68 PROTEIN POISONS 



presumably protein, was filtered out after standing twenty- 

 four hours, and was so small that it was lost on the filter 

 paper. 



The filtrate was then poured into three volumes of 95 

 per cent, alcohol acidified to the extent of 0.5 per cent, with 

 hydrochloric acid. This formed a heavy, white, curdy, 

 fibrous precipitate, which was filtered, washed acid-free 

 with alcohol and then with ether. It was purified by 

 repeated solution in 0.5 per cent, alkali and precipitation 

 with alcohol. Finally there was obtained a fine white 

 powder amounting to something less than 10 per cent, of 

 the original cellular substance, but much had been lost by 

 its partial solubility in dilute alcohol. This powder consists 

 almost wholly of a carbohydrate which is converted into 

 a reducing sugar after prolonged boiling with dilute mineral 

 acid. However, it contained 5.9 per cent, of ash and 0.194 

 per cent, of phosphorus. Solutions of this powder give 

 none of the protein reactions, with the- exception of the 

 xanthoproteic, to which they responded imperfectly. 



Leach prepared the same body, but with a higher phos- 

 phorus content, from the colon bacillus. The cellular 

 substance, after repeated extraction with dilute (1 to 5 

 per cent.) sulphuric acid, was extracted upon the water- 

 bath or over a free flame with from 2 to 4 per cent, of 

 sodium hydroxide. *The alkaline extract, after filtration, 

 was neutralized with hydrochloric acid and poured into 95 

 per cent, alcohol. A light colored, flocculent precipitate 

 was obtained. This turned dark on the exposure to air 

 incident to filtration. It was twice dissolved in 0.5 

 per cent, potassium hydroxide and reprecipitated with 

 acidified alcohol. Each time the fresh precipitate was 

 white or nearly so, but the utmost care in filtering, even 

 in an atmosphere of carbon dioxide, did not prevent its 

 turning dark. The solution in alkali gave the xantho- 

 proteic and furfurol tests, but neither the biuret nor Millon 

 test. Copper chloride gave a precipitate, but picric acid 

 and platinum chloride did not. The solution was accord- 

 ingly acidified with picric and acetic acids, copper chloride 



