In 1965, Betty wrote the following report about the time paper card and tape processing had reached a peak of usage for all kinds’ of information storage, retrieval, manipulation, calculation, manufacturing, and printing. Cards have almost vanished now and become a very strange technology foreign to most young adults. It seems weird in retrospect even to this old webmaster and physical chemist, who was deeply submerged for over a decade, in cards and paper tapes when cards and paper tapes were at their pinnacle of their use by advanced technology, science, commerce, telegraphy, accounting, manufacturing, and many other fields. For more information on card and paper tape technologies see HISTORY OF CARDS AND PAPER TAPE IN ADVANCED TECHNOLOGY. If data cards and paper tapes are unfamiliar to you, you will benefit from reading about them.
Most of the 20th century, where and when computers were not available, some libraries, research facilities, and businesses used small edge-notched cards to store and search for information. Betty describes their use for special collections.
I. Types of manually sorted systems
Librarians have been using card
catalogs for many years now to identify material in the library for any specific subject.
However, despite their almost universal use in libraries, card catalogs with their general cataloging codes and universal subject
headings have been found inadequate for the total tasks of identifying the material the library has.
The outlooks of the cataloger and the reference librarian are different; the latter is concerned with the subject content
of the items in the collection and all their separate parts, while the cataloger tends to segregate each volume as a separate entity,
using its title page information as a means of identification, and general subject headings as description of its overall content.1
Since the usual catalog has followed the preferences of the cataloger, the reference librarian, who is responsible for retrieving material for its subject content, has developed special tools including the special file, to supplement the card catalog.2
2. Ibid,. p. 533.
Other special files have resulted from the segregation of special materials into separate collections which implies that their content demands more care and subject analysis than the general card catalog and the general staff can give.3
3. Ibid., p. 529.
In setting up and maintaining collections of special material and theirs files a number of considerations are involved. First, what is the major purpose of the collection and who is likely to use it? What information will the user want to know about the material through the file? What information will he want from the material itself? (e.g. What information or subjects are important enough to be indicated) And, finally, what is the most convenient way to convey this information?4
4. Collison, Robert L. The Treatment of Special Material in Libraries, 2d ed. London, Aslib, 1957, page 5.
In view of these considerations I would like to describe in this paper some of the possible contributions of manually operated information retrieval systems for special files of the type that might be found in a college library. Before these applications can be discussed the systems must be described at least briefly.
These retrieval systems have two fundamental approaches to the problem of linking subject content with the various items in the collection. One of these approaches uses the subject content as the basic element, the other uses the document. In the first, the subject approach, a card is made for each subject heading and the items in the collection that pertain to that topic are recorded on it. There are various methods of recording the information, but the principle remains the same; usually the documents or items are numbered and these numbers are recorded.
In this first approach one method, known as uniterm, involves writing the document numbers on the subject card, usually in columns so that numbers ending with the same digit are grouped. This is done to facilitate comparison; when material is requested, the subject cards describing the desired material are taken from the file and the document numbers on them are compared. Any number that appears on all of the cards obviously pertains to all of the facets requested. The number of subject cards that must be compared would depend on the question and on the headings used in the systems.
The peek-a-boo or batten cards are more sophisticated version of this same principle. Instead of writing document numbers on the subject cards and comparing them, the cards are punched in the appropriate columns and lines for the numbers when subject cards are matched to obtain documents, they are placed on top of each other above a lighted panel, and where the light shines through coincident holes, the same number has been recorded on the cards. In both systems, once the document numbers have been obtained, a shelf list file of documents must be consulted to get the description of each and its location. Both systems are somewhat limited by the number of documents that can be recorded on the term cards.
In the documents approach, a single card is made for each documents and its subject content is indicated by punches along the margins of the card. These cards have even rows of holes along all of the sides and the punched our cards can be sorted from the remainder by running a rod through the desired hole in the entire stack; the punched out cards fall from the rod while the remainder stay on the rod.
A primary feature of the punched card system is the number of different functions that can be performed with proper planning and punching. The simplest of these is the direct code, where each hole is assigned a meaning and the appropriate documents are punched. This makes sorting a simple matter of selecting the proper hole; however the number of categories that may be used is limited to the number of holes on the card.
More combinations can be obtained by using numerical or alphabetical designations. Numeric codes can be designated by assigning groups of four holes the numbers 1, 2, 4, and 7. Punching each of these digits alone will of course designate that digit, and the others can be indicated by addition, e.g. 1 and 2 provides 3, 4 and 1 makes 5, etc up to 9, thereafter the 1 in the next group of holes designating the tens would be punched. As long as a group of four holes, a field, is allowed for each digit, the system can be extended far beyond the sorting capacity of the operator. This system can be used to arrange the cards numerically; or, with some modification, to select specific number designations which could be used for subject categories. Similar methods can be used for alphabetical designations.
In addition to these coding devices, many combinations of direct codes for large categories with numerical or alphabetic designations as subdivisions can be used to obtain more categories with fewer sorting operations. The extreme flexibility of this system makes it necessary to determine the information which will be required and the most efficient method of indicating the material on the card.
This system, the edge notch, produces different results from the uniterm and peek-a-boo systems; instead of having a list of document numbers to look up, the center space of the card provides room for the bibliographic citation, an abstract of the material, perhaps a microfilmed insert of the document, or a pasted-on clipping. These cards can also be used as a shelf list, since the file represents the entire collection. In large files, the material can be separated into relatively exclusive subject areas to simplify sorting, but these divisions must be carefully planned.
One of the advantages of these manual systems is their low initial and maintenance costs. None of them requires much equipment beyond the investment in cards; the uniterm system requires nothing additional except a place to store the cards; the edge notch system demands in addition only the sorting needles or rods, a munch, and a flat space for sorting the cards. The peek-a-boo system does require a lighted surface for comparing the term cards and a device for linging up and punching the document holes accurately. All of the systems admit additional material fairly readily, and all of them can be used on a small scale. The edge notch system is particularly well adapted for revision and correction; unpunched cards can be stapled to the cards to change the punch pattern, and cards for discarded items can simply be destroyed.
For reference purposes, where information and subject retrieval are most important, these systems have a special value. Instead of setting up rigid items of subject entry which must be arranged alphabetically or in a set classified order as traditional catalogs, these systems have a greater degree of flexibility known as concept coordination. Because subject relationships are determined only when the terms are selected for sorting or for searching for document numbers, a great many more combinations are possible than in a more rigidly predetermined system.5
5. Casey, Robert S., James W. Perry, Madeline M. Berry, and Allen Kent. Punched Cards, their applications to science and industry. 2d ed. New York, Reinhold, 1958, p.391-2
Also, because subjects can be limited and defined by comparison or integration with other subjects in the file, the headings do not need to be as specific as they would in a traditional file.6
6. Ibid., p.22.
An example of this might be a request for French portrait paintings of the eighteenth century. With the proper file, the cards or codes for paintings, French art, portraits and eighteenth century could be used to determine the appropriate material. Few card catalogs would have subject designations specific enough for this request. With a separate card required for each heading and the exploding bulk of card catalogs, the number of subject headings per book is usually strictly limited and few catalogs would have subject headings to represent all of the material contained in the book. This same principle applies to other forms of library material in addition to books.
The special materials and collections which are likely to be used in colleges have several characteristics that would make them particularly good candidates for manually sorted files of the types described here. All of them have many different approaches to the material, author, subject, source of information, and date. These are just a few of the possibilities. Many have reference uses and subject combinations that a traditional catalog cannot even begin to include. On the other hand, some of these collections are of ephemeral material which, because of its limited lifetime, is not worth the expense of complete cataloging. In this last category are many pamphlets and some books which have subject material which is valuable at the time but which will soon be superseded. Because of the detail and expense of cataloging, its cost often outweighs the value of the information, and the item is put in a vertical file under a subject, although it may treat several subjects.
An information retrieval file of such "short term" materials could be extremely useful in locating them through all the possible approaches and could even include a category to insure that they are brought to attention for discard after a period of time. However most of the special materials and special collections discussed here are considered because their many subject approaches are particularly suited for the information retrieval techniques described.
Early collections of newspapers and pamphlets are often of historical as well as subject interest, and systems such as these provide the opportunity for including entries such as date and place of publication as well as the more traditional author, title, and subject entries.7
7. Wynne, Bibliographic files for research in the Yale University Libraries, Bibliographic Society of America. Papers, 49, 1955, page 205.
On the other hand, files of clippings are often necessary to obtain up-to-date material that is available in no other form. The files could include the items themselves or could be indexes to the material. In either case, the amount of material on facets of the most demanded subjects requires elaborate, detailed subject headings in order to identify material and complete cross-references in order to find related items. As with most recent material, the headings and terms are constantly changing and must be revised. Also, much of the material is soon outdated or replaced and should be discarded. One of these information retrieval systems could answer both needs, detailed approaches through subject and other entries and a flexibility enabling discarding. The edge notch system is perhaps especially useful because the center area of the card would allow the actual clipping to be mounted, saving any further steps in retrieving the information.8
8. McGraw, Marginal punched cards in college and research libraries, p.163
In a very small collection, the problem of identifying maps is small because the number of items can easily be handled. However, as the collection grows the problem is complicated by the different formats that are included: globed, roller maps, separator sheets often printed on both sides, and atlases. Most people approach maps from a subject point of view, requesting first of all a geographic area and then a specific type of information: physical features, political boundaries, cities, vegetation, etc. The handling of even this amount of the material would be difficult in a card catalog, but the usefulness of a map is also determined by its date, its scaled, its language, its projection, and the amount of additional material included along its edges. Early maps are also requested by cartographer, publishers, series, etc. All of this material clearly indicates that with a collection of any size of flexible form of indexing must be found, and these retrieval systems seem perfectly adaptable. It might not be feasible to list all of the characteristics given here, but certainly the geographic area, the type of map, the date, and whatever other items that would be most useful could be included.
Illustrative materials in special files fall into two types those used primarily to illustrate all types of objects and those which reproduce art work. In the first category, illustrations of people, material objects, buildings, symbols, etc., there are too many items, they are needed too soon, and they are too ephemeral for full cataloging. In a subject file where the only index is the subject heading on the folder, the number of pictures with more than one item illustrated, causes an extreme problem of cross-referencing.
If, however, an information retrieval system were applied, these multiple entries would present no problem. A subject file such as a peek-a-boo would probably be less effective because it would require a serial number arrangement of the pictures. However, an edge notch system could be used as an index to multiple-entry items, using the same code to file the pictures by subject.
For reproductions of art work the material is less ephemeral than with subject oriented pictures; but the requirements of multiple approaches, such as the example of French portrait paintings, would make an information retrieval system useful. Some of these approaches would be period, media, subject, artist, nationality, style, etc. The same approach could be used for collections of slides used in history and art history courses. An edge notch system could also have a small reproduction of the work on the card along with a description of it.
Although films do not play as large a part in the college library as they do in a school, the special demands of the medium make information retrieval techniques extremely useful, Some of the approaches to the material include:
1. Subject—of both fictional and non-fictional films, including biographical material in
documentaries and newsreels.
9. Collison, op. cit., p.88-9
There are other possible entries, but these illustrate the many approaches which would strain the traditional cataloging system and which might be handled by an information system.
Musical scores represent a similar problem in the number of approaches, At Yale University it was estimated that 70% of the musical materials raised unique problems that were not solved by conventional cataloging.10
10. Shepard, "Problems of music library administration." Music Library Association, Notes, 11 June 1954, p. 360.
Entries for the composer, title and opus number would enable the location of a particular work or composer’s work, but it would not aid the music instructor trying to locate pieces for certain groups of instruments or with instrumental solos or appropriate for certain occasions. Nationality, folk, and ethnic materials require a subject approach, as do certain well known pieces that have a subject association. In addition arrangers, editors, authors of the text, and translators might be the only way of retrieving material.
Musical recordings have all the problems of musical scores with the additional problem of their format. Each record cannot be treated as a separate item like a book or score because quite often the two sides are completely unrelated. Because there is no easy way of browsing -- of discovering the content of the material -- without elaborate listening apparatus, the catalog must be much more complete. All of the entries required for scores would need to be supplemented by information about: the performance, the orchestra, the instrumentalists, the vocalists, and the conductor.
In addition to these approaches to the material certain descriptive material, the number of records, their size, and playing speed, is necessary. For non-music recordings the primary interest is either subject or performed for speeches and literary readings. For these the speaker, the literary work, and its author must be included as well as the subject. The same descriptive material would have to be given.
This has been a discussion of possible applications of manual information retrieval systems for special files in college libraries. It has been based largely on the general information needs of these special files and on some of the basic possibilities of the information systems. Any actual application of one of these systems would require further study of the particular material and information demands of the file and of the technical considerations involved in setting up subject categories and in coding them. From this study I have concluded that there is a unique contribution that these information systems can make to special files that cannot be obtained by any but the most complex traditional card catalog, and then only with a loss of efficiency because of the number of cards and the number of filing systems that would have to be developed.