104 PROTEIN POISONS 



being a protein itself or as being mixed with one or more 

 proteins. Its aqueous solutions give all the protein color 

 reactions with the important exception of that of Molisch. 

 It is worthy of note that the part that separates from 

 alcoholic solution on long standing is inert and does not 

 give the protein reactions, while the solution does not 

 decrease in toxicity. This indicates that the protein is 

 permanently soluble in absolute alcohol. The Millon 

 reaction shows most perfectly and persistently whenever 

 the poison is found. It is generally believed by physio- 

 logical chemists that this reaction is given by all benzene 

 derivatives in which one hydrogen atom has been replaced 

 by a hydroxyl group, and it is also generally supposed that 

 tyrosin is the only oxyphenyl compound in the protein 

 molecule, therefore this reaction is presumed to show the 

 presence of tyrosin. This is interesting in view of the 

 fact already stated that gelatin, which contains no tyrosin, 

 or but little, yields no poison. The fact that the poison 

 contains no carbohydrate, as shown by its failure to respond 

 to the Molisch test, an exceedingly delicate test, is, in our 

 opinion, strong evidence that the cleavage in the protein 

 molecule induced by dilute alkali in absolute alcohol at 

 the temperature of 78 follows along structural lines. If 

 the change were one of simple degradation without chemical 

 cleavage it w r ould be difficult to explain the absolute failure 

 of the carbohydrate test in the crude soluble poison. It 

 seems quite evident from our work that in the process the 

 complex protein molecule is split into several groups, one 

 of which is the poison and another is a carbohydrate, the 

 former being freely soluble in absolute alcohol, while the 

 latter is insoluble in this reagent. It should be stated that 

 the crude, soluble poison not only fails to respond to the 

 Molisch test, but it also fails to reduce Fehling's solution 

 after prolonged boiling with dilute mineral acid. 



The crude soluble poison gives the biuret test beauti- 

 fully, therefore we must say that the poison either is itself 

 a biuret body or is mixed with such a body. As is well 

 known, the biuret test is regarded as the landmark between 



