The relationship of embolic homogeneity, number, size and viability to the incidence of experimental metastasis.