Technical Notes - e1
Remembering ELHILL - e2
NLM Computer Room Circa 1969
ELHILL System Development at NLM
The success of ELHILL and MEDLINE was the result of the convergence of three factors:
The three combined made the service an instant success, and the first nation-wide local access computer service.
It may also be that the fact that the SDC and the NLM Project Officers' visit to the Rheims Cathedral and the lighting of a candle there to the success of the project helped.
The success of the project also had something to do with the marriage of Grace Jenkins and Dave McCarn. [Editor's note: Grace McCarn went on to become the Chief of the Bibliographic Services Division during her tenure at the Library.]
ELHILL System Development at SDC
Building the original AIM-TWX database was a nightmare. We could only add 10,000 documents at a shot, and the addition took two hours and every single Tape Drive (ten) that System Development Corporation could muster. One stretched tape and we were set back two days. It's a good thing that ulcers have been shown not to come from stress; otherwise, all of us involved at the time would be dead.
I installed the program at the Biomedical Center at the Karolinska Institute in Stockholm, Sweden, in January 1972. I was allowed to use the computer between midnight and 6:30 A.M., but had to operate it myself (which programmers in those days had no idea how to do!). Once the main files were loaded we spent three days trying unsuccessfully to connect with a remote terminal. Finally somebody guessed that when you typed the User Id (which was "T2741") you had to add some spaces to make it eight characters long:
About 1968-1970 Davis McCarn and I were experimenting with online interactive search querying as applied to a database of Index Medicus. We were optimistic and zealous about the possibilities of what it could do for the rate of consumption of bibliographic citations provided by NLM services. Some of our glazed-eyed, hand-waving predictions at cocktail hours along Wisconsin Avenue produced amused and indulgent rolling of eyes by more experienced information workers of the period. When finally introduced, the initial consumer reactions to ELHILL-type querying (AIM-TWX) were positive, several times more voluminous, and grew several times as rapidly as had been predicted (or hallucinated) by us neophytes. We were soon transformed into sage old information scientists.
I provided protection, encouragement, requirements analysis, and command language design suggestions to the ELHILL team. The development work was conducted both at Santa Monica and at Bethesda.
In 1972, I was the junior member of a three-person team at the then-System Development Corporation in Santa Monica, California, which developed ELHILL as a replacement for the AIM-TWX system. When it came time to install ELHILL, the first relocatable Information Retrieval System, at the National Library of Medicine, I volunteered to do the work. Actually, I think the other two team members took a step backwards leaving me out front. The installation took two months and ELHILL II came into being and ran for three years.
The following is a poem I wrote when ELHILL II was replaced with ELHILL III in October 1975.
I think these same words apply now as we turn down ELHILL III's flame and wish PubMed the same success and longevity. So, let us wish PubMed God-Speed.
Testing the Prototype at UCLA
One year (or one month) at a time, I pored over the subject volumes of Index Medicus, copying accurately and legibly, one citation per index card, the references that I thought were relevant to the request. I felt like a medieval cleric, pulling these heavy volumes off the shelves, hoping they would yield their secrets to my inquiring gaze. When this task was completed, the cards were given to a secretary to type into a single alphabetic list. It was a chore. And for searches that required the correlation of two or more subjects, the bibliographies were neither comprehensive nor accurate.
Then, later that year, UCLA began its test of AIM-TWX, the new NLM automated retrieval system for a test database of citations from Abridged Index Medicus. I typed in the MeSH headings and a response showing only the number of citations containing the requested headings was generated. The instrument used to transmit the search was a teletype machine; it was in a special area with soundproofing around it, and I was almost deafened by the noise after keying in a search of any length. Transmission was slow (10 characters per second) and you waited for the printout containing the citations to be mailed back. Nonetheless, it was a quantum leap forward in information science technology. I could see that we wouldn't always be bound to the drudgery of manually produced bibliographies and the methods of the past. This was lifting the curtain for a peek into the future and I knew that I had found the right profession to take me there.