The Ohio State University

From: Hannah Thomas

To: SCRT --FOR DISCUSSION

Date: 13 March 1990

For many years, OSUL has been using some non-standard second cutters following certain literary author numbers in RAR. Special schemes have been devised for classing works by or about these authors located within other works, for classing by illustrator (for Thurber only), and for association items. We need to determine the appropriate treatment of such materials, considering this practice in the light of applications outside RAR. Similar types of materials do occur in other locations, not necessarily restricted to literary authors.

I.   SUMMARY OF PAST PRACTICE

1. The special numbers used to date in RAR are all for authors in PR or PS with one literary author (first) cutter. The following special usages were invented:

   Z7	contributions by X in books
   Z719	contributions by X in periodicals
   Z72	books with illustrations by X
   Z82	articles about X in/from periodicals
   Z83	articles about X in books
   Z9	X-iana (association items)

   The numbers invented are not defined for anything else in the LC schedule at the present.

2. The number of authors involved in RAR seems to be quite small. I have found evidence in the catalog of non-standard cuttering for six authors (there may be others, but I don't think so):

         Nelson Algren (97 records)
         Samuel Beckett (28 records)
         Robert Duncan (1 record)
         T.S. Eliot (1 record)
         Jack London (119 records)
         James Thurber (160 records)
         Total to date 406 records

   I have found some notes of Biruta Osis' which indicate intention to use these numbers for Helen Hooven Santmyer, but this has not been done yet. Similar accommodations are also needed now for Chester Himes and William S. Burroughs materials awaiting cataloging. Similar provision will be needed case-by-case for other authors in future, but it will never be a very large number.

3. The attached table (Appendix A) shows the frequency with which each of these numbers has been used to date. The majority usage so far has been for providing access to individual works by X contained within a larger work (78%) and for works about X within a larger work (13%). This should continue to be true.

4. Subarrangement within these numbers has been handled fairly consistently, except in the case of Z9 for Thurber, where earlier items were volume-numbered and later ones triple-cuttered. See examples of each type attached.

II.   PROBLEMS WITH THIS APPROACH

1. Special classification vs special cuttering. RAR has a clear need for special classification. That is, we want the anthology with a Thurber story classed with Thurber, not as an anthology. The need for special cutterinq is less clear --is it really necessary to separate this work by Thurber (Z7) from other works by Thurber (A-Z)? Ditto critical works. If physical separation from the main sequence of Thurber books is desired, sublocation codes would be a preferable way of accomplishing this, if we had them, but special cuttering may be needed since we don't.

2. Potential problem of usurping undefined numbers. What if LC defines them in future for something else? This problem is not limited to Special Collections --OSUL has used non-standard Z8 very widely --but is one basis for resistance to special schemes.

3. The Duplicates problem. The other reason for reluctance is the problem it poses for handling duplicates. If a second copy of the anthology is received, Duplicates Unit must know not to add it to the Thurber-classed record. This problem will arise whenever what is described is not what is classed, and can be avoided in most cases by describing the portion being classed. (A second potential problem with duplicates is that OLIS might pose difficulties in future with 2 LCS records created from the same OCLC record. Again, this problem would not be limited to Special Collections. )

4. Application outside RAR. We don't want to be locked into a practice which only works only for one location or only for literary authors. Several locations have need for cataloging works-within-works.

DISCUSSION/RECOMMENDATIONS

The appropriate treatment is not the same for all categories. Over 90% of cases are in first 4 categories. Describing these works as in-analytics will sidestep the duplicates problem, and the use or non-use of special cuttering then becomes a separate issue. The other 2 cases (Z72 and Z9) are harder. Categories:

1. Author of works contained within a book (Z7)
2. Author of works contained within a periodical (Z719)
3. Subject of text contained within a book (Z83)
4. Subject of text contained within a periodical (Z82)
5. Involved in the creation of a work in ways other than main-entry responsiblity (e.g. illustrator of whole or part) (Z72)
6. Associated with a copy of a work in ways other than creative role (e.g. dedicatee, owner, character in novel/play) (Z9)

Works within serials

b & d:
Works by or about X within serials are now handled with in-analytic records. See Examples 2 and 4. The bib record describes the part, with an “In:” note citing the whole item in which the part occurs. This is correct handling and does not cause problems in the classification of other copies.

Works within monographs

a & c:
Works by or about X within books have been handled differently, with records for the container book. The bibliographic records describe the larger work, with a "Contains: " note to indicate the component part, plus author-title added entry or subject entries for the specific texts by or about x.
See Examples 1 and 5.

This is the situation causing a problem for Duplicates. When they try to add a second copy of the anthology, the classification does not correspond to what they have.

A more logical approach would be to handle these also as in- analytics, with a record for X's text, plus notes and added entry for the container work. There is no apparent logic behind the decision to treat texts within serials as analytics but texts within in books not so. Describing the entire book while classifying for one part is inappropriate; classification should reflect what is described.

In-analytic cataloging

In-analytic cataloging is perfectly legitimate, though not frequently used. The practice is explained in AACR2 Chapter 13 and accommodated in OCLC Books Format. OSUL has largely avoided this type of cataloging because of LC's statement (LCRI 13.5B:CSB11) that LC does "not employ in-analytics except in very special cases". The practice is avoided except in special cases (such as rare-book cataloging), because this degree of specificity is usually not of interest to other users of OCLC. However, if the work is significant enough to justify collecting by a rare-book library, it is special enough to justify cataloging. It cannot be handled in the normal routine of general cataloging, however, and will need to await handling by Special Collections Cataloging (which can be a very long wait indeed). The obstacle to this level of cataloging is not technical feasibility but priority setting.

Analytic records should be coded "a" ( component part, monographic) in Fixed Field Bib Lvl, not "m" ( monographic) as OSUL has done in the past. Bib Lvl "a" is defined as: a bibliographic unit that is physically contained in another bibliographic unit (host item), such that the host item must be identified in order to locate the component part. Examples of component parts within host items are: one article in a serial, one paper or chapter in a book, one band on a phonodisc, one map on a sheet containing several maps.

Access to the larger work should be provided by means of a note and an added entry. In the past, we have used a general 500 note and general author-title added entry, but there are fields specifically defined for this situation. We should use a Linking Entry Complexity Note ( Field 580) beginning "In:", and a Host Item Entry (Field 773).

Field 773 is not indexed on OCLC. This is correct, since we do not need an OCLC title search for, e.g., an anthology to retrieve the record for a single story in it. We should try to get this field indexed on LCS, however, to let a search for the anthology title reveals the whereabouts of this copy of the anthology. This is the same field used to link collection and sub collection records in manuscript cataloging (a closely parallel whole-part situation), and we would like access to Field 773 in those records too. If a decision is made not to index 773 on LCS, we should consider substituting author-title added entries to provide this access in the case of analytics.

Works classed for illustrations

e:
Works containing illustrations by X but main-entried under another entry are now handled by describing the whole and adding notes about the illustrations & entries for illustrator. See Example 3. Classifying the RAR copy by the illustrator is causing problems with duplicates, similar to that described above.

This problem will persist whenever the classification does not correspond to what is described in the central portion of the record. There are no perfectly satisfactory options for handling this case. The possibilities are:

1. Use in-analytic records describing illustrations and class accordingly (as for texts, above). This is conceivable, but would be complicated, since the illustrations of interest to OSUL may be scattered rather than a discrete portion of larger work. Drawback: It seems undesirable to create a bib record on OCLC for, e.g., Thurber drawings on dust-jacket, as if they constitute an independent work.

2. Continue as we are doing and continue to warn Dups to watch out for special Z72 numbers. With one exception, Z72 has been used only for Thurber. We don't anticipate wide use of this number in future in RAR, but could in CGA. Drawbacks: goes against the principle of classing what is described; not easily applicable outside RAR. Advantage: general cataloging workflow can do the descriptive record, Special Collections Cataloging needed only for call number.

3. As ii above, except construct the number with the prefix SPEC, to make it more obvious to Duplicates Unit that subsequent copies are not to be added to this record. Drawbacks: separates these materials from others in shelf list sequence; SPEC not normally used for books.

4. Lump such works together with category Z9 below.

Association items

f:
This is the most complicated situation and cannot be solved by analytics. There is no satisfactory way for the main entry to match the classification. Copy-specific added entries or added subjects are needed for access to the descriptive record. The best we can do is have a call number so weird looking that Dups knows it’s not real.

These records have been handled two ways in the past. The earlier practice was to treat Thurberiana items collectively as an artificial set and assign each new item a volume number under the general call number PS3539H93Z9. See Examples 6 and 7. Cards were made in the RAR catalog for the individual "volumes", but this was before LCS/OCLC cataloging and there are no FBRs. There is an MSET short record on LCS for the" set", with 39 volumes enumerated in the holdings file, but no indication of the contents of individual "volumes".

The eight records created since 1985 were done instead with a individual record for each item, using a 3rd cutter to create a unique call number for each item. See Example 8. The problem with this approach is the Duplicates problem --when the description and classification does not match, adding second copy is a dilemma.

The most desirable approach to description would be to have a record for each item, with needed added access points. This could include tracing (on LCS only) as an artificial "set" (a named internal collection). There is no perfect solution to the classification problem, since we do not want to class for the bibliographic item.

The main problem with the old approach was that in the past we had no way to give individual access to the component "volumes" of an artificial collection. We now have a way, which might provide a model for individual records classed together with "volume" numbers in the call numbers (now used for "works in parts"). The Dups problem persists, but the call numbers should look odd enough to alert Dups.

Another option would be to use a prefix (SPEC) to remove the Z9 numbers from the regular sequence. These would also look odd enough to alert Dups.

In either of these approaches, Special Collections Cataloging would need to provide the call number, but the remainder of the record could be completed in the regular cataloging workflow.

If neither of these two is acceptable, then it is best to leave them in the undifferentiated artificial collection (e.g. "Thurberiana"}, and give up on providing item-Level access to the individual component titles on LCS.

Application outside RAR

In other locations, works-within-works by or about literary authors (e.g. playwrights) would be exactly comparable to the RAR situation and should be handled in the same way. Works by/about non- literary-authors (e.g. actors, directors) are a less exact parallel. Similar treatment should be given as far as describing the component work in an analytic record and classing for the component work, but special cuttering schemes should be avoided, even if a decision is made to use them for literary works. In all locations, handling of the Z72 situation (class by secondary responsibility) and Z9 (association) need further discussion.

All locations will need to use careful judgment is deciding when cataloging at the analytic level of detail is appropriate and cost-effective.