830 THE VARIABILITY OF COWS' MILK, i., 



F a t. — The percentage of fat in the samples of milk was 

 determined by the method of Rose (1887), as modified by Gottlieb 

 (1890). Milk, mixed with ammonia and an equal volume of 

 alcohol, is shaken up with ether; petroleum spirit is then added, 

 and the mixture shaken up again. After standing, the mixture 

 separates into aqueoiLs and ethereal layers. The levels of the two 

 liquids are read off, an aliquot part of the ethereal layer is 

 evaporated down, and the fat left is weighed. The relative accu- 

 racy of the method depends chiefly on the precision with which the 

 levels can be read. In the present case, the volume of the ethereal 

 layer was about 50 cc, and this volume could be determined to 

 0*5 cc, or about 1%. The difference between duplicate estimations 

 might thus amount to double this value, if the errors in the two 

 estimations were to fall in opposite directions. The actual range 

 of variation observed in six control estimations was 2-3^, a value 

 sufficiently close to the calculated maximum variation. The devia- 

 tion from the mean is half tliis value, and the results are given in 

 the Table to 1 in 100. Ten cc. of milk were used for each estima- 

 tion. The milk was delivered from a pipette delivering a known 

 weight, as described previously. 



L a c t o s e. — The percentage of milk-sugar was determined 

 by the method of Salkowski (1912). In this method, milk is 

 almost saturated with ammonium sulphate, and mixed with an 

 equal volume of a saturated solution of ammonium sulphate. The 

 liquid is filtered, and the specific rotation of the clear filtrate is 

 determined, [ajp being taken as 52 '53'. Jahnson-Blohm (1913), 

 and Rosemann (1914) have shown that the presence of ammonium 

 sulphate, in the concentration used by Salkowski, lowers the optical 

 rotation of a solution of pure lactose by about 2%. Scheibe(1901) 

 showed, liowever, that, in the case of milk, optical methods in 

 general gave higher results than gravimetric methods. This is 

 due to the fact that, in precipitating the proteins, the volume, 

 through which the lactose is distributed, is reduced, and its con- 

 centration is raised. Kretschmer (1913) has shown that the 

 results obtained by Salkowski's method are about 1% higher than 

 those o))tained gravimetrically. 



