I'm still concerned about what looks like post hoc reasoning. From what I can see the hypothesis was that there should be some pattern related to astronomical/astrological configurations. The data was then trawled to find any pattern. In order to even to begin to assess the statistical significance of this (even forgetting about the approriateness of the different probability distributions) you need to determine the total number of possible patterns that would have been accepted as implying a correlation, otherwise you are just self-selecting for significance, i.e. if there are enough "valid" patterns available, then the probability is high that one of those would turn up in the data at an apparently high significance.