My long-out-of-print first book was entitled, perhaps too rashly, A New Approach to Psychical Research ( London: C. A. Watts, 1953). When I reviewed the evidential situation at that time it seemed to me that there was too much evidence for one to dismiss. Honesty required some sort of continuing interest, even if a distant interest. On the other hand, it seemed to me then that there was no such thing as a reliably repeatable phenomenon in the area of parapsychology and that there was really almost nothing positive that could be pointed to with assurance. The really definite and decisive pieces of work seemed to be uniformly negative in their outcome.
It is most depressing to have to say that the general situation a quarter of a century later still seems to me to be very much the same. An enormous amount of further work has been done. Perhaps more has been done in this latest period than in the whole previous history of the subject. Nevertheless, there is still no reliably repeatable phenomenon, no particular solid-rock positive cases. And yet there still is clearly too much there for us to dismiss the whole business.
It is in response to this estimate of the evidential situation that I want to develop here the thesis that repeatability is essential to the idea of a natural science; the notions of repeatability and of a law of nature are
Reprinted, with revisions, from The Humanist, volume 36 ( 1976), pp. 28-30, by permission of the author and editor.