u n m e r a n d s m o k e . . . . 1

2 . W i l l i a m s , T e n n e s s e eSummer and smoke . . . . . 1

3 . W i l l i a m s , T e n n e s s e eSummer and smoke . . . . . 1

Figure 12. Brief Displayof Analyic Records

302/ LRTS . 42(4) . Hoffman

(1991), for exarnple, concluded that sub-ject searches are not only rnost used butare also rnost likely to lail. \\/ithout inves-tigating in detail the several reasons forfailure, it is easy to see one ofthem \\/henmany works are packaged together in onecontainer and each work deals rvith a dif-ferent subject, searchers are likely to missimpoftant information. The prirnary rea-son for this is that multiwork items are of-ten assigned only a heading that surnma-rizes the subiect content o{' all thecomponent works. ?ha Analytic Spirit:Essays in the History of Science (CornellUniversity Press, 1981), fbr example, wasassigned the summary heading Science-History. In the contents {ield, the titlesand authors of all 15 essays, ranging {roma discussion of Lavoisiei's theriry"of thegaseous state to the supernova of 1054,are listed. \\/ill a reader interested in theCrab Nebula look under the broad head-ing "Science"? Probably not.

Librarians at libraries where usershave a strong interest in astronomy, re-membering Ranganathan's fourth law("Save the t ime oI ' the reader"), miqhtwant to make it easier lor readers to ac-cess interesting materials by adding es-say-specific subject headings to a booklike this. As long as the essays are listedonly in a 505 field, or even in a 7XX field, itis irnpossible to tell which subject headinggoes with what essay. In such a case, the"In" analytic seems indicated For the es-say by L. Pearce \\/illiams, "The supernovaof 1054," a separate bibliographic recordmight be created, complete rvith author, ti-tle, specific subject headings, and link tothe mother record.

The reader searching by subject willnorv retrieve all relevant "standalone" asvvell as component works. The search isprecise and exhaustive, requires only onepass, and the retrieved rnaterials rvill beneatly and unequivocally collocated onthe screen.

CoNcrusroNl AND RECoNTMENDATToN

There are three major methods employedin libraries to catalog cornponent rvorks-contents notes, added entries, and "In"analytics. Under the third method, sepa-

rate bibliographic records for all cornpo-nent works are created, and these arelinked bv a common kev to their host re-cords, the collections and anthologiesthat contain the works. If all collectionsand anthologies on a library's shelves areconsistently cataloged by this rlrethod, asearch for a specific work will:

1 Retrieve all manifestations of thatwork,

2 Retrieve only that work without {alsedrops,

3. Not require a second pass, and4. Clearly collocate all retrieved rvork

titles

If librarians plan to o{fer their readersreliable analytic catalogs that per{brrn wellon these four points, then the choice ofre-cord forrnat will have to be given seriousconsideration It appears that "In" analyt-ics linked to their host item records, arnethod so far neqlected in most onlinecatalogs, rnight be ihe choice ofthe {uture

Woms Clrrp

Anglo-American cataloguing ntles 1988,Prepared under the direction of tlie JointSteering Committee for Revision ofAACR,; committee of the American Li-brary Association, the Australian Com-mittee on Cataloguing, the British Li-brary, the Canadian Committee or.rCataloguing, the Library Association, theLibrary ol Congress, 2d ed , 1988 rer' , edMichael Gorman and Paul \1'. \\/inklerChicago: American Library Association

Borgman, C L 1996. \\\ are online catalogsstill hard to use? Journal of the AmericanSociety for Information Science 4'l: 493-503

Carlyle, A 1997. Fulfilling the second objec-tive in the onlir.re catalog: Sclremes lbrorganizing author and u'ork ret'ords intousable displays Library resources Ct tech-nical sercices 4l: 79-100

Hagler, R. 1997 Access points fbr works Pa-per presented at the International Con{er-ence on the Princioles and Future Devel-opment of AACR, Torotrto, Ontario,October 23 Available at: http://lrrsrnlc-bnc ca,/jsc/r-access pdf

Hoffman, H H 1976 Descriptixe catalogingin a neu /ig&t Ne\r?ort Beaclr, Calif :Headu'ay Publications

Ho{ Iman, H H,andJerue lL Magner 1985.

LRTS

Future outlook: Better retrieval throughanalvtic catalogs Journal of academic 7i-D r0rlAnsn lD -t I: IDI-bJ

Hou'arth, L C i997 Content versus carrier.Paper presented at the International Con-ference on the Princinles and Future De-velopment of AACR, Toronto, Ontario,October 23 Available at: htto://urnr'.nlc-b nt .ca,/j sc/rcarrier pdf

Larson, R. R 1991 The decline of sublectsearcl t ing: Long-term trends and pat t i rnsofindex use in an online catalog. lou.rnal ofthe American Society for lnformation Sci-ence 42:197-215

Lubetzky, S. 1963 TIre function of the mainentry in the alphabetical catalog-oneapproach Report ofthe lntetnational Con-

ference on CataloguingPrinciples, Paris, Oc-tober 7961 Quoted in Fattahi, RahmatollahRelevance ofcataloguing principles to theonline environment: An historical and ana-

. 42(4) o Notes on Operations /303

lytic study Ph D diss , Univ of Neu'South \\'ales, Sydney, 1996

McCallum,S H 1980 MARC formatfor ana-Iytics (draft) \\Tashington, D C : Library ofCongress, Netu'ork Deveiopent Office

Poulsen, C. 1S90 Tables ofcontents in librarycatalogs: A quantitative examination ofan-alytic catalogs Library resources ds techni-cal seraices 40: 133-38

United States Educational Research lnforma-tion Center 1978 Thesaurus of ERICdescriptors, Tth ed.

Verona, E. I97t Statement of pr incip les:Adoptetl at the lnternational Conferenceon Cataloguing Principles, Paris, Octo-ber 1961, annotated edi t ion, London:International Federation of Library Asso-ciat ions

\\reintraub, T. S , and\\'. Shimoguchi 1993 Cat-alog record cortents enhancement Libraryresources b technical serxices 37: 167-30.

304/

Mendigoinero: Scoling thePeok of the Bosque LibroryBocklog

Morcelino Ugolde ond Kothryn Etcheverrio

The Basque Shtdies Collection, housed at the (Iniaersity of Neaada, Reno, isuoorltl-renou;ned as the most extensice collection of materials both by an.dabout the Basqu.e people. Efforts to keep pace uith the catalogingbacklog ofthis multilineual collection haae met with limited success. This article de-scribes the Basque CatalogingProject, administeredfrom 1993 to 1996 and

funded by an HEA Title II-C gran.t, as it uaorked to produce catalogingfor10,000 monographic titles. Atten.tion is gioen to prepa,ration for the project,special considerntions in catalogingthis unique collection, and recommenda-tions for the future

Collection development efforts by theBasque Library continued to llourishthrough annual appropriations of approx-imatel-y $f0,000 from the university Ii-braryt materials budget. Since 1986, theBasque Library has acquired rnore than1,000 new and retrospective titles annu-ally, receiving 2Vo of tie librarys total ma-terials budget (Basque Studies Prograrn1987). In 1997. the number ofnew t i t lesadded reached 1,300, with almost of all ofthese titles going directly into anuncataloged backlog.

Tnn Gnowrnc BAcKLoc

\4'hy had the backlog of the Basque col-lection become so large and unmanage-able? A variety offactors beyond the con-trol of the university and its BasqueStudies Program rnay have created orcontributed to the situation, including

-Urthnic studies programs cannot flour-ish without a strong library collection.From its inception in 1967, the BasqueStudies Program at the University of Ne-vada, Reno, has understood the impor-tance ofbuilding a strong Basque collec-tion to support scholarly researchactivities. The lbundation of the BasqueStudies Library rvas tu'o respectable pri-vate libraries totaling more than 4,300works. The purchasJ ol these librariesfiom Europe was f'acilitated by RobertLaxalt and the late Ton Bilbao. RobertLaxalt made contacts ivith the Veyrin fam-ily in the late 1960s, and Jon Bilbao spoketo the Gofri f'amily tn 1972 (BasqueStudies Program 1983). Fortunately, thematerials in these libraries comple-mented each other. One ernphasized thenorthern Basques in France and the otherfocused on the southern Basoues ofSpain, especially the Navarrese region.

MARcELINo Ucelor (ugalde@unr edu) is Basque Studies Librarian, Universitl,ofNevada, RenoKATHRYN Ercsrvrnnt'r (kathryne@unr edu) is Basque Catalog Librarian, University of Nevada,Reno. Manuscript received January 9, 1998; accepted for publication April 14, 1998,

LRTS .

publishing trends in the Basque country,limlted e{Torts related to international co-operation in cataloging, and the conse-quent lack of catalog copy for {breignmonographs in the format usable byAmerican libraries.

\\/hile publishing rnay have stabilizedor declined arnong other linguistic or eth-nic groups, literiture publ=ished in theBasque provinces in monographic formhas experienced a boorn in recent years.Otero-Boisvert (1992) points out that theBasque culture, long iuppressed underthe Franco regirne, just recently beganproducing rnore public evidence of awrit-ten culture and literature, while othercultures r,r'ere already under the strong in-fluence of a r'vell-established publishingindustry. Afler Francot death, publica-tion in the Basque language became a po-litical statement made by those with na-tionalistic and linguistic enthusiasm forthe Basque homelind. This, coupled witha dramatic increase in the learnine of theBasque language by the young people ofthat region, may have been influential increating a 63 SVo increase in new Basquelanguage publications in 1985.

In an eflbrt to keep pace with thispublishing boom, Iibrarians at the Uni-versity of Nevada, Reno, responded withan ambitious acquisitions program. Thelibrary's policy was to acquire virtuallyeverything published by and aboutBasque people. The size of i ts mono-graphic collection grew {rom 13,997 ti-t les in 1980 to 25,642 t i t les in 1990, an av-erage annual increase of approximately107o during those years

The University of Nevada, Reno Li-brary along rvith other academic and spe-cial libraries, rvas unable to acquire thequality and cluantity of catalog recordsrepresenting foreign language mono-graphs that are usually available for do-mestic acadernic titles The problem iswell docurnented Grover (1991) found inhis 1983-1985 study that although LatinAmerican rnaterials lvere widelv acquiredby large acadenric and public l ibraries, ad-equate cataloging {br the rnaterials r,r'as of -

ten not provided. Results of the study ofblbliographic control by Leazer and

42(4) . Notes on Operations /305

Rohdy (1995) lend support to the,theorythat far less control exists over {breignrnonographs than dornestic rnonographs;moreou"i. bibliographic records availablefor foreign monographs tend to be lowerquali ty than those ol 'rrronographs in gen-

eral (Sercan 1994). Furtherl the avai l ibi l-

able catalog copy compared to Grover(199r)

One poss ib le reason beh ind the lackof catalog copy lbr foreign rnonographsis the slowdo*, ' t of onci ambit ious-ef-forts to achieve international coopera-

tronic versions o{ records that necessi-tate ineflicient rekeying of catalogingrecords ofother nations and languagesFor example, in the case of the BasqueLibrarv, r ' rhere appt,,* imately 6074 of i tsco l lec t ion is pub l i shed in Spa in , a too lwith l imited use{ulness has been the

transf'erable.

lLC) records, the rest being member copy.Clearly. a cataloging project that would ad-dress the needs of this unique collectionwould include provision foi professional

306/ LRTS . 42(4) . Ugalcle and Etcheaerna

TABLE IPREPRoJECT SAMpLE SuencH Rnsulrs

T}?e ol OCLC Record Classes D, E, F Class p Basque Lanqrraqe Total

DLC

Member

Total found

No matcl.r

Total

22/20Ea

40/36Va

6A56Eo

49/44Eo

IIt/t00Eo

19/l2%o

49/32Vo

68/44Vo

85/56Vo

I53/I00Vo

6/6 57o

6/6 5Vo

I%L3Vo

78/87Vo

90/I00Vo

47/I3Eo

95/27Eo

l lA40Vo

2IA60Eo

354/I00Va

sta{f skilled in original cataloging in a mul-tilingual environ;ent, as weii as classifiedstaff with the ability to rvork with a highpercentage of member r-ecords, rvhich of-ten require additional tirne to add LC callnumbers and subject headings.

_ Searching tracked during the course ofthe cataloging prolec.t resuited in sl isht lyhigher hiiraGi tlian those shorvn in" theprior random sarnple (overall hit rate40Vo),bt lou'enough to justify the staff-ing needs described above.

EARLY EFFORTS ToCereroc rHE CoLLEcrroN

Providing bibl iographic access for thismult i l ingual c..r l lect lon lras ahvays been achalleng!. Despite all ellbrts ind goodintentions ol'the technical services stafl'

fort than the technical services stafl'could supply.

In order to provide access to the col-lect ion. a bihl iographer handled localprocessing and assigned cal l nurnbers forihe rnaterlal. He atapted lrom the LCclassification system temporarv callnumbers using the classilicaiion nLrnberportion and then added either the au-

thor's name or a significant word {iomthe title. He was aisisted in the localproccessing of various material Ibnnats{br many years by students and classifiedstaff.

In 1983, two catalogers rvere hired fora year through funding from the BasqueAutonomous Governrnent in Spain toachieve as much cataloging as pbssible.After one full year of hard u,ork, a signifi-cant portion of'the Basque Library re-ceived {ull-level access. It u'ould beeleven years before another serious e{Iortwas made to reduce the tremendousbacklog of Basque material.

A CererocrNc GRANT

cation (an HEA Title II-C grant), the uni-versity hired a project direitor. two cata-logers, three classified stall ar-rd studentworkers. The r;rimarv focus u'ould be tocatalog 10,000 rnonographs in the BasqueStudie-s Library. lt rias anticipated t'hatthe project sta{f u'ould, in a period ol hvoyears, produce more than 10,000 biblio-graphic records hy creating original cata-log-copy or by edit ing exist ing"lut l- levelMAchine Readable Cataloging (MARC)records

Other objectives included cornpletiono{'related authority vvork and prornotionof increased use of the nervly cataloqedtitles and the collection as a u'hole rvitlinthe Basque scholarly cornmunity.

LRTS .

THe Hunnlss

LaNcuacn TneINIttc

The probability of hiring catalog librari-ans with the necessary skills in the Basquelanguage was predicted to be very slim,and such turned out to be the case. Thelength ofthe project did not allow enoughtimi for the catiloeers to learn the lan-guage to the extenidesired, but trainingwas tailored to make them bibliographic-ally proficient.

dinal number translations, and tide pageconvention descriptions. Additional toolsincluded a photocopy of a name authoritylist of Basque authors, with pseudonymsproduced by the International Society ofBasque Stu&es, and a photocopy of thesection about Basque from Allen (1975),which presented a grammatical structureand background for the language.

A two-week intensive language train-ing session began as soon as the catalogingstaff arrived. The focus of the two-weeksessions was to discuss the handouts andto produce some sample cataloging for avariety of works: current publications,older backlog monographs, and rare booktitles. The catalogers quickly becameaware of the differences between anon-Indo-European Ianguage (Basque)and Indo-European Ianguages (Spanish,French, English). After the two-weektraining, the instructors made arrange-ments to meet with the cataloging projectstaff two hours weekly or when needed toanswer questions and solve problems dur-ing the {irst year of the cataloging project.During the remaining term of the project,language assistance was provided to thecataloging staff when necessaly, usually afew times a week.

42(4) . Notes on Operations /307

BIBLIOGRAPHIC CONTROL

and not transferable to the PRISM Catmesoftware, all data had to be keyed.

DpscnrpTWE CATALOGING

the face are shown in figure 1.

LI, TOUEE BISAYE PIP L COLI,S

EL PIIS VISco POP CINLOS ESPEL YNITGI I

VICTOP I.EI.XI GII,ZTI,TNO

EYSXIL IPYNXTX

Figure l. Examples of the Basque fireface.

308/ LRTS . 42(4) . Ugalde and

Characteristic in the Basque font isconsistent capitalization of all the wordson the title page, while common usage inBasque calls for the same rules of capital-ization as in Spanish and French. This ex-asperating style, combined with the highlysl,nthetic and post-positional nature ofthelanguage, makes the title page and cover ofa Basque publication sometimes appear asan unlocked puzzle. Vl/ithout sorne knowl-edge of Basque vocabulary grarnmar, andsyntax, B.asque on the title page is nearlyindecipherable.

Interestingly enough. the Basque lan-guage is free dfdiacritics. Other languagescommonly used in and around the Basqueculture and found in the collection, such asFrench and Spanish, are not Those atwork cataloging these materials f'am iliarizethemselves with both key equivalents forPCs using PassporVPRlSM/Catme plussoftware, L weil L the three-digit numeri-cal codes representing diacritics commonto those languages, which display in curlybrackets in place of'the diacriiic in the In-novative Interfaces system display.

AurHoRrry \4/oRK

Name authority, particularly in a multilin-gual environment, always presents chal-lenges for catalogers. LC has led the wayin many cases by establishing names witha preference for the Castilian spelling ofnames over the Basque spelling. Recentagreements. however, grant the Basquelanguage co-official status rvith Castilian inthe Basque region o{'Spain (Pais Vasco,Spain, Gobierno Vasco, SCP 1983). Now,most modern Basque writers pref'er to usethe Basque orthography ol'their names,reinforcing their cultural di ll'erences frornthe rest ofspain and France (see table 2)

Etcheaerria

TABLE 3Besgur \4/nrrnns AND PSEUDoNyMS

Real Name Pseudonvm

Nikolas Ormaetxea Orixe

Pierre Topet

Jose Luis AlvarezEnparantza

Emilio Lopez Adan

Esteban Urkiaga

Etchahun

Lanesoro orTxillardegi

Beltza

Lauaxeta

TABLE 2Bnsgun Nnuns

Name authority is further complicatedby the use ofpseudonlnns or nicknamesamong Basque people, a culturally ern-bedded custom datine back to the thir-teenth century not merely used by liter-ary figures (\.\'hite 1992). Basque peoplewere known by their house name ratherthan their surname (Lhande 1908, 14-16;Caro Baroja 1971, 211-13; Manterola1980, 537-600), a practice so pervasivethat most young men did not knorv theirsurnames until they serwed in the militaryand found surnames on their mihtary doc-uments. Some prominent writers who arewell-known by their pseudonyms areshown in table 3

Since 1975, pressure to pubhshBasque works in the unified Basque dia-lect, Euskara Batua, has compelled pub-lishers to use the single Basque surnamein preference over the old double Spanishsurnames (\d'hite 1992, 1I) \\/riting ca-reers that span the two eras often requirereferences in authority records.

Tools eNn Rnsouncrs Usro gy rHr Srarr'A valuable source in decoding andcross-referencing pseudonynns and estab-lishins authors' dates is the monumentalmulti,rolume work, Eusko Bibliographia,compiled bv Bilbao and considered themosf comprehensive bibliography of'Basque materials in the world. Initially,the set was 10 volumes spanning theBasque bibllographic years from the be-ginning of history through I975. In 1985,a three-volurne supplement was pub-lished by the Basque University Press,covering the years 1976-1980. From1987 to 1991, a bulletin was irregularlypublished, organizing information fiom

Spmish Form Basque Form

Angel Lerchundi

Jos6 Miguel

Domingo Aguirre

Echeverria

Anjel Lertxundi

foxe Miel

Txomin Agirre

Etxeberria

LRTS . 42(4) . Notes on Operotions /309

DECHEPARE, BERNARDUM

t t ^ - - ^ - , n ^ - : ^ i r i a a n e r n n m i r r r mD f t t 9 v a c

S a n c L i M i c h a e l i s V e L e r i s . I B u r d i g a ] a e :( 2 8 h . ) .

L IZARDI , XAB IER { seud6n imo de Agu i r r e ,

PELOTA (General)

RF rna rd l r t r DeeheDare RecLo remFranc i scum Morpa in , 15451

J o s 6 M a r i a ) .

A c k e r , P . : " F r o n t o n B a s q u e , " E c h o d e P a i - r s , 1 9 0 4 ( J L . 2 5 ) .

Adema , G . : - l / o t i ce - . u r - l e j eu de paune au reboE , j eu naE ionaT deR e c d n P C R a w o n r F . 1 8 9 4 .

Figure 2. Examples liom Ersfto Bibliographia

1987 to 1989 as a means {br the Associa-tion of Basque Bibliography to keep cur-rent with Basque bibliography Recently,the Basque University Press has com-pleted a second supplement to EuskoBibliographia. listing entries Ibr the years1980-1985. Bilbao dedicated most of hislife to Basque bibliography, travelingextensively throughout his lif'etime and re-cording entries for monographs, serials,pamphlets, leaflets, journal afticles, gov-emment docurnents, and manuscripts Ba-sically, everyrthing by and about Basqueshas been included Bibhographic entrieswere organized by the standard author,title, subiect, and series titles {brmat Allentries were inter{iled and alphabetized.Subject headings were developed usingSpanish subject headings, including placenames, mostly as a sign oI'the time periodBilbao was working in, during the Francodictatorship. Sarnple entries are shown infigure 2.-

The bene{its of havine access to thismultivolurne masterciece *'ere nume.o,ts.On many occasions the bibliography helpedthe catalogers in name authority rvork tolink sumames rvhen only authors' pseud-onyms were present; in addition, it helpedin establishing dates lbr certain authors.

A list that focuses strictly on pseud-onyms for literary {igures is Entrasaca deu.n diccionario de seu.donimos by Amezaga(1991). Though dated, the list, giving au-thor, prof'ession, century author flour-ished, and corresponding pseudonym orpseudonyms, is reliable.

Place names pose another problern areafor the Basque region in tenns of name au-

Herriko Autonomi Elkartearen AdministrazioOrokorr4 HABE 1993), reveals that theBasque fonn of a narne became o{ficial aboutas often as the Spanish fbnn. Catalogers re-l'erred to this list liequendy.

Another useful source to aid in deci-phering title page and prel'atory rnateriali ^t A i4 a n u ni ol E,, ru pb n, La n gt t a ge s fo rLibranans (f975). The chapter on theBasque language offers specific infbrma-tion on qrammar, syntax, surnatnes, andwhat to Jxpect on the title page, lbllorvedby a glossary of publ ishing terms.

SuulFlrsf lNc

the Basque Cataloging Proiect, a surveyofthe Basque col lect iorr shelf l lst shorvedthat in DP402, for exarnple, signifyingSpain, local history and description, othercities, towns, etc., 84 out of the existing220 Cutter numbers had been createdlocally. Similarly, ir-r the literature sched-

3f0/ LRTS . 42(4) . Ugalde and Etcheaerria

ule, where PH5339 represents all rvorksof literature by lndividual Basque au-thors, an irnpressive I95 out <,1'308 au-thors represented in the UNR online cat-alog (\{'ol{PAC) were established at theUniversity of Nevada, Reno

Beyond published sources, projectcatalogers had exclusive access to anumber of useful documents "Lan-guage information: Enghsh, Spanish,Basque" was compiled speci{ically lbr theproject stafl by the Basrlue Studies De-partment. It was a quick reference tool thatpulled together such inlbrrnation as pub-lishing terms, grammar, slmtax, and majorpublishing houses rvhose works were foundmost frequently in the collection

A SuccEssFUL CoNcLUSroN. . . .Wuer ts rlr Sronn?

As ofFebruary 29, 1996, the closing datefor the project, staff'had produced 3,425

Newly cataloged titles were primarilypromoted via the online catalog. E{Iortsto infbrm Basque scholars rvere madethrough the BSP Nervsletter publishedbiannually. Further ease of access wasachieved when the University ol Nevadamade the online catalog available via the\\brld \\,ide \\'eb.

log project rvas, rvithout question, success-

{ul in meeting its goals, but follorvup in theform of continued maintenance and sus-tained cataloging {br the Basque collectionis paramount to ward off huge backlogs ofuncataloged rnaterial.

Also important to reducing backlogsfor this special collection, as well as otherspecial collections, is a {resh attitudeabout cataloging from catalogers. \\/e mayneed to discardthe ideal of"oerf'ect" cata-loging to conquer a backlof. Along withan altered ideal, catalogers could saveprecious time by becorning more accept-ing of cataloging from other institutions.\\/hen editine the valuable {'ew mernberrecords availible, catalogers for this andother special collections do rvell to set aspecific ceiling on the amount of timethey will expend on editing records

Reflection on this project also pro-duced some thoushts on successful siaff-ing. First, projectitaff'were s'ell suited totheir task once language orientation tookplace. One might believe that being profi-cient in a language is necessary to catalogin that language, but cataloging experi-ence and demonstrated language abilityof some kind misht be sufficient Admin-istrators rvere oleased with the level oflanguage ability catalog staff held oncethe project rvas up and running, and addl-tional language assistance r'r,as availablewhen needed. Less than satisfactorv.however, was the r,onliguration ol staft.\\/ith such a low hit-ratebf OCLC OnlineComputer Library Center, Inc. recordsagainst the collection, hiring of one addi-tional cataloger u'ould have yieldedbetter results, i.e., rnore original catalog-ing produced. Sirnilar projects wouldbenelit lrom being sure to have suflicientpro{'essional catalogers to produce cata-loging Ibr a special language collection.

In the future, adequate bibliographicaccess to Basoue materials u'ill continueonly when cooperative efforts are estab-lished betrveen institutions. The Univer-sity of Nevada, Reno Library supplies themajonty of bibliographic cataloging re-cords of Basque language rnaterial toOCLC. Because eight to ten foreign librar-ies as well as a handlul of U.S. academic li-braries collect titles on Basque subjects, co-operative cataloging might be the answer to

LRTS .

keeping up-to-date with nerv Basque lan-guage material, as well as reducing thebacklog. A cooperative cataloging plancould create rnore bibliographic records{br other institutions to use selectively andreduce duplication of international cata-loging efforts (Byrd f993)

There is no Basoue studies collectionin the United StateJas cornprehensive asthe one at the University of Nevada,Reno. Fortunatelv. there is an institutionin the Basque country ofnorthern Spain,the library ofthe Sancho el Sabio Founda-tion, that manages a strong Basque collec-tion that is similar to and slightly largerthan the Basque collection at the Univer-sity o{' Nevada, Reno Un{brtunately,Sancho el Sabio uses IBERMARC, theMARC fbrrnat developed by the Spanishstate and adopted by the Basque countryas well as the rest of Spain, which differsenough frorn USMARC or UNIMARC toprevent IBERMARC catalog recordsfiom beins converted to the UNIMARCfbrmat foi OCLC. Further, OCLC hasdecided to stop allocating time and re-sources to developing sof'tware to convertvarious MARC versions to UNIMARC.At the present time, institutions in theBasque country are not r'villing to investtime or resources to convert their caterlogrecords to the UNIMARC fbrmat. Thi idilemma could leave the Basoue StudiesLibrary in an unlirltunate pbsition tbrproviding adequate access to its collec-tion in the future

More serious rvork on cooperation be-tween cataloging agencies atihe interna-tional level is paramount The benefits of

ture, the Basque Studies Libraryrvill haveto rely on cooperation and collaboration,instead of grants and other fundingsources, to sustain its cataloging needs.

WoRKs CrrED

Allen, C. G 1975 Basque In A manual of Eu-ropean lartguages for librarians, 77 4-99London, Nes' York: Bou'ker

Amezaga, Elias 1991 Entresaca de un

42(4) o Notes on Operations /311

&ccionario de seudonimos Rexista

internacional de los estudios tascos 36:

4L3-26Basque Studies Program, Unir 'ers i ty t - r f Ne-

r id^. R*r ' 'o . 1983. The BasqueStudies Pro-

gram, 1967-1982 Basque studies programneusletter 27( May): 3-8

Basque Studies Program, Universit)' of Ne-rid^, R"rlo. 1987. The Basque Studies Pro-gram. 1982-I987. Bnsque stwl ies progronlneusletter 36 (November): 3-5.

Bilbao, fon. 1970. Eusko bibliographia l0tols San SebastiSn: Editorial Auiiamendi,Estornes Lasa Hnos

Byrd. Jacquel ine. 1993. A cooperat ive catalog-

ing proposal for Slavic and East Europeanlanguages and the languages of the lbrmer

Soviet Union. Cataloging b classificationquarierly l7: 87-96.

Caro Baroja, Iulio. 1971 Los xascos lvladrid:

Ediciones ISTMOEtcheverria, KathD.n 1996 Basque Studies col-

lection cataloging project final performancereport, HEA Title llc, PL. 89-329 Reno:

University of Nevada, University Libraries

Euskal Herriko Autonomi Elkartearen Ad-ministrazio Orokorra, HABE 1993Euskal Herriko Udalen Izendegia In 1994Asenda, 32-42. Donostia: HABE.

Groter , Mark L. 1991. Cooperat ive catalogingof Latin American books: The un{ulfilled

promise Library resources d: technicalserxices 35: 406-15

Joachim, Martirr D. 1993. Issues and problemsin cataloging tl-re languages of the t'orld. In

Lan guage s of the u orld, C atalo gin g is sue s

and-problems, ed. Martin D Joachim,1-14. Neu'York: Has'orth Pr'

Lambrecht, fay H. 1995, International coop-

eration in cataloging: Progress and con-straints. ln Adaaicei in libiarianship , vol

19, ed I rene Godden,217*38 San Diego:Academic Pr

Leazer, Gregory H., and Margaret Rohdy.I995. Tire bibliograpliic control offoreignmonograpl.rs: A revieu'and baseline study

Library resou'rces G technical serxices 39:

29-42Llrande, Pierre 1908 Autour d'un foyer

basque, Rdcits et iddes Paris: NouvelleLibrairie Nationale

Manterola, Ander 1980 Etxea In

Euskaldunak: La etnia xasca, vol 3, ed

Jos6 Miguel de Barandiar6n, 537-600. San

Sebastidn: EtorNorth American Basque Organization 1997

Hizketa (Spr ing)7: 3

Otero-Boisvert, Maria. 1992 The publishingindustry in dre Basque Countr,v: An introduc-

312/ LRTS . 42(a) . Ugalcle and

tton. ln Euro-librarianship : Shared resources,shared respon.sibilities, pt I, ed. fusuntaPisani, 155-71 Nes' York: Harvorth Pr

Pais Vasco, Spain Gobierno Vasco ServicioCentral di Publicaciones lg83 Easkzzlherriko autonomia estatutoa Gasteiz:Gobierno Vasco, Servicio Central de pub-licaciones

Sercan, CeciliaS 1994. \\/here has allthe convgone? Lat in American impr ints in t le

Etcheaercia

RLIN database Librarrl resources b tech-nical ser t ices 38: 56-59

Tovar, Antonio. 1957. The Basque language(translated by Herbert Houghton). Phila-delphia: Universitv of Pennsylvania Pr

\\ralter, Michael 1993 Central Asian catalog-ing: Problems and prospects. Cataloging- l :classification quarterlll 77 : ] 49-57.

\\/hite, Linda 1992 Basque on the title page.Paper presented at SALALM Conference,

June 3

STATEMENT OF OWNERSHIP AND MANAGEMENTLibrary Resources d: Technical senices,Ptblication No. 3ll-g60 is published quarterly (April,

Jull', october, and January) by the Association for Library collectitns and Technical Serurces,American I ibraryAssociation, 50 E. Huron st., chicago, IL 60611-2795. Annual subscriptior.rprice, $27.50. American LibraryAssociatior.r,50 E. Huron St., Cliicago, IL 60611-27g5, orvner;Association lbr Library Collections and Technical Services. Americin Library Association, 50E. Huror St,, Chicago, IL 60611-2795, publisher; fennif'er younger, 221 Theodore M.Hesburgh Library., university of Notre Dame, Notre Dame, IN 46s56-8629, e&tor. periodicalspostage paid at Chicago, Illinois. Printed in U.S.A. As a nor.rproiit organization authorized tomail at special rates (section 423-12, Domestic Mail Manualf, the purpose, functiolr, and non-prolit status of this organization and the exempt status for Iederal income tar purposes have notchanged during the preceding hvelve monthi.

EXTENT AND NATURE OF CIRCULATION

/313

Union Listing ond the InterlibroryLoon Conneclion

Kothryn Ryon-Zeugner ond Mory W. Lehmon

Analysis of unfilled interlibrary loan lending requests at the Uniaersity ofN ot r e D am e sh ou: e d th at 3 5 Vo u e r e fo r x olllme s n ot ou) n e d b y th e lib r ary. Th eItbrary had not maintained accLlrate holdtngs records in the Indian.a Un.ionList of Serials since l9B3. lncomplete and iniccurate holdings statements re-subed in poor fill rates that cou.ld only uorsen uith time. A project to u.pdateunionlistholdings u;aslaunchedinthe springof 1996. The project is expectedto go onfor fiDe rJears, but the fill rate already begins to reflect the resu.hs ofthe corrections made

Ilistorically, the library rvorld hasstressed the value of union lists propor-tionate to the oroliferation of serial titlesduring this ientury As budgets gettighter, however, the ongoing e{lbrts tomaintain union lists ofien f'all bv the wav-side. Yet the increasing emphasis on con-sulting union lists to determine volumeholdings belore sending an interlibraryloan (ILL) request to another l ibrarydemonstrates what rvould seern to be anobvious connection betrveen holdinesdata and l i l l rates. Fi l l rate is del ined isthe percentage of the total number ofrequests received that are supplied by aninstitution, e.g., if 50 requests are sup-plied out of 100 received, the {ill rate is507o hnproving the ILL lending fill rateis greatly dependent on the existence ofaccurate holdings data In this article, wehope to reach th-ose .-ith the power of thepurse, whose approval is required to allo-cate money and staff fbr the creation andmaintenance of union lisrs.

Most ILL units can vouch lbr the f'act

that their use statistics rise annually. As-sociation of Research Libraries (ARL)statistics show a 6l%o increase in ILLlending between 1986 and 1996 (ARL1998). Notre Dame lending statisticsshow a startling l45%o increise for thesame period. Additional money and staffare rarely available to cope with risingworkloads. Learning to handle the rvorkwith increasing efficiency is the only an-swer when other help is not {brthcoming.Accurate, up-to-date union lists can becritical to overburdened ILL staf{.

LENDTNG Frr-r, Rerr

Prior to beginning this project, a ten-yearliterature search was done on such tooicsas lill rates, union lists, costs, and stitis-tics, as related to interlibrary loans. Any-thing that Iooked remotely useful rvas col-lected. In general, the literature proveddisappointing Nothing could be found onthe cost of unfilled ILL requests, or onthe effects ofthe use ofunion lists on ILL

KetHnvr.- Rvar--Znucxrn (Kathryn M Ryan-Zeugner l@nd.edu) is Coordinator of InterlibraryLoan, and Meny \\i Ltulr,lx (Mary\\'.Lehman 6@nd edu) is Head, Serials Cataloging,Hesburgh Library University of Notre Dame, Notre Dame, Indiana Manuscript received Feb-ruary 9, 1998; accepted for publication April 14, 1998.

3I4/ LRTS . 42(1) . Ryan-Zesgner and Lehman

u NFrLLED rrt" Jrl3liJtoor REeu ESrs(JuNn tOSS, IULyAND Noverr.rsnn 1992)

Reason Unfil led Perceltaqe ofTotal

Volume not orvned

In circulation

Not or.r sheli

Nolcirculating

Other (in process, bindery,, reserve, not as cited, poor condition)

JD

22

T4

10

19

costs or lill rates. Review ofthe literaturegenerated rnore questions than answers:

l. \\/eaver-Meyer (1995) found that 607oof ILL borrorving requests were filledby the first two libraries on the lenderstrinq, and 87 Vo .rere filled within thefive. No one knorvs how rnuch ofthis isdependent on union list use, or howaccurate those lists are.

ation rvhere 457o oI incoming requestscannot be {illed {br various reasons.

In the early 1980s at Notre Darne, thelending fill rate hovered around 507o, whichseemed lou,. In 1985, librarians in the ILLdepartment did a survev to determine the."ionr {br unlilled requests (see table I) Inan attempt to raise the {ill rate, staffcontin-ued to se:uch daily for items that did not ap-pear on the shell until the four-day OCLCtime limit rtut out. A check of the other rea-sons requests rvere unfilled showed that allwere beyond the abilities of the ILL unit tomrrect The fl)utine rechecking of theshelves rnanaged to inch the fill rate up toarornd1\Vo And there it shrck In 1992, li-brarians in the unit did two additional surveyswith results that rvere identical to the first.

Tnr fruorarsn UNroN LtsrAT NOTRE DANIE

The survey results clearly shorved the rna-jor obstacle to increasing the {ill rate u'asthe357o ofreouests thatrvent unfilled be-cause Notre Danre did not ou,n the vol-ume. To the extent that ILL borrowersconsult union list holdlngs data, the prob-Iern could be ameliorated if Notre Dameupdated its serial holdings inthe IndianaUnion, Li.st (IUL).This union list was firstpublished in paper in 1973, then micro-Iilm, and in 1983 u'as added to the OCLCUnion List database Althoush the accu-racy of the t-'riginal holdings dita reportedby Notre Darne in 1973 is unknorvn, thelack of 'upkeep over the years can beassumed to have had a detrirnental ell'ect.

The history of union lists goes back tothe middle oI' the nineteenth centuryrvhen the first one rvas cornpiled in ltaly in

There is debate about who shouldrnaintain union lists A recent discussionon the ILL-L listsen/ carne uD rvith a va-riety of answers: catalogrng depart-ments, serials departments, state librar-ies, re{'erence. and student assistantsNo figures are available on the costsIbr handling an unfilled request.There are no {ieures that determinen'hich is rnore iost-e{Iective-main-

an OCLC Online Computer LibraryCenter, Inc service that allorvs deliv-ery of unmediated ILL requestsdirect ly into the OCLC ILL syJtem.

These are beyond the scope ofthis paper.Sorne believe that {ill rates are sipnifi-

cant indicators ol ' operational act ivi t ies.Althoueh {ill rates are not included in theannual ARL statistics, they do dernon-strate the successlul use of time sDent inl i l l ing lending requests. Time is inon"y.Libraries nrust qive rnore attention tocosts. Spendingtime on requests {br mate-rials not owned q'astes lnoney. Imaginethe tirne wasted every day in an I LL oper-

LRTS .

1859. The first one in this country listedthe holdings of Baltirnore libraries and waspublished in 1876 (Hanson 1968) The-current

situation oftight budgets and esca-lating serial prices should rnake these listsrnore appealing, not Iess Climbing priceslead to lurther incentives for resourcesharing, but e{fective resource sharingcannot be done without accurate. uD-to-date union l ists. Besides saving t imi.union lists assist ILL operations in spread-ing the rvorkload more equitably.

Part ofthe oroblem that led to a lack ofrnaintenance i.s that union listing is seen asa cooperative project done for the benelitof other n-rembers of aconsortium, notthelibrary doing the rnaintenance. Accurateholdings beyond the title level didn'tseern ofsulTicient bene{it to outrn'eigh thecosts ol producing them. In the-earlyI980s, there was no pressure on ILL staffto use union lists. If iists were used, it wasto locate a title, not a particular volume orissue The situation has changed, andtheir use is now crucial to ILL operationsand expected by lenders

Be&use of ihe increased reliance onand use o{' ILL, ILL departments havecome to rely much more heavily on unionl ist ings to speed ILL requests Membersof the Notre Dame ILL departmentwereconcerned that inaccuracies in the lULrvere contributins to the low {ill rate atNotre Dame To demonstrate the inaccu-racy of'our holdings to the library adrninis-tration, ILL staff compiled a list showing a{'eu' titles from the IUL, what the list saidNotre Dame had, and what the actualholdings rvere In most cases, holdingsrvere {ar less than the union list indicated.

The lack of current accurate serial hold-ings on the IUL cost Notre Dame severaloliers of reciprocal agreements lbr lreephotocopies rvith other libraries These li-braries made the agreement contingent onaccurate holdings being available on a un-ion list, and u'e could not sign in good con-science u'ith our holdings in such disarlay.

ln the early 1990s. as pressures grerv inILL operations, the ILL department be-gan getting notes from some librariansrvho could tell that rve rvere not ah'r'ays us-ing union lists u,hen selecting potentiallenders. The pressure lvas grou'ing on all

12(4) . Notes on Operations /315

ILL ooerations to use union lists, and thisrtur going to have a substantial effect onour {ill rate. \\'here the list was in error, italmost invariably shov'ed holdings greaterthan actually owned. The more that otherILL units relied on the union list, the morerequests Notre Darne u,ould receive foritems not orvned. Borrorving operationsrvere rl'illinq to consult union lists becausetheoretically it saved them tirne, and pro-vided their clients u'ith f'aster service.Lending operations were interested intheir lill raies. arrd in the time saved notsearching for iterns not owned.

At this point it rvould have been help-{ul to have had sorne research on the aver-age cost o{'handling an unfilled ILL re-quest. An important consideration whenrnaking libriry budget decisions isrvhether it costs more to maintain an ac-curate union list even with a higher fillrate, or to check the same title over andover because of a I'aulty union list. TrvoARL studies on ILL costs (Roche 1993,and Tackson 1998) rvere checked to seervhether they had come up with any spe-cific infbnnation on the cost of handlingan unl i l led lending request. Unfortulnatelv. the authors of both studies fo-cused .rnly on determining the cost offilled ILL requests.

\\'hile the ILL departrnent was ad-dressing the need fbr a union list updateproject, some unanticipated questionsarose fi'om our non-ILL colleagues. Theyasked rvhether it was f'easible to use theInternet to obtain borroq'ing request in-fbrmation rather than go to union lists.Accessing directly the catalogs of theu'orld dai ly lbr hundreds of requests.horvever, is not nearly as elficient assearching through OCLC. The questionrevealed the lack of understanding of theuselulness ol union l ists, and the hurdles{aced in assigning sta{fresources to main-tain serial holdings

NornE Del'rc UNroNLtsr Upp.lrn PRoJEcr

\\'hen the Indiana consortiurn, INCOLSA,an-ansed in 1983 lbr oCLC to batch loadNotre Dame's union listing data, the li-brary's holdings r* ere represented by about

316/ LRTS . 42(4) . Ryan-Zeu,gner and Lehman

14,000 local data records (LDRs). The onlymaintenance on these holdings in OCLCover the lbllorving | 2 years waJto delete anLDR when all holdines of a listed titlewere withdrawn. Earlyin 1996, the librarypurchased lrorn OCLC a printout thatlisted all records towhich LDRs were thencurrently attached. Unfbrtunately, OCLCcould not provide a listing of all the serialbibliographic records tLat had NotreDame's holdings symbol attached.

A member of the librarv s svstems staffwrote a short program to-identily all re-cords coded as bibliographic level S in thelocal NOTIS svstem-to determine the ti-tles to be union listed. \l/hen the programwas run, the resulting printout containedjust over 31,000 titles. The printout,sorted lirst by OCLC record nufrber andthen by local system number and title, isbeing annotated with the date the LDR isupdited or neu'ly created and serves as themaster record ofwhich holdings have beenupdated. Unfortunately, OCLt's printoutis sorted alphabeticallybytitle; it could notbe sorted into OCLC record number or-der, which would have eased working withthe two printouts side by side.

ings necessitates a reworking from Level 4to Level 3 during rekeying into OCLC. En-

The NOTIS printout of serial titles tobe union listed was accurate only as ofthefirst day that union l ist ing resumed. Sub-sequent serials processing that affects theaccuracy oI union listing must be takeninto account. In order that other staffmembers have ready access to the infor-mation and do not incur search charqes inOCLC to deterrnine whetheror howlhe l i -brary has union listed a title, Level 3 datarvithout captions are entered in the localsystem on a copy-specific basis. Staff

members invarious serials workllows needonly notify union listing stalf of'the localsystem's record number when there is ad-ditional work to be done in the union list.

The updating project is warmly en-dorsed bv the librarvs ILL office. Out-dated information is being corrected, andthe library is reporting to the broaderscholarlv communitv via OCLC aboutholdinei for thousands of titles never be-fore union listed. It has rnade a positivedif i ference; the stat ist ics show it .

Uuroru Lrsr CONTRTBUTOR SuRvEy

In reading the literature on union lists,some thlngs became obvious. Everyonewho writes on the subject praises thern asgood and necessary-fbr ILL, for re-source sharing, for cancellation decisions,etc. Little is written, however, on theproblems of maintaining union lists. Atthe birth of any union list, there is muchenthusiasrn-but as Bloss (f985/86, 143)states, "the popularity of union list revi-sion was second only to shelf reading "

\\/ith the knowledse that union listmaintenance was lorv on evervone's orior-ity list, the original contributors to theIUL were asked about their perceptionsof its accuracy and completeness. Also,they were asked vvhether they consultedunion lists in ILL borrowing and kept uptheir holdinss once the IUL was availablein the OCLC database. Usingthe originallist ol'contributors found in the 1973 edi-tion, we sent a briefsurvey to librarians at56 Indiana libraries. Thirty-three librari-ans (59vo) responded Of those. 25 (75Ea)indicated that they used the IUL in plac-ing ILL requests.

Participants were asked to give theirirnpressions of the listt accuracy andcurrency. Only l0 had the courage to at-tempt an answer. Sorne of their responseswere: "lull of erroneous data-quicker totake a shot in the dark"; "difficult to say,but it beats an OCLC 'dha' display";"better than flying blind"; "I feel it is accu-rate-\,e do depend on it"; "spotty";"50/50 currency,T5/25 accuracy"; "I don'tknow"; "since we're not updating ours, weshouldn't exDect others to do theirs."

Asked rvhether the respondentt insti-

LRTS . 42(4) . I'Jotes on Operations /3I7

TABLE 2UNFTLLED INrnnI-reneny LoAN REeUESTS

iM,rncu, Apnrl, Me,y I997)

Reason Unfilled of Total

Volume lot ou'ned

In circulation

Not on shelf

Noncirculating

20

32

t0

10

tution had kept its union list holdings in-formation up-to-date, the vote split evenlythree ways: Il yes; 11 no; and ll partly

Those who answered "no" or "partly"on whether their institution had kept itsunion listings current were asked whetherthey had any plans to update, and if not,why not Some of those answers were:"too short of staff'; "lack of time"; "neveruse this"; "we have let it fall through thecracks"; "not a priori ty at this t ime."

One thing was apparent from the ques-tionnaire results: the smaller libraries werethe ones rnost likely to be maintaining theirholdings Perhaps this is because they havesmaller collections to maintain arrd fewerchanges to input annually, making laborcosts less criticirl. Another factor for thelarger institutions with large serial collec-tions to consider was the OCLC storaqecosts. In 1983, the initial load into OCLtcost Notre Dame $360. The charee is now 2cents a year per local data record. In 1997,this came to $372 for Notre Dame Not bad,given 13 years o{'inflation.

In these days of tight budgets, all activi-ties are scrutinized, with costs weighedagainst results: union listing activities are noexception ILL costs more ifyou receive alarge nurnber of unfillable requests, but fix-ing the union list has costs, too. \\/ithout anycost data, \ve are suggesting, nevertheless,that it is less expensive to fix the union list.This is only a guiss, but there is an underly-ing rationale. Once the list is up-to-date andmaintained properly, there should be fewerunfillable lending requests, and ILL costsfbr unlilled requests should come down. Ifthe union list is never fixed. its accuracv will{uther deteriorate over time, the filf ratewill further decline, and the cost for unfilledrequests will rise. Experience shows that

requests {br the same titles are receivedover and over again, e.g , titles ofwhich No-tre Darne owned live volurnes twenty yearsago, but no longer receives. Fixing and usinga union list can put an end to the erroneousrequesrs.

I

CoNcrusrou

After only three rnonths, the Notre Dameunion listing maintenance project beganto dernonstrate its value to ILL opera-tions Our fill rate rose {rom 554o to 627o,and a survey ofunfilled requests showed arather dramatic drop in the percentagesfor "Don't own the volume" (see table 2).These fiqures also demonstrate that ILLoperations around the country do use un-ion lists when selecting potential lenders.The project is now in full swing and is ex-pected to take five years to complete.

WoRKs CrrED

Association of Research Libraries. 1998. Asso-ciation of Research Libraries statistics, edSpencer R Graf . Online. Available: http://fisher lib virginia edu/neu,arV 18 May.

Bloss, Marjorie E 1985/86 And in hindsightthe past ten years of union listirtg. Seri-

als Librarian 10, no 1/2: 14l-48Hanson, Eugene R, 1968. Uliol catalogs. In

Encyclopedia of library and informationscience Neu,York: M. Dekker, 3l: 415

Jackson, Mary E 1998 Measuringthe perfor-ntance of interlibrary loan operations inNofth American research and college li-brarias \\/ashington, D C.: ARL.

Roclre, Marilyr M 1993 ARL/RLG interlibranlk)an cost study \\'ashington, D C : ARL

\\'eaver-Meyer, Pat, and Yem Fong. 1995 hi-terlibrary loan and document delivery: Thedebate over union lists Libraru adminis-trat ion and nnnogemcnl 9:204-6

318/

Book Reviews

Morgoret Rohdy, Editor

MARC Manual: Unilerstand,ing anilUsing MARC Record.s. Deborah JByrne. 2d ed Engleu'ood, Colo.: Li-braries Unlirnited, f998. 263p. $34(rsBN r-56308-r76-8) LC 97-3596r.

Cataloging @ith AACR2R anil US-MARC for Books, Computer Files,Serials, Sound Record,ings, Yid.eo-recordings. Deborah A. Fritz. Chi-cago: American Library Association,1998. 580p $60 (ISBN 0-8389-0728-8)LC 97-32939The second edition o{'the MARC Man-

ual, which rvas {irst published in lggl, isupdated throughout, rvith nen' chapterson MARC lbnnat integration and thecommunity infbrmation forrnat. Byrne in-tends this book to be useful to all practic-ing librarians-not just catalogers, butalso administrators, library vendors, andl ibrary science students. ihe succeeds inher goal of explaining rvhat MARC(MAchine-Readable Cataloging) is, rvhyit is irnportant, and horv it rn'orks.

Introductory chapters, "MARC The-ory and Development," 'MARC ForrnatStructure and Content Designation," and"Patterns in the MARC Fonnat,"precedethe bookt longest chapter, "Major MARCBibliographic Codes." This chapter is a{ield-bv-{ield oresentation of the MARCformairvith "description, Potential Uses,Caveats" for each {ield. Byrne emphasizesfonnat integration: chapier 3 is

^devoted

to the topic, and revisions throughout thisedition focus on .specific fcrrmat integra-tion-related changes in MARC coding. Inaddition to the lengthy treatment of theMARC bibhographic lbrmat, there areseparate chapters on the authority for-mat, the fbnnat fbr holdings data, and theforrn at lbr corn rnunity infonnation

Byrne Ibllou's this detailed informa-

tion on MARC content designation rvithseveral chapters on the use o1'MARC re-cords: "MARC Records in ElectronicForrn," "MARC-Based Online Systems,"and-MARC Use in Dif l 'erent Types of Li-braries " Chapters 7 and 8, "MARC Data-base Processing" and'MARC DatabaseProducts," rvill be rnost useful to librariesautomating their catalog {br the {irst time.The processing services that vendors canprovide (e g , de-duping, holdings consol-idation, correction of {iling indicators,srnart bar-code processing, and authoritycontrol) are covered in detail and accom-panied by a discussion of the infbrmationthat libraries rnust provide to vendors, thedecisions theywill need to make, and op-tions they should consider. Byrne con-cludes the section s'ith a discussion ol thepricing o{ MARC database systems

The layout of the book is attractiveFolder icons distinguish chapters, sec-tions, and subsections; a magnifying glassidentifies USMARC definitions; and ! inthe left margin indicates a MARC {ield orcode with major changes since the book'sfirst edition Unfbrtunately, there is nooverall explanation of these syrnbols

Despite the author'.s expressed con-cern Ibr timeliness, the book lacks explic-itness and consistency as to its currency.In general, changes rnade through Up-date No. 2 (March 1996) of the USMARCBibliographic Format (1994) are sholvn;however, several earlier changes are notincluded. For example. the coJe l 'or Corelevel description (approved in June 1994)is not included in the discussion of En-coding Level The 856 field (Electroniclocation and access) was {irst approved in

January 1993. Byrne includes it s'ith thecomrnunity inlbrmation lbrmat, but notin the lield-by-field presentation of the

MARC bibliographic codes, where it is es-sential information {br rnanv users o{'thisbook. The M arch 1996 c'hanses in the l i rstindicators lbrX00 l ields alelncluded, butthe 1993 deletion ol the second indicators0 and 1 for lields 700-730 is not. Byrnedoes not mention the 1991 USMARC For-mat for Cla,ssification Dafc, which pre-dates the community inlbrrnation {brmat( I993) .

In the MARC Manu,al, careless editingis apparent in several instances. UndeiType of Record code (p 63), the defini-tion of Non-rnusical sound recording isgiven for the code g (Pro.jected rnedium),while the de{inition o{'Projected mediumand the code i lbr Non-rnusical sound re-cording are ornitted Though she includesthe current definition o1' the 740 field,Byrne retains its pre-1993 name, AddedEntry - Variant Title (p. 125) In chapter12, rvhich is neu' fbr this edition, Byrnenotes that "the cornmunity informationforrnat is the newest lbrmat (first pub-l i shed in 1993) " ( f 233) Chapter I i be-gins with a statement that the holdlngsformat "is the youngest of the USMARCformats, being published in 1990" (p.229) This sentence. a.ccurate in the l99Iedition of the book, should have beenomitted from this editiort

Byrnet book u' i l l be cornpared to\\'alt Crarvlbrd's MARC for Library Use(2d ed , 1989), rvhich appeals to the sameaudience Crau'lbrd includes seDaratechapters l<>r each t lpe ol bihl i ,rgiaphicrnaterial fbr which a MARC lbnnat ex-isted when he rvrote the book, whileByrne presents al l o{ ' the malor bibl io-graphlc codes in one lbrmat-integratedchapter, lbcusing on the MARC {ieldsrather than on the tr'oes of rnaterial. Bothbooks have separatl^chapters on the Au-thorities and Holdines lbrmats. Crarvfordincludes chapters on intra- and inter-record Iinkinq fields, on non-Roman text,and on exteniions of the USMARC lbr-rnat developed by the bibliographic utili-ties, as rvell as a more extensive glossaryByrne provides a chapter on the commu-nity inlbrrnation format and use{ul chap-ters on MARC processing and products.

\\rhile there are many rnanuals (e g ,Maruell's Handbook for AACR2R, 1997)

LRTS . 42(4) . Book Reuieuas /319

that explicate the Anglo-American Cata-loguing Rules, 2d ed. revised (AACR2R) ,and at least trvo (Byrne and Crawfbrd)that describe and explain the MARC lbr-mat, Deborah Friti's Cataloging uithAACR2R and USMARC for Books, Com-puter Files, Serials, Sound Recordings,\/ideorecordings is the first to correlatethe cataloging rules rvith MARC contentdesisnation. The book is based on theauthbr's more than eight years' experi-

pose of this book is to bring together-inone place-inforrnation about the catalog-ing rules, various sources ofrule interpre-tations, and the MARC coding standards.It cannot claim to be a completely cornpre-hensive tool; it is lirnited in scope to {ive

$pes of material only; it is not meant to re-place the rules or interpretations or coding'manuals;

and it does not include e,rery ruleor MARC tag But this work u'ill givethe cataloger facing a MARC record on acomputer screen, or a blank tern-plate/rvorklbrm, a place to statt'' (p. uii).^

The book includes onlv the naris oltheMARC format that relateio the catalogingrules for books, computer files, serials,sound recordings, ani videorecordings.Thus there is no history o{ MARC, no de-scription of the Directory structure, nobackground discussion of' forrnat integra-tion, and no infbrmation about catalogingmaps or archival materials. The book con-sisti of unbound, three-hole punched 8l,- inch bv II- inch sheets. Most of the in-forrnation is packed into tables and lists,Fritzt solution to the problem of showingthe complex patterns and relationships inthe MARC forrnat and the rules for biblio-qraphic description."

in the lirstlntroductory chapter, Fritzl ists basic cataloging tool i and-describesthe relationship of these tools, the cata-loging rules, and the MARC Ibrmat to thisbook. For each updatable tool she speci-fies the latest update used. Here she erlsodescribes the coiventions used in the book;many of the explanations are repeatedwhere appropriate to rnake the book easier

320/ LR?S . 42(4) . Book Reaiews

to use. The second chapter is a detailed dis-cussion ofthe cataloging process and typesof cataloging, Irorn straight{brward copycataloging to creating original records

Bibliographic description is the fbcusof chapters 3-7. Fritz introduces eachchapter with a list of "Cataloging Steps"for the type of material and then pres-ents, in tables arranged by MARC lield,detailed guidelines for searching, match-ing records, editing records, catalogingdifferent editions, and original catalog-ins. Guidel ines for both USMARC andOe LC records are included TheAACR2R rules are clearly paraphrased,but if the rule is too complex to be safelyparaphrased, Fritz refers the reader tothe cataloging rule itself She places herown hints for cataloging and coding inbrackets to distinguish them fiom thosetaken from offlcial sources. The MARCtables, or "cheat sheets" as Fritz callsthem (p. 3), are quick reminders of basicinlbrmation for each field: repeatability,indicators, subfields, end punctuation,LC and OCLC input standards, catalog-ing rule numbers, prescribed sources ofinlbrmation, and related lields Eachchapter includes a l ist o{ 'speci l ic tools,beyond those listed in chapter I, that areuseful in cataloging that type ofrnaterial.Several chapters include additionalinforrnation tiilored to the tr,me of mate-r ial. The "Books" chapter inciudes discrls-sion of CIP (Cataloging-in-Publication)and large-print books. In "Serials," theediting sections are divided into "same is-sue" and "First-Later issues" to provideguidance when the i tem in hand i inot theone described in the bibliographic re-cord "Videorecordings" concludes u,ith aVideo Viewinq Noteskorksheet.

Chapter 8l "Tags," constitutes almosthalf the book. All lixed and variable fieldtags lrom the Leader to the 830 {ield, ln-cludinp the 049 {ield for OCLC localholdings and 09X for locally assigned callnumbers, are listed. The tables of fixedfield codes, arranged by USMARC char-acter position, include OCLC and Biblio-file labels, a blank space for local systemlabels, and a helpful column of relatedMARC fields. Each variable f ield is pre-sented with the applicable AACR2R

rules and Library ofCongress rule inter-pretat ions (LCRIs). Chief source, re-peatabi l i ty, LC and OCLC input stan-dards, indicators, subfields, and endpunctuation are l isted in "cheat sheet"tables with each field.

Chaoter 9 covers choice and fonn ofnuln" *^.td title access points, based onchapters 2L-25 of AAChZn, b,rt limltedto those access points applicable tobooks, computer f i les, serials, sound re-cordings, ind videorecordings Eachtype of access point is linked to applica-ble cataloging rules, LC rule interpreta-t ions, and MARC taes.

The chapters on bibliogr"phic de-scription and the "Tags" chapter includemany of the same details, such as MARCindicators, subfields, input standards, andend-of-field punctuatio-n, first in the con-text ofthe cataloging process and then inrelation to the parts of the bibliographicdescription. In addition, Fritz providesready-reference access, in appendixes, toMARC indicators, end-of-field punctua-tion, and sources of infonnation for thebibliographic description

The subject of these two books is thesame, but their approaches to the topicand their potential uses are very different.Deborah Byrnet MARC Mantr.al, a thor-ough introduction to the MARC format,can be used as a reference tool or a text-book. In Cataloging uith AACR2 andUSMARC, Deborah Fritz assumes aknowledge and understanding of MARCand focuses on the relationshio betweenMARC coding and. AACR2R catalogingrules As a tool designed for the practicingcataloger, this book would be even moreuseful-if it were available in electronicform, {br installation on cataloqers'workstations.-,f udtth Hopkins (ulijh@acsu..buffalo edu), State (Jn.ioersity ofN e u Y o r k a t B r f f i l o

Research Misconduct : lssues, lmplica-tions, and Strategies. Ed. EllenAltman and Peter Hernon. Contempo-rary Studies in In{brrnation Manage-n'rent, Policy, and Services. Greenwich,Conn.: Ablex Publishing Corp., 1997.206p. $37.50 (ISBN 1-56750-34I-r)LC 97-1806r.

LRTS . 42(4) . Book Reaiews /321

Ideallv. research is conducted in an en-vironment based on honesty and trust.Peer review and replication of resultsrnake the research process sel{'-correcting,so that the validity ofresearch results is as-sured. In this book, Altman and Hernonnote that this ideal process sometirnesfails. Thevpresent a detailed discussion ofresearch mlisconduct, with a case study offaculty and student perceptions of mis-conduct and an experiment using a falsi-fied research paper. The book is unique inits library perspective; the editors fbcusthroughout on the implications of re-search misconduct for library collectionsand services, advocating that librarianstake a proactive stance to prevent distri-bution of false or misleading infbrmation.

Altman besins u,ith an exarnination ol'the dilficultiei in definins research mis-conduct. Common elernents ol rnost del i-nitions are plagiarism, f'abrication of data,and falsification ofresults \1/hile the rnostnewsworthy misconduct o{ien involvesmedical research, the problem occurs inall areas ofthe natural sciences, social sci-ences, and humanities. By the tirne an ar-ticle survives the peer lsvisu' process, ithas earned a reoutable stance in the aca-demic community. Peer revieu,, hou'ever,is not infallible; reputable journals havepublished falsi{ied documents. The ex-tent to which misconduct occurs is un-known, because few people rvillingly ad-rnit to fraud and because the exposure ol'fiaud in scientific and scholarly researchis usually handled with discretion.

Peter Hernon and Laura R. \\taltersreport on a study done at Tufts University,where thev used a ouestionnaire to ex-plore facu[ty and student awareness ofre-search misconduct. Their results indi-cated that neither laculty nor studentsusually considered the possibility of re-search misconduct when using library re-sources, though students u'ere somewhatmore likely tiian faculty to trust printedresources in library collections Facultymembers tended to assume that studentshad developed sufficient critical thinkingskills through their acadernic experi-ences, but most students indicated thatthey did not question the validity of thematerial they use.

uted to library school students, Iibrarians,professors, and deans to read and critique.the part icipants were then told ol ' thefalse-natur6 of the document and de-briefed in focus groups about theirthoughts and reactions concerning thepaper. Finally. the authors distributed thepaper to iournal editors, inftr lrned thernofin" f"tiiti"d data. and requested [eed-back on whether the paper appeared suit-able lor publ ishing. a{ter revision.

A common theme of the student reac-tions was the helplessness they f'elt upon

ical thinking skllls and information liter-acy to be ab-le to use both print and elec-tronic resources effectively; and \4taltersshows how bibliographic instruction canincrease awareness about research tnis-conduct. Researching critical receptionby tracking reactions to a research docu-ment through citation searches and re-views encourages library users to developa luller frame of reference and a healthyskepticism about research results. Theylear^n that misconduct is diflicult to detectand that the retraction may not be linkeddirectly to the falsified docurnent.

Moit library users believe that informa-tion found in a library must be correct, butAltrnan points out that {actual correctness

322/ LRTS . 42(4) . Book Reaietos

rnation they provide. She discusses meth- authors ol this book believe that librariansods that have been used in some medical can and should participate actively in es-libraries to notify users about retractions, tablishing p.<rceii.t.es

-and standards {br

e_.g., stampingastatementneartheti t leof docurnenting research rnisconduct.the article indicating that a retraction is The booli includes several useful ap-available at the library's information desk, pendixes, revieu's of publicly discussedor u'riting the citation of the erraturn or cases of scienti{ic misconduct, a selected

INDEX To ADVERTISERS

AIAA 2d coverALCTS/ALA Editions 327Archival Products 271Blackwell'.s 253LibraryTechnologies 3d coverOCLC Forest Press 254,328OCLC Asialink 4th cover

t,5zJ

IndexVolume 42, 1998

Compiled by Edword Swonson

General Procedures Used in Compiling the IndexThe {bllorving fires of entries are included:

a authors-of articles. reviews, and lettersb. titles-of articles and of articles about which letters rvere published

tion are arranged as words.Subiect headings are based on: ASlS Thesau.nrs of lnformati,oy Sc_ience and Librari-

an.ship', editedby Jlssia L. Milstead (Medford, N.J.: Published fbr the American Soci-ety for Infbnnation Science by Learned Information, Inc., 1994).

Paging of Volume 42:Pages I- 84 = Number I (January)Pages 85-144 = Number 2 (April)Pages 145-252 = Nurnber 3 (July)Pages 253-328 = Number 4 (October)

Bibliographic records' 7l-73 (r)

Book revieu,s: 21-43Books and nonnrint media reviewed

ASIS'97 Proceedings ofthe 60th ASISAnntLal Meeting: Digital Collec-tion.s: Implications for Users,Funders, Dexelopers, and Main-tainers (Schwartz and Ron'ig,eds.): L4l-44

The Bibliographic Record and Informa-tion Technobgy, 3d ed. (Hagler),

7r-73Cotaloging ttilh AACR2R nnd

USMARC for Books. ComPuterFiles, Serials, Sound Recordings,\lideore cordings ( Fritz) : 318-20

Catskill: An Interactir:e MultimediaCou.rse on AACR2 and MARC(Mortirner, Lochhead, andHvland): 250-5f

A

AACR 2, s e e An glo - American C atalo guin gRu.les, 2d ed.

Acquisition o{ library materialsManager.nent ol: 113-25

AltaVista (Search engine): 258-71Analytical entries : 292-303Anglo-American Catalogu.ing Rules, 2d

ed.: 318-20 (r)Computer-assisted instruction:

250-5r (r)Approval plans: I33-39Atkins, \\/inston: 2+7-48 (r)Atkinson, Ross:7-20

B

Basque rnaterials: 304-12Bensman, Stephen l.: 147 -242

324/ LR?S . 42(4) . Index

Describing Music Materiak: A Manualfor

-Descriptire Cotaloging of

Prinled and Recorded Music. Mti-sic \/id.eos, and Archiaal MusicCollections, for Use uith AACR2an.d APPM,- 3d ed. (Smtaslia):24647

From Print to Electronic: The Transfor-mation of Scientific Commuiicn-tion (Crawford, Hurd, and\\/eller): 68-69

MARC Manual: Understanding andUsing MARC Records, 2-d ed.(Byrne): 318-20

Outsourcing Library Technical Sen:icesOperatioru : Practices in Acacbmic,Pu.blic, and Special Librarie-s (\\'il-son and Colver, eds.): 67-68

Plannin g and Implementin g Technic alSertices Worksttttioni (Kaplan,ed.): 69-71

Promoting Presensation Aroareness inLibiaries: A Sourcebookfor Aca-rlemic, hblic School, and SpeciolCollections (Drewes and

'Paqe,

eds.): 247-48

Qualitatiue Research for the Informa-tion Professional: A PracticalHandbook (Gorman and Clay-ton): 140-4I

Research Misconduct: Issues, Implica-tions, and Strategies (Altman andHernon. eds.): 320-22

U nderstandin g lnformation Retrieaallnteractions: Thectretical andPractical Implications (Hert):248-50

Brackbi l l , Lise:25&71Budd, John M.: I4o-4I (r)Burbank, Richard D :246-47 G)Byrum, John D.: 89-l0l

cCalhoun, John C.: 2l-43California State University Dominguez

Hil ls:21-43CannCasciato, Daniel: 7l-73 G)Cataloging backlogs: 304-12Chan, Lois Mai: 45-60Choice Reviews: 2l-43Collection management: 7-20Communications

in science: 68-69 (r)

Computer filesCataloging ofi 89-101

Contracting outoftechnical serwices: 67-68 G)

Corporate namesSubject access to: 272-8I

D

Descriptive catalogingComputer-assisted instruction: 250-51

\r)Rules, sae also Anglo-American Cata-

loguinq Rules, International Stan-dard nlbliographic Descriptions

Diedrichs, Carol Pitts:-I 13-25Digitized library materials

Congresses: I4l-44 (r)Dissertations

Subiect access to: 282-gIDunklb, Clare B.: 67-68 trt

E

Eden, Brad:69-7f (r)"Electronic Journals in the Online Catalog:

Selection and Bibliographic Control-"t2H2

Electronic journalsCataloging of: 126-32Selection of:126-32

Electronic resourcesCataloging of: 89-101

Etcheverria. Kathn n: 304-12"Evaluation of Thiee Record Tvres for

Component \.\/orks in Anall,tic On-line Catalogs" 292-303

F

FictionLibrary holdings of: lO2-12

Fox, Ann Sandberg-,see Sandberg-Fox, Ann"From ISBD(CF) to ISBD(ER): Process,

Policy, and Provisions" 89-101

G

"Gauging the Reception of Choice Re-views through Online Union CatalogHoldings" 21-43

Ga;'nor, Edward, 68-69 (r)Goetz, Diane Mzine-, see Yizine-Coetz,

Diane

H

Hansbrough, Mary : 282-91Hof{han, Herbert H.: 292-303Hopkins, Judith: 318-20 (r)Hypertext Markup Language: 258-71

I

"The Lnpact of Subject Heading Assign-ment on Circulation of Dissertationsat Virsinia Tech" 282-91

"In" analytics, 292-303"In Memoriam: Iohn Oldlck" 74-75"Index, Volume 41, 1997 (Corrected)"

76-79Indiana Union List: 313-17Information retrieval: 248-50 (r)Information technolory: 7l-73 (r)Infoseek (Search engine): 258-7I"Instructions for Authors" 80-81Interlibrary loans: 313-17International Standard Bibliographic De-

scriptionsfor computer liles: 89-101for elecfronic resources: 89-f01

K

Kirchner, Terry L :320-22 (r)

L

Lancaster, F. \\/.: 6I-66Lehman. Marv \\'. : 313-17Library colleciions

Evaluation of: 147 -242. 24345Louisiana State University. Serials Rede-

sign Project: 147 -242, 243-45

M

"ManagingTraditional Materials in an On-Iine Environment: Some Definitionsand Distinctions lbr a Future Collec-tion Management" T-20

MARC {brmats: 3t8-20 (r)Comprrter-assisted instruction: 250-5 I

( f l

"Mendigainera: Scaling the Peak of theBasque Library Backlog" 304-12

Michel, Dee: 248-50 (r)Miller, Heather S.: 133-39"The Monographic Series Approval Plan:

LRTS ' 42(4) ' Inder /325

An Attempt to Reline Purchasing ofBooks in Series" 133-39

Monographic seriesAcquisition of: 133-39

Music materialsCataloging of: 246-47 (r)

N

Nardini, Bob:74-75"Notes on Operations" l13-39, 292-317

oOCLC Onhne Computer Library Center,

Inc.: 45-60Oldick, John: 7 4-75 (about)

Outsourcing, see Contracting out

P

Presewation of library materials: 247-48

\r)"Public Libraries and Adult Fiction: An-

other Look at a Core List of 'CIas-

sics"' 102-12Public libraries

Fiction collections in: I02-I2

aQualitative research: I40-4I (r)

R

Redundancyin subject access: 6l-66

"Redundancy and Uniqueness of SubjectAccess Points in Online Catalogs"6r-66

"Re{'erees 1998" 257Research: 320-22 (r); see also Qualitative

research"Reviewers 1996" 82"Reviewers 1997" 83"Rising to the Top: Evaluating the Use of

the HTML META Tag to ImProveRetrieval of \\'ord \\/ide \\'eb Docu-ments through Internet SearchEngines" 258-71

Rohdy, Margaret: 67 -7 3, 140-44, 246-SL,318-22

Ryan-Zeugner, Kathryn: 3 f 3-1 7

326/ LRTS . 42(4) . ln6ls,

s

Sandberg-Fox, Ann: 89-101Sapon-\\zhite, Richard E.: 282-91Schnridt, Karen A.: 250-5I (r)Science

Communications in: 68-69 (r)"Scientific and Technical Serials Holdings

Optimization in an Inefficient Mar-ket: A LSU Serials Redesign ProjectExercise" 147-242

Scientilic rnaterialsEvaluation of: 147 -242, 243-45

Scientilic s erials : I 47 -242, 243-45Seeds, Robert:126-32Selection ol library materials: 2l-43Senkevitch, Judith 1., 102-12Serials Evaluator (Computer program),

147-242,24345Serials

Un ion l i s ts : 313-17Simpson, Pamela: 726-32"Specificity, Snydetic Structure, and Sub-

ject Access to \\zorks about Individ-ual Corporate Bodies" 272-81

Subject accessRedundancy in: 6I-66

Subject headingsValidation: 45-60

Swanson, Edrvard: 7 6-79, 323-26Srveetland, James H.: 102-12

T

Technical rnaterialsEvaluation of: 147 -242, 243-45

Technical serials: 147 -242, 243-45Technical services

Contracting out: 67-68 (r)\\/orkstations fbr: 69-71 (r)

Thornas, David H.: 14144 (r)"Tou,ard a Computer-Generated Subject

Validation File: Feasibility and Use-{ulness" 45-60

Turner, Thornas P. : 258-7 L

UUgalde, Marcelino: 304-12"Union Listing and the Interlibrary Loan

Connection" 3I3-17University of Nevada, Reno. Basque

Studies Collection: 304-12University of Notre Dame: 313-17

vValidation

oI'subject headings: 45-60Virginia Tech Universily : 282-9IMzine-Goetz, Diane: 45-60

w\\/hite, Richard E. Sapon-, sea Sapon-

\\/hite, Richard E.$'/ilder, Stanley J. : L47 -242, 243-45\\/ilson, Mary Dabney: 272-81\\/orkstations

fbr technical services: 69-7I (r)\\brld \\ride \\/eb

Indedng oI':258-71

xXu, Hong: 6l-66

Y

"'Year's \\brk' to'State of Art"' 5Younger, Jennif-er: 5

zZe:ulgne\ Kathryn Ryan-,see Rynt-Zeugner,

Kathryn

Great ALCTS Books from ALA Editions!

Guide to Managing APProval PlansAcquisitions Guidelines #l I

Susan Flood. EditorThe best source available for essential information on

managing all different types of approval plans'

$1 8.00Pbk.' ALA Members $16'20ISBN 0-8389-3481-l ' 7)e.' 1998

Cataloging CorrectlY for KidsAn Introduction to the Tools, Third Edition

Sharon Zuiderveld. EditorCompletely updated and expanded one-stop guide for cataloging

library materials for children, sets the standard for accurate,accessible cataloging.

$20.00Pbk.' ALA Members $1 8'00ISBN 0-8389 -3416-5' 142P.' 1997

Guide for Written Coff]"tion Policy StatementsALCTS Collection Management and Development Guides

Second EditionJoanne S. Anderson, Editor

An effective, step-by-step plan to create written policy statements'$15.00Pbk.' ALA Members $13.50ISBN 0-8389'3455-2 '40P. ' 1996

AHrditions

Burro THE AsreN LANGUAGECOLTECTIONS YOUR USERS WANTwrTH THE OCLC AsnLlNK AND

TncHPno sERVTcES

DevelopingAsian language collections is easy with the oclc Asialink senriceStaffed byAsian language experts,Asialink offers every'thing you need to establish,maintain and expand Chinese, Japanese, Korean and Vietnamese (CJK\)collections in your library.

. TheAsialink Collection Deuelopment seruice lets youdetermine the specific subiect matter and format ofmaterials that Asialink staff will acquire on your behalf,based on your users'needs. You'll also get significantsavings on the cost of books, including higher-than-aroetage discounts from book vendors.

. Asialink.Sets get you off to a quick start or allow youto easily expand existing collections. Books are avail-able in sets of 25,50 or 100 pre-selected fiction andnonfiction titles Both iuvenile and adult sets ar€available. Title lists are updated annually, so it's easyto add to youf collection regularly.

. FullLevel OCLC cataloging is providecl for all titles andmaterials may be physically processed upon request.

For help cataloging titles not acquiredthroughAsialink, turn to the OCLCTechPro service TechPro providesexpen cataloging for materials in allbibliographic formats and in many languages(including CJK\), plus optional physicalprocessing. TechPro can help you quicklyand efficient$ eliminate backlogs andkeep your cataloging up-to-date, givingyorr users prompt access to materials.

For more information onAsialink orTechpro, contact Marcia stout at ocLC.

r-800-84&587g,ext 4386 614-764-4356 Fax Gl4_7la_7277E-mail: marcia stout@oclc.org.

ffiilltlufww.oclc.0rg/oclc/menu/col.htm

If you have been searching for an easy way to authorify control your

library's cuffent cataloging,try Illl's Autltority Express seryice.

vrthAutltorist Expresg a library uses the Internet to transmit a file of

newly catatoged bibliographic records to LTI (via ftp). LII immediately

pfocesses the records through its state-of-the-art authofity control system.

Then, at the library's convenience, it logs into Lil's ftp server to retrieve

fi.rlly authorized catalogrecords, along with linked LC name and subject

authority records.

AutboriSt Express. Keeps authority control current at an affotdable price. Integrates easily into existing workflows. l,owers cost by reducing staff time spent on catalog maintenance. Provides t hour turn around for up to 1,OOO catalog records

"Autbority Contralfor tbe 21st Century"

Uenlrnv T ecnxotocrEs, ftttc,23OO Computer Avenue, Suite D-19 Willow Grove, PA 19090

(215) 830-9320 Fax: (215) 830-9422(800) 795-9504 email: LTl@LibraryTech.Com

Visit our website at:www.LibratvTech.Gom

ClassifyingKids' Collections?

Subject Headings for Cbildren(second edition, 1998) provides asirnple, authoritative and up-to-datesource for Library of Congressiuvenile subject headings and theDewey numbers that go with them.

In this two-volume set, you get:. All headings on LC bibliographic

records for children's literaturein one convenient place.

. Main headings with their associ-ated subheadings.

. Recommended Dewey numbersfor all headings, with Biographyand Fiction titles identified.

The ke1'word index:. Lets you find'entfies under all

significant words in the mainheading and the first word ofsubheadings.

. Gives you the power of onlinesearching with the economy andconvenience of the print format.

Recommended for all school andpublic libraries that use LC juvenilesubject headings.

SubjectHeadings forCbildren. Volume 1: A Listof Subject Headings Used by theLlbrury of Congress with DeweyNumbers Added. Volume 2:Keprord Index. Edited by LoisWinkel. 2nd edition. 1998.2 volumes. Paper.rsBN 0-91060&58-X. $SO.OO

To order, writeOCLC Forest Press5555 FrantzRoad,Dublin OH 43077-3395,or send a toll-free fax to:l-888-339-3921 or use the orderform at www.oclc.orglfp.

ffillllil$35SS'Publisher of the DeweyDecimal Classification@A diuision ofOCIC Onl:ine Compaterlibraty Center, Inc.