Is Benford's law a good instrument to detect fraud in reports of statistical and scientific data? For a valid test the probability of "false positives" and "false negatives" has to be low. However, it is very doubtful whether the Benford distribution is an appropriate tool to discriminate...