Published Cataloging & Classification Quarterly Vol. 29(4) 2000

ABSTRACT

In October of 1999 there was a discussion on Autocat (an e-list for cataloguers) of the advantages and disadvantages of classifying Internet resources catalogued locally, which grew to include a discussion of cataloguing electronic resources generally.

This article reviews the background of applying bibliographic description techniques to electronic resources, and summarizes the Autocat discussion.

While some librarians see classification as primarily a method of assigning a shelf location for a physical item, many others see classification as a valid subject approach for all the materials in the collection, or available to the library's patrons through the library catalogue.

KEYWORDS: Classification, Cataloguing, Electronic resources, Internet, Autocat, Browsing

Bibliographic Description Techniques Applied to Nonbook Materials

For over a century librarians have been developing a method for describing books in libraries. There is a clear line of development from Italian born Anthony Panizzi, chief librarian of the British Museum from 1856, to the development of the ISBD (International Standard Bibliographic Description) by IFLA (International Federation of Library Associations), and its adoption in Anglo-American Cataloguing Rules, second edition, 1978.

For most of that period, library collections consisted primarily of books. When, as a Canadian representative to the Steering Committee of the IFLA Section on Cataloguing, I first suggested the need for an ISBD for nonbook materials (just after the adoption of the ISBD for monographs), several European librarians indicated that a nonbook ISBD would be of little utility, since library collections contained so few nonbook items. Today, some library collections are more than one third nonbook it has been reported on Autocat, an electronic discussion list presided over so ably by co-list owner Judith Hopkins.

A few of us can remember when microforms first began to form an important part of our collections. More can remember when sound and video recordings become common in libraries. Most can remember when CD-ROMs were first commonly acquired by their libraries. Now many libraries are beginning to include remote electronic resources in their catalogues.

With the advent of OPACs (online public access catalogues), not only are new functions available (learning whether a book is on loan, placing a reserve), but through an active url (universal resource locator, an Internet address), the catalogue terminal may be used as a reading device for Internet resources.

The practices used to describe book materials transferred without too much difficulty to most of these new formats. The introduction of the general material designation (gmd) following title proper (with "text" being optional for the traditional book) made it clear what was being described. The inclusion of new ISBD areas in MARC (machine readable cataloguing) has been more problematic. The single ISBD area for media specific information between edition and imprint has been given multiple MARC field tags, so that numbers between 250 (edition) and 260 (imprint) will soon be exhausted as nonbook forms proliferate. Ignoring the fact that many systems can only display notes in MARC tag order, the assigning of MARC field tags to new types of notes, many of them associated with nonbook materials, is being done with no attempt to follow prescribed note order (unlike the original MARC tags which adhered to unit card, later ISBD, order). The standard description began to break down even more with the inclusion of Internet resources. With no physical item actually in the library, the collation (MARC field 300) was omitted. The fact that the special material designation (smd) as part of the collation was important to more narrowly define the gmd was ignored. The opportunity to further define "[computer file]" as "remote Internet database", or "website", or some other specific designation was not taken, leaving many patrons confused about what was being described.

In North America it has been the general practice to classify books, and then to add to that classification number (sometimes including a subject Cutter) a main entry Cutter or work mark which creates a unique identifier for the work, the result being known as the call number. Initially a year was added to distinguish later editions, but subsequently most libraries using LCC (Library of Congress Classification) began adding a year to all monographic call numbers, as libraries using NLM (National Library of Medicine Classification) had done earlier. (Many smaller libraries using DDC - Dewey Decimal Classification - add letter workmarks rather than Cutters, and do not have unique call numbers.) The term "call number" dates from the time when patrons noted the number from the catalogue, presented it at a circulation desk, and the book was fetched for them. Now that most libraries have their stacks open to the public, patrons will often note only one or more classification number from the catalogue, and in fact use a classed approach to finding material by consulting the shelves. This always had the limitation of their not finding material in special collections such as reference, or in circulation. Now it has the added disadvantage of patrons not finding nonbook materials usually shelved separately, and particularly remote Internet resources. (No one seems to have suggested the old solution of a block of wood on the shelf with a label attached.)

Practice has varied in whether material not being shelved with books is assigned a call number and "shelflisted", that is, being fitted into the classed file of the collection. The terms "classification number" (a subject number, sometimes including a subject Cutter, assigned from a schedule), and "call number" (the classification number plus Cutter or work mark and usually year), are often wrongly used interchangeably. This has led to "call number searching" being used when actually "classed searching" is meant, even in OPAC index names.

Libraries have reported in Autocat discussion that only from 1% to 3% of their OPAC searches are classed searches. Libraries have not done extensive instruction in classed searching, nor created easily browsable indexed classed files with an appropriately labeled search key. They have tended not to see shelf browsing as classed searching, nor to be concerned with the materials not found using that method.

Autocat Discussion of Classifying Internet Resources: Pro

In October of 1999 there occurred a discussion of the classification and cataloguing of Internet resources on Autocat. There was a fairly clear division between those who saw classification as call number assignment, and felt no need of it for material not physically present in the collection, and those who saw classification as a means of subject organization. The latter saw classification as important for all library materials, placing material in a systematic array, with mammals from Aardvarks to Zebras together, as opposed to being scattered alphabetically.

The gauntlet was thrown down by Julie Moore Crowley of Stetson University School of Law, October 14th, 1999.*

"At Stetson Law, I currently treat Internet titles accordingly:

" MARC 090: LC Call no. " Location: Internet

"I must admit that I am surprised by all of the responses saying that they are dropping the classification of Internet records. If the Internet records are not classified, then the patron/librarians will never run across those titles in a browse by call number. This completely compromises that avenue of access!"

Some quickly responded in support of Julie's position (we tend to use first names on Autocat).

Gene Kinnaly, a cataloguer at LC (Library of Congress) reported on LC practice (in a message identified as not an official statement of LC). (Gene, along with Aaron Kuperman, law cataloguer, have both helped give a human face to LC though such frank and informative messages on Autocat.)

"LC practice is to assign an LC classification number to the Internet resources we catalog. We don't shelflist Internet resources, since we don't have a physical item sitting in our collections under that call number. I think complete call numbers - classification number and 'book' number - for Internet resources are confusing for that very reason.

"But I think classification is useful. Now that we have a system (LC ILS) in which records can be retrieved, sorted, and displayed by call number, all our records which have at least a classification number are included in that display.

"By assigning classification numbers to materials accessible via your online catalog but not physically housed in your stacks (Internet resources), and if your OPAC allows browsing by call number, then the patron can 'browse' not only the materials you have in your collection but also those Internet resources on the same or closely-related subject. Since this same patron can access the Internet resource through the link provided in the bib record, having this record show up in a call number search is a method by which we can provide greater access to information."

Tim Knight, Systems Cataloguer, Great Library. The Law Society of Upper Canada, also drew a parallel between shelf browsing and classed searching.

"Do library users associate a 'Call number' search with browsing the shelves? As librarians we naturally make this connection. One library site I saw recently (can't remember which) actually labeled their Call number search something like 'Browse the Shelves'. Perhaps helping users make this connection would raise searches above the 1.5% cited earlier and encourage users to browse at the terminal rather than running off to the stacks.

"Classification browsing in the online catalogue is also the only way that users can bring together all of the various formats now available in our libraries, i.e. books, CD-ROMs, CDs, cassette tapes, video tapes, etc. Because these formats are usually placed in different locations in the library browsing them all on the shelf at once is not physically possible. How often do users miss relevant information in other formats because they only browse in one location?

"So now add in Internet resources. Yes, these are not physically in the library, but they are considered important enough that we want to alert users and provide access to them. The OPAC offers the ability to browse all of these resources along with our physically available material together on 'virtual' shelves if you will. Should we not collocate these resources, relevant to our particular user group, along with the other materials we have to offer in our library and facilitate discovery of relevant information in all formats?"

John Riemer, Assistant Head of Cataloging, University of Georgia Libraries, also spoke in favour of classifying Internet resources (as forwarded to Autocat from CORC-L by Judith).

"I see a number reasons classification is/can be of value in records for electronic resources:

"Classification is discipline-based, while subject headings are topic- based. In a title like Journal of Personality, specificity in subject heading assignment would theoretically result in the single heading 'Personality$vPeriodicals'. There would not be any keyword 'psychology' in the subject heading(s) to search against.

"If people like to browse a list of electronic resources, alphabetically arranged either by title or by LCSH string from within the records, it may be too overwhelming to give them everything the library has. If the records contained LC classification numbers beginning 'BF ...' we could isolate the subset of resources falling into the discipline of psychology and present them with that more manageable alphabetical list.

"If OPAC users learn to 'online shelflist browse' at the terminal, they can discover in a single setting resources that reside in a variety of physical locations: regular stacks, special collections like reference, one or more oversize sequences, those in circulation and those that aren't, those in both open & closed stacks locations, etc. Further, if we think of classification as a subject indexing language instead of a mere location device, users can discover the electronic resources that fall into the same areas as the physical objects with the same classification assignments."

"I wonder if our users won't discover classification number searches as the virtual equivalent of browsing the shelves, as they get used to finding information on the Internet."

"Exactly. If your OPAC allows patrons to browse by call number and display the results arranged in call number order, then you DO, in essence, have a 'virtual shelflist', a list of bibliographic records arranged in the order in which the corresponding materials are shelved in your collection. Take it one step further, and begin your call number search not with a complete call number but with a classification number, and you pull in all those bib records to which you assigned a classification number and not a complete call number. In this way, you could have a virtual shelflist which reflects print AND nonprint resources which your library physically possesses and those Internet resources to which your library provides access.

"In addition to our regular book cataloging with complete LC call numbers, LC currently assigns classification numbers to Internet resources, tangible computer files, and many microforms (and there may be other formats as well). So in the LC ILS, an OPAC call number search of PE1625 (a general search for English dictionaries), retrieves a hit list of 678 records. The first five records show the classification number only; the rest have complete LC call numbers. Four of the first five are tangible computer files (computer disks or CD-ROMs) and the fifth is an Internet resource.

"Does this distinction matter to the patron? Maybe. But, all of these materials, whether physically housed within the Library or not, are available to the patron. When the patron points to one of the records on the hit list and says 'I want to see that one', it shouldn't matter what format the material is in (and if it's an Internet resource, hopefully access is just a click away)."

Tina Gunther, Cataloging Technician, Biola University Library, spoke of the importance of classed retrieval, granted the small percentage of classed searches.

"Call number searches may be a small percentage of online searches, but I see a lot of patrons getting classification numbers from their searches in other indexes. Once they know what part of the stacks they want, they head off to browse the shelves to choose material from what is available in that area."

A search by that classed number in a classed index would obviously then produce much more material than a trip to the shelves.

David Moody, Cataloging Librarian, University of Detroit Mercy, added:

"I've been thinking about that 1.5% for classification number searches. That's not insignificant. That's 15 out of every 1000 searches--and how long does it take for your library to do 1000 searches?

"More importantly, these are likely to be quality searches for known information. How many subject/title/author searches are just stabs in the dark, and have to be repeated several times before they are 'right'? But a patron looking for a classification number probably knows exactly what they are looking for--or else they wouldn't try the search in the first place!"

Autocat Discussion of Classifying Internet Resources: Con

Daniel CannCasciato, Head of Cataloging, Central Washington University Library, pointed out the small use currently made of class searching, a fact already alluded to in posts quoted above.

"In our most recent statistical summary, from the online catalog, of 14,687 searches, 186 were call number searches. That's 1.27 % of the searches. AND, of those a number were actually subject searches somehow entered into a classification number search. 20 more were from a class assignment that required students to use the classification search. (Somewhat ironically, they were doing an accession number search.) The actual total is closer to 120. So, while classification is an avenue to finding materials, it doesn't seem to be one that is a well beaten path, at least online. For physical items in open stacks, (where the patron does browse, unknowingly and without much guidance, by classification number), then it is very much used. So, for those of us who do not classify certain materials, there is a basic reason for it. Yes, in the long run it could prove to have been narrow sighted. On the other hand, I have a much easier time supporting the decision to participate in SACO and NACO than I do to classify materials that are physically not present in the library."

Those who defended their decision to not classify Internet resources spoke primarily of the time, effort, and cost, and of the confusion caused library users by the fact that the material could not be found on the shelf by that number. One who stated that position cogently, even though she does class Internet resources, was Angela Murphy-Walters, Catalog Librarian, Sherrod Library, East Tennessee State University.

"Although we do the kind of call number checking that Julie supports, and have made the decision to classify Internet resources cataloged in our database, I disagree that choosing not to classify these materials would undermine a catalog or compromise call number searching to a significant degree. As Julie wrote,

Classification is an avenue to finding materials, and it is an avenue that is being corrupted if you do not classify every type of material in the collection.

"but Internet resources are not in the collection. To many patrons, the classification number is simply a means of finding an item on the shelf. Seeing a call number associated with an Internet resource could cause a lot of people to head for the stacks, and others to be simply confused. Yes, there are ways around this. But if, in fact, call number searching is only about 1.5% of the total number of searches done in our catalogs, and since all the wonderful things that come from finding a call number and browsing the nearby shelves for related materials won't happen when a patron clicks on a URL, I understand that the time and effort involved in classifying Internet resources could be hard to justify."

In another post Angela noted:

"... in many smaller libraries, including many school and special libraries, there are not enough resources to accomplish everything. .. . In those circumstances, taking the time to classify resources that are not to be found on the shelves may not be worth the trouble."

Pam Deemer, Assistant Law Librarian, Cataloging Services, Hugh F. MacMillan Law Library, Emory University, made practical points:

"I would say neither subject heading access nor classification alone will provide complete access to titles on a subject. Both must be used and combined because of different ways people perceive 'aboutness' and assign subjects and classification. As far as online resources are concerned, whether to classify or not depends on staffing, budget, patron demands and expectations, source of cataloging, and systems. Another very hard thing to buck is tradition. ... Triage is the name of the game when it comes to assigning classification, especially retroactively and with very small staff. How predecessors decided to handle classification of certain collections causes problems for the current librarians. If a decision was made to use acquisition numbers for videos, microforms, and audiotapes because they were not collections to be browsed, the current cataloger may be faced with the choice of retrospectively classifying things or doing current cataloging.

"Then there is the factor of long-time patrons being used to certain situations, such as serials shelved in alphabetical order. If they are influential users, such as one's law professors, change is very hard to effect."

Gordon Pew, Cataloger, U.S. Court of Appeals Library, 1st Circuit, reported that his library does not even catalogue Internet resources, and would not class them even if including them in its catalogue.

"I'm jumping in to say that my library does not catalog material that is only available online -- primarily because of Court concerns about Internet security: we don't want to encourage accessing online resources through Court PCs. However, if I were cataloging these materials, I would not classify them in the same way I do physical materials. As other posters have noted, it is misleading to the patron to assign a call number to something that can't be found on a shelf. I would favor giving online resources a location code (such as INTERNET or ONLINE) and maybe even using the URL as a kind of classification number (after all, it is a locating device). But to classify the online version of the U.S. Government Manual in JK421 would be incorrect, IM(NS)HO. It isn't on the shelf with other JK421 materials. However, the use of the same subject headings, as well as collocation by title in the catalog, will make the library user aware that it can be used as a library resource."

Some have even abandoned "shelflisting" as advocated by Julie, i.e. adjusting the Cutter to achieve alphabetic order within class number, for all library materials. Don Spicehandler, Head of Technical Services, School of Visual Arts, NYC, reported:

"Currently we do -not- do any 'proactive' shelflisting, even for shelved items, unless we know that the incoming item is an added copy, added volume, etc. In most cases, we accept call numbers blindly, and this is whether the copy is LC or not. The catalogers do look at class numbers quickly, just to make sure they are close to what they are 'supposed' to be. Cursorily. And so our shelf order is far from perfect. Just like us. Real shelflisting is simply not an option for many smaller libraries, those with one professional cataloger who's on a zillion committees and has the same number of responsibilities beyond cataloging. This is sad but true and I assume, not that uncommon."

Autocat Discussion of Classification of Internet Resources: Other Aspects

Some pointed out uses of classification for Internet resources other than subject searching. Among them Jenna L. Oliver, Cataloging, Renton Technical Library, The Boeing Company.

"Well Julie, I do not think that classifying internet materials is a waste of time. Aside from the fact that if someone (patron or librarian) searches by call number they might well miss some material (and faculty and grad students do use the call number searches for browsing!!) - because we didn't classify it....

"Classification is often used as a collection development tool!! And in my experience, it has been a very valuable tool. I fought hard in my former (university) library (like a dog with a bone) to keep class numbers on online resources - whether Internet, CD, what-have-you. While some said it was a 'waste of time', others saw the value of keeping the records consistent and tracking the material, if only for collection development purposes. (Today's free online resource is tomorrow's for-profit commercial venture.)"

John, quoted earlier, also points to other uses for class numbers in records for Internet resources.

"When accreditation is done on campus, I often get asked if there is any way to determine the number of library books and serials the students and faculty have access to in a given discipline. Given the expense of acquiring and the effort to organize electronic resources, it is quite helpful to classify them, such that a report can be run to determine how many OPAC records have class numbers falling into the desired ranges from the schedules."

Manuel Urrizola, Serials Cataloguing Librarian, University of California, Irvine, described the collection development benefit:

"One should not overlook the collection development value of classifying electronic resources (many of which costs thousands of dollars). Many of the collection development software programs I am familiar with use classification as a means of evaluating one's collection. I can understand not Cuttering, but I think electronic resources should be given a base call number (which can be suppressed from the public but still bring up records on a call number search)."

Mark Braden, Catalog/System Librarian, Occidental College Library, made the basic distinction between classification and call number, drawing attention to our lapse in calling the classed search a call number one.

"1) Classification. I consider it important to include a Class number ... 2) Call Numbers. Now, for the way that most Innopacs are configured, the only search by Classification is the Call Number search. In my mind, Classification is not Call Number. If you have a separate index for the Classification Number in a bib record, that's a good thing. If you can only display and search Class Numbers by the CALL NUMBER index, then that's bad--call numbers are the location of the piece in your collection (on the shelves). As many have said, there is no 'shelf location' for Internet Resources."

Perhaps the most definitive post summarizing comment so far, and stating other values of classing Internet resources, was made by Charles Pennell, Head Catalogue Department, North Carolina University Libraries (formerly of Memorial University Library).

"I agree with most of what I have seen here concerning classification of Internet resources. It IS true that assigning class numbers to these materials WILL cause some undergraduates to wander aimlessly in the stacks and that patrons do not rely heavily on call number browses in the OPAC, preferring to restrict this practice to the stacks. On the other hand, this may be another of those self-fulfilling prophecies, as at least North American librarians have conditioned patrons to think of call numbers primarily as parking places and to restrict assignment of a resource to a single class number. In a large academic library, you need only look at the QA76.5 or E98-99 areas to see how unbrowsable this practice within the LC classification has become. ILS (integrated library system) vendors respond to our unidimensional use of classification by giving us poor call number search capabilities. Our DRA system still cannot browse call numbers backwards and can only go forwards within the range provided by the call number requested. If you enter QA76, then you can't see QA75 or QA77. No wonder only 1.5% of OPAC searches are by call number!

"Obviously, there are plenty of reasons for increasing classification access to our growing collections. If your library provides any type of new receipt notification service to faculty and students based on classification number profiles, anything (Internet stuff, SuDocs, unclassified microform, unclassified periodicals, etc.) not classified is going to be left out. If you are analyzing holdings in a discipline for an accreditation review or a proposal to establish a new academic program, unclassified items are going to be left out. Continuing to assign only single class numbers where more than one may be justified also affects how patrons will use the OPAC for subject access, how vendors will develop their OPACs, and how libraries will harvest resources for management use and publicizing recent acquisitions. This is certainly an area where we could use OPACs to overcome some of the limitations which are forced upon us by our need to physically park materials in one place only.

"NC State is wrestling with the classification of electronic, microform and other non-LC classed materials, not because we want to change how we shelve (or not shelve, as the case may be) these materials, but because we want to provide automated SDI-type services for patrons who want to know what is available in their area(s). Our MyLibrary service (http://my.lib.ncsu.edu/) uses call numbers to provide customized new material notification and lists of available reference resources. We have partially extended this to electronic journals and databases, but since these are not classified here we have had to add discipline information to MARC 699 fields. To more fully realize this service we would like to add classification numbers, but would not like this to display to the public (the wander aimlessly factor). We are looking for other places where class numbers might be embedded for class browsing without affecting public display. We could, for example, point the indexes to the 050, (055, 060, 082--add your favorite here) 090, although these have never been controlled here (we have not edited or removed call numbers from other libraries while cataloging). Since we are a MARC holdings site, we could embed this info in the MFHD and solve the problem of control, except there are no classification fields (05x-09x) provided for in that format, only the local call number in 85x-86x. The 85x and 86x display in our OPAC, which would be counter-productive. So, I think that there is interest out here for increasing classification access to our collections, for adding the ability to assign multiple classes to a single work even if it IS only parked in one of the classes, and for developing an underdeveloped part of the OPAC and management report structures of ILS. Perhaps Internet resources will prove the impetus for this?"

Let us hope Charley is correct. The former Catss defined 089 for classed catalogue numbers, and it was carried over into MARCit as a locally defined field. OCLC has so defined 699. MARC21 still ignores this very vital need.

Julie also raised the question of single or separate records for Internet resources.

"I am equally dumfounded at the number of responding catalogers who are lumping the paper and the Internet records together on the paper record. It is confusing to have a 300 tag that looks like paper -- with the volumes and/or pagination, cm., etc., but it is actually a computer file on the Internet. This is blatant disregard for AACR2R98: 'If an item is available in different formats . . . give the physical description of the format in hand. Optionally, make a note describing other formats in which it is available.' Forgive me if I have misinterpreted, but I take this option to mean that it is merely a note to give the user more information. This is not giving the catalogers a rule that it is OK to lump all of these different formats onto one record. What is the difference between that and taking a video, a sound cassette, and several differing paper editions of a title and lumping them all onto one record?! I realize that every library has its own quirky policies. (We have our own, too!) However, I am saddened to see such disregard for such basic principles as classification and cataloging different editions/formats."

In a further post, Julie added that another problem with combining print and electronic resourses in one record was determining whether they were in fact the same edition with the same content.

"For instance, one electronic journal might include all the 'works' in the original which are substantial articles but omit the 'works' which are letters to the editors, etc. Another might include elaborate graphics or video or other extensions of the articles which it would not be feasible to include with the print edition.

"A semi-related question I have worried over is the best means to relate a print serial which has ceased to the web site (not a true serial but an on- going entity none-the less) which succeeds it."

My answer to Julie's last question above was that full format integration should allow the use of 780/785 (supercedes and succeeding notes and entries) to link records for print serials which become websites with the succeeding website. As remarked earlier, and as Julie agrees, having holdings united for serials is an advantage for keeping all formats (print, microform, electronic) of a serial in one record. The use of a 530 (additional physical forms available note) in the record for each format would obviate this advantage for monographs.

Gail A. Spears, Catalog Librarian, College of Law Library, Georgia State University, responded to Julie's post by posing some questions, but not taking sides:

"One record or separate is the question when it comes to cataloging internet resources. That is the sixty four thousand [dollar] question these days. Yes, we know what the various agencies have called for in the way of national standards. But wait, what about local standards? Don't they count for something also? I sure hope so. We might be able to follow the general intent and outline of national standards, but we also have to think of our users. Ok, some (librarians) think that one record for the various formats for one title leads to confusion. Others say that separate records are the way it should be, especially if there is some variation in either of the formats of the title. Ok, so we all have our opinions. But what about the user? Is it really confusing to view one record which has multiple formats or does it make it simpler for the user to fetch? Do we think users really want to scroll down a screen or two to make sure that he/she has not overlooked the title they might wish to access regardless of format? These are just a few of the simplest questions that might run through my mind."

Mark also expressed his reaction to the question of whether Internet resources should be added to the print record for the same work, or should have their own records.

"Separate Records. Having read the excerpt from the CONSER Manual, I feel as though there is some conflict. Microforms often have the exact same content as the paper format of the newspaper. However, we commonly build a separate record for the microform version. Why, now, do we change practice and consider electronic format (Internet resource) something which doesn't merit a separate record, if the content is the same? I'm disappointed with the outcome of that CONSER discussion.

"But, I must admit that I disagree with other CONSER decisions about Internet resources, especially serials. And, even esteemed colleagues such as the University of California have apparently decided to place electronic on one record with the print."

Erica R. Powell, Director, Technical Services, Calder Memorial Library, University of Miami School of Medicine, also posted in opposition to the single record for multiple formats. She does not agree with Adam Schiff (see below) on the probable development of rules in this area.

"In our library, I catalog an Internet resource just like any other item. My background has always been in the cataloging of nonprint materials. ... In the Interim guidelines for the Internet, the policy states that if you use 1 record then you must combine all the information in that one record. Therefore, you should have two 300s, two 362s, etc. I know that our patrons would find this very confusing and I suspect that down the road, the rules will state that you must use individual records."

While we have read little in the literature about the work record concept since the Toronto meeting, that seems to be what is sought in combining text and computer file in one record. Patrons are certainly served by having all holdings of a serial combined, but other advantages of combining formats in a single record seem to have other solutions, such as a note in each format record recording the existence of the other, as already mentioned.

Mary Ann Van Cura, Associate Director of Technical Services, Thomas M. Cooley Law School Library, quotes with approval the CONSER suggestions:

"The cataloging of Internet resources has been a challenge. CONSER established a working group to recommend treatment of electronic resources. They have prepared some recommendations, which some of you may find useful. Although the recommendations have not yet been approved, they are practical, and will be considered for adoption by CONSER libraries down- the-road. Note: They may appear to contradict a long-held practice of creating separate records for different physical formats. In a time of limited staffing resources, they seem to have proposed a middle ground. See what you think!

"The full text of the working group recommendations, 'Single or Separate Records: What's Appropriate and When' is available at this web site: http://wwwtest.Library.ucla.edu/libraries/cataloging/sercat/conserwg/

"This excerpt provides the proposed criteria for cataloging electronic titles on a single record versus on a separate record:

"For serials that do warrant cataloging, define the characteristics of those that could be appropriately described on a single record (for libraries which have opted to use this approach) and the characteristics of those that would best be handled by separate records. Again, consider the subject, scope and purpose of the site, its similarity to the original version, etc.

"A. When to use a single record A single record is recommended in any of the following situations: Equivalent content. The content of the paper and electronic versions are the same but the titles differ. The content is equivalent but the presentation differs (e.g., articles are added to an issue by the publisher as soon as they are ready, rather than releasing complete issue.) The titles of the print and electronic versions change simultaneously. GPO single-record copy is available.

"B. When to use a separate record A separate record is recommended in any of the following situations: Resource exists only in electronic form. Content of electronic resource differs significantly from print resource. Resource undergoes a change in format, usually from print to electronic. Resource is a database or web site whose content is equivalent to more than one print source (e.g., Web of Science). Resource is a database or web site whose content includes significant new material beyond existing print sources. The original text cannot be definitively identified."

This CONSER proposal has firm supporters. Mary Grenci, Serials Catalog Librarian, Knight Library, University of Oregon, stated:

"Unfortunately (in my opinion) we do not put multiple formats on one record here at the University of Oregon. The one exception is for titles received in both remote electronic and paper formats. In this one case, we note the existence of the electronic version on the record for the paper. The only reason we are doing this is because CONSER has given a semi-official 'okay' to the practice, I argued strongly for it, and when I asked public service staff what they thought they agreed that it was preferable to separate records (I showed them examples of each).

" ... I can say that I don't think it's all that important if patrons don't look at the notes in the bibliographic record that explain the various formats held. As long as they can tell from the holdings screen what vols. are held and where they are, then that's what really matters. For our purposes (as librarians, and in particular for collection development purposes) it is important to have the notes and to be able to easily tell what we own in each format. It would be helpful for patrons to know this too, so they would know if they have the proper equipment and such, but they usually don't much care.

"I guess what I'm saying is that what really matters is whether or not patrons know where to go to get something. This information can be in the bibliographic record, or it can be somewhere else--it doesn't matter, as long as the patron sees it. Using separate records means that in most cases the patron will not see all of this information because they simply won't look."

Adam L. Schiff, Principal Cataloger, University of Washington Libraries, whole heartily supports the single record concepts, and predicts that rules will change in that direction, in contrast to Erica.

"We have adopted the one-record approach for monographs as well as for serials when we own a tangible version of something and there is also a one-to-one equivalence in its remote electronic version. (Yes, we can debate what one-to-one equivalence is, but we generally follow LC's guidelines about this as spelled out in their Draft Interim Guidelines for Cataloging Electronic Resources, (http://lcweb.loc.gov/catdir/cpso/elec_res.html). I also like the University of California's Task Force on Electronic Resources report on cataloging of electronic resources (http://tpot.ucsd.edu/Cataloging/HotsElectronic/tfer.html). Among its recommendations is to 'use a single record approach whenever the electronic resource would serve as a suitable substitute for most users.' We only use a one-record approach (LC calls this a delineation) if and only if we also own a tangible (print, microform, CD-ROM, etc.) version of the electronic resource. If we don't own the tangible version, then we do a separate record for just the remote electronic version. Our policies for doing a single record require that we add an 007 for the electronic version, a 530 note, an 856, and a 740 added title entry if the title of the electronic is different from the title of the version described in the 245. In some cases we modify our local bib. records and in other cases we upgrade a record on OCLC. Some examples of single records for monographs in OCLC: #40415131 A reading group toolbox for the works of Russell Banks, #38246248 International directory of art libraries, #40108464 Estimating historical snag density in dry forests east of the Cascade Range, #20380090 Catalog of the Diptera of the Australasian and Oceanian regions, #10577390 Species profiles : life histories and environmentalrequirements of coastal fishes and invertebrates (Pacific Northwest). Chinook salmon, #10579955 Habitat suitability index models. Clapper rail, #39029134 Volcanoes / by Robert I. Tilling. 1998., #34333394 Philosophy in cyberspace.

"It is all well and good to say that we should do separate records for all electronic resources, but the facts are that users probably prefer to retrieve one record for a resource rather than multiple records for multiple manifestations. Web-based online catalogs can do a very nice job of handling single-records for these things. As an example, search our catalog for the titles listed above (http://catalog.lib.washington.edu/search) to see how this works in one particular catalog. Very few libraries have the staffing to do original or copy cataloging for all the electronic resources that we would like to include in our catalogs, particularly if most of the work has already been done when the tangible resource was cataloged.

"Finally, although it is correct that AACR2R currently does not sanction the creation one bibliographic record to represent multiple manifestations of the same work, this is likely to change. CC:DA (ALA's Committee on Cataloging: Description and Access) is on record as favoring a rewriting of the rules so that they would permit using one record for multiple manifestations and it is likely that rule revisions will eventually get proposed and passed to this end."

It seems evident that until North American Librarians create indexed, browsable, classed files, and educate patrons in their use, the effort to assign class numbers to materials not requiring a call number for shelf placement will not appeal to some cataloguers. If it is wished to assign multiple class numbers and suppress them from public view as Charley suggests, than a MARC field for that purpose in needed in both MARC21 and in local systems. The two tags which have been so used are 089 and 699.

In the meantime, it will primarily be those libraries who have other uses for class numbers (such as collection development, bibliography production, and SDI services) which will assign classification numbers to Internet resources.

The present examples of single records for text and electronic resources are not composite records. Even with the addition of an 007 (physical description fixed fields) for the electronic resource, they are records for the print or other tangible form, with the electronic version noted. To be a combined record, there would have to be a compound gmd, e.g., "[text & computer file]", a second 300 with an smd such as "website", and a differing title for the computer file in 246 with identifying $i information. The 530 (additional physical forms available note) and 740 (added entry for related or analytical title) imply that the computer file is a related work, which in fact it is. One could equally well add the 530 and 740 to each of the separate records for the differing formats. Considerable further development is required to have a "work" record covering two or more manifestations. As has been suggested, there seems little reason to single out Internet resources for single record treatment with print versions, while other physical formats receive separate records.

* In all cases only portions of posts are quoted to help reduce redundancy among posts. Only internal omissions are indicated. The posts have also been passed through spell check. U.S. Congress persons may remove misstatements from the Congressional Record, but we Autocat posters' bloopers are there for all to see in the Autocat archives, which may be searched for the full texts of the quoted posts.

The archives may be searched on the WWW. Go directly to the following URL: http://listserv.buffalo.edu/archives/autocat.html