19,6 Perkins: Chlorine Dioxide and Compounds 731 
and of the pair and surrounding octet in F and the heavier 
atoms . 6 
Explanation 2 is less objectionable than 1 because atoms 
can exist under certain conditions with incomplete shells. When 
we consider, however, the high temperatures necessary to pre- 
vent such atoms as 0 and Cl from completing their octets by 
combination, and the marked reactivity which these atoms show 
in the nascent condition, it seems that C10 2 , though reactive in 
some ways, is not in the same class. Nor are there indications 
that CIO., contains an extra electron. Therefore explanation 2 , 
while valid in ordinary dissociation at very high temperatures, 
does not apply to this case. 
Explanation 3, namely, that five electrons are shared in 
C10 2 seems to be the most reasonable. This is not inconsist- 
ent with the remarkably successful theory of pair sharing if 
we assume that there is a strong tendency to share in pairs, 
but that under certain conditions one or three electrons may 
be shared. To explain CIO. on the basis of this assumption it is 
necessary to show why this substance, containing the inherently 
unstable three-electron bond, is formed in reactions which would 
be expected according to the shared-pair theory to give CLO s 
or C1,0 4 . 
The question of the structure of CIO. thus resolves itself 
largely into the reasons for the instability of C1,0 4 and C1,0 3 
which brings us to one of the serious limitations of the octet 
theory. That is, the octet theory accounts for not only many 
compounds which exist, but also an enormous number which 
apparently do not and cannot exist. Langmuir has recently 
8 The only convincing explanation of the properties of N 2 , CO, CN, and 
NO which has come to the writer’s attention has been Langmuir’s assump- 
tion (loc. cit 903) that in these molecules a double kernel is formed sur- 
rounded by a single octet. The odd electron in NO is in all probability 
very close to the center of the line joining the two nuclei, which is prac- 
tically the central point of the octet. If each kernel even in NO definitely 
contains two electrons, as Langmuir’s theory would indicate, the odd 
electron is one of three binding electrons situated between the kernels. 
It is also conceivable, as is pointed out on page 739, that under these 
conditions a group of four is (by magnetic or unknown forces) stable 
around each nucleus, only one electron being shared by the two groups. 
In any case the conditions are entirely different from those in a single 
kernel surrounded by an octet. The determining forces in C10 2 must be 
very similar to those in neon and argon, whose kernels can take up or lose 
an electron only under the influence of a powerful impact, and then only 
momentarily. 
