462 
Proceedings of the Royal Society of Edinburgh. [Sess. 
junction, too, the determination of the volumes is rendered much more 
difficult than with a form in which no such junction occurs. To avoid 
the junction, and at the same time have a series of observations for one 
immersion of the dilatometer, two other forms were tried. In both forms 
the dilatometer was made entirely of quartz-glass, and had the open limb 
comparatively long and wide. In one of them, contacts connected as 
before were fused into a thin glass rod, which was inserted in the long 
limb of the dilatometer, and the mercury, as it rose between the rod and 
the walls of the dilatometer, gave galvanometer deflections. This proved 
unsatisfactory, owing mainly to capillary effects. In the other case, a fine 
wire was stretched in the long limb of the dilatometer, and the change in 
its resistance with the rise of the mercury was measured. Here, however, 
the temperature variations made the resistance measurements practically 
useless for this purpose. 
Finally, then, I had to have recourse to a form which had early 
suggested itself, but which did not lend itself to more than single measure- 
ments for a given pressure. The figure * shows the form with which the 
majority of the observations were made. 
The dimensions were approximately as follows : — 
Part. 
Length. 
External Diameter. 
Internal Diameter. 
cm. 
cm. 
cm. 
AB 
2-6 
0-8 
>0-50 
BC 
5-5 
0*4 
<0-10 
CD 
1-9 
0-6 
0-34 
AD 
10-0 
DF 
9-0 
6-5 
0-25 
FG 
9-5 
0-6 
0-34 
EC is made with small diameter, to prevent too great conduction of 
heat through the mercury. H and H' are two stout platinum wires 
insulated from one another in FG by quartz-glass tubing except at F, at 
their lowest points, where GF becomes narrower, and connected with a 
battery and galvanometer circuit. They are held in position by being 
bound with thread to the excrescences M and M'. At the commencement 
of an experiment we have water from A to K and mercury from K to E, 
the volume EF being taken approximately equal to the volume KB. 
When the volume AK of the water has increased to AB, the level of the 
mercury is at F, and this is shown by a deflection on the galvanometer. 
The method therefore bears a certain resemblance to that used by Richards 
for compressibility. 
* Fig. 3d. 
