Smithsonian Collections Blog

Highlighting the hidden treasures from over 2 million collections

Collections Search Center

Wednesday, August 3, 2016

SELGEM: Designing IT applications in the 1970s

This is the last of a three-part-series on SELGEM, a pioneering computer systems used to manage museum collections in the United States. Read the first part here; and the second part here

You might think that total confusion would reign if specifying requirements and assigning data Category Numbers was not strictly controlled by “the system.”  However, the end users cared deeply about information accuracy and the quality of final products, such as specimen labels, catalog cards and resulting publications.  The staff were highly motivated to following the data standards and produce high quality products.  Museum staffs were generally acquainted with structured data because of their traditional work cataloging museum specimens and object, and in research data analysis.
A standard data capture form of the Botany Type Register Project. Page 13, from:  Shetler, Stanwyn G., Mary Jane Petrini, Constance Graham Carley, M. J. Harvey, Larry E. Morse, Thomas Kopfler, and collaborators.  1973.  An Introduction to the Botanical Type Specimen Register.  Smithsonian Contributions to Botany 12. Smithsonian Institution Archives. 
The design of a new SELGEM application in the early 70’s was simpler than design of a new information system in 2016.  There were three reasons for this:  Smithsonian operated a single Honeywell mainframe computer; possessed a single general purpose software “database” tool; and there were only two or three data entry procedures available.  In addition some experience acquired in one application, such as cataloging mammal specimens, could be applied directly to another application, say cataloging mollusk specimens.  While every discipline has some unique data elements, there were recurring groups of data elements for taxonomic information, geographic information, collectors, etc. that you could build on.  Best practices learned designing the earliest SELGEM applications were applied to the design of succeeding applications.

Stanley A. Kovy Director of Information Systems Division (ISD)
and early supporter of the SELGEM system.  Standing at the 
computer console of the upgraded Honeywell 2015, Nov. 1971.
Smithsonian Institution Archives. 71-2785-12. 
For a first time application, usually the system owner or sponsor would assemble sample input records, data sheets, current output products and samples of desired outputs.  The staff member would discuss the proposed SELGEM application with an experienced SELGEM person, “the expert.”  Current workflows were reviewed for possible changes and improvements.  Often there was a very specific output desired.  From these discussions they would begin to prepare a data dictionary, define data standards, and rules for data entry; and SELGEM Category Numbers would be assigned.  They would consider immediate objectives and future objectives; as well as possible improvements and changes to the workflow in the future.

While Category Numbers could be assigned sequentially without any gaps (001, 002, 003, 004, etc.) there was no technical reason to do that.  Theoretically, at least, you had 999 Category Numbers to select from.  It was generally believed that a more flexible design was achieved by providing gaps when assigning Category Numbers.

For the scientific name you might have this data structure:
071 genus
073 subgenus
075 species
077 subspecies
This format was frequently adopted even when the systems owner had no intention of entering data for subgenus or subspecies.  So the documented design might look like this:
071 genus
075 species
This design met the immediate requirements, and provided a flexible logical data structure for the future, with subgenus and subspecies not even defined.

For truly new applications without any precedent, such as an archives catalog, a prototype file of a few records could be created within a day or two.  This “prototype file” could be used for what we now call “proof of concept.”  It could be reviewed by other staff members for comment and approval.  The file could be used to create sample reports, or evaluate how to best structure particular data elements for the most flexible data processing.

Or a proposed application, such as an index of reprint publications, might be very similar to previous applications.  Once you have created several bibliographic applications and produced reports for several years, you understand the basic structure and processing requirements of this type of application.
Deborah Bennett and Tim Coffer, museum technicians, sort trays of shells for the mollusk inventory in the National Museum of Natural History's Division of Mollusks.  SELGEM was used to support the inventory and in preparation for the collection move to the Museum Support Center in Suitland, MD. Smithsonian Institution Archives. SIA2009-3232. 
Initially the SELGEM technical knowledge was in the hands of the pioneers, but as the knowledge of the SELGEM system was acquired by an increasing number of museum staff members, new applications could be quickly designed to meet new requests.  SELGEM processed all the information for the “great count,” the Smithsonian collections inventory program (1979-1983). SELGEM was a truly powerful information processing system in its day.

David Bridge, Volunteer
Smithsonian Institution Archives

No comments:

Post a Comment