418 
exception of helium, but for helium values of B for temperatures 
below the Boyre point are still comparatively uncertain. 
The most important results yielded by the present investigation 
can be summarised as follows. The experimental results concerning 
the second virial coefficient for hydrogen above —— 100° C. (the 
observations reaching + 100° C.) are consistent with the above 
assumptions of Suppl. N°. 245 $ 6 (rigid spheres with constant 
doublets). Below — 100° C. hydrogen exhibits deviations from this 
behaviour which finally become considerable. Below the Borre point 
(the corresponding region of observation is from — 180° C. to — 230° C. 
for H,) hydrogen is found to correspond with argon, and also with 
helium in so far as the experimental data for helium at present 
available allow of any definite conclusion. It appears therefore that 
between — 100° C. and — 230° C., as far as B is concerned the 
thermal behaviour of hydrogen also approaches that of a monatomic 
substance and eventually becomes the same, as was found by EvckEn*) 
to be the case with its caloric behaviour. This conclusion is supported 
by the results for the coefficient of viscosity. 
It was also found that, as far as the second virial coefficient is 
concerned, the thermal behaviour of oxygen between 0° and 200° C., 
as deduced from AMAGAT’s observations*) corresponds with that ofa 
system of rigid spheres of central symmetry, each with a doublet of 
constant moment at its centre. 
For nitrogen, on the other hand, within the same temperature 
region (O° to 200° C., Amacar’s observations) important deviations 
were found from the behaviour of rigid spheres of central structure 
each with an electric doublet of constant moment at its centre. With 
nitrogen in that temperature region, the dependence of 4 upon the 
temperature corresponds to that deduced from the assumption that 
the var pER Waars quantities aw and fy are constant (Suppl. 
N°. 24a § 3); but then, however, the values given by BESTELMEYER 
and VALENTINER for 4 from 81° to 85° K. differ greatly from this. 
§ 2. Method. Logarithmic diagrams were employed for the com- 
parison of the experimental values of 5 with those deduced in 
Suppl. N°. 24 from various assumptions (cf. Suppl. N°. 23, Math. 
Enc. V 10, Nr. 382). For this purpose log By was plotted as a 
function of log T upon transparent squared paper to a scale of 
1 mm. = 0,005. Here, following Suppl. N°. 23, By represents the 
+) A. Eucken. Berlin Sitz.-Ber., Febr. 1912, p. 141. 
2) Cf. p. 428 note 1. 
