The analysis of a factorial series of insecticide tests

When a set of insecticidal toxicity tests yields parallel regression lines for the relationship between mortality probit and log dose, the potencies of the materials or conditions under test may be compared purely in terms of log L.D. 50's. The purpose of this paper is to suggest that, when tests have been made with all combinations of several different factors, standard methods for the statistical analysis of factorial experiments may be adapted to the examination of the relative potencies.
Data obtained by Potter & Gillham (1046), in a 2' factorial experiment on alternative storage conditions for insects before and after spraying and the adjuvant action of terpineol in a pyrethrins spray, are used in an example of the computations. Details are given of the test of parallelism of the regression lines, the factorial analysis of the log L.D. '50s, the estimation of the mean effects and interactions and their standard errors, the significance tests, and the preparation of summary tables.