( 945 ) 
to the ordinary temperature in between. To do this if is necessary 
to allow the gas to escape from the dilatometer and then allow 
it to return when the second low temperature has been 
established. For measurements with oxygen the lowest temperatures 
are given by a bath of liquid oxygen, from — 217° C. to — 183° C., 
temperatures between — 183°C. and -— 164°C. are obtained with 
a bath of liquid methane, and temperatures between — 164° C. and 
— 120°C. with a bath of liquid ethylene. In this way one must 
begin with a new quantity of material at least three times. Hence 
some standard points on the temperature scale are given by the 
nature of the bath itself. 
On the other hand a change of apparatus on proceeding from one 
region of temperature to another would be a decided advantage from 
the point of view of obtaining greater accuracy. The expansibility 
of the liquid and the density of the vapour increase rapidly as the 
critical temperature is approached: it is clear that if one did not 
wish to be confined to too small an interval in using the same 
dilatometer over this region, one would choose varying diameters for 
the stem and for the appendix. For this reason we constructed a 
series of dilatometers with different stems and different appendices 
calculated for a series of temperature intervals. Dilatometers that 
had to be used at temperatures nearer the critical had appendices 
and stems of greater diameter than the others. Furthermore, by 
constructing our dilatometers of two parts united by a graduated 
capillary we tried to provide ourselves with as many controls as 
possible, for by this device we should be able to check direct 
measurements made with each of the individual apparatus with one 
or more of the measurements that were made with another. 
In our first experiments we avoided the complications which we 
have just described and which are encountered as soon as one 
attempts to obtain for all the data for the diameter the greatest 
accuracy possible considering the constancy of the cryostat tempera- 
tures and the degree of accuracy with which these can be measured. 
For this reason we made all our measurements both of vapour den- 
sity and of liquid density with the same dilatometer, whether they 
were at very low temperature or in the neighbourhood of the 
critical. Hence this dilatometer was so constructed that it could 
withstand the critical pressure. The great advantage in using a single 
dilatometer for all the determinations lay in the fact that as soon as 
it was in its place in the cryostat, one had only to pour in the 
various liquefied gases necessary for the different baths to be able 
to traverse the whole range of temperature, and that without having 
