238 PROCEEDINGS OF THE AMERICAN ACADEMY 



The 15.1 grs. of tribromdinitrobenzol which entered into the reaction 

 should have lost 5.96 grs. of bromine. 



Percentasfe of the theoretical amount of bromine removed from tri- 

 bromdinitrobenzol in the form of sodic bromide : — 



I. II. 



99.8"6 98.49 



These numbers agree as closely as could be expected, when the 

 necessary sources of error in the process are considered, and prove 

 that all the bromine removed by the reaction is converted into sodic 

 bromide. 



The second point to be determined was from what source the atom 

 of hydrogen was obtained which replaced the atom of bromine. For 

 this purpose we investigated the oily product of the reaction, and, 

 after we had found that nothing definite could be obtained from it 

 by distillation under ordinary pressures, or with steam, and that sodic 

 hydrate or other alkaline saponifying agents gave most unpromising 

 results, we tried heating it with strong hydrochloric acid in a sealed 

 tube to 130° for 20 hours, as the bromdinitropheuylmalonic ester 

 which was dissolved in the oil would be converted by this treatment 

 into the corresponding phenylacetic acid, with the properties of which 

 we were familiar, and which therefore we thought could be removed 

 easily from the other products of the reaction. The tubes after heat- 

 ing contained an aqueous solution and a small amount of viscous 

 matter, which gradually changed into an amorphous solid. These 

 were separated by filtration, and the filtrate evajaorated to dryness on 

 the water bath, leaving a crystalline residue, which was extracted with 

 cold water to leave behind the bromdinitrophenylacetic acid present. 

 The extract after being evaporated again on the water bath gave crys- 

 tals, w^ich made up about one half the amount of the residue from 

 the first evaporation. They were freely soluble in water or alcohol, 

 nearly insoluble in ether, and could be sublimed at temperatures below 

 their melting point forming glassy white thick needles or prisms, 

 which melted at 185°. These properties prove that the crystals are 

 tartronic acid, as they are identical with those given by Conrad and 

 Bischoff for that substance.* We tried also to confirm this inference 

 by an analysis, but the small amount of the acid which after some 

 unsuccessful experiments remained at our disposal, little more than 

 0.1 gr., prevented us from purifying it thoroughly, as sublimation, 



* Ann. Chem., ceix. 222. 



