19, 1 Perkins: The Octet Theory of Valence 15 
From what has been said one would expect the common va- 
lence of these atoms to be about 8, and the maximum to be 8 in 
- Sa and all atoms below it in Table 1. The actual valences, how- 
ever, have been successfully explained by Langmuir ** on the 
basis of the stability of certain partly completed shells. The 
shells of Ni, Pd, Er, arid Pt, can have a stability remotely resem- 
bling that of the inert atom shells, but only when rearranged in a 
form (the 8 form) not stable except when surrounding a kernel 
more highly charged than the kernels of these respective atoms. 
Therefore, some of the atoms somewhat below Ni, Pd, Er, and 
Pt in Table 1 tend to lose electrons until they have a pseudo- 
kernel of the form B-Ni, B-Pd, 8-Er, and B-Pt. This is made 
possible only by the rearrangement just mentioned, and there- 
fore does not affect any atoms above these in the table. 
There are probably slight electronegative tendencies in some 
of this large number of atoms, that is, forces tending toward 
completion of certain stable arrangement of shell electrons. 
Any such forces are so weak, however, that we have no evidence 
of them except, perhaps, in a few compounds like Na,ZrF,, 
[Co(NH,),C1JCl,, [Pt(NH,),Cl,]Cl,. 
It is usual to make somewhat larger groups of the atoms, but it 
seems to the writer that the grouping in Table 1 shows most 
plainly the relations between the structure of atoms and their 
chemical properties. The partial relation between such groups 
as the chromium and sulphur groups has already been pointed 
out, and it is easily seen that the partial resemblance of f-Ni, 
B-Pd, B-Er, B-Pt to the inert atoms causes a number of partial 
similarities, such as those between the copper and lithium groups 
and the zine and beryllium groups. 
EXAMPLES 
vs = gigi i 
Ti 2==5, Qc Cr ==-- QO ) <°22 4 
- a ea : 
. = sie a 
i os " + 
K+---------- a NF i gt 
Rts Oo ts Grea RAs YS, 5 
1 Os ie Le 
i. 
oe a o-- 
*Tbid. p. 876. 
