TESTING 239 



It should be noted that the smallest difference between 

 the liquid part and the ice is .2 per cent and the greatest 

 1.4. Probably the variation in frozen cream is much 

 greater. 



221. Sampling whole milk. — When testing a single 

 batch of whole milk, it must be agitated thoroughly before 

 a sample is taken. In taking a composite sample in the 

 usual way in creameries, it is a question whether it pays 

 to take the time to agitate the milk after it has been poured 

 into the weigh-can, for it seems that if the portion placed 

 in the composite sample is lower in fat than the average 

 of the batch of milk from which it is taken on one day, 

 it may be a little higher the next day. Thus at the end of 

 fifteen days, which is the usual time in creameries between 

 tests, the composite test would be the same whether the 

 daily batches were thoroughly agitated or not stirred at 

 all. In this relation Potts ^ first discusses the limit of 

 error in reading in relation to the error of sampling. He 

 says it is impossible in our present test-bottles for 

 whole milk to read closer than .1 per cent, that a sample 

 may really test .05 per cent higher or lower in fat than it 

 is read, and that likewise another sample may test .05 

 per cent lower or higher than it is read. Therefore, there 

 might be a difference of .1 per cent due to reading the 

 first sample too high and the second sample too low. In 

 like manner in testing cream in the six-inch test-bottles 

 graduated to read to 50 per cent, it would be necessary 

 to allow .5 per cent as a reading error. The following 

 figures show the results of different methods of obtaining 

 composite samples in comparison with a daily sample and 

 test. Composite sample A was taken with a little dipper 



1 Potts, A. E., Sampling of Milk, Thesis in Cornell Univ. 

 Library, 1913. 



