A stratified binomial marker model for bone-marrow repopulation experiments.

The paper considers bone-marrow repopulation experiments with injected mixtures of two types, A and B, of genetically marked donor cells. The covariance of the proportions of type A erythrocytes and lymphocytes is analysed as the sum of two components, under a stratified binomial model allowing the proportions of type A cells to vary in postulated strata of the mixture and with the assumption that the genetic marker does not influence cell development. The ratio of the two components is not experimentally estimable, but each of them has an interesting "demographic" interpretation. Possible inferences about certain "two-cell probabilities" are derived, and the experimental findings that necessitated the stratified model are illustrated.