Dr. Garfield will present an autobiographical account of his involvement in chemical and pharmaceutical information systems, tracing his personal transformation from laboratory chemist to information entrepreneur. Beginning as a consultant to pharmaceutical firms, he developed information services such as Current Contents, Index Chemicus, and the Science Citation Index. As these databases grew, they led to systems for monitoring and forecasting science technology. More recently, he has been developing systems for historiographic analysis.
Just one week ago, I returned from the International Online Meeting in London. ISI, the company I established in 1954, was celebrating the 50th anniversary of my 1955 paper in Science proposing the idea of the Science Citation Index.1 When this invention came into my head, I was a dedicated member of the academic not-for-profit community. Private entrepreneurship was the last thing on my mind. From 1951 to 1953, I was enjoying life as a researcher at Johns Hopkins University. Then the First Symposium on Machine Methods for Scientific Documentation led to the idea of the Science Citation Index. A letter from a retired executive led me to Shepard’s Citations.2 It would take two more years to translate the idea into a publishable paper. In it I made the speculative proposal designed to catch the attention of the NSF and friends at Chemical Abstracts. Little did I know the resistance this simple idea would encounter
Not being commercially minded at the time, it never occurred to me to apply for a patent for the invention. While it might not have been patentable in those days, it might well be today. But the patent would have done nothing but feed my ego and empty my meager pocketbook. On the other hand, a paper published in one of the most prestigious journals of science gave a much greater boost to my ego. That paper would become one of my two most-cited papers out of a large number published over a long career
My ambivalent life as an academic and entrepreneur often left me conflicted. Indeed, if the whole truth be told, the most aggressive competition I ever encountered was from the American Chemical Society (ACS) and not the many dozens of for-profit publishers I would meet. For a long time my company was described best by the old saw – “We are not-for-profit but it wasn’t planned that way.” I seem to have a predilection for such enterprises. Like developing a new drug, you have to put aside short-term interest and hope for a long-term return.
But that is not the story I was asked to relate to you today. If we are remembered at all, it is usually not for our failures. So let me tell you something about the successes. The key events in my life as an entrepreneur were often serendipitous. And I was either lucky or smart enough to take advantage of those unexpected happenings.
In 1954, I arrived in Philadelphia to work as a documentation consultant for Smith Kline & French Labs. I was considered an expert on punched-card methods and SK&F needed an efficient system for retrieving information on the rapidly growing literature and case reports on the new wonder drug Thorazine or chlorpromazine.
Soon after coming to SK&F, I also developed a number of chemistry related products such as the Index Chemicus which have survived to this day. In fact under other people’s leadership at Thomson Scientific, it has become a formidable service for the pharmaceutical industry. But it was a losing venture for two decades carried by other profitable products like Current Contents. A more rational businessman would have killed the product years earlier. Indeed, four of my vice presidents resigned en masse because I refused to do so. I called them the four horseman of the apocalypse.
A fundamental discovery I made in 1957 was that one could use an algorithmic procedure to convert chemical nomenclature to molecular formulas. I tested the procedure on the Univac I computers both at Penn and the Franklin Institute. I could have applied for a patent on that discovery, but I wrote a doctoral dissertation instead.3 You can access the full text at my website, as well as most everything else I have ever written at www.eugenegarfield.org . There is an amusing aspect of that event, though it didn’t seem funny at the time. My original thesis manuscript was about ten pages and laid out the basic steps of the algorithm. Indeed, the journal Nature agreed that the idea was original enough to publish a short communication entitled “Chemico-Linguistics: Computer Translation of Chemical Nomenclature.”4
Nevertheless, my substitute faculty adviser at the time was shocked that I would expect a dissertation of such brevity to be approved. So I expanded the text with an appendix which provided a structural linguistic analysis of the Geneva System of Nomenclature. While it is rarely cited, some gracious colleagues have described this work as the seed into which the multi-million dollar chemical information industry blossomed.5
In telling you this part of the story, I have jumped ahead somewhat. All the time I was working on that Ph.D., from 1955 to 1961, I was operating my consulting business. In 1955, I had started a publication, known to many of you, called Current Contents. It did not at first cover chemistry and the life sciences. Rather, it was a service for management and social scientists. By coincidence, when I launched that contents page service (it was then called Management’s DocuMation Preview), Bell Labs was contemplating an almost identical service for its research staff. So they negotiated a contract with me for 500 subscriptions to a customized version. The regular issues of Management’s Document Preview were wrapped up in a special Bell Labs cover. That was just too much for me to print on my multilith machine. I had to use a local photo-offset commercial printer. When the time came to pay his bill, I was almost broke. And Bell Labs made me wait six weeks for my first check.
That was how I accidentally discovered the Household Finance Company. I had learned at a half dozen local banks that a purchase order, even from Bell Labs, was not sufficient collateral for a $500 loan at 6% per annum. HFC was happy to lend me the money. While 18% per annum was steep, I had the privilege of paying off the loan immediately when the check from Bell Labs arrived. The interest came to about $8.50. I was able to pay off the printer immediately and start the new cycle. Indeed, I soon learned to expand my borrowing capacity by visiting half a dozen HFC branches in the state. They were oblivious to my activities because in those days computers were not yet used to keep track of people like me.
A few months later, I ran into a few entrepreneurs in Washington who thought that MDP could be sold with a concentrated direct mail effort. They also recommended the name Current Contents instead of the mouthful Management’s DocMation Preview. However, they mistakenly believed CC could be sold to highly paid management executives, not realizing that such people are entirely different animals from researchers. A $5000 direct mail campaign produced about a half dozen subscriptions. My partners quickly vanished and took their tax loss.
Then a more positive event occurred. I received a phone call from Charlotte Mitchell, librarian at Miles Laboratories, who asked me to produce 100 copies of a contents page service covering medical and pharmaceutical journals. A similar in-house service was already provided at Lederle Labs and at Merck Sharpe and Dohme. I produced the early Pharmaco-Medical version of Current Contents using the early Xerox plate maker in my converted chicken coop in Thorofare, N.J. When I was able to sell enough group subscriptions to several pharmaceutical firms I once again needed to use the commercial printer. This time I knew where to go for financing on a larger scale. HFC eventually honored me with a Gold card.6
Within a year or so, Current Contents of Pharmaco-Medical, Chemical and Life Sciences was off to a good start. One day I received a phone call from Jacob Gershon-Cohen a Philadelphia radiologist who was the inventor of thermography.7 While visiting one of my customer drug companies, he had seen Current Contents. He wanted to receive his own subscription but didn’t want to receive it as a gift from the drug firm. So I asked him what he thought it was worth. He blurted out $2.00 per week. So we settled on an annual subscription fee of $100 per year. That is how the original industry price for one copy of Current Contents was established. Months later, I got another call from Dr. Harold P. Rusch at McArdle Labs at the University of Wisconsin, Madison. He said that he wanted to order 25 copies of CC to distribute to the faculty, but thought $1500 per year was kind of steep. He suggested a 50% discount and I agreed immediately. That is how the academic price for Current Contents was established. With this new pricing model, we began a direct mail campaign. The next important serendipitous event at the time was the discovery that the subscription could be charged to one’s NIH grant! You might say we had done a great job of market research but it has also been described as seat-of-the-pants decision making.
From then on CC grew like topsy and eventually became a highly successful series of products used throughout the world. At one point, CC/LS reached a circulation of 20,000 print copies. Remarkably, in spite of all the electronic advances that have been made, the print version is read by thousands of researchers worldwide who still find the pocket-sized TV Guide of the literature their weekly meat and potatoes. In fact, CC got an early boost from the editor of a medical journal who described how he got the “meat out of 700 journals “ 8
Later on, we started a chemical edition of Current Contents. The American Chemical Society decided to compete with us. In fact, they liked the title Current Contents so much that they decided to call it just that. Fortunately, I had trademarked the name. But in the interim my copyright attorney Arthur Seidel, an avid and loyal ACS member, asked me if I wanted to be the first guy to sue the ACS. Instead, he wrote a very polite but firm letter to Walter Murphy. That is why the ACS decided to name their new service Chemical Titles.
The last serendipitous event I shall mention to you today is the chance letter I received from Joshua Lederberg who asked me in 1959 what had happened to the idea of the Science Citation Index that I had described four years earlier in Science.9 This was shortly after he had received the Nobel Prize. This encounter eventually led to my discovery that even a private company then could receive an NIH grant. A committee of eminent molecular biologists and geneticists was formed and eventually ISI produced the experimental Genetics Citation Index. By the first year, however, Congress decided that only non-profits could receive grants. So a contract was negotiated to finish the project after NIH transferred the funds to NSF. The first multi-disciplinary SCI, which had to be produced in order to extract what was of interest to geneticists, had not been published yet.
I tried in vain to get NIH, NSF, or Chemical Abstracts, to publish this product. Once again, I took the plunge. Even HFC could not help me this time. Financing came at a much steeper price. A group of Wall Street investment bankers loaned us $500,000 to see us through the long period of drought while the new SCI was being cautiously and skeptically evaluated by the library and science communities. The investors got a 20% interest in ISI for that convertible debenture. Their loan was paid off rather quickly but they had to wait a long time to see their investment pay off. Indeed some did not wait long enough and sold their shares to Robert Maxwell in 1985. That is another story too long to relate today.
Not many years after its launch, some government agencies began to appreciate the science policy potential of the multi-disciplinary SCI database. NSF has used the SCI’s byproduct bibliometric data routinely as part of their biennial Science Indicators reports.10 There have also been countless studies of publication and citation output by country, institution, discipline, journals, etc. And last but not least, the ISI Journal Citation Reports and its annual journal impact factors have become the bete noire of academe and the scientific publishing industry. Publishers love it and hate it at the same time. If their journals achieve high impact, they advertise that fact. And if they are judged to be of low impact, they will argue it is because ISI has not yet included them in CC or SCI, ignoring the fact that nothing prevents them from being cited in the many journals already indexed. The scientometric data are also used in the UK Research Assessment exercises and elsewhere in grant making decisions.
In 1966, we reported that Nobel Prize winners published five to six times the average author but were cited 30 to 50 times the average 11. This has led to many inquiries as to whether the Nobel committees used our data. As early as December 1965, the Chief Librarian of the Royal Swedish Academy of Sciences stated that the “SCI is very useful and of great value -- especially when we make the preliminary investigations for the election of Nobel Prize winners.” The irony of this is that I did not know about his letter until a few months ago when I discovered he had written to one of the Four Horsemen of the Apocalypse. That letter would have been worth its weight in gold had I known back then when I was asked if the SCI was used by the Nobel committees.
To continue this odyssey let me tell you briefly about the Index Chemicus Registry System. In 1960, when Index Chemicus was started, Chemical Abstracts was three years behind in indexing. You can imagine the angst in the pharmaceutical industry not to be certain whether a new compound was really new. From my experience in indexing US patents for the steroid project of the Pharmaceutical Manufacturers Association I had figured out how to index chemical compounds without going through the tedious process of drawing structural diagrams to obtain the empirical molecular formula, sometimes called the empirical formula. I told a group of drug companies I could provide them a formula index within one month of publication. Even some of my friends in the organic chemistry department at SK&F did not believe it was possible being loyal supporters of Chemical Abstracts. Everyone had overlooked the fact that molecular formulas for new compounds were routinely provided in most published papers and could be extracted by a clerk. Thus, while a molecular formula is not unique one can easily separate homologs when searching the literature.
The monthly appearance of the Index Chemicus formula indexes was a real spur to CA to get its act together. So they went to NSF for a multi-million dollar grant to set up their registry system. Like Halliburton, this was a no-bid contract. And it was a major turning point in the development of the CAS system. Unlike my earlier efforts, it not only became a profitable enterprise, it was planned that way! In contrast, it took ISI 20 years before the ICRS system earned a profit! It survived in part because of support by the pharmaceutical industry which recognized its unique features, not the least of which was the integration of citation indexing and reaction indexing. But it survived primarily because a stubborn chemist who had failed in the lab, would not give it up!
When I changed the name of Eugene Garfield Associates to the Institute for Scientific Information in 1960, I did it for several reasons. It evened the playing field in marketing our services in competition with Chemical Abstracts and the government (National Library of Medicine), especially abroad, where ISI was perceived of as a non-profit institution. But more important, I was responding to the post-Sputnik challenge of the Soviet Union’s VINITI, that is, the All Union Institute of Scientific and Technical Information. We boasted that we could accomplish, with one tenth the staff, what it took VINITI to do with thousands. You can understand why I felt quite sympathetic when I heard Burt Rutan on “60 Minutes” last month describe his competition with NASA for the space tourism market!
2. Adair, W.C. “Citation Indexes for Scientific Literature?” American Documentation 6:31-32 (1955).
3. Garfield E. “An algorithm for translating chemical names to molecular formulas,” Philadelphia: Institute for Scientific Information, 68 pgs. (1961). http://www.garfield.library.upenn.edu/papers/thesis.pdf
4. Garfield, E. "Chemico-Linguistics: Computer Translation of Chemical Nomenclature." Nature, 192(4798):192 (October 1961). Reprinted in Essays of an Information Scientist, Volume 6, pgs. 489-491 Philadelphia: ISI Press (1984). http://www.garfield.library.upenn.edu/essays/v6p489y1983.pdf
5. Cooke-Fox G.H., Kirby, G.H., and Rayner, J.D. “Computer Translation of IUPAC Systematic Organic Chemical Nomenclature. 1. Introduction and Background to a Grammar-Based Approach,” Journal of Chemical Information and Computer Science 29(2):101-105 (1989).
6. Garfield, E. “How it all began: with a loan from HFC,” Current Contents No. 3, pgs. 3-5 (January 21, 1980). Reprinted in Essays of an Information Scientist, Volume 4, pgs. 359-362. Philadelphia: ISI Press (1981). http://garfield.library.upenn.edu/essays/v4p359y1979-80.pdf
7. Garfield, E. “Jacob Gershon-Cohen, M.D., D.Sc.in memoriam.” Current Contents No. 4 (March 17, 1971). Reprinted in Essays of an Information Scientist, Volume 1, pgs.152-153. Philadelphia: ISI Press (1977). http://www.garfield.library.upenn.edu/essays/V1p152y1962-73.pdf
8. Brown, E.A. “How I Get the Meat out of 700 Journals a Month. Medical Economics 39(6): 128-143 (1962).
9. Lederberg JL. Private Communication (May 9, 1959) http://www.garfield.library.upenn.edu/lederberg/050959.html
10. National Science Board. Science and Engineering Indicators 2004. Two volumes. Arlington, VA: National Science Foundation (volume 1, NSB 04-1; volume 2, NSB 04-1A) (2004)
11. Sher, I. H. and E. Garfield. "New Tools for Improving and Evaluating the Effectiveness of Research." Research Program Effectiveness, (M.C. Yovits, D.M. Gilford, R.H. Wilcox, E. Staveley, H.D. Lemer, Eds.) New York: Gordon and Breach, pgs135-46 (1966). Proceedings of the conference sponsored by the Office of Naval Research, Washington, D.C. July 27-29. Reprinted in Essays of an Information Scientist, Volume 6, pgs. 503-513 (1984). http://www.garfield.library.upenn.edu/essays/v6p503y1983.pdf