ANOTHER LOOK AT SOME OF MENDELS RESULTS

The data of Gregor Mendel have been widely discussed since Fisher pointed out that chi square tests reveal them to be too close to expectation. Although many explanations have been advanced, Mendel's data remain enigmatic and, in an effort to uncover some clues as to the reason, further analysis is presented here. Five of his experiments involving heterozygosity at two loci have been compared to the results of a computer simulation of those experiments. This approach easily permits answers to questions that are difficult or impossible to handle with standard statistical techniques The experiments are those in which Mendel states specifically the protocol for the crosses, the numerical results for each, and the tests made to verify his results. The results presented here suggest that the chance of getting his numbers in those five experiments is considerably less than the figure of about one in 700 that is commonly given. Furthermore, none of the commonly stated explanations for the closeness of Mendel's data to expectation seems to be entirely satisfactory for these experiments, and, in particular, the suggestions that Mendel either omitted results or repeated experiments that were incompatible with his expectations seem very unlikely in view of the computer simulation analysis.