The accuracy of the casern number of milk as the basis for a chemical method for the detection of mastitis has been investigated. The number was determined for 247 samples of milk from the individual quarters of sixty-two cows, and the samples were examined bacteriologically for mastitis streptococci. Of the 247 quarters 49% were found to be infected—mostly subclinically—involving 74 % of the cows.

The total nitrogen content of the samples varied from 362· to 687·4 mg. per 100 ml., and the casein nitrogen from 153·0 to 554·0. The casein number varied from 42·3 to 83·4. The casein number averaged 79·2, 80·4, 79·5 and 80·3 in the samples from the uninfected quarters of the Shorthorn, Friesian, Ayrshire and Guernsey cows respectively; and 71·7, 74·1 and 70·3 in the samples from the infected quarters of the Shorthorn, Friesian and Ayrshire breeds respectively.

For the diagnosis of mastitis from the casein number, a figure of 78·0 and less was taken as indicating an infected quarter, and on comparing this chemical method with the bacteriological method, out of a total of 243 quarters, the chemical result differed from the bacteriological for 21 quarters—representing 8·6 % of the total.

Reasons are given for considering that this estimate of the error is excessive and that in practice the casein number would provide a valuable and rehable method for the diagnosis of mastitis.