MARAC 1999 Presentation

Presented at the Mid-Atlantic Archivists meeting, 

Pennsylvania State University, 1999 

In this presentation I will apply functional analysis to a natural history museum as a means of appraising which records have continuing value and determining where those records are likely to be found in the institution. This builds on Helen Samuels’ approach to documenting modern colleges and universities in her 1992 Varsity Letters. In keeping with the program theme, "Archives under Construction," I will also suggest that when you build recordkeeping systems on a foundation of legacy data, you must expect that the data has neither been reported consistently nor at the same level of detail over time. Therefore, you must reinforce your system accordingly by keeping both your information system and your users aware of these inconsistencies.

I would argue that one of the major in difficulties identifying, locating and using the archival records of an institution, is that of adjusting for legacy data, maintained in legacy systems. Legacy systems need not be confined to those in outdated electronic formats. Legacy systems can be any body of data with an established practice of use and interpretation where a change in that use or interpretation would require a substantial effort and substantial resources. Legacy systems are those information systems which were too important to discard, yet were too large, too complex or just too time-consuming to convert to the present information system. Maintaining a Rolodex after you’ve gotten a PDA would be an example of a legacy system – you’ll probably convert the information you use most frequently, but perhaps not include everyone’s email, particularly those involved with you on a project that’s been on the back burner. But note that you are now maintaining two information systems, probably organized slightly differently and with different levels of detail. One day there will be some administrative mandate – or some technological crisis – at which point you will abandon one of the systems, most certainly losing data in the process.

While archivists sometime consider themselves manifestations of Shiva the Destroyer, we are not in the business to lose data or records. Of course, we can’t keep everything, even discounting electronic records, nor should we, thus making it all the more important that we can continue to use what we’ve kept. In the course of understanding and documenting the business of an institution, such as a museum, you will need to identify those record types which support the primary functions of that institution. In a museum, one of the key types of information to be retained is the accession record and its surrounding complex of information that is used to identify the individual object.

Museums are defined as collections of objects assembled and maintained within a specific intellectual environment. In order to maintain that environment, information about those objects in the museum’s collection must be available. In museums, institutional memory is preserved in the accession record. The accession record, the basic element of a museum’s traditional recordkeeping system, records the transactions by which the object is acquired and describes the object so that it can be identified throughout its life-cycle. Past practice has maintained this information through a series of paper forms and associated indexes. In order to maintain the public trust, museums must be accountable for their collections. This need for accountability drives collection development policies, and mandates that viable and long-term information management systems be maintained. The museum’s mission statement identifies maintaining information about collections as one of the important business processes of the organization, particularly as it supports the other fundamental goals of promoting scientific and artistic research and providing public information.

Maintaining the consistency of legacy data is a particular concern when the evidential qualities of those records maintained in those systems are invoked in order to authenticate institutional information. Examples of this mission-critical institutional information could include documentation of the exact terms of an endowment, the status of ownership of a collection about to be deaccessioned and sold, or specific collection data, such as the location and date of the discovery of a new species of mammal. Because the forms used to collect data and the procedures used to retain and organize the data for access have mutated over time in most of our organizations, it is critical to maintain institutional awareness of those changes. These changes in the structure of the information are known as schema shift to your IT colleagues. There may also be cases where parts of the institutional record have been lost – departmental correspondence files discarded or collection information irreparably damaged by fire, flood or simple neglect. A user’s understanding of the total series will certainly be skewed if data is missing and that gap is not evident. The hazards of inattention to the gaps and inconsistencies in legacy data are particularly serious when you migrate your paper records to an electronic format. If the correspondence files of your second director are missing, users might wonder at the inactivity of that administration as they search the correspondence database, unless you have other sources to triangulate and reconstruct the institutional history.

Data loss from conversion of legacy systems is not so much the errors in transcription, but the loss of the anecdotal knowledge of how these records are to be used, based on personal knowledge of those specific records and yes, thank you, Nicholson Baker, the physical appearance of the records. Sarasan’s 1981 frequently quoted article "Why museum computer projects fail" states that systems fail because they did not have a mechanism for incorporating the continuity supplied by the collective memory of curators and support staff – that implicit domain knowledge that sustains hand-built paper systems. [1]  Without this oral tradition, users of these earlier legacy systems would have had a hard time either locating the object or being aware of related information about it.

A common problem of legacy data is the difference in the level of detail reported. Information system designers refer to this as a difference in the granularity of the information. Location data is a particularly good example of differences in granularity in natural history accession records. Location data can be as specific as "Road cut, west side, about 1 mile north of Legrand, Marshall County, Iowa, first hill north of Iowa River" (Acc.no. 28,480, 1976) or as general as "Vertebrate fossils, collected by Dr. J.L. Wortman and party in Albany Co., Wyoming, during the summer of 1899." (Access no. 1246, 1899).

That last, by the way, was the accession record, in its entirety, of the fossil bones that were picked from their surrounding matrix and reassembled into Diplodocus carnegii, the Carnegie Museum of Natural History’s signature dinosaur. This is also the type specimen of that particular dinosaur, that specimen on which the original description was based and to which all others must be compared. It’s yet another instance of a cultural artifact of iconic stature with an incomplete archival record. Prior to the wide-spread use of GIS (Geographical Information Systems) by collectors – and that is to say most natural history collections made prior to 1990) – collectors and the curators that used their specimens and the related data worked through a complex of resources, including the collector’s notebooks and contemporary atlases in order to determine just where the collector actually found those plants, animals, or their fossilized remains. Location information is a critical element of the evidence that specimens and their related documentation present to the user of museum recordkeeping systems. This data is then used to construct distribution maps that show the range of the organism and to serve as a specific base line proving the existence of at least one instance of that creature in that place at that time. And all of this is being done in an atmosphere of incomplete and inaccurate information.

In order to emphasize the changing nature – and the changing granularity of accession record data, I’ll show you a few examples of accession records from the CMNH collection. Remember that accession records are the basis for the legal claims museums make of their legitimate acquisition and ownership of specimens – and represent the distillation of a range of other archival records, including correspondence, field notebooks, sketches, photographs and now email.

[Accession no. 1] Database template for Accession no 1 illustrates the difficulties encountered when transforming existing and inconsistent data into a consistent format. And we begin with a Y2K problem where the accession date is ambiguous, unless you "know" that the date is likelier to be 1896 and not 1996. Databasing of the accession information was done through 1996, but omits the descriptive notes.

Considerable processing – and time – elapses between these stages in the life cycle of the specimen record. David Bearman [2] correctly separates the act of accessioning from that of acquisition in the stages of the life cycle of a museum object as "Considered, Acquired, Accessioned, Managed, Conserved, Documented, Studied, Interpreted and Deaccessioned." Natural history collection records are based on objects that are incompletely known at the time of acquisition and are often incorrectly identified while in the museum setting. Art museums particularly have to contend with the problem of inaccurate identification as it appears more newsworthy, if not equally intellectually scandalous to mis-identify a Rembrandt than to put the wrong head on a dinosaur. Bearman notes that "the problem of the collections management information system is to preserve for future use everything that was once explicitly known about a cultural object in order to reduce redundant effort and improve the quality of decision making vis-a vis the collection." [3]  

Not to belabor this point excessively, but I trust that you now have a sense that accession records illustrate the difficulty of maintaining records of continuing value to an institution that contain legacy data.

Correct interpretation of legacy data is essential when documenting the past practices of historical institutions such as museums, historical societies or other types of public archives. These past practices can include the changing administrative organization of the institution, the responsibilities and activities of individual components of the organization and the specific recordkeeping systems used to document items in the collection.

Functional analysis has been suggested as a technique to be applied to the business of an institution in order to identify, and ideally capture, those records which completely and correctly represent that the business to future users of the institutional archival record. Rather than asking what a specific office does and focusing on collecting those records generated by the "significant" offices within the organization, functional analysis asks what is being done by the institution as a whole, then determines where records which document that activity are to be found. Often, records documenting those functions will be found in more than one place. Establishing these patterns of function will allow comparative analysis of similar types of organizations, regardless of the complexity of their organizational structure. Historians of non-profit organizations, such as universities and museums have noted that these organizations are following the lead of business in re-structuring the administrative organization more frequently than in the past. I see this as another argument for functional analysis as the foundation for appraisal to determine the continuing value of a record group rather than basing the appraisal on the place of issuing office in the institutional hierarchy. And in the course of a functional analysis, particularly when looking at capturing the historical record, you may discover, as you would in most natural history museums, that assuming a strictly hierarchical organization of a top-down management structure would have you collecting high-level financial data, but probably missing most of the records of public programs, temporary and permanent exhibitions. Museums are less hierarchical than columnar, with individual discipline-based departments operating autonomously. It is also important to remember that natural history is a cumulative science, so many older records have current and continuing value, thus making the "archival" status of these records a matter of semantic debate. Using the issuing office based technique, you would probably miss much of the research, except for initial proposals with their signoff pages and final reports to the granting agency and practically none of the specimen-based collection information.

In order to correctly interpret data describing collections or individual items in a collection, future users of the data need to know the metadata – what level of granularity/detail was expected, what descriptive standards were used and what known gaps exist in the record series.

Metadata is particularly important when paper-based data is migrated into an electronic recordkeeping system. Inadequate/inaccurate, incomplete or inconsistent metadata will put all the data at risk for misinterpretation and jeopardize the ongoing validity and usefulness of the data, particularly as those individuals who once knew the peculiarities of the legacy system are no longer available to advise on its nuances.

Understanding what natural history museums are in the business to do is essential to determining what institutional records should be kept. One place to start would be a typical museum mission statement which lists the following responsibilities: