Most Museums keep a list of citations for the objects in their collections. This list, usually but not necessarily in bibliographical form, serves multiple functions: It collects evidence that documents an object’s cultural and/or aesthetic significance; it aids scholars and researchers interested in objects; and it records the activities of the museum’s public relations and educational arms, while serving as a repository and source of information for them. The collection and maintenance of this resource is central to the societal and intellectual obligations of most museums. Indeed, the mandate to collect bibliographical, usage and citation information is often implied by a museum’s statement of purpose, its so-called "Mission Statement." Declarations mandating museums to "foster understanding," to "promote scholarly study" and to "provide public interpretation" of the collection are typical, and oblige museums to document the use made of the collection, to collect analyses, information and publicity published about it.

The accumulation of citations is one part of an information gathering process that incorporates functions of several separate museum divisions. The collection of citation data is most closely tied to the activities of the Registration, Exhibitions, and Curatorial divisions. This activity is also linked, however, to the operations of the Library, Education, Public Relations and Rights and Reproductions departments, and sometimes even to the Museum Store. [2]

In many museums, especially when intellectual and registration personnel is overburdened or their departments are understaffed, the time-consuming manual procedures customarily used to collect and display citation and bibliographic data, have, at best, been applied haphazardly and insufficiently — lacking structure, consistency and efficacy. The causes for this state of inefficiency owe directly to the manual and compartmentalized nature of the activities of the departments that use and collect this kind of data. Departments of Public Relations, Curatorial Activities, Object Registration, Rights and Reproduction, Exhibitions, Education and Library all maintain information that would better be used if it were available to everyone. To be useful, then, each division must either posses its own copy of (or have direct access to) the information produced by others. In manual systems, more often than not, the ability to access files maintained by someone else is hampered by numerous factors. Physical displacement of information, and the true or false perception of proprietary ownership of data are two important inhibiting factors. The inaccessibility of this data in many cases may simply be due to a lack of procedures that will ensure that the information will be collected consistently and systematically.

An automated module integrated into a museum’s collection management and documentation software could help ease the process of collecting, structuring, disseminating, and otherwise making this valuable and useful data available.

Opportunistic incorporation of routine activities of several museum departments can be merged and orchestrated by automation into contributing components of a larger whole. More cohesively integrated, this information will be more readily available and will tend to be used to greater advantage.

Citation and usage data may manifest itself in several forms:

Citations lead the researcher from the small world of the object to the macrocosm of its place in culture. Citations may include references to articles about objects, to books discussing them, to exhibit catalogues, to text books and other books in which objects appear as illustrations. They may also refer to posters, to postcards and to calendars illustrating objects, to newspaper articles and clippings, wire service transcripts, press releases, gallery brochures and other promotional items, documentation, provenance data, plus a wealth of other scholarly and commercial products. Other types of citations may include references to the appearance of objects or their reproduction in films, television shows, photographs, or in various media and electronic formats, including CD-ROM [and potentially ephemeral pages on the World Wide Web]. Some of the above citation types clearly are texts generated by the museum and hopefully held within the museum’s own documentation system, but others will certainly be available outside of the museum walls.

References such as these are traditionally recorded in a paper file system by appending the appropriate citation information to the end of an ever growing list held with the object record, and by accumulating the various articles and miscellaneous texts in a folder or bin that has been filed by accession number or some other simple scheme. Label copy documenting the local and lending display history of museum objects, is also frequently thrown in with the same lot. Sometimes data is collected by event, as when the divisions of Exhibitions or Public Relations maintain clipping or publicity files.

It is easy to see that access to the collected data is hindered by the very process of collecting it. The labor intensive method of gathering the citation data fosters ever increasing inaccessibility to and isolation from the information. This data is difficult to retrieve. No wonder so many staff members see this task as a tedious unrewarding endeavor. They know that the concept is important, but they see that the entire process is self-defeating. It is hard work, and only a few people can use it. In manual citation systems, procedures enacted to increase access (for instance, cross-referencing) tend to produce an ever more thickly woven web of maintenance requirements and impose increasingly intricate and parochial access techniques upon users.

Simple tasks often impose insurmountable procedural issues. When a scholar requests bibliography on a museum item, a curator or a secretary using traditional manual files must consult the list of references, decide which are appropriate and then manually copy those into a letter. To save time, the entire card is occasionally photocopied and mailed — even though museum policy may forbid or discourage this practice. Inefficient manual systems make it impossible or prohibitively laborious to search for citations that refer to groups of items or which may not be discovered by the typical path of investigation — one that begins with the object record as the entry point. Often, entry of citation data is retarded by the press of other obligations and by current business. In these cases it is generally impossible to fill such requests and too embarrassing to provide only old information. How many curators stock-pile requests for information about objects under their purview?

Museums using docents and volunteers to give gallery tours and lectures on its holdings need a process that researchers can use easily to explore the literature on their objects and themes. Inexperienced volunteers and undergraduate students will often depend on the services of the museum librarian to undertake the preliminary phase of their research, imposing unnecessary demands on his time. Museums without library staff place even greater pressures on volunteers and docents. An automated citation file that can be selected by object, by exhibit, by culture, and by any number of other significant classifications such as may have been used to classify objects, and that can optionally limit itself to resources on hand and to resources of a particular type would considerably ease the investigative processes for all.

Admittedly, development of the citation index is one of the more arduous and demanding chores of collection information maintenance — yet among scholarly museum professionals, it is one sub-system often requested and one that promises to expedite one of the more time-consuming chores of curatorial responsibility. [3] To be successful, citation maintenance demands the concerted effort of museum professionals from the scholarly, educational and public relations divisions of the museum. Separately, they must routinely contribute their products and their discoveries to the curatorial or records division. Curators must also acknowledge the importance of the development of the citation archive by informing the collection information manager of their products and discoveries. Unfortunately, bibliographies developed in the process of studying a museum object often remain in the private possession of the researcher and are never properly linked to the object record — electronically or otherwise. These, too, should be collected whenever possible.

Whereas some entries into the citation index derive from serendipitous discoveries by museum personnel, many other citations may be added to object data through systematic procedures that analyze the museum’s own publications, and from requests received by the museum for permission to use or publish its works. Today citation indices are no longer limited to references to books and articles, but may include non-print media such as films, commercial slides, and CD-ROMS, among other means of communication and transmittal of information. Even commercial products such as facsimile reproduction and adaptation of three dimensional objects, or the manufacture of jewelry based on museum objects need not be excluded from this record.

The museum has an excellent source with which to gather this kind of information. The data enters the system through the process by which potential users request permission to publish or employ images of museum objects. The museum’s Rights and Reproduction division administers these requests, grants permissions, establishes conditions, and may even oversee contractual arrangements with vendors and purveyors of museum merchandise. It collects documentation of the authorized usage, invoices users and collects or waives fees and royalties for use.

At the same time, Rights and Reproductions (R&R) works with the photographer and with the museum’s collection of negatives and reproductive images. The R&R department is often given the responsibility to care for other forms of media. To accomplish their ends R&R must have access to files listing available media suitable for reproduction or playback. It must be able to consult the registration records to determine any unusual stipulations and conditions governing the right to reproduce or use objects, and must have access to curatorial opinion regarding any special circumstance that may apply. It may also wish to survey the museum’s past history with a specific publisher or client, review outstanding debts, and monitor (or even control) the frequency of use of certain objects.

One of the by-products of the business of Rights administration is information suitable for recording in a citation index. Some museums have established manual procedures intended to capture and improve citation data produced by the Rights and Reproductions operations. [4]

At first appearance the operations of Rights and Reproductions seems to resemble a traditional supply and rental business. Transactional data structures uniting clients, events (sales/rentals), inventory and supply, and billing can be assembled for this enterprise in a traditional manner. The request for supplies and inventory may proceed from custom orders or from stock. Items rented have return dates; outgoing and incoming condition must be tracked. Fees received may be posted to accounts within or without the institution. On this level there is very little difference between administering a Rights and Reproductions department and running a video rental store. However, when more deeply scrutinized in the context of the museum’s mandate, transactional event such as these embody issues beyond the sale or rental of tangible matter and the barter of legal rights. They document the business of scholarship, they build the resources of learning and they assemble the building-blocks needed erect a Wissenschaft of culture.

Information that is merely transactional and ephemeral for a business: the record of who bought what, is a significant product for the museum. Businesses keep this data as long as the client and market exists. Museums must keep it for an eternity. Capturing this information, therefore, though a byproduct of photo sales, is vital to the museum’s mission. Further, it may be just as significant to the development arm of the museum as it is to the department that oversees object documentation. [5]

How may data become candidates for inclusion in the citation file? They may enter under any one of three situations:

Both Historical (1) and Contemporary (2) citations can yield definitive (though sometimes partial or informal) references. They are definitive because they already exist; they can be partial or informal because they may not have been collected systematically or rigorously. Future Citations (3), however, cannot be collected definitively. They cannot be considered set until the publication or usage actually exists; until then anything and everything can change.

The citation database is linked informally to potential sources among those staff members who feel they should submit their findings to it, and formally to those activities of the Registration Department that collect data on current and past citations (Exhibit and Catalogue Research) and formally to the Rights and Reproductions Division that collects potential future citations.

As collected, citations therefore may exhibit different levels of completeness, varying from the most summarily cited items to those that might satisfy the cataloguing criteria of research librarians.

Although the citation index manifests itself as a list, as a list it is a gateway to stores of information held within and without the museum. The citation list, is, itself, a set of pointers. Each citation may or may not point to an object held by the museum. Citations may point to nearly anything: book, article, letter, document, sound recording or transcript of same, label, catalogue, brochure, calendar, film, journal, slide, advertisement, poster, facsimile or trinket ("tchotchke"), and even (though rarely) to another object that may be within or without the museum. These physical items are sometimes filed or stored in files that are linked to the accessioned object record, or, as with books, catalogues and periodicals, stored in their own context.

Some museums may elect to save every known reference to their objects, including the advertisements, bookmarks, pins and postcards, while others may choose to save only those they consider the most significant or useful items. Each museum, accordingly, must establish its own policy toward the collection of citation data and usage objects. In many cases only some items will be considered appropriate to hold, while others may be marked for disposal. Items are disposed of in different ways: An elementary school textbook may be donated to a local school. Other items may be thrown out. In other cases, museums may find that items relating to their objects are, in themselves, worthy of being collected, and will even want to accession objects related to their own. Some items will be considered appropriate targets for researchers to find when doing scholarly work, and others will not.

Traditionally, citation files favored the needs of scholars and curators, but with the opening of opportunities to market museum images, the need to collect a wider variety of examples is becoming ever more apparent. As our society finds significance in the products of popular culture, the objects made to bridge the gap between so-called "high" and "low" cultures may assume a collectible significance. Originally museums collected ancient replicas of Athena Parthenos because they provided clues to the appearance of the lost original. Now, I suspect, they are also valued as documents illustrating a lively trade in ancient tourist souvenirs and mementos. Have museum’s begun to collect Statue of Liberty ashtrays and thermometers yet?

When viewed from the perspective of the accessioned object, each object must potentially lead to a list of citations and to a list of other kinds of references to itself. Each object has its own history of usage and scholarship. Ideally, it should be possible to produce a "History of the Object," a compilation of all actions and activities that concern it. When seen from the vantage point of the citation, each individual citation must potentially lead to one or more objects, often with specific qualifications that indicate how each object is linked to each reference item. In computer terms, this means that a "many-to-many" relationship must connect items to citations and citations to items. Certainly, just as some objects will have no citation, some "citation" records will not immediately (or ever) be linked to objects; they may be linked to nothing at all or just linked to events or one kind or another. [7]

From the above sketch, it would seem clear that the electronic management of citation data (and by "citation data" is also meant "usage data") — linked to standard object management software — can help create an efficient and self-documenting system that harvests data from several essential museum enterprises:

Click thumbnail, above, to see image in current browser.
Click, below, to open new browser for image.
 

Accumulating citation data on accession cards was appropriate in a scholarly climate that focused on the object-maker relationship. In today’s exhibit-minded and process-oriented arena, "thematicized" thought tends to be ethnically-, culturally-, regionally- and politically-focused. Museums are now charged with the responsibility of collecting and presenting the diverse forms and kaleidoscopic perspectives of our cultural and aesthetic past and present (our heritage and that of our descendants). In this ambience, the old methods are too limited to function well, if at all.

In medium to large museums, the architecture of paper record management is mono-dimensional. When internal and external demand for citation services is large, these tasks can only be conducted efficiently through a data architecture built upon electronic foundations.

The following discussion outlines a strategy and suggests data architectures that may help coordinate many of the functions and meet the needs cited above. The features discussed are based upon hypothetical requirements of a museum for whom the development of and need to access bibliography and object citation data is important, that administers a lively business in granting rights to publish data and images from the collection, and that maintains a photographer on staff to help fill the needs of both the publishing public, and the requirements of museum departments — Curatorial, Conservation, and Public Relations being the most demanding. In this hypothetical museum the activity record (the history of what is going on) is considered important and is valued as a resource, useful both to the museum and to the community it serves and to which it answers. [8]

To make the most use of collected citation data, each item should be entered as a record in an independent citation file, and not held as a list within a given object record. Each such record should be fielded in such a way as to make it compatible with accepted library cataloging practice for various media, and should generally conform to the MARC format or the best practical approximation of MARC that the current system can allow. The best solution would be to make the museum’s automated library file (if it has one) serve as the cataloging envelope of the on-line citation engine. (See diagram: Citation and Usage File.) [9] In this way searching through the library catalogue's OPAC will lead researchers to museum holdings.

Although citations are usually considered "different" than objects, under some circumstances, as when reproductions, commercial and other objects are based on museum holdings, they may be catalogued as accessioned objects. In these cases the citation index should be able to point to any kind of object record, be it library object or accessioned object. An intermediary "citation link" file should connect to each object record and point to the appropriate kind of record(s) and/or file(s) containing the citation information as well as to the record itself. [10] (See diagram: Citation Link File.)

Because the citation file serves double duty as a finding tool and as documentation of interest in the collection, it is important to integrate museum resources (the library) with the citation database whenever feasible.

Typically, when they exist at all, automated museum library catalogues are independent from object collection catalogues and function by themselves. If the museum library has such an independent automated catalogue, procedures could be defined to insure that there will be suitable cross documentation of their respective contents. When new items that pertain to specific accessioned objects are entered into this catalogue, the new library records should be directed to point to the objects they document. When citation records are added to the library catalogue first, suitable complementary records should be able to be batch loaded into that portion of object database devoted to citing object-caliber and related citation materials.

If, on the other hand, the library catalogue is electronically linked to the general object database, the citation information should be recorded with the general holdings data. This file then will doubly serve as the library holdings and the citation file. It will be necessary, however, in this case, to distinguish between items held and items only cited. If the library database contains a disposition field to indicate object status: on loan, reference only, etc., this field may be used to indicate that the implanted citation record does not signify a physical holding in the library and/or the museum collection. [11]

If the library system is manual, to avoid duplicate cataloguing, the electronic citation file within the object database should be made capable of producing a standard set of catalogue cards. The citation catalogue should therefore contain at least a minimum set of fields compatible with the library cataloguing system and with other standard cataloguing systems, such as that of the Library of Congress, etc. If the museum’s library system is manual, the citation index should be compatible with standard library cataloguing practices or otherwise be suitable for expansion into a full-fledged library cataloguing system when so desired. Accordingly, there should be fields that the museum can use to designate the call number and/or placement location of the physical item that backs up the citation — when available.

Although the citation file should include fields appropriate to general library cataloguing of printed and non-print media, it must permit entry of full, abbreviated, and even partial references; full for all items, but especially for those held in the library, and abbreviated or partial for those items entered for documentation purposes and collected opportunistically from formal and informal sources. Partial, temporary and incomplete citation records should be flagged as in need of revision or confirmation so that they eventually can be corrected, even though it is possible or likely that they will remain incomplete forever.

To establish consistency in citation syntax and form, entry screens should provide an optional prompted entry format for historical and current additions to the citation file. Such entry screens should contain very few required fields. However, authority files and hierarchical lexicons or thesauri for the names of authors, publishers and manufacturers, journal names, and locations of publication should support data entry. Authority files should be able to be summoned during data entry and during data queries. Adherence to authority rules, however, should be optional. Authority violations (including new terms) should be so marked so they can be selected and queried for review. The citation data exhibited should be a subset of the full cataloguers entry screen. Brief citation entry screens should be available as well as full citation entry screens. Reports developed out of the citation database should be made to include data fields that have been filled and (for users) ignore those that have not, unless, of course, the reports are produced to support data maintenance operations.

As mentioned above, each citation item should be linkable to one or more museum objects through a file whose records establish the nature of the link. Fields in this entity link file (if it may be called that) may include date of entry, occasion for entry (Exhibit, research, publication, etc.), and a link to the name of the person (or corporation in the case of photo orders from publishers) who requested the entry (authority controlled). This entity link record may indicate whether the entry is provisional, and whether the published item is to be received and when it is expected to arrive. The system should be able to create link file records quickly and to replicate their contents easily, changing only the citation reference number, the object reference number (accession number), and/or the media number (e.g. negative number), if appropriate. The data in a link record may be templated so that users can select sets of values to be input at the same time. Creation of citation records and the entity relation to the object should be experienced as a single activity. Guards should be in place to prevent duplicate citation records, though duplicate links to objects should be permitted. This record, or one attached to it, can be used to administer the assignment and administration of rights to reproduce items. This file is used to establish links between objects and citations, but it also explains the circumstances of the connection to the activities of Rights and Reproduction, so it must also be linked to the invoicing module used to administer the collection of fees for distribution of rights and licenses. See further, page ff.

The success or failure of any citations database will ultimately depend upon how well it is integrated into the working methods of its users. Some users may approach the database with the express purpose of developing a bibliography, while others may wish to obtain citations as a consequence of their participation in some other activity. When a request for an object or theme bibliography is received, the designated museum staff member given that responsibility will want to produce a report directly from the citations file. When a curator is researching museum objects, entering or correcting data, she may wish (almost serendipitously) to obtain a bibliographic citation list.

Consequently, users should be able to produce lists of citations directly from the object record, any activity record, and independently through citation reports. When queried in either way, the hit list of citations produced should be able to be limited by type. The user should be able to select or limit the kind of citation he needs. He or she may be given an option of ticking off specific items from a list that identifies

• the motive or occasion for entry (exhibit, research, iconography, conservation),
• the date or date range needed (nothing before 1932, for instance),
• the type or types of entry required (books, journals, pamphlet, postcard, manufactured items),
• the type or quality of reference (scholarly discussion, mentioned in passing, illustrated or illustration only, commercial product, etc.),
• the location(s) of the physical item (offprint in curatorial index, in library), and, of course,
• the standard fields such as Author, Publisher, Journal, etc.

To ease the seeming complexity of working with some of the above categories, several of the field types and their conventional values, notably Type of Reference, Type of Entry and Occasion for Entry, might be prompted by the program itself, so that examples are supplied during data entry and during queries. These examples might be entered as values in a single multi-valued field — one that swallows the distinction between Type of Reference, Type of Entry, etc., parsing these out to the appropriate fields as needed.

Users should be offered the option of displaying citation lists one record at a time or in tabular form, the default display order should be reverse chronological order, with options available to display items chronologically or to list them alphabetically by author. Journal names and common references (i.e. major monographs, museum catalogues) should be authority file controlled and should show onscreen in accepted abbreviated form with the option of displaying (or reporting) the full name and authority file data on demand. [12]

If a citation report is produced while viewing an object record, the user should be able to print his selection and then return to the place (object record or process) he was before the bibliographic request was made. In other words, he should be able to branch to the report function and then return to the query mode at the location previously set. In this way, if the user was browsing a set of objects that appeared as a result of a query, that set would still be active after obtaining the citation report. The user may be given an option of allowing the report writer to execute its task invisibly — the user will always be shown the object record while the report-writer operates in the background.

One feature useful to anyone accumulating a bibliography through the citation file would be a "line item veto" over what will print or be saved to disk. If a user could de-select (or select) specific items (or all items at once) from a list, the task of bibliography creation would be considerably eased. [13]

For most work, the single-most useful reportage function will be a standard citation report. As noted above such reports may be produced from an object or from an event, as when one is asked to provide the bibliography for an exhibit. Users should also be able to issue a command to print a report of the citations attached to the results their current query. In this case the report would be able to produce references for more than a single object. A customizable default report format might first list objects in accession (or other) order, and under each object citations in reverse chronological order. When a query has prompted a citation report, the query producing the report should be recorded in the report header or in some other conspicuous location so that readers can understand the context of the bibliography they are viewing. Printed reports should be able to be sorted and selected by the usual fields, including by object file fields: "Show me all citations to objects in the collection that were donated by museum trustee Donatus du Cheval." "What did the museum publish on its new accessions between 1939 and 1945?" Users should be able to limit the report to scholarly references only or limit it otherwise by other criteria: "List the publishers who requested rights to reproduce painting X within the last year?" "For which of our impressionist paintings have we not published a postcard in the last five years?"

Scholarly databases are different than management information systems. Data placed in object management systems tend to be about current events, and consequently are certifiable by the inputter. Scholarly systems, on the other hand tend to record opinions and assertions, and must consequently employ methods that will document the sources of the information entered.

For scholarly purposes, then, the citation system will assume the central role of authentication and documentation. Any scholarly or curatorial assertion should be linkable through it to its source. If such connection is necessary, it will also be required to identify which specific part of the general citation reference applies to the "footnoted" item. This means that the link between data and citation has to occur in two parts, and the link itself must be through an intermediary file or some other similar structure. Because any information recorded in the object record that conforms to the general requirements of assertions and opinions may be multi-valued — meaning that there may be more than one such assertion in a record — the data structure must be made to support multiple links to citation records from the same location in the object record. [In the data structure used for this paper, the complexity of links from object to citation are not indicated. Instead, refer to the attribution entities created for the Academic Image Cooperative prototype. See next paragraph.]

The data structure needed to maintain appropriate varieties of linkages may be fairly complex. Any single data value of a set that may be multi-valued must be able to link to a variety of sources, including multiple times to the same source. Multiple values in the data set may also link to a single or to multiple sources. A single assertion (for instance, an attribution to a maker) might need be linked to more than one citation entry when that attribution can be found in more than one source. Therefore each such attributed value should be attachable to more than a single citation and specific citation location. As a corollary, it would seem obvious that in the object file each assertion must be potentially part of a multi-valued set or linkable to multiple intermediary file records. In the end, any attributed statement may have multiple assertions, and each assertion may have multiple citations. [14]

Methods appropriate to achieve the above mentioned structure will vary according to the database engine and the preferences of the programmer.

When records are created, the method of linking citations to assertions should permit the user to branch into the citation file from the object record. The citation file then should be able to be queried to find the correct record. On return from the citation file to the object file record, the correct linking fields (from the object record to the intermediary file and from the intermediary file to the citation file record number) should be automatically inserted into the appropriate record belonging to the link. Sometimes the citation record will prove to be non-extant; in these cases the user should be able to create the record on the fly, entering just enough data to establish its identity and to create the needed link. Any required proofing or additions should be able to be carried out later. This means that each citation record must carry a field that indicates the degree of completeness of the citation record. [15]

Scholarly papers, as the rule, are authored by a specified set of individuals, usually one, but sometimes more. In either case, their names are attached to their product as a publication. In contrast, information systems are always fluid, dynamic and evolving. Authorship is either carried on directly or by proxy for a curator in charge of a series of objects or exhibits. Each object record must also carry data regarding the present author(s); this means the proximate author who enters the data and the supervising author who created the data. Each such record should be suitably dated. Revisions should not replace this information, but add to it.

Obviously, with so many levels of data input being documented, unless extreme caution is taken in the design of data presentation, the complexities of access may overwhelm any database user. The typical user must be presented only with the latest version of object data, but be given the tools with which to do on-the-spot research into the history of data. It is crucial that object data screens be kept simple and uncluttered or that basic versions of object screens be made available to those who must administer the collection as object managers.

To avoid cluttering the object record visually, citation documentation should not appear in the object record itself, but the user should be given an opportunity to bring up a view or a window of the "footnote" if needed. Like a footnote, any attributed assertion should leave a marker in the object record to make its presence known. [16]

Every citation record linked to an object record, be it linked in a general, non-specific manner, or linked to a given assertion, should be included in any request for a citation report on the object and should appear in any automated query from the object record that is intended to produce a bibliography for that item.

Every object record should offer the potential for a link to the citation file. By executing a one to many request for citation records the list of linked citations should appear. This list must also include any citations made to the object record in support of assertions of attribution.

The citation record structure provides an excellent opportunity to record non-standard citation materials such as didactic labels, scholar comments, brochure texts, catalogue entries, correspondence, press releases and short newspaper articles. Providing a text field or image field (as appropriate) for these items will help unify access to such disparate resources. Most system designers will want to isolate the text document from the citation record by placing such texts in related files.

The administration of rights to publish and use museum objects offers an opportunity to capture potential citation data and to collect information about planned uses for museum objects and resources. At the same time it should be possible to collect data on people (primarily authors, publishers, scholars, historians and students, but sometimes donors, too) who exhibit interest in the collection. Indeed, for some collections, retrospective entry of old invoices and logs of photo orders may enable researches to investigate — in ways never before practical (or even imaginable) — the history of interest in objects museums own. [17]

Future citations (or, better put, potential citations) may be developed out of the process of filling orders for photographs or from receiving requests for permission to use images of objects from the collection. An invoicing module made to fill such orders and bill for them also can be used to capture citation references.

Requests for photographs and images come to the museum in many ways and forms. An order for a photograph may or may not be accompanied by a request to publish it. Photographs may be sold or provided as a courtesy to scholars or to newspapers. All photograph requests, even those from staff should be made to pass through the standard ordering process. In this sense, a curator who needs a photograph of a museum object is just another user. If necessary, the invoicing utility can charge the curator’s photo order against a stipulated allowance or designated budget. Even if curators’ photographic requests need not be tracked, having their requisitions appear in the same system as all other photographic requests has a good chance of adding a touch of order and sanity to the high-pressured work environment of museum staff photographers.

Typically, photo orders follow a simple procedure: An order is placed, permission requested, permission considered, conditions determined and permission granted. Some of the events do not occur in a specified order, and some of the steps in the procedure may take place verbally. For example, sometimes a purchase order is received, and then the photos are sent, while at other times the photos will be sent and the user billed. In any case payment is received, or waived, deposited or transferred to the accounting division of the museum for further handling. Rights and Reproductions may administer the notification of late payees or this function may be transferred to another museum division.

Prices and charges should be read off a tariff database in which services are coordinated with costs and into which new service/cost combinations can be entered as necessary. The same database can contain categories for discounts and for waiver of fees.

The photo (read: media) request form records the name and address information of the requester, identifies the project (which becomes the incipient citation entry), the date needed, and the object(s) needed. Added to this will be the negative (media) number(s) to be used. Lookups based on the object numbers provide access to stipulations regarding usage, label description from the object file group, and the credit line. Museums may wish to provide for the waiving of reproduction fees based on a variety of criteria, including professional courtesy to other museums or journalists, worthwhile scholarly products, and upon the likelihood that the use qualifies as a Fair Use under US Copyright law. Honoring Fair Use increases the chance that the scholar will use a museum-supplied photograph and not an inferior one taken from alternate potentially inferior sources. Offering courtesy photographs and the pro-active granting of Fair Use and distinct and potentially unrelated events. Optionally, an object report can be included (or sold) with the photo when sent. [18]

The information collected above should be distributed among several discrete files. For example, the requester will be added to an agents or name file and will be linked to the activity file (ordering photographs) by a field that specifies his role. The project information is recorded in the citations file. It will eventually be turned into a full-fledged citation. The photo data will ordinarily be taken from an archive file of negatives and media available (as opposed the the file of physical media). Stipulations, limitations and object data will come from an object file. The order will be placed in a transaction file that will administer the business end of the agreement that will query an inventory file and create a work order if it is necessary to create a product.

Cost will be determined according to a pre-built schedule (the tariff database cited above) that may reside in its own file. An invoice/contract will be produced as a report that takes data from all relevant sources. When payment is received or when the purchase order is registered, the order to send the item is given.

The photograph or transparency will then be taken from inventory (or commissioned) and sent to the user. Photographs and slides are typically sold. Large format transparencies are rented and must be returned. Transparency rentals will require a return date and inspection reports. Each order should contain the text of museum photographic use policy which should be written in such a manner that the purchaser understands that it carries the weight of a contract. A similar contract (perhaps the same) should accompany images custom ordered from the kiosk facility described above. Depending upon the history of the ordering institution, some orders will be billed and some will require pre-payment. Those orders that require pre-payment should be entered into the system as provisional.

Stipulated conditions applied to the use made of an image may derive from legal requirements accepted as a condition of acquisition, imposed by the owner of copyright, or from current museum policy regarding use of objects in general, or of one object in particular. An incomplete list of potential museum requirements follows:

The invoice will serve as a contract limiting the usage of the supplied image (media) by including the relevant stipulations. Two sets of stipulation statements will be appended. One statement will be derived from general museum policy and the other will come directly from the object record. It is better to have the museum policy statement print out on the invoice than to have it pre-printed on custom forms. If policy should change, only the text of the policy statement needs be altered.

It is advisable that institutional image use policies not attempt to abridge the First Amendment rights of authors. For instance, while a museum might wish its attribution to be reproduced with the object, it is not considered ethical to force all writers to adhere to a museum attribution. In the same spirit, museums should not award or withhold rights to reproduce museum objects based upon the expectation that the author will disagree with museum attributions.

If a scholar wishes to publish an altered attribution, the museum may wish to solicit that information in advance and record it in the citations file — providing a good opportunity to record the museum’s opinion regarding the attribution.

Each institution will charge and collect fees in its own way, but (as already noted) usually there will be a set schedule of fees for different kinds of services. The fee schedule may include the costs of original photography when needed. Even though the user pays for the original photography, most museums keep the negatives or original images produced (a custom that will have no meaning when objects are scanned digitally and digital copies are sold). Fees may also slide on a scale depending upon the kind of usage envisioned. Scholarly usage fees are frequently waived, so each invoice must be set up to waive standard fees, as every permission request form should provide opportunity for scholars to describe the nature of their project for the purpose of receiving preferential fee treatment. Commercial usage fees tend to be higher. Not all production projects divide clearly into scholarly or commercial; there are many shades between these two extremes. R&R staff should be trained to identify projects that may be worthwhile recipients of the benefit of waived fees.

The museum may wish to exercise an option to collect a deposit when transparencies are lent. Some work may be charged back to the consumer from a sub-contracted agent, so some fees will not conform to a prebuilt schedule. Photography fees may also be charged against museum funds that have been especially tagged — usually grant funds, but sometimes acquisition funds, so the transaction system must be able to assign costs to a fund accounting system. Some museums must pass on a royalty to holders of the copyright on the image and this data should be included in the object record in such a way so that it is automatically inserted in the invoice, calculated and posted to a "royalties due" account — probably best created within the fund accounting system. Copyright holders may include the artist or estate of the artist whose work is being reproduced, and/or the photographer whose image transmits the work. Royalties are usually distributed according to a previously agreed upon schedule which may differ for each copyright holder. Whenever possible, as a service to potential licensors, the museum database should list all individuals and agencies who hold rights to museum objects. This information should be provided as a service to those who wish to reproduce an image or object. Copyrights expire and eventually enter the Public Domain. Copyright royalty data should contain an expiration date past which no benefits will be allowed to accrue.

When a museum entertains a busy Rights and Reproductions operation, efficiency can be increased if the R&R department has access to a database of media on hand, and has access to mechanisms with which to commission new photography. This last leg of the automated system is by no means mandatory, but because the photographic resources of the museum will include many other kinds of images than those that reproduce its holdings, many departments may wish to treat this repository as an essential museum asset worthy of systematic cataloging and inventory control.

The database of object documentary media will cross reference accession numbers to an inventory of media available. Usually a code or some other indicator distinguishes each media type and format. Often a museum will have more than one negative or image of an object and these images

• may have been taken over time,
• may be in different condition and quality, and
• may be linked to an event, such as an exhibit or an activity of the conservation department.

In most cases it will be possible to designate one negative, probably, but not automatically the most recent one, as the standard negative to use to fill an order. Frequently, however, as for three-dimensional objects, there may be a number of views available and one or several will have to be selected. Other image variants may include object details, objects in situ, objects in states of completion, in various conditions as well as object documentation made during treatment and conservation. Scholars sometimes will request that a photograph taken during conservation be made available for publication. It is clear that each image (i.e. negative) must be identified by how it documents the object. [See Image Production Data Model for the Academic Image Cooperative.] Sometimes more than one phrase will be required to describe a photograph: Upper left corner in raking light. All but digital images and prints made from extant negatives need also to be classified by their condition. Negative and media repositories may be stored in different locations throughout the museum. Usually the conservation department will wish to maintain physical control over its own products. Photographs taken at the same time tend to be stored together in the order they were photographed or in the order in which they were originally filed. Clearly, any database that serves as a guide to the collection of media must include physical location data. [19]

The Rights and Reproduction department should have access to an image database (a database of information about images) that can be used to produce catalogues of documentary images and objects available for research and publication. This database of objects and media will usually encompass photographic media, slides, transparencies, photographs, negatives, x-rays, etc., but other types of reproduction may be recorded as well: casts and molds, sound recordings, video disks, electronic images, video tapes and movie film are all current means of documenting objects in collections. In the near future CD-ROM disks may be dedicated to documenting and presenting entire exhibits [and, at the date of this revision it might be wise to be able to document relevant pages on the World Wide Web]. Sound recordings may be taken of the sound of objects as well as of commentary by artists or scholars and critics. Taped museum and exhibition guides (including video tape) may also be listed and administered through the Rights and Reproductions system. [20]

If no negative, transparency or photograph exists to execute an order, the system must be able to generate a work-order for the in-house photographer or a purchase order for an outside source to produce one. The costs incurred from the independent commission should be invoiced to the client, either at cost or at a premium — according to museum policy. The programmer should develop a system that logs all requests for images as pending. The work-order will be produced once authority has been received to go ahead, in which case the pending status is removed and the work-order is activated and released. The orders left pending, i.e. the requests that never received authorization, should be kept and used to produce a list of items that eventually may be needed. Pending requests for images may be candidates for entry into a citation file if it is understood that not every proposed use comes to fruition. Citation reporting (discussed above) should be able to distinguish between real and potential citations.

When a photograph is to be supplied that can be taken from inventory, a standard inventory control system should decrease the number held by the number supplied. Each object should be assigned a number that indicates how many images should optimally be kept in inventory at any one time. For most objects this number will be always be zero, so when the inventory module is introduced, the collection as a whole can be defaulted to zero inventory and increased by happenstance for the museum’s most demanded works. Whenever an order is received to produce prints for an item that currently or traditionally has an inventory of zero, the order to the photographer might include making a few more prints than needed. These extras should be then placed in inventory. When inventory on any object falls below its designated restocking value a tickler report (run automatically at set intervals) should produce a list of negatives of the objects that need more prints. If a procedural control system has been implemented, these items can be posted to the photographer’s to-do list and given a relatively low priority — work to be done during slack times. The inventory control system should allow inventory stock of photos to be sent to a category designated as "trash" in order to allow for disposal of damaged or old photos or to make up for missing ones.

Museum photographers are usually charged with the registrarial tasks associated with maintaining the collections of negatives and media, and are responsible for assigning new negative or other identification numbers. Each institution has its own system (traditions) of handling negatives, and these negative numbers may differ in format according to the system, the event and/or the photographer who produced them. The number system should be able to record negative numbers conforming to the archives and to the form currently in use and be able to sort them correctly in lists. Because negative numbers often follow markings on the film, or are assigned after media has been developed, they should usually not be assigned in advance.

The process of building a photo archive customarily requires two sets of photo identification numbers:

• a photo log number, assigned by the photographer when the image is exposed, and
• a final number that may partly be derived from the media itself. [21]

When single exposure media is used, e.g. 4x5 or 8x10 negatives and transparencies, the photo log number may be the only number assigned, or the negatives might be renumbered when they are deemed suitable for use.

When 35mm or roll stock is used, one of several numbering systems may apply. Kodak 35mm bulk roll frames are numbered in sequence from 1 through 44, sometimes with half-frames labeled A and B. 35mm roll film may be cut into strips of four to six full-frame images and stored as individual strips or assembled into pages placed into ring binders. 2 ¼ inch negatives have fewer images per strip. When 35mm film is stored in books, a typical identification numbering system will include book number, sheet number, strip number and frame number. If strip numbers start anew at the top of each page the strips cannot be rearranged at a later date without having to be renumbered. If they are not uniquely identified, they may easily be replaced in incorrect locations. If they are numbered continuously, however, each image will always have a unique identification number composed of strip number and imprinted negative number. Slides made from internegatives should take the identification number of the negative. Other slides should have unique identifications. In neither case should they be assigned numbers based on the accession number of the object they reproduce, though having that number on the slide is always useful, and storing them in accession number order is convenient, though there is good argument to adopt a geographic/historic/stylistic arrangement as used in university slide libraries. If the slides are collected for sale, the former arrangement is more convenient; if they are to be used, the latter is better.

The photo-log kept by the photographer often contains technical data germane to the taking of the photograph. Some institutions may consider the exposure, lighting and film-media data recorded in this list metadata of sufficient importance to link to the media database, others may not. [Scanning metadata is also generally saved.] Understanding conservation photography often depends upon knowledge of how the photograph was taken, including the type of photograph or image, so this data will almost always be vital to the documentation of the image produced. Some digital image files can be automatically numbered by the date and time the files were first created. If this system reliably will produce unique file numbers, it may be used for image identification.

Historically, the assignment of identifying numbers in media collections is much more chaotic than the assignment of accession numbers for objects. Each change of photographer or each change of camera or format may initiate a new numbering sequence or scheme. Some museums may just assign negatives numbers that derive from the accession numbers of the objects photographed. This is a poor practice because it does not allow the available media to be uniquely identified. When rights to reproduce a work are sold, it is not the work that is being reproduced, but one of perhaps many images of that work. (Ceci n’est pas un pipe.)

Collections of negatives are often accompanied by contact sheets or other parallel documentation of the negative storage system. These books provide an easily accessible way to view many images very quickly, and at this writing [1994] are much more practical than viewing thumbnail electronic images.

Negatives produced systematically using common formats tend to be filed by format and size chronologically, cross indexed to object numbers. Some museums file negatives in accession number order and mix all media types together (except strip film). Not all negatives can be linked to objects, however, and not all negatives are captured by campaign or through the orderly process of purchase orders for photographs. The identification system for photograph negatives and other documentation systems should be flexible enough to accommodate themselves to any future numbering and filing scheme, including the identification methods of electronic image archives.

To many museums the photograph collection is itself an important historical archive. These negatives and/or the photographs therein contained must therefore be identified by fields that acknowledge their archival status and historical significance — just like accessioned objects. Relevant fields will certainly include the date of photography, the name of the photographer, a link (when possible) to the object photographed (or objects: one to many), the copyright owner and copyright date information. Media and format data help ease searches and are useful for filing tasks. (Because documents are often works of art, and works of art often documents, there is ample reason to describe them in the same data file.)

The citations engine that provides access to published objects should link back to the negative and transparency that was used, correspondingly it should be possible to pick any negative and find out its history of use and publication. Public Relation photographs; documentation of exhibits, gallery and case arrangements; building campaigns; old buildings; donors; staff; lecturers; museum-sponsored festivities and receptions; visits of dignitaries; building dedications and groundbreaking ceremonies and newsletter photographs should all be gathered together in the photo collection — available to all, and not sequestered into their associated departmental offices. Museums also collect photographs to document archaeological projects, sites, reconstructions and comparative materials from other collections. It is also important to distinguish the inventory of photographs for sale from the collection of documentation photographs, even though there may be considerable overlap.

From the above, it is obvious that many images will not be linked to accessioned objects at all, but to events or other situations. Names of people, their roles and the names of events (including exhibits) should link back to the photo documentation collection.

When a photograph does not document an accessioned object, it must be afforded a comparable opportunity to be identified: Name files used to identify the makers of objects and agents germane to the operation of the museum should also be used to identify the makers and human subjects of the photographic archive. In short, the descriptive fields appropriate to a photo (media) archive, and the way it connects to museum objects and events, very closely parallels the structure of an object file. It makes use of similar or identical authority lists, and must identify people, subjects, dates, places and events. In the long run, if a photo or media archive will have many items that are not represented by records in the object file, it would be best to establish a photo archive file to identify and classify images. The photo archive file and the object file are parallel entities and are individually linked to a media identification file that provides only the data needed to identify the type, origin and physical characteristics of the photo/media collection. The media identification file could be linked further back to a photo log file supplied by the photographer from his notes. (See above.)

When a photograph is requested for use in a publication, a temporary or provisional citation entry should be inserted into the citation file. The citation file must carry a field to indicate that the entry is provisional. Data for this entry may be taken from the original request correspondence or from a form sent to and returned by the requester. The information pertinent to the citation normally includes the provisional title, author, publisher and expected date of publication or whatever corresponds to these entities for the intended use. It could also include a field for media or object type. The record should be marked as provisional and dated with the date of request and the proposed date of use. The provisional citation record may be used in invoices and work-orders and other matters relevant to the request for rights to publish and/or use the image. Even though the data is provisional, the citation record number uniquely assigned to the request will be permanent. When the citation is eventually corrected or completed, any historical research into the record will produce the corrected entry. [22]

The administrative data regarding the collection of fees and the awarding of rights to publish must be kept out of the citation file. This data may be included in the file that links object and citation or, perhaps better, it may be gathered together in a Rights & Reproductions file appended to the entity link file cited above (page ).

Because the record of usage data input into any single Rights and Reproductions management database record may tend at times to be more varied than the standard author, title, publisher cluster of fields, such requests should be collected as associated multi-valued sets in which terms like "author" and "publisher" are values placed in a role field, the name of the publisher or author in (for want of a better name) a role player field that would be associated with a role player address/phone number series of fields. "Title" might be a value placed in an object name type field that would be associated with object name (title) and object type field (book, article, postcard). The values placed in these fields should default to the most commonly used set: author, title, publisher, etc., and the values themselves should be able to be chosen by set, this, to facilitate commonly recurring combinations of field values.

If the request to publish is for an accessioned object, and if the object record contains label text or a means of generating label text, this data should used for invoices and obtained by a lookup from the list of object identification numbers. Because the resulting use of this data is in a document of legal force, the current label or object identification should be copied over into the Rights and Reproduction transaction record—not looked up each time the record is accessed. This data consequently will be held fixed (in duplicate) as part of the invoice data. If this data is eventually purged, all that will remain are the links that unite object, citation, citation type, and negative. The purpose of this procedure is to establish a firm identification of the item licensed that will not change even if someone changes the data in the object record. But the object record, no matter what happens to it will always be linked to the citation. The fixed data lasts for the duration of the transaction record and can be reproduced without change for the benefit of the purchaser or licensee. In addition, the current label copy establishes the historical dimension of the transaction, so that the prevailing object identification remains attached to the invoice data.

The object input field on the invoice, therefore, should link to the object through the object accession number. The field set should repeatedly create new records so that multiple accession numbers and multiple media identification numbers can be allied into a single order. To facilitate finding the appropriate accession number, users should be able to branch to the object file, execute a query to find the appropriate accession number (or numbers) to be input automatically into the appropriate field in the form that will record the transaction and create the invoice.

Once the accession number is set, the system should automatically branch over to the database of media, executing an automatic query to pull up a list of all media associated with the given accession number. These should be presented sorted, first by media type, then by format and then by reverse date. It would be useful if only the main sort divisions would first appear, and these then could be exploded into their individual components when necessary. Specific media examples predetermined to be standard or recommended should be flagged. At this point, the database operator should be given opportunity to select one or many of those proposed items to be included in the order. The negative identification numbers (from the media file), the object label (from the object file), the name of the view chosen (from the media file) and perhaps the date and other identification data pertaining to the photograph should be included in the invoice. At this point general and specific stipulations should be merged into any printed contract or invoice. Many of the operations described in this and in related paragraphs may be carried out in the background. [If ported onto the web, only a subsection of the features described in these pages should be made available to the public. It is easier for the public (including the commercial public) to order images and rights to images, if they can query an on-line catalogue from which a rights request form is accessed. See the AMICO rights form at this url: http://search.amico.org.]

The field on the input screen used to register the accession number of the object requested should be one element of a set of fields, another component of which identifies the collection type (usually the object file). If a request is received for a photograph of an event, a non-accessioned object, for instance, the façade of the museum, or a visual document in the museum's photo archive, the R&R personnel should be able to enter "photo archive collection" (or some other designation) in the collection type field and then search for the corresponding photo archive number in the identification number component. Once the collection type is entered, a search for the appropriate number should cause the request form module to branch to the indicated database to accept a search and bring back the appropriate number and title information as detailed above. In short, the user should have the ability to search the object collection file to get to the media record, or search the media file or media archive file directly. In either case, if an accession number exists (if the search is for an accessioned object) this number must be inserted into the transaction record. For accessioned objects the negative number alone should be considered insufficient because of potential errors. Requiring an accession number and a media number, redundant information (when looking from media number to accession number) may be a useful safeguard.

When the request for a photograph is not for an accessioned object, the descriptive text used in the media archive file to identify the media must be inserted as the description of the photograph. The reader will notice that the above procedure differs from standard photo orders in that it views the order for an image of an object as an order for an image made from a collection of negatives. (See above.) The purchaser is provided with the actual negative number used to fill the order. (Purchasers of other media forms would be supplied with appropriate identifications: tape numbers, etc.).

As a rule, most museums that license the use of their images for specific, defined purposes (i.e. excluding site-licensing schemes) will request the return of documentation or at least a notice indicating that the item has been published or produced. Under certain circumstances, however, the request for proof may be waived. The transaction record should, therefore, include a field describing whether the museum expects to receive confirmation of publication, and what form that confirmation should take: an example, an off-print, a tear-sheet, a still from a film, etc. [23]

When confirmation is received, the provisional entry must be corrected and the flag field that indicates expectation should be modified. The Registrar, or, better, the Librarian should be assigned the task of certifying the citation, correcting it, and (if appropriate) passing the data into the library database, thereby making it permanent. The ultimate disposition status of the submitted documentation should be indicated: Should it be placed in the library collection, curatorial file, photo collection, disposed of or accessioned into the permanent collection. Logging submitted books and articles can be simplified by supplying publishers with a return label bearing the appropriate Rights and Reproduction record number.

Publishers do not always fulfill their obligations to supply examples of published materials. Therefore, periodically, perhaps by tickler report or on demand, a report should be run to catch provisional entries whose expected date of publication have be past by a definable increment. Follow-up mail-merge letters might be sent to the publisher of books when no example has been received in reasonable time. In addition, on-line resources such as OCLC or RLIN [or even amazon.com] may be checked for confirmation before follow-up letters are sent.

When the object citation index is built by the multi-edged process of retrospective entry and entry from current operations, several benefits accrue. Development of the citation index is no longer a task that must be maintained separately from photographic sales. Photographic sales, no longer viewed as an independent activity, will be perceived as necessarily linked to the resources of the collection catalogue and the library. Because both citation maintenance and Rights and Reproduction can be integrated with the library reference system, citation entries will tend to adhere to a uniform style sheet. In all, what had traditionally been seen as a series of independent operations will have evolved into segments of a multi-faceted activity in which the goals of separate departments are achieved with greater precision, proficiency and integration. There may even be increased employee cohesion stemming the mutual dependence of employees who operated heretofore relatively independently and from the mutual realization that each member of the staff operates within a sphere of information created and supplied by others.

In the end, all users can be provided with a versatile resource that can help fulfill the museum’s mandate to make its collection known, to promote the study of its objects and to enhance the lives of its public through knowledge of its collection. What has been conceived principally as a "back-room" activity, withdrawn and distant from the public goals of the museum, can now be drawn into the circle of activities through which the museum presents its public face. Not least importantly, by automating the accumulation of and access to citation and usage data, the museum will be better able to meet its fundamental obligation to pass a record of its activities and contributions to posterity.

Robert Baron Home Page | Museum Management: Papers | Guides

© 1994 - 2002 Robert A. Baron, Arts Information Consultant.