Holdings are multipurpose data that libraries use to order and acquire materials, display information to users, and manage physical and sometimes virtual items. This workshop focuses on how the various functions interact, what further benefits libraries will expect from automation of their holdings, and what they can do in the meantime to make their holdings more useful. It will give special attention to the USMARC Format for Holdings Data, its features, advantages and difficulties, and its particular role in conveying holdings information. It will explore possibilities for adapting the format to solve the special problems associated with multiple versions and title changes.

Introduction

Welcome to "Do Holdings Have a Future?"

It's wonderful to see so many people here to listen to someone talk about holdings. It certainly suggests to me that holdings, if they don't have a future, must have a very lively present: one on which, perhaps, we can build.

We can't overlook the fact, however, that some of the audience may be here because holdings in automated systems can be a problem area for staff, precisely because they are at the hub of so many activities and are met with in so many forms. Staff responsible for them want to know whether there is a way to reduce duplication of work, to steer between rigid rules and utter permissiveness, and whether each person who works with holdings is obliged to go it alone in finding this balance.

Some of these conditions may apply to you:

• Holdings work is little understood by and little supported by your administrators;
• Reference staff and users have very high expectations but often don't understand what they're viewing--or don't look at holdings when they should;
• you have very little guidance available in working with holdings;
• when you change online systems, you wonder how it will affect your holdings;
• you wonder why you and your colleagues have to input holdings for the same material into different files in different ways with little coordination among you;
• your online system does not implement certain standards;
• there are real practical problems for your library in following certain standards.

The standards issue is one I am going to stress throughout this workshop, because of the vicious circle that is established when standards are not available or not regarded. Data that is inconsistently formed, or uses a standard in conflicting ways, may be a real stumbling-block when your library wishes to upgrade or add features to the system. A library might be forced to say, "Don't implement the standard for us; it wouldn't work for the way we have done our data." On the other hand, if you are consistent in the way you don't follow standards, and you can distinguish the parts of your record--at a minimum, the holdings from the notes--you probably will be able, with the aid of extra programming, to migrate your non-standard holdings into a standard MARC holdings format, perhaps making use of the free-text options in that format.

Many of us, in fact, are starting out with serials systems that do not complete support our standards. Even if we can't code completely for the standard and aren't able to use all of its functionality, we have a responsibility to have our holdings at least in conformance, not in contradiction, to the demands of this standard.

There would be nothing I would like better to come out of this workshop than these a sense that we might use our common interest and strength and skill in a quest for five things:

1. more consistency in holdings data
2. cooperative development of a better and more workable model to use, a better standard itself (developed cooperatively like the bibliographic record)
3. cooperative and competitive urging of better implementations in our systems (why not put the serials system at the top of the list of desired features, rather than last?)
4. better guidance in the form of interpretation and documentation
5. and archives of content such as publication patterns and the publication data itself that we can draw upon. (If we want it....we need to work for it!)
   

Why the topic?

In the year since the topic was proposed to NASIG, the picture has continued to change. It's something of a pendulum swing throughout the decade. Initially as more and more holdings were visible online from all over the globe, things seemed very hopeful for holdings.

Then starting about 1994-95, we saw many less favorable signs:

1. So much emphasis on those new virtual items in the public catalog that are not "held" by the library; so much talk of de-emphasizing local collections and turning our catalogs into finding aids for global information;
2. We heard constantly that the MARC format for holdings was little spread and little supported. That report was somewhat over-exaggerated;
3. The MARC Holdings Format Interest Group at LITA disbanded and simply became the MARC Formats Interest Group;
4. A consultant at LC wrote an article about how LC was designing a non-MARC holdings serials system and made the claim that "no library uses the MFHD for check-in and claiming"; which would be news to many libraries that do.

However, over the past couple of years...

1. The Web Catalog spread. New looks for the OPAC and new ways of navigating give us the chance again to assess how we allow search and display of information to users. New emphasis is put on making the holdings screen informative, and this is true even for electronic resources even when we do not give holdings per se for those items.
2. Not one but several vendors have announced new implementations of the Holdings Format. As always, these vary in completeness.
3. LC gave up its separate system and bought a system, Endeavor Voyager, which does utilize the MFHD.
4. The interest in remote searching of holdings has revived discussion of standards for holdings structure and content.
5. The newest NISO holdings display standards (Z39.71) have been re-ballotted and approved. They should be published this fall. [Note: available as of February 1999]
   

All of these trends have induced people to take a look at what holdings are and do, and what libraries want from them.

Holdings are at the hub of library serials use and serials management, just as central as the bibliographic record. A holdings record can do all of the following:

• serve as a basis for check-in and claiming
• record bound units, with barcodes for circulation
• generate a spine label
• display a summary holdings statement to users
• become part of a Z39.50 retrieval from a remote site
• serve as a report to a union list
• answer a reference question

The portion of reference questions dealing with serial holdings is in some surveys reported to be over 40%.

It hasn't been easy for the holdings format to do all that, for vendors to program it, or for us to code it.

History of the Standards

Timing has been unfortunate in a number of ways. Holdings are among our newest bibliographic standards, all arising during the 1980s. First came the display standards. These were developed for a paper and microfiche environment, including union lists. The first was a now superseded summary display standard, Z39.42. It allows open entries, does not include captions, and records only at the highest level, for example, the volume, which was counted as held if the library held 50% or more of it. There is no provision for recording of supplements and indexes. I emphasize that there is data conforming to this standard being loaded even today .

Probably everyone here has heard something about how this summary standard, Z39.42, conflicted with the next detailed holdings display standard Z39.44, and how they had to be resolved. In the end, a combined summary and detailed standard was published. Its summary holdings level defined "holding" a volume differently. The volume is now counted if the library has any of it. There are captions in Z39.44. There are even display options, mindful of all those libraries with older data: Enumeration and chronology could be reported together, or separately.

At the new fourth level of holdings, every gap in holdings now had to be recorded fully. The presence of a volume is a guarantee that the library owns the entire volume (or at least did so at the time of the report).

Meanwhile, another set of standards had to be reconciled, and these were the monographic standards--Z39.57. The new combined standards are being put together as Z39.71. The news was that the ballotting which took place early this year was successful but there were some comments raising questions which need to be worked out before the final version can be published.

Z39.71

The new standard is impressive. It has flexibility where the former standard was extremely rigid.

1. It gives more space to instructions for itemization, that is, the presentation of holdings in terms of individual items, which is the way many libraries want to present them today. (Why? because that way we can present, as well, the special statuses, notes, etc., that apply to the individual volumes as volumes.)
2. It has eliminated some of the excessive detail and repetition that occurred in the old standard whenever there was a gap in holdings. Captions, for example, can be omitted in the part of the holding that comes after the hyphen.
   
3. There can be an open holding at Level 4, the most detailed level.
   
4. The presentation at either level can include a "separate" presentation of enumeration and chronology, rather than adjacent presentation, that is,

   v.1-2 (1991-1992) as well as

   v.1(1991)-v.2 (1992).

   In the earlier standard, the first presentation was an option only at Level 3.
   

The USMARC Format

The standards for holdings display, which fit in a modestly sized, slender book, should be distinguished from the Holdings Format, which is in a big binder. A concise version of it is available on the MARC web site at theLibrary of Congress; the whole is available at the Library Corporation's web site. However, for the purposes of this web site, we have developed a special version with a more explanatory approach

USMARC Format for Holdings Data Handbook

Web Visitors: This Handbook is the heart of my presentation, so don't overlook it in speeding through this text!!

The USMARC Format for Holdings and Locations was in its final form just about the same time as the publication of Z39.44. The publication date is 1986. With the second edition in 1994 it became the USMARC Format for Holdings Data.

The MARC Holdings Format was developed by libraries in the field, members of the Association of Southeastern Research Libraries, or ASERL, who wished to have a means of communicating holdings to each other for purposes of resource sharing. It was created by serialists for serials. SOLINET played a leading role in its development. You often hear about how serials were kind of shoehorned into the bibliographic format which was made for monographs--but the MFHD was made with serials in mind.

Did the aforementioned display standards come into the picture? Yes they did; that is, the two groups consulted, but the Format is kept deliberately out of the business of prescribing a display. Those who wish to display holdings in terms of physical pieces are perfectly capable of doing so using the MARC Holdings Format. It is capable of generating many types of displays-but the success is dependent on the programmers who implement the format in a particular system.

There are several things that could be said to characterize the holdings format

It is very complex. It contains various pieces of data that have to work together. If a piece is missing, not only that piece of information but perhaps the entire display or the entire functionality is lost. The unique feature of the format is the "paired fields" construction of the actual holdings data, with the captions in one field which is linked by means of a shared number in a subfield to a series of data fields containing the corresponding enumeration and chronology. The contents of both the caption fields and the enumeration-chronology fields are then displayed together.

$a v. $b no. $i (year) $j (season)

$a 3 $b 1-3 $i 1997 $j spring-fall

The display is not mandated; it can be manipulated by computer to various forms of presentation. Here is the NISO-like display discussed above:

v.3:no.1-3(1997:spring-fall)

(note that the "captions" for year and season are suppressed from display by putting them inside parentheses; the parentheses are NOT there to insert the chronology data inside parentheses, although that is also done!)

Other systems might use a local display standard:

v. 3, no. 1-3 spring-fall 1997

Other characteristics of the Format:

It uses a fairly small range of fields in comparison to the bibliographic standards, using only those beginning with zero, five, and eight. Access fields such as the 856 were originally designed for the Holdings Format. Like the bibliographic standards, the MFHD has been extended by some vendors with proprietary fields and subfields which might begin with 9 or some other number or letter.

Some difficulties exist for workers with the Format.

As noted, it has been subject to neglect, with few revisions and even fewer interpretations, because it is thought of as a carrier for local data. Standards have not been enforced, and incomplete, non-standard implementations lead to non-standard data that later cannot be accommodated correctly. To gain ground, we must once again focus our attention on the Holdings Format and make a case for the functionality that we need.

Also, despite the significance of the local, it has a strong and very stubborn bias toward those aspects of local data which are important in the national and global arena, for interlibrary loan, union lists, and the like. It does not allow for a lot of management data. But even on the global front it is somewhat behind the times, because it leaves out some crucial data such as availability of individual volumes, requiring that to be accommodated elsewhere in the system to be sought out and combined with holdings data by remote search engines. Also, very little of the essential, complex coding can be done on an automated basis.

The format has some ambiguities. I admit to not understanding the concept of expansion of a compressed holding (for example, Vol. 53-56 is a compressed version of Vol. 53, v.54, , v. 55 v.56, and each of those is a compression of v. 53 no. 1,2,3,4, etc.). You can explode it once it is compressed, but can you do it accurately, by automated computer algorithm? Even knowing how many issues were supposed to be published in a volume doesn't tell you how many issues actually appeared, what their designations looked like, and how many were combined issues. Also, though physical pieces can be coded, I do not feel it is easy to show unambiguously to users, staff, and interlibrary loan or even to yourself which of your data represent physical pieces unless you code them all that way and are able to make a blanket statement to that effect.

Another problem is that there is often a divided responsibility, divided workflow, and distinct procedures for different flows of holdings work. In some institutions, the check-in staff are responsible for coding in new annuals and irregulars, which often get barcoded on receipt; while the periodicals get their final coding when received back by binding staff or by branch librarians. In contrast to the level four coding of the annuals, the periodicals often get an open-entry summary holding. In some cases, the latter can be considered Level 4 (as permitted in the newest standard), because the staff is conscientious about noting gaps. But in others periodical information is not kept up to date. Staying on top of changes in status and reflecting these in coding is a difficult task for any library.

The programmer's job in making the Format work may also be fairly characterized as difficult; a vendor-librarian discussion on the MFHD held at ALA in 1995 brought this out very clearly. Even the basic premise of the paired fields is complex. NISO standards prescribe a display whose elements are in what we know as 362 order, "enumeration-chronology, enumeration-chronology:"

v.1:no.2(1996: Feb.)-v.3:no.5(1998: May)

Even Option B--the other optional order--does not have the data in the same order as the MFHD:

v.1:no.2-v.3:no.5(1996:Feb.-1998:May)

The MFHD also prescribes input in the form enumeration-enumeration, chronology-chronology, as Option B does: but it also requires data at one level to be expressed in one segment; it cannot be repeated. A statement with incomplete ranges and levels as above is not possible in coded form unless manipulated programmatically, since the first level of enumeration (a) and the second level of enumeration (b) are not repeatable. Data would have to be input in the form

$a vol.-vol. $b no.-no. $i (year-year) $j (month-month).

At this point, very few vendors seem to be manipulating the data so that the data of the second level ($b and $j) can display interspersed with the first level as it is in the Option B example above, let alone the Option A display above that. Should they? It isn't clear. Some enumeration, in particular, has internal hyphens. It would be dangerous to interpret a hyphen as always representing a connector between the first element and the last element of a range. If holdings communication were at the same level of sophistication as bibliographic standards, we would need to reserve all of those elements of holdings punctuation, as we do for internal punctuation in the bibliographic standards. Until these matters settle out, it is best to accommodate complex data by using the free-text field options in the Format.

The foregoing has presented some problems. So it's time to bring in the advantages.

Benefits of automated holdings

If you have a good implementation of the Format, it provides for some things you are SUPPOSED to have, and other things you really need, such as:

1. a way to make a copy- and location-specific report, or combined report, at your volition
2. easy ways to navigate or jump among screens of related data
3. easy copying of information from one place to another with automatic resequencing
4. minimal need to recode when you have to move data or insert data
5. ample space for notes and the means to enter commonly used ones via codes, or shorthand
6. clear, easy-to-interpret interface
7. automatic compression of holdings if you have the publication pattern data (automatic compression of holdings should be possible at the volume level even without that data!)
8. correct linkage with other modules; access to information and editing capability for all related data from inside your MFH program without having to back out
9. provision for ordinal numbers, which appear before their captions rather than after
10. the ability to use free text holdings and/or notes and coded holdings together with proper display and ability to suppress the former in favor of the latter
    

You should be able to express almost any kind of enumeration or chronology in coded form, or if not, in free-text form. Programming can be created to transfer the data into almost any display. Someone should test the capability by inputting test data. This sounds elementary but is crucial. Buy a system on the basis of what exists, not what is promised: another time-honored maxim.

You should have enough space, at long last, to give any amount of detail you need about the copy at a particular location, the individual piece, or the relationships between one holdings record and another, one bibliographic record and another.

You should be able to display all the data that exists to the public if you want to. There is no excuse for giving you the option to show only check-in data or only summary data, so don't accept that. It should be as easy for the public to navigate and identify what information is being shown as it is for you.

In short, you need the ability to display what you have in the optimal way to help your users.

Holdings pitfalls

It is our responsibility to avoid some of the possible sins against data consistency:

Don't mix caption data and enumeration-chronology data in the coded fields, even if you end up with something that looks right. That mixture is fine for the free-text fields but not for paired fields.

Be careful with indicators and with link and sequence numbers. It is better to leave indicators blank than to code them with the wrong information. Be especially careful with link nos. if you combine free-text with coded holdings. If you give link no. zero to a free text field, you may be telling the system not to display any of your coded holdings!

Provide a training program and make sure that if you have various units doing the coding in several different workflows, they are all using the same guidelines and methods.

Make notes in a standard wording. If you use macros or codes, you have a better chance of having consistent wording: something that can be globally changed when you need to do that.

The Future of holdings

There are so many frontiers in holdings. One of them is the Web. Web linking between citations on remote databases and individual local holdings records is already being offered by some commercial companies, just as they offered "hooks-to-holdings" in the non-Web catalogs of the past. This kind of linking is possible in-house as well. If there is a separate MARC holdings record that can be accessed over the Web, we may be able to use its control number that in a Web address which could be accessed via library created finding aids for such materials as newspapers or serials in special collections or subject areas.

In terms of public service, we need to make sure that we add value to the greatest extent possible to the information we give the public about holdings. If we find that users aren't looking at holdings screens, we ask ourselves, Do we give them enough information? the right information? It clearly isn't that they don't care about holdings information! So what can we do to enhance that information?

For instance, we have a location in a holdings record. By clicking on the highlighted location, could one bring up in most systems a stacks map? a building address? library hours?

Are there sufficient user-friendly notes? The bottom-line question is, "Would a library user leave the catalog record for a title without recognizing that the library has the needed piece? Surveys of public service staff at the UNC-Chapel Hill Libraries have brought in strong endorsement of holdings notes.

One of the most acute problems we have is the coordination of holdings that are fragmented both horizontally and chronologically across titles and formats. There is a strong need in catalogs to tie related records together. In large databases, it is of limited use to relate records simply by assigning headings in common! Close relationships of microformat to printed or electronic format (with identical content), of earlier title to later title, need to have reliable pointers to each other. In today's catalog, one of the best places to do this is on the holdings record. It is in looking at a holdings display that someone is likely to say, "I see v. 12, but I don't see v. 13-15; don't we have them?" or, "But I don't want to read it on [the Web] [microfiche] [etc.] What else is possible?"

The following is a series of four diagrams showing a title which has undergone two title changes and is held in two formats. The first title is held in microform and print; the third in print and electronic format; the second in all three formats. The special relationships highlighted are the horizontal (across formats) and the chronological (across title changes).

1. the traditional organization of related titles, each separately cataloged, each linked to a holdings record.
   
2. the increasingly popular Multiple Versions (Mulver) solution. In this approach, the primary format (usually the print) is cataloged, and the additional formats are noted, rather than fully cataloged, on the record for the primary format. In the holdings record(s) for the additional format(s), bibliographic information, e.g., of the type usually found in 533, is included; but the secondary format(s) do not receive full cataloging. This approach does nothing to solve the title change (chronological) problem.
   
3. the Link Record solution, inspired by the Toronto Conference on AACR2 Revision. In this approach, a "work record" is established as a layer between the various titles and formats and the combined holdings records. The work record is a very brief MARC or MARC-like record which contains links to all the separate titles, including both title changes and the identical title in different formats, as long as each is linked to another record in the cluster by a 776, 780, or 785. Holdings records consist of a record for each format, but within that format record, the records for all title changes are combined. Subfield $3 can be used to distinguish the holdings that belong to a particular title, and permits programming to extract and display the holdings for just that title.
   
4. the "Activate the MFH" approach. This approach is a modification, somewhat simplified, of the solution proposed by Melissa Bernhardt (Melissa Beck) in Cataloging and Classification Quarterly, v. 9, no. 2 (1988). As in the link record solution, records for the title changes and the various formats are retained in the system, but they are linked via their holdings records by means of special programming. The first program tests the control numbers embedded in each 780, 785, or 776 to see whether they find a record in the database. If they do, that linking field is marked with an "ownership subfield;" if not, a "non-ownership subfield" is placed. If the field is a 776, it also sets a "format" subfield using the 007 field of the linked record. For all the linking fields that contain the ownership subfield, another program then places a note in the holdings records, containing the words: More holdings: identifier for each type of link (Earlier title: Later title: Other format (Microfilm): and finally, the linked title (the subfield $t, or $a and $t, of the linking field. If a non-ownership subfield is in the field, there is no highlighting in the bibliographic record and no note in the holdings record. The program could go one more level by testing each earlier or later title for the presence of 776 fields with ownership subfields in them; these fields then set notes also: Earlier title (Microfilm): Later title (CD-ROM):
   

If you're familiar with cataloging in the USMARC format, you'll recognize the bibliographic field tags and the way they signal relationships between serial records. Arrows on each diagram show how linking fields build bridges representing the different relationships. Comments are welcome on these diagrams. Are there other solutions to propose? Modifications? Let us know!

• We ask that the holdings record provide so much--basis for check-in, prediction, and claiming, physical piece management with binding and barcode, display to users, "hook" from citation, report to a union list, response to Z39.50 search, spine labeling.

  Can it really do all of that, and do it well?

• Even if it can, is it just too complicated for staff to code correctly or vendors to program?
• How can we get the utilities to commit to participate?
• When the information needed for management of specific copies and user display conflicts with the information needed in a national report, how do we provide for both needs?
• Remote retrieval needs availability data usually found only in item records, not accommodated in MFHD (even in the new item fields). Management needs transaction tracking. Should we try to incorporate more transaction-oriented data in MFHD?
• What is the verdict in the field on the NISO holdings display standards?
• What is the optimal level of coding to satisfy local and global requirements?
• How much tolerance is built in to the system to accommodate violations of standards?
• To what extent can we still improve the standards? How do we proceed?
• Who, or what body, should be carrying our suggestions for improvement forward?

1. 004 field (link to bib record). This field is non-repeatable. In some systems (ours for example), repeating it is the only way to show item status on records linked to one barcode (e.g., in- analytics and bound/withs). Are such violations of standards harmless, and how do we decide? [Note: MARBI has disallowed repeatable 004. The only approved method of tying items together is to link at the item level, ordinarily by duplicate barcode.]
2. Receipt/acquisition status, General retention, and Special retention are often used by systems to set local notes specifying the status and retention of individual copies, inevitably directing a library to input a copy-specific code. To derive composite reports, could a program rank code combinations and accept highest value as institutional default, e.g. Currently received/Permanently retained?
3. Completeness. Prefer an institution-specific (rather than copy- specific) code for each MFH record unless copy-specific data is needed for local use?
4. Copies. Institutional policy on what constitutes a second or third copy in multiple-volume titles varies. Do we need consistency (what is the data used for)?
5. Compressibility/expandability (853/854 1st indicator). This functionality has been very little developed by vendors. Lack of benefit, combined with lack of testability, is a discouragement to coding, and in a true vicious circle, lack of data gives little motivation to develop the functionality. Should we have a discussion about what exactly the functionality requires and then modify the Format instructions for input to make the requirements for coding clearer?
6. Suppression of paired fields in favor of free-text holdings (863/864 2nd indicator 2 or 3). Another missing feature in implementations. Do we need to make a case for it?
7. Enumeration/chronology. International holdings display standards at the highest level (Level 4), as well as one of the two summary- level (Level 3) options, require holdings in the form enumeration (chronology)-enumeration (chronology). The MFHD requires coded input in the form enumeration-enumeration//chronology-chronology, which then must be manipulated by computer (or replaced by free- text fields) to get the Level 4 display. Unless systems provide this manipulation, a NISO statement:
   v.1:no.2(1993:June)-v.3:no.2(1995:June)
   appears as:
   v.1-3:no.2-2(1993-1995:June-June).
   How can we either change the Format or provide guidelines to avoid this garbling?

Barry B. Baker, ed. USMARC Format for Holdings and Locations: Development, Implentation, and Use. New York, Haworth Press, 1988.

Bernhardt, Melissa (Melissa Beck). "Dealing with Serial Title Changes: Some Theoretical and Practical Considerations," Cataloging and Classification Quarterly 9:2 (1988), 25-39.

Goldberg, Tyler, and Eric Neagle. "Serials Information in the OPAC: A Model for Shared Responsibility," Serials Review 21:4 (Winter 1996), 55-63.

McMillan, Gail. Coded Holdings Manual: Applications of the USMARC Format for Holdings at University Libraries, Virginia Polytechnic Institute and State University. 3rd ed., 1993.

McNellis, Claudia Houk. "A Serial Pattern Scheme for a Value-Based Predictive Check-In System," Serials Review 21:4 (Winter 1996), 1-11.

Rosenberg, Frieda. "Managing Serial Holdings," in Tuttle, Marcia. Managing Serials. Greenwich, Conn., JAI Press, 1996, p. 237-256.

Van Cura, Mary Ann, "A Step Beyond Shared Patterns: A Shared Holdings Record Database," Serials Review 17:3 (Fall 1991), 72-76.

Wallace, Patricia M., "Serial Holdings Statements: A Necessity or a Nuisance?" Technical Services Quarterly 14:3 (1997), 11-24.

For holdings documentation, see http://www.loc.gov/marc/. Also see http://www.loc.gov/z3950/agency/ for information on the use of holdings in remote retrieval.