Summary The records used to prove 13 bulls in Kansas Dairy Herd Improvement Associations were examined by the analysis of variance to see whether there were significant differences among the daughters of each bull when the daughters were grouped according to their maternal grandsires. For 9 bulls out of 13, the variance in milk production between groups exceeded the variance within groups. The mean square for the sire's index for milk production was larger between groups than within groups in only 5 out of 13 cases. In 6 out of 13 cases the mean square for butterfat percentage was larger between the groups than within the groups. As concerns the sire index for butterfat percentage, the variance between groups exceeded the variance within groups in only 4 cases out of 13. Using linear regression to correct daughters’ records for differences in the records of their dams gave results similar to those secured by the other two methods. The data from these 13 bulls gave no indication that “nicking” is generally important enough to need much attention when proving sires. Differences between groups could easily have been due to chance variation in the sample of inheritance transmitted by the bull, or to differences in the environments that affected the different groups of daughters when their records were made.