254 PROCEEDINGS OF THE AMERICAN ACADEMY 



ends, which, when the cooling took place rapidly, appeared in cir- 

 cular groups of little needles. These two modifications differed 

 entirely in crystalline habit and color, and resembled strongly the 

 yellow and red forms of anilidotrinitrophenyltartronic ester, the 

 discovery of which had led us to undertake this work; but whereas 

 the two esters showed a difference of 21° in their melting points 

 (red 143°, yellow 122°), the yellow and red forms of the trianilido- 

 dinitrobenzol melted at the same temperature, 179°. To be sure, 

 the yellow form turned red at about 140°, but there were no signs 

 of melting, and we do not feel that such a change from yellow to 

 orange-red is definite enough to have much significance. The yel- 

 low form was also much less stable than the corresponding form of 

 the anilidotrinitrophenyltartronic ester, so that we have not ob- 

 tained it absolutely free from the orange modification. The best 

 way for preparing it that we found was to crystallize the orange 

 form from a mixture of benzol and alcohol containing much benzol, 

 when a portion of the substance usually appeared in the yellow 

 form. On the other hand, a single crystallization of the yellow 

 crystals from a mixture of benzol and alcohol containing but little 

 benzol was sufficient to convert them completely into the orange 

 prisms. As it was possible that the yellow crystals might be a 

 compound containing alcohol or benzol of crystallization, instead 

 of an isomere, we heated to 100° 0.3190 gr. of the best we could 

 obtain, which had been air dried, but found that the loss was only 

 0.0004 gr., showing that this is not the explanation of the occur- 

 rence of this form. 



The presence of a small amount of impurity seemed to be favor- 

 able to the existence of the modification ciystallizing in plates, as, 

 if a little tribromdinitrobenzol was present, crystals were obtained 

 of this form although usually a little more orange in color than that 

 made from pure material, and these would undergo several crys- 

 tallizations before they were converted into the orange form. A 

 sample, which we obtained accidentally, was even more stable, and 

 owed this stability to an oily impurity, the presence of which was 

 indicated by the low melting point, and which we finally succeeded 

 in separating, but only after a great number of crystallizations, 

 when the substance passed into the orange form; the amount of 

 this impurity, however, was very small, as shown by the following 

 analyses : — 



I. 0.2522 gr. of the substance gave on combustion 0.G004 gr. of 

 carbonic dioxide and 0.1086 gr. of water. 



