OP ARTS AND SCIENCES. 33 



by the direct addition of fluoride of silicon to the base, and also by the 

 strong resemblance in their properties, the differences observed being 

 such as might well occur among substances belonging to the same 

 class. 



Upon considering, in general, the way in which the fluoride of sili- 

 con could attach itself to a base, we have been able to find only three 

 probable methods, which we will proceed to discuss as applied to our 

 compounds, (a.) By replacing the hydrogen of the amido group, 

 forming a substance which would be at once an anilid and a fluosili- 

 cate. This method would seem at first sight the most probable, es- 

 pecially since A. Harden * has found that chloride of silicon gives 

 with aniline SiCl 2 (NHC G H.) 2 and aniline chloride ; but this mode of 

 union is impossible, since both chinoline and dimethylaniline, which 

 contain no hydrogen attached to their nitrogen, form compounds of 

 this class, (b.) By the action of the fluoride of silicon on the benzol 

 ring, forming a substance analogous to pararosaniline fluosilicate. 

 This hypothesis, which is improbable on account of the ease with 

 which the substances are broken up by water, is rendered entirely 

 inadmissible by the formation of the dimethylamine compound, which 

 contains no ring, (c.) On the supposition that the fluoride of silicon 

 combines with the base to form a sort of salt, this view is the only 

 one compatible with our results, and its correctness is confirmed by 

 the observation that all the substances tried formed salts with one 

 exception, tribromaniline, and that this was the only one which did 

 not form a silicotetrafluoride ; further, the stability of the silicotetra- 

 fluorides keeps pace with the stability of the salts of the bases, those 

 like aniline, the two toluidines, parachloraniline, chinoline, and dime- 

 thylamine, which form stable salts, giving silicotetrafluorides, which can 

 be sublimed without decomposition, whereas diphenylamine and di- 

 methylaniline gave compounds decomposed by heat. That the salts of 

 diphenylamine are unstable, being decomposed by water, is well 

 known, and although we have not been able to find any published 

 statement about the salts of dimethylaniline that would imply they are 

 unstable, our own work has furnished the proof that the fluosilicate at 

 least is less stable than that of aniline, as only the products of the 

 decomposition of the fluosilicate were obtained, when water was added 

 to the tridimethylaniline disilicotetrafluoride. 



We infer, then, from the arguments given above, that the nitrogen in 

 the silicotetrafluorides is in the quinquivalent condition, and think it 



* J. Lond. Chem. Soc, 1887, i. 40. 



VOL. XXIII. (N. S. XV.) 3 



