p m " :' Joumal 01 EducII!ion叫 M叫ia & Library Scic忱的 VoJ. 27, No. 2 (l 990), pp, 1拘 1'" UNDERSTANDING MARC : ANOTHER LOOK Rog Ch8ng G:rwloging Coordi1lator Rene Raatjes Cataloging Librarùm The U1l;Vf' 叫'y Library Wes/e r1l IlIi1lois Un;versity Mocomb, lllùlOi又 U.S.A Abstract MAR C forma! has been widely used and discussed in our prof.，髓。" Hοwever， !here appear!o have a wide spread misunders!anding of i!s real s!r叫ur.， and at! ribu!且 This anicle discuss 恥的吋s for u話'" underll!and i! a liule mOre, Also, il presen!s !he generaJ mis<(li~ & t.i brary Sc iences 'n : 2 ('"付 ntcr I咬狗) to terms with the MARC fonn泣. at least on a practical working level. l n rece nt years, however, the influence of MARC has extended beyond the cataloging uni t. As more libraries mQve toward the use of circulation controI systems, online public access cata logs (OPAC), and var io us local automated systems, the need to have bibliographic records conform to the “ MARC standard" has grown increasingly important. Li brarians are becoming aware that MARC records furnish the major source of data for building online catalogs and are beginning to realize that their choice of library automation system is dependent on t he abi lity of that system to support the MARC standa rd .1 The time has arrived when a fuller understanding of the MARC format can place Iibrarians in a more secure position from whi ch to participate in database decision making and maintenance. Never has it been more important to take another, close r Iook at MARC Popular Misconceptions About MARC Although the MARC record fonnat has bee n w自dely used and discussed, most librarians have only a limited understanding of it. Even if t heir library uses a bibliographic da tabase system, it has not been. necessary for librarians to know exactly how records are processed or in what shape they are stored in the machine or on archival tapes. Public service librarians have viewed MARC as just another catalog {ormat replacing th e old 3'" x 5" catalog card by a record display on a computer screen. For mast cataJoge峙. the extent o{ their required knowledge o{ MARC has been the input format which appears on their computer screen and which they often erroneously believe to be identical with the MARC record forma t. The example below shows a display of the input format designed by OCLC to accommodate the data elements required to make a complete bibliographic record. It uses codes known as tags and indicators to identi{y various elements of bibliographic data 寄: Ch"n刊g & R"a恥，~ : U"de .，.tandin~ MARC : AnOlher Look '" (refe rred to as fie lds) 、叭的in most fields, it uses subfield codes known as delimiters, to further narrow the identification of data elements 010 gh84.1258 C謂 “協扭曲785 (pbk) 050 1 HQ814 l聞 10 Wakzak. Yvette 245 10 Divorce : b the child's point of 叫ew / c Yvette 、lialczak with Sheila Burns 260 0 London : 11 Sa n Francisco : h Harper & Row, c 1 9.倒 A more sophisticated but still incomplete understanding of MARC recognizes that the MARC format also conta Îns fixed fie lds which accommodate data such as language of text, presence of a bibliography, source of cataloging, etc. In an OCLC display these appear at the top of the record. as shown below Type: a Bib lvl: m Go叫抖'" Lang: 自tg Source: d 1IlW;: Rep E耽'"' Conf pub: 0 Ctry:目'" Dat tp: s 卸νF/B : 10 J:>e,;c: a Int lvl Dat-純 l曲4 010 gb84.1 2!:姐 Another misconception involves the understanding of the term “ fu ll MARC record" , which 的 often thought to be one that di叩lays all the tags required by the data in the record. rather than one which represents the standard USMARC format In actuality, few librarians have ever seen the real standard record fonnat known as LCMARC or USMARC, since it was designed for the co mputer, not for the human eye. The display formats shown above are only the visible part of the MARC record. The other parts of the real MARC record are the leader and the directory, both invisible in online systems, but “ vÎtal to 這6 吵~ 1" Joumal uf Educa !lonal M..d旭& Library Scienct喝 27 : 2 (Winler 1唉1II) communication and some forms of processing ......" You may never see a leader or a directory on line, but those elements make USMARC processi ng efficient and fl exible. 2 In the real MARC record, the tags are nol included with the indicators attached to a particular field. Nor does the USMA RC leader match the same fields as the fix ed fields of the familiar input format . The visible input formats used by th e va riou s bibliographic uti 1ities (OCLC MARC, UT LAS MARC, etc.) a ll have the sa me general st ructu re as USMARC, bu t vary frorn eac h othe r in their use of ex tended nan-USMARC fields. FOf exam ple, OCLC MARC uses an 049 tagged field to s how item holdings, while RLl N MAR C uses 95X tagged fields for th e same purpose Most bibliographic da taba se sys tem s use the formats shown above to display a bibliographic record in place of the conventional 3'" x 5'" card format. 50me system s, how ever, o ff er library users the option of viewi ng the sa me record online in a 3"'x 5' card format or, as in the case of OCLC. may also produce ca rds offline for use in the Iibrary's ca rd catalog , As accustomed as librarians a re to the traditional catalog card , th叮 are beco ming increasing ly comfortabl e with the tagged format, which offers more room and flexibility in displaying and storing a bibl iog ra phic record 、Vhy ßother Trying to Understand MARC Despite some of the ir misconceptions, librarians have for yea rs managed to make practical use of MARC. 50 why bother tryi ng to learn what the MARC format really is? Walt Crawford, in the introduction to his book , MAR C for Library Use. summariz es the reasons that today's librarians need to increase their understanding of the MARC record Many lîbrarîans create and use MARC records wîthout ever unde r standîng the nature of MARC îtsel f. Whîle no such understandîng is ，.~"~門;."! n幫手 ChanR & R叫j，，" : Und""nanding MARC : Ano!h"r 自刷k r叫uireJ for calaloging, librarians need 10 know more about MARC as tneir uses of computers "xpand. A thorough und"rsta吋ing of MARC w.11 help when dealing with vendors of services, when considering online catalogs and other aUlomated systems, and when considering possible local development of aUlomaled syslm es. 3 '" In recent years more and more libraries are implementing some type of automated system. Because of this trend, the need to understand the MARC record foramt takes on new impor- tance. Most librarians have heard enough abOUl MARC to know that i1 represents the standa rd for machine-readable record formats As computer applicat自ons become more common in libraries and the opportunity to share bibliographic records inc reases, librarians are becoming more conscious of the need for standardization of bibliographic records. We need to be assured that our automated systems are in accord with whatever standards MARC has establîshed . When faced with the responsibility of choosing an automated system, the lîbrarÎan must assess th e capabÎI Îty of that system to accc阱， store, an d process MARC records. Since most venclors claîm t ha t thei r systems a re MA R C compatible, it is important that librarians know enough about MARC to be able to verify suc h claims The StructuTe of MARC A complete description of the MARC structure is beyond the scope of this paper. Readers who are interested in the cletailed s peci fi cations of MARC should refer to the MARC documentations of Lîbra ry of Congress or of national bibliographic systems such as RLIN , WLN, UTLAS, or OCLC and to W. C rawford's MARC for Library Use. However , a g e n e ral descr ip tion of the real structure of MARC may be helpful in visualîzing the ove rall p lcture The MARC format is divicled into three main parts: the lea der, :可 ι lH j州"n ，， 1 "f Eoj "，-n川的論 I ~'l叫.n &. 1..1>...r) S山-"....~'n :l(W州n l'~叫 the directory. and the variable fields. The va rîabl e fields a re, in lum, subdivided inlO twO g roups: the variable control fields and th e variable data fields. The following ex扎mple îllustrates the overa ll format of a MARC record LEADER RECORD Dl RECTORY VARIABLE FIELDS 01 2:'“ 567ft. Use rs of the MARC record do not see lhis data On Ihe screen. For exa mpl e, what dU' uscr sees in the beginning po喝ition of an OCLC record is the rccord îdentîfication number. Thi s should not be co nfuscd wilh thc record length. which occupies t he beginning position in u USMARC record 扒)lIowing the de sig nati on of re地ord lenglh art.' three l.ch.lfacler data fields for coding the r(..cord statlls (new, T{'vised. del l'll'd , t' IC.), the type of reco rd (Iar明uage maten圳， muslc、 map. etc.), and the blbliographic levl'l (monograph, sc ria l, etc.). The品l' Ihree fidd討“long with characrer 17, the encoding level (c!egret. of complctencss of rt-!mrd) can be 間en online in Ihe fix t.'1ARC provides a common ground for sharing d..ta; without ∞mpatibility. a library is forec1osing 包uch sh..ring.5 D.S 恥1cPherson offers related advice When a library evaluates an automated sySlem , concern about record formats may take a back seat to other criteria such as system features and purchase price. ln the long term. however. l的e of a system that does not meet e"，叫 ing standards may prove extremely cos ll y.6 '" We sometimes hea r that a library is using a local system to download its own bibliog raphical records from a national bibliog raphic database system, by connecting the system to the printing port of t he terminal and sending each of its records off the printing port. It is true that the local system can capture all data elements of each record as it appears on the computer screen or print-out. But that record is not the same as the one on the national system archival tape. Therefore, the local system needs to employ an additional program (separate from the program that processes the standard MARC record) in order to be able to process the records down loaded from the terminal We a1so hear at tîmes 出at certain Iîbra ri es are using micro computer packages such as dBASE m + or RBASE 5的o to ca阻10g special materials, and that they are creatin g the自 r records in MARC format. It is quite possible that such database management systems could be used to produce a true MARC record. However, it would require extensive programming efforts to achieve so complex a record format, because while both dBASE III + and RBASE 50個 can handle fixed field records quite eas i1 y, they cannot dea l effective1y with variab le length records. In fact, any re lationa l database management system wou ld be unsuitable for handli ng variable length records since the relational record characteristica Jl y places records in a flat talbe form. Although samples of the output '" Joumal of Ed uca lÌonal Media & Library Scicnc". 'zl : 2 {Wimer 19裕的 record format show that MARC tag s are used to indicate data field s such as author, t itle, and subject, the data in these fields is often truncated when space requi rem ents exceed the flat tab le limits Some of the system vendors point \0 t heir tagged display format as evidence that the ir systcm ca n process and autput MARC records. Such assertions are not uncomman but should not be taken at fa ce value. As such tÎm間， the Ii brarian must insist on asking the essential qu estion: Can th e Jocal system reproduce a comp!ete MARC record fron、 the records stored in the syste m, if in the future the records had 10 be transferred \0 another system ? lt is easy \0 understand why the Technical Standard s for Libra ry Automation Committee (TESLA ) of ALA became con cerned several years ago aboul the MARC compatibility of automated lib叩門 systems being marketed. As a result, it launched a compatibility survey of various vendor's products. The survey ~:~/ indi阻ted that there was a generally strong vendor comm it ment to the MARC forma t. "However, t her e were e nough nonstandard practices reported to indicate that MARC compatibility cannot be assumed and the customers should qu estion prospective vendors carefully in a number of a reas 叮 Standardization and Data Communication The codes and data fields within the MARC structure are not the same as the MARC structure itsel f. To be MARC compatible, both the structure and the inlerior codes must follow a set of standards. The MARC structure standard was set by the American National Standards Institute and is known as ANSl Z39.2 1979.8 Although the ANSl standard did not specify the standa rd for tags, indicators, and data-element identifiers (de limiters), those defined by LC have been accepted as standard practice, for example, 100 for main entry-personal name, 245 for title, etc The need for strict adheren ce 10 the standard is related to the i注2再 正步品 ι可an. l011Tnal 。而 Ed"心lional Mt'd ia & I" lv-ary S<:iences '!:1 : 2 (Winler I院期 fîeld s, but it witl s imp!y leave them there as part of the recorcl; it will oot do anything with them. Other tags Qr subfields oot defined by USMARC or OCLC MARC will be rejected by the sys tem Because each bib l iograph自 c utility has adopted a some wha t different set of extended tags and subfield code s, bibliographic communicati on among these systems requires additi onal process ing. For a discussion of alternative wa ys to reso\ve co nfli ct in commu nicati on amon g sys tems, see R. Renaud's “ Resolving Conflict in MARC Exchange."9 Computer programs for reading a MARC record are much easier to develop than those for co n st ru cting a MAR C record. Many vendor s wîll use MARC record s produced by LC, OCLC, and RLIN, etc. as input data , but they will not reconstr uct MARC records for other systems to use. ln so me cases, the local system will use a totally different str ucture to store records. For example, the LCS system in Il linois has a much simplifi ed record format , although its stru ct ure still maintains the framework of leader, directory, and va riable fields. A s pecially constructed program would , aga間， be required to process th ese record s Issues Concerning MARC There is no question that MARC is a hi ghly co mpl ex record format. 1'0 de ve lop a program for reading the MARC record would not be an easy task for a novice programmer. The easiest record for a programmer to work with and for a comput er to process is one co ntaining 1imited data eleme nt s which can be en te r ed into fix ed rather than va ri ab le fields. ' For example, a record consisting of libra ry staff nam es and telephone numbers req uires a very simple format. 80th the nam e and the number can be treated as fixed field data On the other hand, a book record w ith titl e and author data is a difficult record format , even though it a1so involves only two data Chang & Raa!jω Und.,rs!anding MARC : Ano!her L歐泳 '" elements: the title and the author. A book title ca n vary greatly În length from very short to very long. Simîlarly, there can be a s ingle author, no author. or more than one author indicated. The length of the author's name îs also widely varÎab le. Since a large portion of the MARC record co nsÎsts of variable fi el血，此時間叮 to foresee the difficulties of developing processing programs. The presence of numerous subfields, es pecia l1 y the comple x: s ubfi eld s în the 049 field of OCLC MARC present add ed difficulti由 for processîng. To co mpli cate matt e rs even further, many of the tags and subfields can also be used repeatedl y within the MARC structure. AII in all, the large number of tags , indica岫間， and subfield codes combi ned with variable-length data elements make the MARC format a highly comple x: record structure to deal with Is there a good rea son for the complexity of MA B,C? We may reply that it is the nature of the bibliographic record that makes h the machzne record structu re so complex-and umthe users' ，車'r' information needs that, in turn, dictate the nature of the bibliog raphic record. We librarians require a record format that will accommodate a ll needed data elements related to a bibliograph ic record. MARC, with its flexibility to accommodate multiple author s, multiple subjects, and all types of subject headings，的 a format designed to fill this need lt is inacc4rate to maÎntain that in developing a machine- readable record "We put the card catalog in electronic form 的 The MARC record is not limited to the traditional access points found on a cata log card, but a ll ows the record to be manipulated in numerous additional ways. For example, catalog records may be accessed by LC card number, ISBN, key words, et已， none of which is accessib le in the card catalog. lt is true that MARC arranges the variable data elements in the approximate order in which th叮 appear on a catalog card , e.g. , the call number (0間， etc.) comes before the main entry (100, etc.). and the main entry comes before the tit!e (245), and so on. The designer of the MARC format most ι司 152 Joumal ,,{ Edu<,a,ion,, 1 M~d他& Library Sc i{'n cL"S z7 : 2 (Win,er I吱~) likely reasoned that librarians are accustomed to this order of displaying bibliographic elements. However, the input and output format of a record can be indepenclent of the format that is stored in the computer. Name entries, whether main entries (lXX) or added e ntries (7XX) can be accessed together, despite their location in different fieJds. SimiJarly, series data, whether they occupy 4XX fields or 8XX fields , s hare identical acce時 The order of these tags does not prohibit the programmer from reordering them when they are processed. I-I owever, o nce t he meaning of t hese tags, indi ca tors, and subfield codes are se t, they should be standardized so that all 叮stems can easily process each others records How to utilize the data eleme nts within a MARC record is really up to the local system. As far as computers are co ncerned, any fi eld can be selec ted to be indexed for qui ck retrieval. Any of the data elements ca n be extracted from the MARC structure for building any type of data model: network, hierarchic訓， or relationa l. The flexibility of the MARC record all ows for the c rea tion of s pecial ized types of data files. For exampJe, a subject authority fiJe can be created by selecting data entered in the fields tagged 6XX The complaint that “ The MARC record does not provide adequate subjec t access to the very materials it has been used to access" 11 reflects a misunderstanding of MARC's pote ntial. The MARC format prov自d es t he fields for any num ber of subject headings an d sub-headings. It is up to ot hers to use them for providing adequate access. The \ocal ca taloger must accept the responsibility for inputting whatever subject entries are deemed necessary for adequate aι:cess to any bibliog ra phic record. The MARC fo rma t itself can not be blamed for the fa 自 lure to make use of its capabilities Conclusion MARC record s have long served as the key data source for 、郎， '、 ChanK & R叫阿 U"ders， and ， nl! MAR C : AnOl her [胸k 153 library automation systems. These systems must not only be able to meet the bibliographic information needs of the library冶 users ， including the librarians themselves, but they must also embody the standardization o f formal which Îs a prerequisite 10 data s haring and system communication. Although librarians are making 自 ncreasmg use of MARC for both technical and public services, their concept自on of ÎIs real nalure remains cloudy. Because MARC is made for “ machine-eyes" , it is nOI easy for those librarians who are not yet computer-literate 10 fully understand its nature and its potent岫1. ln the automation age MARC is essential to the library profession. If librarians persist in conti nuou s ignorance 01 the MARC format. the future electronic catalog could be totally at the mercy of system d的igners a nd data processing personnel. The main principles of library service might be severely compromised in favo r of conven- ience of data processing. As G. Patton suggests, the best results will come from the active participation of system experts who understand library func tÎons and librarians who have a fundamental understanding of computer systems. 12 The key to the excelle nce of future library operations is library automa lÌo n. Many library systems that are being developed loday will have a long.term e ffect on library services. One of the key eleme nts in system dev e!opment is the MARC record. To be able to take an active role în the decision-making related to li bra ry automation systems, one must have a good understanding of the nature and possibilities of the MARC format. Li brarian s as a profession must accept the c hal lenge of understanding MARC and how it relates to the future of library service Notes 1. Rober' A. Wahon & F. R. BridR吧!“Auwma， ed Syslem MarkelJ>lacc 1987: M訓uri\y and Compotli\ion弋 L， brary Jou ,."u!. 113: 6 (Avri1 1. (988), l' 強 2. WaIT Crawford, MARC fo ,. L ,br“ ,.y Use: Un duSllmd"'K Ih" US MAR C 『- ,,.. Jour吼叫 01 Educa 1iOIl"] Mc“為 & I. ， b開 'Y Sαe"c... 'z7 : ~ (W叫“ 1蹺，的 Fo ,,",uls (While Plains. N,y': Koowl叫“e llldustric$ I'uhli~~tions. 1984), p.7 3. /1",1, p. I 4. Mieh..le 1. Dalchi ,.., "MARC Formal 01\ Tape: A T Ulorial,Rin From Tuþ /0 F'rodurl, ed. Barry 且Baker (Ann Arl酬， Ml<~h' Pienan Press, 1985). p.25. 5. Crllwford. 1).52 6. Domthy S. McPhenon. KMARC Compalibihly: A TESLA Survey 0' Vef、“n: I"/Q，.，..nl的同 Tuhn6岫αω耐心b，帥~s.. 4 (5叩tember L泊的 1) ， 241 7. lb,d. 8. For a discu肥ion 01 the "Arneriun National SllI.nda叫~ for ß ,hliograllhic lnformation Int..rch~nge 01\ MlIgnellc Tap'"白5tt Crawford. p. 29. 9. Robert Renaud. 曰恥rol叫ng Co nflict in MAR C Exchange, K 1 .. /0""01 /0" 7 ..... h""/o，盯 帥d L.. bronn, 3 (Seplem悔穹，，.‘). pp 必.，.， 10. Kevin Htgany , MMylh~ 01 I 且 b.8.Y Automal ;on,M J., brary JOU""Il~ 110 (Ocl. 1. 1985) p.471 11. lbul, p. 48 12. Glenn 1'.1100. 可怕他raction; l..ene帽 10 Ih.. F..d ilor" , ub悶ry R t:StW1US & T，r1"U個t b何時，或 32: I (hn. I伺8). p.9.