Reference has been made in the preceding article (1) to the error in using the mean percentage mortality as typifying the underlying probability for an event, such as death, in a population, unless the deviation of the mortalities by an authentic subdivision lies within the errors of random sampling of that of a binomial, or Bernouillan, distribution.
In the experiments heretofore reported, if heredity has played a major part in determining the resistance of an animal, it follows that the behavior to an infection of the offspring from any mating will be determined in the main by the interaction of the factors received from the respective parents. Under such conditions, the litters constitute the logical subsets into which the population should be divided for an analysis of the type of distribution.
A method for such an analysis has been previously described by the author (2, 3). A much more complete exposition has been presented by Snedecor (4).