On statistical estimation in fission track dating

Statistical methodology is described for obtaining point and interval estimates of age in fission track dating experiments, particularly for (a) the external detector method, which is reviewed briefly, and (b) the population method, which is considered in more detail. Some mathematical background to models, based on the underlying physical process, is given. This is used to suggest appropriate statistical methods, to assess them and to help understand sources of variation in the data. For the external detector method, a model is required for numbers of tracks given the amount of uranium; the usual Poisson model is derived as a consequence of independent thinning of a bivariate point process. For the population method, uranium heterogeneity does not necessarily invalidate the estimate but does represent a source of variation that must be allowed for in its error; thus an interval estimate of age might be relatively uninformative.