Brief Communication
A Second Example of Mutually Isolated Medical Literatures
Related by Implicit, Unnoticed Connections
Don R. Swanson
Graduate Library School, University of Chicago, 1100 E 57th, Chicago, IL 60637
The biomedical literatures on migraine and magnesium
are mutually isolated to a high degree-with a few excep-
tions, they do not cite or refer to each other and are not
cocited. They are, however, related by unnoticed connec-
tions that become apparent when certain subsets of the two
literatures are brought together. These connections suggest
that magnesium deficiency may be a cause of migraine, a
thesis examined in a companion article published in a bio-
medical journal. A previously described trial-and-error
search strategy was used in discovering this second example
of logically related noninteracting literatures, a discovery
that lends significance to an earlier published example
based on Raynaud’s disease and dietary fish oil. The frag-
mentation of science into specialties suggests that unin-
tended and undiscovered relationships within the literature
cannot be uncommon. The goal of this. research is to develop
methods and strategies for finding such relationships.
In earlier articles, I developed the argument that, in
principle, new knowledge might be won by assembling
pieces already published but never before brought together
(Swanson, 1986a, 1986b, 1987, 1989). A prototype of one
possible structure underlying such a process was sug-
gested, a syllogism in which the two premises, “A causes
B” and “B causes C,” are separately known by two differ-
ent groups of researchers in different specialties, neither of
which knows about or cites the work of the other. If no
one person knows about the work of both groups, then the
implicit connection between A and C might be completely
unknown, but discoverable through reading the publica-
tions of both groups.
How to discover such pairs of logically related but mu-
tually isolated or noninteractive literatures is the central
problem addressed in this series of studies. An example
based on the literatures of Raynaud’s disease and dietary
Received October 9, 1987; revised December 9, 1987; accepted March
24, 1989
0 1989 by John Wiley & Sons, Inc.
fish oil was previously given, and a search strategy was
described (Swanson, 1986b, 1987, 1989). A second ex-
ample, based on migraine and magnesium and reported
here, lends weight to the first example, to the search pro-
cess used, to the common logical structure of the two
cases, and to the surmise that there must exist many such
pairs of specialized literatures related by implicit, unno-
ticed connections.
This study leads to a medical hypothesis, the plausibil-
ity of which I examine in a companion article that reviews
in depth the relevant medical literatures (Swanson, 1988).
Because that article is published in a biomedical journal,
the arguments put forward are available for critical analy-
sis by physicians and physiologists, and have profited from
suggestions by medical editors and referees. I now take up
problems appropriate for an audience of information scien-
tists -problems concerning the form and structure of the
argument, the corresponding citation pattern within the lit-
erature, and the search process used in identifying and
finding that literature.
To understand these problems it is helpful first to exam-
ine the main biomedical conclusions reached in the com-
panion article. Those conclusions, summarized as follows,
are based on an analysis of 65 journal articles on migraine
and 63 articles on magnesium (Swanson, 1988). For our
purposes here I assume that the biomedical aspects of the
analysis are not problematic. The conclusions about mi-
graine are numbered la through 1 la, and those on magne-
sium, lb through llb. A close relationship between each
a/b pair is apparent, a relationship in most cases similar to
the ABC syllogism mentioned earlier. Each pair is consis-
tent with, and suggests, the hypothesis that magnesium de-
ficiency may be a causal factor in migraine headache. The
following references are selected from the 128 articles ana-
lyzed in Swanson (1988) and will be further discussed.
1. (a) Spreading depression of cortical electrical ac-
tivity is thought to be implicated in the early
phase of a migraine attack (Gardner-Medwin &
Skelton, 1982).
JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE. 40(6):432-435, 1989
CCC 0002-8231/89/060432-04$04.00