Abstract and subjects
Measuring the double isotope atomic abundance ratio as function of time of a homogenized stock of plutonium using mass spectroscopy provides a means to estimate the half-life of Pu-241, denoted here as t(1/2,241). After a logarithmic transformation, estimating t(1/2,241) along with a justifiable associated uncertainty is reduced to the analysis of a linear relationship, as illustrated in this paper using 15 published data pairs (time, log(double isotope ratio)) of Wellum et al. (2009) that span approximately 31 years (greater than two half-lives). However, as noted by Wellum et al. (2009), the published 15 data pairs exhibit inconsistencies that indicate possible underestimation of individual experimental uncertainties. Similar inconsistencies often arise in multi-experiment comparisons of the same estimated quantity, typically because some components of uncertainty such as individual experimental biases are difficult to identify and assess. It is therefore an important and common problem. In such cases the experimental data must be supplemented with other information to make plausible uncertainty estimates. We therefore analyze the data pairs under several different assumptions regarding total experimental uncertainties and show quantitatively that the best estimate of t(1/2,241) and of its uncertainty depend on the assumptions regarding experimental uncertainties. It is unlikely in this context that the 15 data pairs and associated estimated experimental uncertainties could guide one toward a very certain choice among the reasonable sets of assumptions regarding total experimental uncertainties. Thus a definitive recommendation cannot be singled out. Fortunately, the best estimates and associated uncertainties arising from different yet tenable assumptions regarding experimental uncertainties are all in reasonably close agreement. And, one of those best estimates we provide (with approximately 95% confidence limits) is (14.329 +/- 0.006[fit] +/- 0.029[bias]) years, which uses similar data stratification arguments as in Wellum et al. (2009) but a completely different approach. Furthermore, this estimate of t(1/2,241) agrees closely with the value recommended in Wellum et al. (2009) of (14.325 +/- 0.024) years. We conclude that the value of t(1/2,241) supported by the available data is robust, despite evidence of some non-ideal behavior, and that alternative means of estimating t(1/2,241) and its uncertainty yield reasonably similar results. (c) 2013 Published by Elsevier Ltd.