MEDLARS I & GRACE: The Early Mainframe Experience

By Nicole Contaxis

Providing access to bibliographic data has long been a part of the National Library of Medicine (NLM) mission. Through a variety of means, NLM has publicized and shared medical research nationally and internationally since its inception. In 1879, it published the first edition of Index Medicus, a list of current medical literature, and in the 1940s, it began to experiment with early mechanized systems. The advent of computing brought a series of new techniques that dramatically altered the scope of the NLM goals and actions, and these changes began in earnest in 1960 with the formal proposal of MEDLARS—the Medical Literature Analysis and Retrieval System.

MEDLARS began with a request for proposals for a “study to investigate the feasibility of using electronic digital computers for the publication of the Index Medicus and also as a basis for the construction of an efficient reference and bibliographic service.” MEDLARS was thus designed to kill two birds with one stone. It was meant to assist in the publication of Index Medicus and other medical bibliographies as well as provide for on-demand searches of the computerized data. In 1962, the planning and designing stages were completed and construction could begin.

Getting the system operational and maintaining it required a significant amount of labor and planning. For example, data on journals and journal articles had to be input into MEDLARS using punched paper tape, and then this data was added to the magnetic tapes that were used to store the majority of the bibliographic data. New bibliographies and on-demand searches would be created using these magnetic tapes.

Paper tape being loaded into tape reader, taken from MEDLARS 1963–1967 National Library of Medicine

Illustration of the MEDLARS magnetic tape, taken from The Principles of MEDLARS National Library of Medicine

At this point in history, room-sized mainframe computers that employed batch-processing dominated computing. MEDLARS, which used a Honeywell-800 mainframe computer, followed this model. Batch-processing means that the computer was given a series of jobs and then the users would return, sometimes days later, for the results. People had to schedule their time around when the computer was available, and it could take several attempts to have a particular job, or program, run successfully. It was a time-consuming process. However, from the point of the view of the medical researcher, the process of submitting a search request was relatively simple, even if results were not always received quickly enough. A search request would be mailed via a form that detailed the researcher’s interests, needs, and contact information. While researchers did not need to pay for searches, they were required to respond to the library about the quality of search results they received. These responses helped inform strategic and technological changes.

John Smith in the computer room, c. 1965 National Library of Medicine

Example form that researchers would fill out in order to perform a MEDLARS search, taken from MEDLARS 1963–1967 National Library of Medicine

Of course, these bibliographies and search results would need to be printed, and this was not a straightforward task. Because there was no commercial equipment that could fulfill NLM’s needs, including printing Greek and diacritical characters at a high level of typographical quality, NLM designed and built GRACE, or Graphic Arts Composing Equipment. Operational between 1964 and 1969, GRACE was, for a time, the fastest computer-driven photocomposer in the U.S. and was studied by NASA and commercial companies in order to inform future designs. It was replaced with a commercial model in 1969. Although it suffered from malfunctions throughout its operational history, GRACE was ground-breaking and is now housed at the Smithsonian.

Coinciding with the creation of the MEDLARS system, NLM began providing its bibliographic data via magnetic tape to other medical libraries throughout the United States and eventually internationally. These medical libraries became MEDLARS Centers that could process MEDLARS searches in the same way that NLM did. Search analysts and librarians from these MEDLARS Centers would be sent to NLM to receive training. In this way, MEDLARS was not simply a computer or a piece of software; it became an international information-sharing network that relied on human co-operation as well as unique computing power.

Throughout its operation, its ability to fulfill its design goals depended on the amount of demand placed on the system. By the late-1960s, MEDLARS was no longer able to handle the workload and provide the services that users had come to expect. It could take two weeks or more to fulfill a MEDLARS request. This increased workload has been traced to expansions in scientific literature, medical research, and medical specializations throughout the 1960s. At this point, NLM began to plan for MEDLARS II, which would replace MEDLARS in 1975.

This blog post is first in a series on the history of computing and software development at the National Library of Medicine. Tune in next month when MEDLARS II will be discussed along with the continued growth of NLM and the medical library community.

Nicole Contaxis is a National Digital Stewardship Resident at the National Library of Medicine.