OF ARTS AND SCIENCES. 175 



the case, unfortunately at an expense of time and labor entirely out 

 of proportion to the value of the results obtained. 



Our experience with sodic ethylate induced us at first to try the 

 action of sodic methylate in the cold ; but as we soon found that the 

 same products were obtained when the action was assisted by heat, 

 we have usually proceeded in this way, since, when boiling, the re- 

 action runs to an end in ten minutes, whereas in the cold it takes 

 about two days. Finally we adopted the following method. 10 gr. 

 of tribromdinitrohenzol dissolved in about 20 c. c. of benzol were 

 mixed with a solution in methyl alcohol of the sodic methylate from 

 1.7 gr. of sodium, and the mixture boiled for ten minutes on the 

 steam bath in a flask fitted with a return condenser, when a clear 

 red solution was obtained, which on cooling deposited white crystals. 

 These were filtered out, the filtrate allowed to evaporate to dryness, 

 and the residue and crystals extracted three or four times with small 

 amounts of hot alcohol, as in this way a viscous impurity was re- 

 moved, the nature of which will be considered later. The purifi- 

 cation of the less soluble portion offered great difficulties, which was 

 the more surprising because it was but slightly soluble in any solvent, 

 so that the separation of it from the viscous product was very easy, 

 and by crystallization from hot alcohol we soon obtained a substance 

 showing the constant melting point 237°-238° ; but the results of its 

 analysis did not agree with those calculated for bromdinitroresorcine 

 dimethylether, as is shown by the following comparison: — 



Found. 



32.78 



3.21 



10.64 



25.02 



As these analyses seemed to indicate that the substance was not 

 pure, we crystallized it again several times, using glacial acetic acid 

 as the solvent ; and after a new set of analyses had yielded no 

 better results than those given above, we recrystallized six or seven 

 times, this time from acetone, but with no better agreement of the 

 percentages derived from analysis with those corresponding to the 

 formula. As the substance crystallized very well, we were inclined 

 to think at first that we had a compound different from the brom- 

 dinitroresorcine dimethylether, probably with a higher molecular 

 weight. Accordingly, we tried to determine the molecular weight 



