METHODS OF ESTIMATING CARBOHYDRATES 43 



tions were made volumetrically and without use of the 

 polarimeter, and so glucose and fructose are classed together 

 as hexoses. The method of clarification adopted, too, was not 

 thorough . However, as the very interesting results obtained 

 depend for their significance on fairly large carbohydrate 

 fluctuations, they are probably substantially accurate. 



Davis and Daish, in continuing the mangold experi- 

 ments, began with a critical examination of the methods 

 employed in carbohydrate estimations. They tested the 

 tables of Brown, Morris, and Millar, employing pure and 

 carefully dried sugars, and obtained agreement with the 

 tables to within 1 milligramme on the weight of copper 

 used, which the authors regard as the probable degree of 

 accuracy of their method. 



In doing this they prepared their asbestos much in the 

 manner recommended by Munson and Walker as described 

 in C. A. Browne's " Handbook of Sugar Analysis/' Special 

 attention was drawn to the seriousness of the errors arising 

 from neglect of this precaution. Davis and Daish, however, 

 do not seem to have been aware of Munson and Walker's 

 procedure, and arrived at it independently. 



Their manner of weighing the copper as cupric oxide, 

 after heating for half an hour in a Gooch crucible protected 

 from reducing gases by a large outer crucible of porcelain 

 or nickel, is both accurate and very convenient. It was 

 adopted by Wilson and Atkins in testing Kendall's (1912) 

 sugar tables. The agreement obtained with the tables was 

 very close, though Kendall estimated the copper by means 

 of his modification of the iodide volumetric method (1911). 



With regard to volumetric methods, Davis and Daish 

 criticize that of Bertrand adversely both on the score of 

 accuracy and rapidity. That of Ling and Rendle (1905), 

 however, was found to be accurate to at least 0*3 per cent., 

 and also quite rapid. 



