On the Compressibility of Water. 351 



mained so ; but the cementation was unaltered, excepting 

 at the surface, where it had become a little concave. 



Being satisfied that the piezometer, as first constructed, 

 would not show all the compression, I determined to make 

 one differently modified. The object was, to avoid the fric- 

 tion occasioned by the collapsing of the leather upon the rod, 

 under such great pressure. The drawing in Plate XXXII, 

 (PI. 3,) Fig. 3, shows another modification of the piezome- 

 ter, made since I have been in this country. This proves 

 my suspicions to have been correct; since, under the same 

 pressure, it indicated nearly double the compression shown 

 by the former, [d) 



This instrument is constructed as follows. Fig. 3 being a 

 section of it. It is simply a small tube, A, closed at the 

 end, B, and water-tight. At the upper end, C, the water is 

 allowed to enter through a small aperture, E, closed by a 

 very sensible valve, opening inwards. The tube is flattened 

 at D, in order that it may yield to the expansion of the 

 water, when taken out of the press. 



(d) It seems scarcely credible, that the whole difference between the re- 

 sults of these two sets of experiments could be owing to the friction of the rod, 

 employed in the first set, against its stuffing box. The friction, on this sup- 

 position, must have been sufficient to counteract an excess of pressure on 

 the outer end, of fifty-seven pounds. Nor is it probable that any error, 

 which may be supposed to have arisen from the neglect to ascertain the tem- 

 perature of the compressed water in the first experiments, would be sufficient 

 to account for the difference. The temperature of the water in the cannon 

 was undoubtedly raised, by a diminution of specific heat, analogous to what 

 takes place in other cases of compression ; but, unless the experiment was 

 completed in a very few moments, its sensible heat must have been reduced 

 nearly to a level with that of the surrounding air. We are not informed 

 what time was allowed for the equilibrium of temperature to be restored, 

 nor what the temperature of the water was at the commencement of the 

 experiment; both which circumstances are necessary to an accurate esti- 

 mate of the error which might have arisen from the neglect to ascertain the 

 temperature after compression. If, however, we suppose the previous teni' 

 perature to have been 50°, and the rate of expansion for different tempera- 

 tures to be the same under all degrees of pressure; the elevation of the 

 water in the piezometer ou compression, as is easily inferred from the inves- 

 tigations of Deluc, must have been no less than 26°, to produce an increase 

 pf 2-^g- in bulk. Is it not more probable than an error was committed in 

 estimating the pressure ? The valve, C, Fig. 1, is not particularly described; 

 but it would require to be constructed with peculiar accuracy not to indicate 

 too great a pressure, in consequence of the water insinuating itself between 

 the top of the cap and the part of the valve which rests upon it, before it 

 sensibly lifts the valve. The lever too, by which the valve is held down, 

 unless made of great strength, would prove a source of deception, by its lici> 

 bility to spring. 



