The question of testing for synergic action of treatments of Vibrion septique infection in mice is considered. Henderson and Gorer's data are used as the basis of the analysis.

Their six experiments are tested for heterogeneity and one experiment is found to be discrepant. This is then removed and only the remaining five used in the later working.

A x2 formula is derived for testing the significance of the evidence for synergic action. Sulphapyridine shows such interaction in effect with both antitoxin and antibacterial serum. These last two show no interaction.

The most efficient design of such experiments is then considered, and it is shown that the death rate in the single treatment tests should be kept at least as high as 90 %, and that whereas the numbers of mice assigned to the single treatment tests should be equal, that assigned to the joint treatment should be 1·5 times as great.

A graphical method for testing the significance of the results is described. These graphs also allow of an opinion being formed as to the number of mice which would be necessary to detect an interaction of given magnitude.