MEDLARS and GRACE: How the National Library of Medicine Computerized Medical Literature

Sharing what medical research has found has been part of the National Library of Medicine (NLM)'s mission from the start. In 1879 it published the first edition of Index Medicus, a list of current medical literature, and in the 1940s it experimented with early mechanized systems. Computers changed the scale of its ambitions, beginning in earnest in 1960 with the formal proposal of MEDLARS, the Medical Literature Analysis and Retrieval System.

John Smith with a team of computer specialists at NLM, about 1965. National Library of Medicine.

Two birds with one stone

MEDLARS began with a request for proposals for a study of whether digital computers could be used to publish Index Medicus and to support "an efficient reference and bibliographic service". The system was meant to do both: help publish Index Medicus and other bibliographies, and run searches on demand. Planning and design were complete in 1962, and construction began.

Running it took a great deal of labour. Details of journals and articles were entered on punched paper tape, then transferred to the magnetic tapes that held most of the data, and new bibliographies and searches were produced from those tapes.

Paper tape loaded into a tape reader. National Library of Medicine.

How information was laid out on the MEDLARS magnetic tape. National Library of Medicine.

Batch processing

Computing then meant room-sized mainframes working in batches, and MEDLARS ran on a Honeywell-800. Users handed the computer a series of jobs and came back, sometimes days later, for the results; time had to be booked around the machine, and a job might take several attempts to run. For a medical researcher, though, asking for a search was simple, if not always fast. They mailed a form describing their interests and needs. Searches were free, but researchers had to tell the library how good the results were, and their answers shaped later changes.

A MEDLARS search request form. National Library of Medicine.

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

GRACE, the library's own typesetter

Results and bibliographies had to be printed, and no commercial equipment could do it to NLM's standards, including Greek letters and accented characters in high-quality type. So NLM designed and built its own: GRACE, the Graphic Arts Composing Equipment. In service from 1964 to 1969, it was for a time the fastest computer-driven photocomposer in the United States, and NASA and companies studied it for their own designs. It broke down often, but it was ground-breaking; a commercial machine replaced it in 1969, and GRACE is now at the Smithsonian.

GRACE receiving maintenance from a contract technician. National Library of Medicine.

A network, not just a machine

From the start NLM also sent its data on magnetic tape to other medical libraries across the United States and later abroad. They became MEDLARS Centers, running searches the way NLM did, and sent their search analysts and librarians to NLM for training. MEDLARS became an international network for sharing information, built on people's cooperation as much as on computing power.

It could only do its job as long as demand allowed. By the late 1960s, with the growth of scientific literature, medical research and specialties, MEDLARS could no longer keep up; a request could take two weeks or more. NLM began planning MEDLARS II, which replaced it in 1975; the next post in the series tells the story of MEDLARS II and MEDLINE.

Sources

• Nicole Contaxis, National Library of Medicine, "MEDLARS I & GRACE: The Early Mainframe Experience," Circulating Now, the first in a series on computing at NLM; rewritten in hubnx's own words. Photos and illustrations from the National Library of Medicine.