This essay was originally published in the Current Contents print editions May 23, 1994, when Thomson Reuters was known as the Institute for Scientific Information (ISI)

While most users think of indexes as merely a means of recovering information, index compilers have long been aware that indexes--especially those for scientific literature--serve a vastly more important and often unstated purpose. That is, they reveal connections between ideas or concepts that were not considered before. And so, from the earliest days, professional information scientists considered the library and its many indexes as an extension of the laboratory. The literature can be used creatively to develop and even to test new and interesting hypotheses. One particular manifestation of this is in meta-analysis, which brings together many different studies on a subject and--from analysis of small deviations between the studies--endeavors to draw new conclusions about the subject.1

In a previous essay, I referred to the work of physicist and library school professor Don Swanson. In some of his most recent work, he has experimented with a systematic trial-and-error strategy for linking what he calls noninteractive literatures--that is, bodies of work that do not cite each other and are not both co-cited by any other publication. His experiments have led to rather impressive results, so let's take a brief look at his work and indicate the role of the Science Citation Index® in this type of search.

The Strategy

The application of citation indexing that we will explore in this essay involves a strategy that Swanson designed to find pairs of medical literatures that are logically (scientifically) related but noninteractive. The strategy consists of two processes: exploratory and exclusionary.

Exploratory Process. The exploratory process begins with some scientific problem, such as a disease for which neither cause nor cure is known, and seeks to identify indirect (hence unintended) literature links that are suggestive of possible solutions. Traditional descriptor-based indexing is useful for this process.

Subheadings, such as those in Medline®, serve as intermediate topics that bridge the two noninteractive literatures (see Figure 1). For instance, a researcher might notice that the topic of emaciating or degenerative diseases is frequently accompanied by the subheading of growth hormones in the index. The subheading of growth hormones could then be considered as an intermediate topic.

The next step in the exploratory process is to find titles that relate directly to factors that influence the intermediate topic. In the case of growth hormones, such a factor is arginine. Once a relationship is established indirectly between two primary topics--such as degenerative disease and the use of arginine--through an intermediate topic--such as growth hormones--a hypothesis can be formed. 2 I will come back to this example again.

Exclusionary Process. The use of the SCI® is of paramount importance in performing the exclusionary process. According to Swanson, SCI is an effective tool for showing that the two literatures are "mutually isolated and noninteractive to a high degree."2

As Swanson reiterates in his articles, SCI provides "a means of extracting new knowledge from the body of published information."3 This is a feature of citation indexing that I and others have always stressed. While many citations are routine in the sense that they cite common methods and established ideas, others are unique and bring to the cited author and others new perspectives and knowledge not previously anticipated. That is why I often describe citation indexing as a posteriori indexing. 4

Success Stories

Raynaud's Disease and Fish Oil. The first trial-and-error search by Swanson was designed to find a then unknown cure for Raynaud's disease--a vascular condition that can be caused either by exposure to cold temperatures or by emotional upset.5, 6 In exploring the literature, Swanson found that the subheading "blood" is important in the search of the literature on Raynaud's disease. Turning to further exploration of the literature--but this time for articles about factors that influence the characteristics of blood disorders common in Raynaud's disease patients--fish oil appeared to be an important topic.

Working on the hypothesis that fish oil would relieve the blood disorders associated with Raynaud's disease, Swanson then used the SCI to assure that there were no existing links in the literature between Raynaud's disease and fish oil. Satisfied that he had identified two noninteractive literatures, he studied the two literatures to determine the medical defensibility of his hypothesis. Two years after the publication of the hypothesis, it was corroborated by a clinical test at Albany Medical College.7

Migraine and Magnesium. In a subsequent study that linked magnesium deficit to migraine, Swanson identified 11 intermediate factors--including type A personality and vascular tone. 3 At least eight different groups of researchers have published one or more reports that discuss confirming evidence for the hypothesis. 8

Somatomedin C and Arginine. As I mentioned above, another study links characteristics of degenerative diseases (particularly the low levels of circulating somatomedin C) with the administration of arginine. 2 Even though both literatures deal individually with growth hormones, wound healing, and other intermediate topics, the somatomedin C and arginine literatures are remarkably isolated from one another. Swanson suggests that a clinical test of the effect of arginine intake on blood levels of somatomedin C may be worthwhile as it could benefit elderly patients with degenerative diseases as well as patients with emaciating diseases such as AIDS.


As pointed out in other essays this year (see the list of Essays in the 1994 Series), there are a multitude of unique applications for citation indexing. The strategy that Don Swanson has used successfully for linking noninteractive literatures is another productive--albeit unusual--use of the SCI.

In the next essay, we will look at a use of citation indexing that involves journal evaluation and management of library journal collections.

Dr. Eugene Garfield
Founder and Chairman Emeritus, ISI


1. Garfield E. Meta-analysis and the metamorphosis of the scientific literature reviews. Current Contents® (43):28 October 1991. (Reprinted in Essays of an information scientist. Philadelphia: ISI Press®, 1992, Vol. 17. p. 170-4.)

2. Swanson D R. Somatomedin C and arginine: Implicit connections between mutually isolated literatures. Perspect. Biol. Med. 33:157-86, 1990.

3. ------------------. Migraine and magnesium: Eleven neglected connections. Perspect. Biol. Med. 31:526-57, 1988.

4. Garfield E. Citation indexes: New paths to scientific knowledge. Chem. Bull. 43(4):11, 1956.

5. Swanson D R. Fish oil, Raynaud's syndrome, and undiscovered public knowledge. Perspect. Biol. Med. 30:7-18, 1986.

6. ------------------. Online search for logically-related noninteractive medical literatures:a systematic trial-and-error strategy. J. Amer. Soc. Inform. Sci. 40:356-8, 1989.

7. DiGiacomo R, Kremer J, Shah D. Clinical effects of fish-oil fatty acid ingestion in patients with Raynaud's phenomenon. Arthritis Rheum. 31(suppl.):S34, 1988.

8. Swanson D R. Intervening in the life-cycles of scientific knowledge. Libr. Trends 41:606-31, 1993.