The Babcock Test. 59 



good duplicates being often obtained, even when the 

 fat is not completely separated. 



68. Ascertaining the necessary speed of testers. In 

 buying a tester the operator should first of all satisfy 

 himself at what speed the machine must be run to give 

 correct results ; the preceding table will serve as a guide 

 on this point. He should measure out a dozen tests of 

 the same sample of milk, and whirl half the number at 

 the speed required for machines of the diameter of his 

 tester. Whirl the other half at a somewhat higher 

 speed. If the averages of the two sets of determinations 

 are the same, within the probable error of the test (say, 

 less than one-tenth of one per cent.), the first whirling 

 was sufficient, as it is believed will generally be the case. 

 If the second set of determinations come higher than the 

 first set, the first whirling was too slow, and a new series 

 of tests of the same sample of milk should be made to 

 ascertain that the speed in the second set of determina- 

 tions was sufficient. 



This method will test not only the speed required 

 with the particular machine at hand, but will also serve 

 to indicate the correctness of the calibration of the bot- 

 tles. A large number of tests of the same sample of 

 milk made as directed (pouring the milk once or twice 

 previously to taking out a pipetteful for each test) 

 should not vary more than two-tenths of one per cent 

 at the outside, and in the hands of a skilled operator 

 will generally come within one-tenth of one per cent. 

 If greater discrepancies occur, the test bottles giving 

 too high or too low results should be further examined, 

 and calibrated according to the directions already given 

 C53 et seq.). 



