New Zealand radiocarbon reference standards

Abstract

A uniform approach, based upon the statistical behavior of random measurement errors, is suggested for assessing the relative merit of alternative methods of measuring radiocarbon and for interpreting and reporting the results of such measuremnts. The need for an objective, uniform approach is manifest because of the increasing concern with results which are close to the limits of precision or detection of the dating procedures. The standard deviation of the estimated net signal (sample counts) is first examined in terms of its dependence upon random errors arising in observations of background and gross sample counts. Because of the high precision characterizing such observations, it becomes vital to evaluate random errors additional to the Poisson (counting) errors. The ability to discriminate between samples of similar age, and minimum and maximum detectable ages are discussed in terms of the statistical theory of hypothesis testing. An explicit approach results for the uniform treatment of detection, detectability and age limits. The performance characteristic of alternative measurement methods cannot simply be stated in erms of a single 'figure of merit', because f the rather complex dependence upon counting times, sample, background and modern standard counting rates, and non-Poisson sources of random errors. A 'reduced activity' (sample activity/background equialent activity) plot is offered as a means for planning experiments and for rapidly assessing the capabilities of any specific measurement procedure. As a result of the particular choice of variables for this plot, alternative procedures may be represented by points in a fixed, two-dimensional array, where simple translations of the entire array correspond to changes in counting time or sample age.