Recent advances in low level counting techniques.

Abstract

Problems involving low level counting techniques are encountered in many diverse fields including biological and chemical tracer studies, archaeo­ logical and geological dating (I), investigations of natural and induced radio­ activities, and the study of fundamental particles. Such a host of methods and variations on methods have been reported that an exhaustive catalog would constitute little more than an annotated bibliography. We have, there­ fore, limited our attention to a few selected examples which seem to be of special interest because of their versatility or their extreme sensitivity. We prefer to organize our discussion around specific problems rather than around types of detectors since this permits a comparison of alternative methods and provides practical examples of various techniques in actual operation. Alpha counting is arbitrarily excluded. For a discussion of photographic methods (radioautographs) the reader is referred to Norris & Woodruff (2) and refer­ ences cited therein. We will discuss the methods developed for the measure­ ment of natural radiocarbon and tritium, cosmic ray produced Be7, and the K(O content of the human body. DOUble-beta decay experiments are consid­ ered since they seem to constitute one of the most difficult of the outstanding problems and because of their importance from the viewpoint of theoretical nuclear physics. Finally, an account is given of recent attempts to detect the neutrino, probably the most difficult low level counting problem of all and one which seems finally on the verge of successful solution. Principles.-The fundamental principles of low level counting are simple: maximize the number of detected events of the desired kind (signal) and minimize all others (background). The number of detected events is, of course, the product of sample size, over-all counting efficiency, and counting time. However, the striking importance of the net sample counting rate is not always fully appreciated. It can be shown (3) from elementary counting statistics, that if the sample rate is small compared with background (which is usually the case in a truly low level problem) and if counting statistics is the limiting source of error (also usually true), then the proper parameter for the comparison of counting methods is S2/B, the ratio of the square of the net sample rate to the background rate. (Any sample strength can be used in making the comparison so long as the same strength is used in testing all methods.) The method giving the maximum value of S2/B will yield the