810 
Proceedings of the Royal Society 
I. The galvanometer is adjusted, and its delicacy, that is the 
value of its scale-reading for one degree centigrade of difference of 
temperature of the thermo-electric junctions, tested. 
II. A series of experiments are then made, each of the folloAving 
nature : — The vessel of water containing one thermo-electric junction 
is subjected to a certain pressure, known and recorded. While 
under pressure, the junctions are put in connection with the galvano- 
meter, and the deflection noted; we thus get the difference of 
temperature between the internal and* external junctions. The 
actual temperature of the external junction is simultaneously read 
on a very delicate and accurate thermometer. Then the pressure is 
let off, and the heating or cooling of the water, as shown by the 
junction inside, observed on the galvanometer. 
III. After a set of experiments made at pressures varying from 
50 to 600 atmospheres, the galvanometer is again tested, and any 
small change allowed for in calculating out the results obtained. 
In many experiments made on water between 0° and 4° cent., 
with pressures under 1 ton weight per square inch (150 atmospheres), 
we got a faint indication of cooling when the pressure was applied, 
but no appreciable change when it was let off. As the question of 
the change of the maximum density point is only attacked in an 
indirect manner in these experiments, we did not look for exact 
quantitative results. Thus, by comparing several days’ experiments, 
we find that in order that water may neither heat nor cool when 
released from 1J tons per square inch pressure, it must have a 
temperature of from 0*2 to 0*5 degrees centigrade. But we do not 
regard ourselves as justified in stating that the maximum density 
point for water at this pressure lies within these narrow limits. 
The diagram below gives our results in "a convenient form. 
Horizontally we have the initial temperatures of the water under 
compression in degrees centigrade. Vertically the numbers repre- 
sent tenths of a degree centigrade, and show the amount of cooling 
which takes place on release. Line I. is drawn through the various 
points obtained by letting off 1 ton per square inch pressure (150 
atmospheres) at different initial temperatures, and noting the effect 
produced. Lines II., III., and IV. are similarly drawn for 2, 3, 
and 4 tons respectively. 
From these curves it is evident that in the case of 4 tons pres- 
