208 MODERN BUTTER MAKING. 



cream test bottles reading to 30 or 40 per cent, the 

 effect of this small error is much greater. Suppose 

 the reading of fat from the six grams of butter is 

 28 ; this multiplied by 18 gives 504, which divided by 

 six (the weight of butter taken) gives 84 as the per 

 cent of fat in the sample tested. If, however, the 

 reading 28 multiplied by 18 or 504 is divided by 5.9 

 instead of six, then the result is 85.4. 



This shows a difference of 1.4 per cent fat between 

 the result obtained when the weight of butter is 

 taken as six grams and when it is 5.9 grams. 



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

 weighing the butter for a water determination, the 

 effect of the error on the final result will be much 

 less as is seen by the following figures : If ten grams 

 of butter are weighed for a water determination, 

 and the loss of weight by drying is 1.6 grams, then 

 this loss is 16 per cent of the ten grams of butter 

 weighed out. If an error 'of one-tenth gram was 

 made in weighing the butter then the loss, or 1.6 

 grams divided by 9.9 grams instead of ten grams of 

 butter gives a percentage of 16.1 per cent water in 

 the sample tested. In this case there is a difference 

 in the final result of only one-fortieth per cent 

 when an error of one-tenth gram was made in weigh- 

 ing the butter. 



Butter makers in creameries do not often make 

 weighings closer than 1-10 grams on the scales they 

 use and many of these scales are not sufficiently sen- 

 sitive to weigh as fine as 1-10 gram. They will, 

 therefore, have difficulty in getting duplicate tests 

 of the same sample of butter to agree much closer 



