ON THE STEREOCHEMISTRY OF NITROGEN. 191 
Since, in general, no isomerism arises, the process probably takes 
place as follows : 
Fig. 14. 
Be SN 
(abeN +dX) (adeN +X) 
giving rise to the same product. 
But in one case at least—namely, that of the a and £ phenyl-benzyl- 
allyl-methyl ammonium salts—different products are formed. It is true 
that the untransformed addition product might be expected to be an 
isomeride, as in the case of the hydrindamine salts, even in compounds of 
the type N a,bX and Na,bcX, and this may account for the labile amor- 
phous products observed by Wedekind and the writer in some instances. 
Such instances are, however, of quite a different order to the stable 
isomerism of the a and f phenyl-benzyl-allyl-methy] ammonium salts ; 
our knowledge is unfortunately insufficient to enable us to form any idea 
of the reason for the unique position of those salts, and the following 
suggestion as to their mode of production is simply tentative. Since 
they are both stable it is better to regard these both as transformed 
products which might arise thus : 
Fig. 15. 
De ; 
DX Hy Ne Dé} H, 
a ern. OH,N 40.111) (C,H;.C,;H,;.CH,N + C,;H,I) 
The benzyl and allyl groups are regarded as the ones which inter 
change their positions with the iodine atom to form the a compound. 
The 3 compound might arise thus : 
Fie, 16, 
C3Hs 
pe<[- aes 
CeHs CgHs C7H7 
(C,H;.C,H,.C,H,N + CH,I) 
The benzyl radical] is regarded as more mobile than the allyl, which is 
in accordance with observation. In all these cases the addition of the 
