592 MR J. Y. BUCHANAN ON THE 
manometer No. 2, both being at a temperature of 12°5° C., and the following 
results were obtained as the mean of nineteen observations :— 
Mean rise of manometer No.2(A) . : . 41°35 divisions. 
Mean apparent contraction per thousand of water 
in piezometer K, No. 4 (H) . ; : Me reiesifer sm. 25 
But from the results in Table I. we have for the pressure in atmospheres 
P=3'132 x A=3°132 x 41°35, 
= 129°5 atmospheres, 
and the apparent compressibility of water in glass at this temperature (12°5° C.) 
in volumes per thousand per atmosphere, is 
H  5:8782 
M= Po 1995 7 
= 0°04539. 
We see, then, that at pressures up to 240 atmospheres the property peculiar 
to water of diminishing in compressibility with rise of temperature is preserved 
unimpaired, and the amount of change corresponds closely with that found at 
low pressures in the experiments of ReEGNAULT and GRASSI. 
In Table IT. the details of the experiments on the effects of pressure on the 
glass rod are given. The length of the rod from point to point was 75:05 inches, 
at the temperature of the laboratory, 13° C. Its diameter was 0:28”, and was 
very uniform. The weight of the rod was 209°5 grammes. ‘The substance of 
the rod was remarkably homogeneous, there being a complete absence of air- 
threads. 
The micrometers used were, at the east end a photographic copy of 
HLARTNACK’S eye-piece micrometer, and at the west end one of Merrz’s. They 
were both compared, and the value of their divisions, as used, determined 
by comparison with a stage micrometer of Smiru & Beck, obligingly lent to me 
by my friend Dr Witttam Ropertson, who had very carefully verified its 
graduation. It was remarkable as a coincidence that the values of the divisions 
turned out to be identical in both, namely, 0000417”. 
Under A we have the manometric pressures, under B and C the micro- 
metric determinations of the expansion at the east and the west end respec- 
tively, and under D the sum B and C, or the total expansion of the rod. It 
will be seen that while the values of D, or the total expansion, are very con- 
cordant in each series, those of B and C individually are not always so, the 
