( 946 ) 
to alter the measuring apparatus in any way during the whole 
series of determinations. It was also of the highest importance 
that the densities necessary for the calculation of the diameter 
should be obtained with the same accuracy over their whole 
course, while, on the one hand, the largest quantity of gas which 
we could use for each measurement was limited by the dimensions 
of the volumenometer, and on the other hand, the accuracy of the 
measurements at the highest temperature was limited by the degree 
of constancy of the temperature itself. Moreover, there was no 
necessity for obtaining a greater degree of accuracy for individual 
points than that which is possible by this method and which may 
be represented by a unit in the third decimal place. This accuracy 
is sufficient to determine if the diameter for oxygen is to be consi- 
dered as straight, and to deduce its slope with the same accuracy as 
that with which it has been determined for other substances. In any 
case this first step was desirable in order that data might be obtained 
to serve for the calculation of the other apparatus mentioned, and to 
be useful in leading to a greater degree of accuracy. For the degree 
of accuracy desired it was seen that it was not necessary to determine 
vapour densities at the lowest temperatures, for it may be assumed 
that these densities can be calculated with sufficient accuracy. 
The dilatometer is seen in position in the cryostat Cr in the 
diagrammatic plan shown in Plate I, its exact dimensions are given 
on Plate II, fig. 2. The method of joining the glass capillary dy to 
the steel capillary d, of about 0.6 mm. diameter is described in 
Comm. N°. 69 (April 1901) Pl. I, fig. 4. The piece dy is soldered 
to the glass in the manner described in Comm. N°. 27 (June 1896); 
its end surface is ground perpendicular to the stem; instead of a 
leather ring steeped in wax a fibre ring was used as packing. This 
connection as well as all other metal connections and taps are kept 
under test by immersion in oil — a method introduced in Comm. 
N°. 944, June 1905; this is done by means of a tube attached at 
a to the cover of the cryostat in which the dilatometer is placed 
(see. Pl II. fig, 2:and Pi). 
§ 8. Method Il. Volumenometer. The mass of the gas was 
determined volumenometrically. We used the accurate volumeno- 
meter deseribed in Comm. N°. 84 (March ’03). Since then it has 
undergone some modifications for experiments by KAMERLINGH ONNES 
and pr Haas on the compressibility of hydrogen vapour at and 
beneath its boiling point, the results of which will soon be published 
The reservoir for the preparation of mixtures (Comm. N°. 84, PI. II, 
ww 
