26 JOURNAL OF THE 



3.44 per cent., au average of 3.57 per cent. The determinations 



in hay show the modification gives reliable results. In the case 



of cotton seed meal, however, we see that 77 per cent, crude fiber 



is dissolved by the ordinary method, on account of the longer 



contact produced by the slow filtration, being more than 20 per 



cent, of the whole amount. 



For accuracy and, therefore, agreement of results, as well as 



time-saving, this modification seems to recommend itself. 



Rapid Filter. — The asbestos filter gives good results if covered 

 with a layer of glass wool. Care should be taken, however, not 

 to add too much liquid at a time, as this may cause the glass 

 wool to become displaced. Tlie coarse asbestos gives the best 

 results. 



Ladd (N. Y. Ex. Sta. Rep., 1887, p. 378) finds that when the 

 asbestos filter was returned with the residue a larger per cent, of 

 fiber was found, and the two reasons given are that it prevents 

 so thorough contact with the reagents, and that it necessitated 

 lowering of the temperature to prevent bumping. When, how- 

 ever, the substance is first treated with alkali there is no tend- 

 ency to bump in the acid treatment, and consequently no need of 

 lowering the temperature. As to how much the first mentioned 

 cause will affect the results, I have not tested. 



Prevention of Frothing in Boiling. — This can be done by direct- 

 ing a moderate blast of air into the flask through a small glass 

 tube (1.5 mm. diam.). The constant level of the liquid can be 

 maintained by adding boiling water from time to time, or by a 

 reflux condenser. 



Chemical Laboratory, 



Cornell University. 



