TESTING FOR MOISTURE. 207 



mination in butter when butter contains 80 per cent 

 fat and 16 per cent water, and. tlie effect of these 

 errors is twenty times as great for butter testing as 

 it is for milk testing when the lattor contains 4.0 per 

 cent fat. 



The errors which may occur in weighing the 

 sample of butter for testing is multiplied by the 

 ratio between the weight taken and 100. If an 

 error of one-tenth gram is made in weighing 10 

 grams of butter, this one-tenth gram is one-tenth per 

 cent of 100, but it is 1 per cent of 10 grains ; and 80 

 per cent of this 1 per cent falls on the butter fat, 

 while only 5 per cent of it would fall on a milk test- 

 ing 4 per cent fat. 



The actual effect which an error of one-tenth gram 

 in weighing out the butter will have on the final re- 

 sults may be seen by the following illustration: If 

 instead of ten grams only 9.9 grams of butter are 

 weighed into a test bottle, after which the test is 

 completed and the reading of the fat in the neck of 

 the test bottle is found to be 45. This reading mul- 

 tiplied by 18 gives 810, which divided by 10 grams 

 of butter gives 81 per cent fat in the sample test. If, 

 however, an error of one-tenth gram is made in 

 weighing, and the product of 45x18 or 810 is divided 

 by 9.9 grams, then the result is 81.8 per cent fat. 



The variation of one-tenth gram in weighing the 

 sample of butter into the test bottle has made an 

 error of .8 per cent fat in the final result. 



If this same error of one-tenth gram is made in 

 weighing six, instead of ten grams of butter into a 

 test bottle, and the test of the butter is made in 



