of EdMurgli, Session 1882-83. 
227 
smaller compression apparatus, which was constructed expressly to 
admit of measurements of temperature by thermo-electric processes. 
I had therefore to work with the huge Fraser gun employed for 
the “Challenger'’ work, and to use the protected thermometers 
(which are very sluggish) for the measurement of temperatures. 
It was also necessary to work with the gun at the temperature 
of the air, — it would be almost impossible tc keep it steadily at a 
much lower temperature, — so that I had to work in water at about 
12° C. 
The process employed was very simple, A tall cylindrical jar 
full of water had two “ Challenger ” thermometers (stripped of their 
vulcanite mounting) at the bottom, and was more than half- filled 
with fragments of table-ice floating on the water, and confined by 
wire-gauze at the top. This was lowered into the water of the gun, 
and pressure was applied. 
It is evident that if there loere no conduction of heat through the 
walls of the cylinder, and if the ice lasted long enough under the 
steadily maintained pressure, the thermometers would ultimately 
show, by their recording minimum indices, the maximum density 
point corresponding to the pressure employed : — always provided 
that that temperature is not lower than the melting point of ice at 
the given pressure. 
Unfortunately, all the more suitable bad conductors of heat are 
either bodies like wood (which is crushed out of shape at once 
under the pressures employed) or like tallow, &c. (which become 
notably raised in temperature by compression). I was therefore 
obliged to use glass. The experiments were made on successive 
days, three each day, with three different cylindrical jars. These 
had all the same height and the same internal diameter. The first 
was of tinned iron ; the second of glass about J inch thick ; the 
third, of glass nearly an inch thick, was procured specially for this 
work. 
With the external temperature 12° '2 C., the following were the 
results of IJ tons pressure per square inch, continued in each case 
for 20 minutes (some unmelted ice remaining on each occasion). 
The indications are those of two different “ Challenger ” thermo- 
meters, corrected for index-error by direct comparison with a Kew 
standard ; — 
