Published online by Cambridge University Press: 01 June 2009
Samples were tested for fat percentage from every consignment of herd bulk milk from 10 herds over a period of 12 months. The mean monthly within-herd variance was 0·043. Also, triplicate samples were taken from 8 herd bulk supplies for 8 days and tested in duplicate and an analysis of variance was applied to the within-herd fat percentage in order to find the contribution sampling, testing and biological variances. It is shown that when dealing with tanker-collected milk where composite samples are not used the testing accuracy is secondary to frequency of sampling. By combining the variances in the formulae given, the accuracy of different systems may be obtained and compared using either single tested or composite samples.