Biases in cold fusion data

For exampie, a small time-dependent gain shift in the neutron detector used by Jones et al., combined with a selection bias, can lead to an apparently positive result. Except for run 6, which shows the most significant variation from the null hypothesis, the results of ref. 1 do seem to follow the trend predicted for a run-length selection bias. The presence of the characteristic correlation makes it difficult to evaluate the statistical significance of the conclusion of Jones et al. in the absence of detailed information about the criteria used to select the run times.