428 
to lead to a large decrease of the whole of the attraction, but only 
to that of a part of it. This, then, would again lead to the hypothesis 
that at higher temperatures only part of the attraction is to be 
ascribed to the mutual action of the doublets of constant moment, 
another part being ascribed to a mutual action of the molecules 
corresponding to the mutual attraction of monatomic molecules (ef. 
Suppl. N°. 23 Nr. 84d). The answer to the question as to whether 
treatment on these lines would lead to a still better agreement with 
observation than that obtained inc must, in the meantime, be postponed 
till a later Communication. 
§ 4. Oxygen*) The individual virial coefficients for oxygen were 
taken from Comm. N°. 71, p. 148. 
From Fig. 4 it is evident that the oxygen 
points (oo) lie well upon the curve (——) for 
constant doublets, so that in this particular region 
(0°—200° C.), as far as B is concerned, and sub- 
ject to the reserve of note 1, the behaviour of 
oxygen may be regarded as that of a system 
of rigid spheres of central structure each with 
a doublet of constant moment at its centre. From 
the following data concerning the superposition 
of the diagrams (cf § 3c) we obtain the accom- 
panying results: the point log 7 = 2,6, log by = 
= 6,5 — 10 for oxygen coincides with the point 
log hv = 0,204, £, = 9,628 — 10 on the curve 
for constant doublets, hence : 
UY = 7,71.10—14, bya =0,745.10-3, o = 2,27.10-8, me = 9,47.10—-19. 
On the assumption that each of the poles of the doublet bears 
a charge equal to that carried by a single electron, the length 
of its axis should consequently be one tenth of the diameter of the 
molecule. The oxygen molecule should accordingly be about as large 
as the hydrogen molecule, but the moment of its doublet should 
be about twice as great as that of the hydrogen doublet. 
§ 5. Nitrogen’). The individual virial coeffcients deduced from 
1) The lack of agreement between the observations of KAMERLINGH Onnes and 
BRAAK upon hydrogen and those of Amagar (cf. p. 420, note 2) shows how desi- 
rable it is that new observations should extend our experimental data over a wider 
range of temperature and give a control upon the values of B deduced from 
Amacat’s data for oxygen and nitrogen as well as for hydrogen. In the meantime 
