88 The Philippine Journal of Science i9i« 



DISCUSSION 



In the tests on the Japanese soy beans the ferric chloride test 

 was positive in every case, usually strongly positive ; the Amer- 

 ican, Chinese, and native beans tested gave either negative or 

 somewhat faintly positive tests with ferric chloride. It would 

 appear that the beans from Japan are slightly different from 

 those of the other places cited. Climate, soil, or methods of 

 handling could account for these differences in the tests given 

 by the Japanese and Chinese beans, since the latter are soy. 

 Fermentation alone did not bring about a synthesis of the react- 

 ing substance, as we have shown by experiment on Japanese soy 

 beans, numbers 1, 2, 3, and 4. Brand has proved that malt 

 which undergoes a peculiar kind of fermentation forms a com- 

 pound giving the test. Sherman states that moisture and bak- 

 ing or roasting cause the formation of a similar compound in 

 cereal products. He ignores in his statement the influence that 

 the fermentation process which takes place before the baking 

 of the bread on the addition of the yeast may have on the 

 formation of this reacting substance. Among the breakfast 

 cereals tested, grape nuts, which contains roasted wheat bread 

 and malt, alone gave a positive test. , Puffed wheat and shredded 

 wheat biscuit which had been baked and had had moisture 

 present when this baking process occurred gave negative results. 

 This would indicate that moisture and baking are not sufficient 

 to cause the synthesis of a reacting substance similar to maltol 

 and would throw doubt on the surmise of Sherman that moisture 

 and baking effects its formation. Backe, in his statements that 

 the formation of the color substance is due to the action of an 

 enzyme on farinaceous substances, seems much nearer the truth. 

 As soy beans contain no starch, his statement would need to 

 be broadened to include substances such as soy beans. 



A very interesting test is the positive reaction given by four 

 of the coffee-soy-bean mixtures with Jorissen's reagent. As 

 these are commercial products, we have no way of accounting 

 for the negative results in the other two cases except the sur- 

 mise that the compound was probably volatilized by too long 

 roasting. 



The reacting compound in the beans is volatile in steam, can 

 be sublimed, is soluble in dilute alcohol and in ether, and gives 

 a violet color with ferric chloride. In this respect it answers 

 the requirements for the very generally used ferric chloride 

 color test for salicylic acid, and it appears to be a compound 

 similar in nature and composition to the maltol of Brand and 



