122 PROCEEDINGS OF THE AMERICAN ACADEMY. 



As most of these substances were purified by crystallization, too much 

 weight must not be given to these yields, but we can safely infer from 

 them that Reaction II. ran to the largest extent, Reaction I. came next, 

 and Reaction III. took jilace only to a very limited extent. 



These results are capable of a satisfactory theoretical explanation, but 

 before giving this it will be well to consider a parallel case studied by 

 W. H. Warren and one of us,* in which the conditions are simpler and 

 rather, better marked. This is the action of sodic ethylate on sym- 

 metrical tribromtrinitrobenzol CgBr3(N02)3, Br3l,3,5,(N02)32,4,6, which 

 consisted of two simultaneous reactions, first, the replacement of two nitro 

 groups by two ethoxy groups, giving C6Br3NOo(OC2H5)2, and second, the 

 replacement of the three bromine atoms by three ethoxy groups, giving 

 Cg(OC2H5)3(N02)3. These two reactions ran to nearly an equal extent 

 (about 45 per cent of each product), if the solvent was alcohol only.f 

 In attempting to explain these experimental results we must consider 

 the agencies which tend to loosen the nitro groups on the one hand and 

 the atoms of bromine on the other from the benzol ring. We have 

 designated these agencies by letters, and arranged them in the order of 

 intensity, beginning with the strongest. 



A. The loosening effect of the three nitro groups on the three atoms 

 of bromine. This is a strong agency, as shown by the fixct that 

 picrylchloride (in which the chlorine is similarly disposed to the tliree 

 nitro groups) is decomposed easily and completely, even at ordinary 

 temperatures. J 



H. The loosening effect exerted by the three nitro groups on each 

 other. Lobry de Bruyn § has shown that symmetrical trinitrobenzol is 

 converted into dinitroanisol by sodic methylate, even at ordinary tem- 

 peratures ; this loosening effect is therefore a powerful one, but not quite 

 so strong as A, since the action with picrylchloride is complete in a few 

 minutes, whereas the reaction with trinitrobenzol needs several days, if 

 carried on in the cold. 



C The loosening effect of the three atoms of bromine on the threo 

 nitro groups. This is a much weaker action than B, since with the 

 tetrabrombenzol || (Br4l, 2,4,0) the atom of bromine (which stands in the 



* These Proceedings, XXVII. 283. 



t If the solvent was benzol and alcohol, less of the nitro groups were replaced ; 

 but this case need not be considered in this discussion. 

 } These Proceedings, XXXIII. 176. 

 § Rec. Trav. Chim. Pays Bas, IX. 208. 

 II Jackson and Calvert, These Proceedings, XXXI. 132. 



