**Abstract**
| In each of 6 experiments, about 300 III of honeybee *(Apis mellifera *L.) semen was diluted, mixed, and used to inseminate about 20 queens. The mixture for each experiment contained a portion of genetically marked sperm. Since the queens were also genetically marked, one could estimate the ratio of marked and unmarked sperm in each queen by counting her marked and unmarked progeny. The objective was 10 determine if each queen received equal proportions of marked sperm. In two experiments having semen mixtures with about 50% marked sperm, the 95% confidence interval for percent marked progeny was ±9·7% among the 22 queens in group I (semen was diluted 1:1 and stirred for 3--S min) and ±9·0% among the 16 queens in group 6 (semen diluted I:40 and centrifuged). The confidence intervals were similar and both were significantly greater than ±4·S% (expected 9S% confidence interval when sampling a perfect mix with 500 workers from each queen) (P<O.O1). In experiments having a very low frequency of marked spermatozoa in the mixture (1%, 2%, and 4%), at least one marked worker was found among the 500 progeny from each queen (n = 68 queens), even when insemination doses were <0· 2% of the total semen mixture. Nonrandom changes in progeny frequencies were related to insemination volume, time, and sperm handling procedures. Since nonmixing would cause random changes in progeny frequencies, much of the variability may have been caused by sperm competition rather than nonmixing. |