210 MODERN BUTTER MAKING. 



The errors all come on the amount of the sub- 

 stance tested and 80 per cent of this error falls on 

 the fat, while only 16 per cent of it falls on the wa- 

 ter in butter and 4.0 per cent of the error would fall 

 on the fat in milk testing 4.0 per cent fat. 



After understanding the full effect of all the errors 

 of analyses, as well as the difficulty of weighing but- 

 ter into test bottles, the necessity of using extreme 

 care in every manipulation and in keeping the scales 

 and glassware in a most sensitive and accurate con- 

 dition, one must not expect to use the Babcock test 

 for determining fat in butter with the same degree 

 of accuracy as it is used in milk and cream testing. 

 Duplicate results on the same sample of butter may 

 not agree within 1 per cent of each other and the 

 results obtained may easily be 2 per cent from the 

 actual fat contained in the sample tests. All these 

 variations and inaccuracies, being due to the errors 

 of analyses and the fact that the fat is such a large 

 percentage of the butter. 



When fairly satisfactory results are obtained in 

 testing butter by the Babcock test it is probable that 

 the errors of one class have compensated for, or bal- 

 anced the errors of another class in the opposite di- 

 rection, but at the same time it is possible that all 

 the errors may be in the same diT;ection and thus 

 increase the inaccuracies of the results." 



262, Number of grams of butter to use for making 

 moisture determinations. By consulting the Dairy 

 Arithmetic it will be seen that the smaller the basis 

 upon which moisture determinations are based the 

 greater is the per cent of error, caused by an error. 



