228 Proceedings of the Royed Society 
Tin Cylinder, Thin Glass. Thick Glass. 
4° C. 2° -67 0°-83 
4° 2° -61 0°-83 
The coincidence of the first numbers with the ordinary maximum 
density point of water is, of course, mere chance. When no 
pressure was applied, but everything else was the same, the result 
was 
Tin. Thin. Thick. 
5°7 C. 5° 4“ 
It is clear that the former set of numbers points to a temperature of 
maximum density, somewhere about 0° C., under tons pressure 
per square inch. But still the mode of working is very imperfect. 
I then thought of trying a double cylindrical jar, the thin one 
above-mentioned being enclosed in a larger one which surrounded it 
all round, and below, at a distance of about f inch. Both vessels 
were filled with water, with broken ice floating on it, and had 
Challenger ” thermometers at the bottom. By this arrangement I 
hoped to get over the difficulty due to the temperature of the gun, 
by having the inner vessel enclosed in water which would be 
lowered in temperature to about 3° C. by the application of pressure. 
The device proved quite successful. The result of IJ tons pressure 
per square inch maintained for 20 minutes, some ice being still left 
in each vessel, was from a number of closely concordant trials — 
Temperature in outer vessel, . , 1°‘7 C. 
Temperature in inner vessel, . . 0°'3 C. 
The direct pressure correction for the thermometers is only about 
— 0°T C., and has therefore been neglected. 
The close agreement of this result with that obtained (under 
similar pressure conditions) in the thick glass vessel leaves no 
doubt that the lowering of the maximum density point is some- 
what under 4° C. for 1 J tons, or 2° 7 C. for 1 ton, per square inch. 
It is curious how closely this agrees with the result of my indirect 
experiments. 
The fact, that an increasing compressibility of water at lower 
temperatures points to a lowering of the maximum density point by 
pressure, is easily seen from the consideration of the surface which 
represents the density of water in terms of pressure and temperature 
as independent variables. Let temperatures in degrees C. be 
