4 PROCEEDINGS OP THE AMERICAN ACADEMY 



III. 0.1075 grm. gave 0.2780 grm. of carbonic dioxide 



IV. 0.1926 grm. of substance gave 12.4 c.c. of nitrogen at a temper- 



ature of 20° and a pressure of 740 mm. 

 V. 0.221 1 grm. of substance gave by the method of Carius 0.103C 

 grm. of argentic bromide. 

 VI 0.2005 grm. gave 0.0937 grm. of argentic bromide. 



Found. 

 I. II. III. IV. V. VI. 



Carbon 38.51 38.30 38.38 



Hydrogen 3.24 2.83 



Nitrogen 7.17 



Bromine 19.91 19.88 



These results agree about equally well with the numbers calculated 

 from two different formulas, as appears by the following comparison. 



The first of these formulas, in which the two atoms of bromine 

 removed from the tribromdinitrobenzol have been replaced by the 

 bivalent radical =C(C0 2 C 2 H.) 2 , seems at first sight the most probable, 

 especially since W. II. Perkin, dr. and others have observed the 

 strong tendency of sodium malonic ester to react in this way ; but 

 although this is the only formula we have been able to find which ex- 

 plains easily the removal of two atoms of bromine in this reaction, we 

 are inclined to reject it in favor of the second one, in which one atom 

 of liromine has been replaced by the univalent radical -CH(C0 2 C 2 H 5 ) 2 , 

 and the other by hydrogen, for the following reasons: — 



First, the formation of a sodium salt is hard to reconcile with the 

 lirst formula, unless, indeed, sodic hydrate was added to it, t giving 



CTI15r(N0 11 ) li OHCNa(C0 2 C 2 TI,) 2 ; 



but our analyses of the sodium salt, given later in this paper, prove 

 that it contains no hydroxy], and to make this proof more convincing 

 we prepared the suit, for one of our analyses from sodic ethylate, as 



* Tin* water was lost in tins analysis, 

 t Wislicenus, Ann. Chem., ccxlii. 23. 



