BY H. S. HALCRO WARDLAW. 831 



In the present work, the rotations could be read to 0*02', or a 

 variation of about 1%. The vohnne of the Hquid, in the cylinders 

 in which the precipitations were carried out, could be read to 

 0*5 cc. in 100. The total range of variation is thus 1'5%, which 

 would give a possible difference between duplicate estimations of 

 3*0%, if all the errors in the two cases fell in opposite directions. 

 The difference actually found between duplicate estimation.s was 

 2%. The results given are regarded as relatively accurate to 2 in 

 100, and are given to the nearest 0*05%. As this variation is 

 greater than the correction proposed by Kretschmer, this correc- 

 tion has not been applied in the present case. 



Ash. — The percentages of ash in the samples were deter- 

 mined on the weight of milk delivered from a 5 cc. pipette, cali- 

 brated as described. The samples were first dried at 103°C., 

 and then ashed in a muffle-furnace at a low red heat. In order to 

 determine what loss of chlorides occurred in the ashing carried out 

 in this way, control-determinations were made by charring the 

 milk at a temperature below red heat, leaching out the chloride, 

 ashing the remainder, adding the solution of chlorides, and evapor- 

 ating to dryness. The results by the two methods differed by only 

 0'4% of the weights of ash obtained. The values in the Table 

 are given to 0*01%, or to 1 in 70. 



The substances left, after the estimation of fat, sugar and ash, 

 consist of protein, citric acid(0"2%, Scheibe, 1891), and substances 

 containing about 5% of the total nitrogen of the milk. Of these 

 residual substances, the protein forms more than 90% (Munk, 

 1893; Camerer and Soldner, 1893). Numerous attempts were 

 made to estimate the protein directly by precipitation with alcohol, 

 according to the method previously described for human milk 

 (Sikes, 1906; Wardlaw, 1914). With cows' milk, however, the 

 precipitate obtained in this way contains a very high percentage 

 of ash (over 10%), and in the time available for this portion of 

 the work, it was found to be impossible to dry the precipitate to 

 constant weight. The figures given for the percentage of protein 

 and the remaining constituents of the milk were ol)tained })v sul)- 



