It is frequently useful to combine the probabilities resulting from two or more independent significance tests. The following simple method appears to do so directly in an unbiased manner.