B.— CHEMISTRY. 35 



We have already seen that the formation of a co-ordinate link involves 

 the presence of one atom which can act as a donor and another which can 

 act as an acceptor. The donor must have a pair of unshared valency- 

 electrons. The acceptor must have fewer valency electrons than it is 

 capable of holding. This raises the question of the maximum size of the 

 valency group. If we maintain the original octet theory, that the valency 

 group cannot exceed 8, and at the same time hold that every covalency 

 involves two shared electrons, it follows that the maximum covalency 

 cannot exceed 4. The existence of stable compounds such as sulphur 

 hexafluoride shows that this conclusion is false, and hence that one or 

 other of the two assumptions must be abandoned. Some chemists main- 

 tain the octet limit, and explain the existence of atoms with a covalency 

 greater than four by assuming the possibility of a covalent link formed of 

 a single shared electron : they suppose, for example, that in sulphur 

 hexafluoride the sulphur has eight shared electrons, and that two of the 

 fluorine atoms are attached by two electrons each, and the other four by 

 one each. This view seems to me to be untenable. There must be some 

 relation between the mechanism of a link and its behaviour ; if not, it 

 is of little use to discuss the mechanism. Links of single electrons 

 undoubtedly occur in a limited number of compounds of hydrogen, such 

 as [H 2 ] + and the hydrides of boron (B 2 H 6 , &c.) ; but, as we should expect, 

 they are always very unstable. I cannot believe that a substance like 

 sulphur hexafluoride, which is one of the most stable of known com- 

 pounds, and can be heated to a red heat with sodium without 

 decomposition, can contain four such links. I should therefore abandon 

 the limit of 8 for the valency group (as G. N. Lewis has now done), 

 and adhere to the view that in all but a few unstable compounds every 

 covalency involves two shared electrons. On these principles the 

 maximum size of the valency group is twice the covalency or co-ordination 

 maximum. An examination of the structures of known compounds 

 gives strong reason to believe that there is a direct and simple relation 

 between the maximum covalency (co-ordination number) of an atom and 

 its position in the periodic table, and that this depends not on the periodic 

 group but on the period in which it occurs, so that the co-ordination 

 classification runs horizontally, while the normal valency values, as we 

 all know, run vertically. It would take too long to discuss the evidence 

 for this statement, but I may give the conclusions. The maximum 

 covalency of hydrogen is 2 : that of elements in the first short period 

 (lithium to fluorine) is 4 : that of elements in the second short period 

 (sodium to chlorine) and the first long period (potassium to bromine) is 

 6 : and that of the later elements is 8. The maximum number of electrons 

 in the valency group is, of course, twice as great, being 4, 8, 12, and 16, 

 respectively. No physical reason for these facts can as yet be given, but 

 a certain relation can be traced between the numbers and those in the 

 grouplets of the Bohr theory as modified by Stoner and Main Smith. 



The next question is the difference in properties which is to be expected 

 between the normal and co-ordinate covalencies. These are essentially 

 of two kinds. In the first place the co-ordinate links are in general less 

 stable. The stability of a link depends on the work required to break it, 

 or, in other words, on the difference of energy content between the original 



d 2 



