Nutrition Investigations. 343 



and amounts to maintain a constant body weight. The carbohydrate 

 under investigation was dissolved or suspended in water, and mixed 

 with this basal ration. In the earlier experiments the periods were 

 divided as follows : Fore = 3 days on the basal ration ; mid = 3 days 

 in which some preparation was added, the amount being the same 

 each day; after = 3 days like the fore period. Separation of the pe- 

 riods in the faeces was accomplished by marking with soot or carmine 

 capsules. In all later experiments, two days constituted the fore pe- 

 riod, and a day on the normal diet was included at the beginning and 

 end of the mid period, making thus four days, to insure against any 

 of the material under investigation being carried into the faeces of 

 the after period. 



In several cases, the presence or absence of galactans or mannans 

 in the faeces has been verified by testing the hydrolyzed material for 

 mucic acid or mannose-hydrazone. 



For analysis, the faeces, collected and weighed, were rubbed to a 

 thin mud with alcohol, dried to constant weight on a water bath, 

 weighed air dry, and ground finely in a coffee mill. The portions 

 constituting each period were thoroughly mixed, and from 2 to 5 

 grams taken for hydrolysis, according to the yield of carbohydrate 

 anticipated. The samples were boiled on a reflex condenser with 

 100 cc. of 2 per cent hydrochloric acid, for two hours; or longer if 

 thought to contain a carbohydrate which previous analysis^ had 

 shown to require more time for complete hydrolysis. 



The products of hydrolysis, cooled and neutralized, were made up 

 to 250 cc. and sugar determined as dextrose by Allihn 's gravimetric 

 method. It was found that the copper reduction w^as often very in- 

 complete, and that much more satisfactory results came from clari- 

 fying the solutions with charcoal after making up to volume. Not 

 only were duplicate analyses in closer agreement, but in some cases 

 the yield of cupric oxide was two or three times greater than before 

 this treatment. Owing to the complexity and diversity of the prod- 

 ucts of hydrolysis, results are at best only approximate. 



In experiments with dulse, the pentosans were determined by the 

 phloroglucin method. ^ 



The human subjects were healthy, active young women. Their 

 diet was not weighed, but was kept as uniform as possible. All cel- 



1 Cf. table, p. 317. 



^Cf. Official and Prov'sional Methods of Analysis, Bulletin No. 107 (1907), 

 Bureau of Chemistry, United States Department of Agriculture. 



