Composite Samples of Milk. 171 



sample a true representative of the different lots of milk 

 From which it has been taken. 



195. Every time a new portion of milk is added to 

 the jar this should be given a gentle horizontal rotary 

 motion, thereby mixing the cream already formed in 

 the jar with the milk and loosening the cream stick- 

 ing to its side. This manipulation also prevents the 

 surface of the milk from becoming covered with a layer 

 of partially dried leathery cream. 



Composite samples having patches of dried cream on 

 the inside of the jar are the result of carelessness or 

 ignorance on the part of the operator. If proper at- 

 tention is given to the daily handling of the composite 

 samples, the cream formed in the jars can again be 

 evenly mixed with the milk without difficulty. 



196. Fallacy of averaging percentages. A composite 

 sample of milk should represent the average quality 

 of the various lots of milk of which it is made up. This 

 will be true if a definite aliquot portion or fraction of 

 the different lots of milk is taken. If the weights of, 

 say ten different lots of milk, are added together and 

 the sum divided by ten, the quotient will represent the 

 average weight per lot of milk, but an average of the 

 tests of the different lots obtained in this way may not 

 be the correct average test of the entire quantity of 

 milk. The accuracy of such an average figure \vill de- 

 pend on the uniformity in the composition and weights 

 of the ten lots of milk. When there is no uniformity, 

 the weights of the different lots of milk as well as their 

 tests must be considered. The following example will 

 illustrate the difference between the arithmetical aver- 



