732 Proceedings of Roy al Society of Edinburgh. [sess. 
saturated type. The derivatives of the dibasic acids have the 
general formula M 3 'M"'X 3 '', or 
M"' : (X".M') 3 , 
(where M'", M', are tri valent and univalent metals respectively, and 
X" is the radical of the dibasic acid). Those of the monobasic 
acids have the general formula M 3 'M'"X 6 ' , which may be written 
m"' ; ([x' 2 ].m') 3 , 
when the analogy is at once seen. 
With regard to the question as to how these ‘univalent’ 
radicals are themselves united, Kohlschuttcr * has pointed out 
that if Blomstrand’s and Remsen’s views are correct, such com- 
pounds as FeClg , 3KC1 must be written 
C1.C1.K 
Fe^Cl.CLK 
X31.C1.K . 
This assumption of the single bond between the halogen atoms 
does not explain the constitution of such compounds as that pre- 
pared by Jones and Knight, f possessing the composition 3RH 4 Br , 
ZnBr 2 , whereas, by the assumption of a double bond between these 
“univalent” radicals, concordant constitutions can be given to all 
their derivatives at present known, and the existence of such 
compounds as iodine trichloride IC1 3 agrees with this view. 
Whatever is true in this respect of one of the halogens is 
evidently true of all. This is shown especially by a reference to 
Christensen’s papers, J by the comparison of the complex halides 
of thallium, bismuth, etc., and finally by the existence of a series 
of compounds such as 
BiBr 3 , 3XH 4 C1. 
It will be noticed that the series of double halides belonging to 
the type A. 2 can equally well be represented by formulae half 
those used in the table, as for example 
F.Cr : (F 2 .K) 2 . 
It is not impossible that these compounds may be found capable 
of existing in both conditions, but I have chosen the double 
formulae in order to bring out more strikingly the relation 
between them and those derived from dibasic acids. 
* Ber , , 35, p. 483. t Amer. Chem. J., 22, p. 136. 
t Loc. cit. 
