MCLAUGHLIN gives an interesting illustration of the serious error which may arise from the incorrect application to discrete distributions of Fisher's method of combining significance probabilities. Wallis's exact computer solution is one that is more readily available nowadays; but it seems also worth noting an approximate solution applicable to the present context.
About this article
Cite this article
BARTLETT, M. Combination of Probabilities from Significance Tests. Nature 220, 1251 (1968). https://doi.org/10.1038/2201251a0
The Journal of Toxicological Sciences (2006)