1671 
pressure of which was read to half a millimetre on a scale attached 
to the manometer; from this reading the temperature was calculated 
by means of the formula 
369,838 
t= 3 os 
6,98460—log p 
which as shown by Horst *) accurately represents the temperatures 
on the absolute scale according to the temperature-measurements of 
KAMERLINGH ONNEs and his collaborators. In this manner the tem- 
peratures could be determined acurrately to within a few hundredths 
of a degree. 
The density of the liquid air was measured bv means of a 
suitably designed hydrometer of constant weight, the submerged part 
of which could be read with an accuracy of a thousandth, so that 
the densities became known with the same degree of accuracy *). 
The composition of the liquid was not measured by direct 
analysis *), but derived from its density and that of the component 
substances at the temperature of observation *), by assuming that 
no appreciable change of volume is produced by mixing the liquids 5). 
An estimate of the proportion of argon in the mixture was obtained 
by assuming that its mass-ratio to that of oxygen is not appreciably 
modified by condensation or evaporation, an assumption which is 
based on the fact that the boiling point of argon (87°,25 K.) does 
not differ much from that of oxygen (90°,12 K.), at any rate much 
less than from the boiling point of nitrogen 77°,28 K). It is pro- 
bable, that this assumption gives somewhat too high a value for 
1) Comm. 1484. 
2) Following the example of Benn and Kigsirz (Ann. d. Phys., (4), 12, 421, 1903), 
who used density bulbs for their density-measurements of liquid air, a 
series of determinations was carried out, starting from freshly prepared liquid air 
which by evaporation gradually became richer in oxygen. In this manner it was 
found that from + =79° up to 7 = 83° 
f= 0,886 + 0,042 (x — 79) — 0,002 (r — 79)’, 
+ being the normal boiling point. The values obtained in this manner differ little 
from those arrived at by Bean and Kiesrtz. 
3) Analyses of liquid air and of other mixtures of oxygen and nitrogen arising 
from it were carried out by Baty (Phil. Mag. 15), 19. p. 517. 1900) who gives 
a table containing the proportion of oxygen corresponding to different boiling 
points ; the proportion of argon was not determined by him. 
4) for the densities of oxygen, comp. Comm. 117, for those of nitrogen Comm. 
145c, for argon 1314. 
5) The deviations from this rule are not likely to amount to more than 19/, so 
that we may consider this also as the degree of accuracy of the determinations 
of the composition, 
