257 



With skiiu-milk, buttermilk, or whey, in which there is but little fat, 

 it is reconiiuended to treat four portions iu as many bottles and collect 

 all the fat in one bottle, or better, to use a bottle holding three or four 

 times as much as the ordinary test bottle. With sour buttermilk or 

 milk that has stood a considerable time before testing, the tendency of 

 the albuminoids to precipitate on the addition of water maybe avoided 

 1)3^ using a hot mixture of equal parts of sulphuric acid and water in- 

 stead of water alone. 



In testing cream tbe test sample may be divided about equally into 

 two or three parts and treated in as many bottles, the several readings 

 being added. The sources of error and means of avoiding them are de- 

 scribed, and the results of fortj" comparative tests with the gravimetric 

 (asbestos) and the above-described method are given. 



In thirty samples of whole milk the new test gave results in fifteen 

 cases higher and in fifteen cases lower than the gravimetric determina- 

 tions, the average being 0.107 per cent higher than the gravimetric, and 

 the greatest difference 0.3 per cent. In two samples of skim-milk the 

 results with the new test averaged 0.035 per cent higher, iu three sam- 

 ples of buttermilk 0.17 per cent lower, in one sample of whey 0.07 per 

 cent higher, and in four samples of cream 0.02 per cent lower than the 

 gravimetric determinations. " Several tests have been made with both 

 condensed milk and with cheese, which have given a clear separation of 

 fat." In three such tests the new method gave results averaging 0.12 per 

 cent lower than those of the gravimetric method, and in one sample of 

 condensed milk 0.18 per cent higher. 



'' Two samples of milk may be tested in duplicate iu fifteen minutes, 

 including all the work, from the mixing of samples to the cleaning of 

 the bottles. After the milk has been measured sixty tests ma^' be 

 made in less than two hours, including the cleaning of the bottles. 

 * * • The cost of the test will depend upon the price of commercial 

 sulphuric acid. One pound of acid is sufficient for fourteen tests. In 

 a factory where acid is purchased by the carboy it should not cost more 

 than one fifth cent per test, and in no case should it cost more than one 

 half cent per test. The breakage of bottles, if properly made, is so 

 slight that it need not be considered." 



