PHYSICAL PROPERTIES OF WATER,, ETC. 5 



have pointed out the untrustworthiness of the Bourdon gauges, and the uncertainty of 

 the unit of my external gauge. This gauge was amply sufficient for all the purposes of 

 my investigation of the errors of the Challenger thermometers, where the inevitable 

 error of a deep-sea reading formed, according to the depth, from 5 to 20 per cent, of 

 the pressure error ; but, besides the uncertainty as to its unit, it was on so small a scale 

 that an error of 1 per cent, in the reading, mainly due to capillary effects at the 

 surface of the mercury column, was quite possible when the pressure did not exceed 

 150 atmospheres. Fortunately I was informed of the great improvement made by 

 Amagat on the principle of the old Manometre Desgqffes, — an improvement which 

 has made it an instrument of precision instead of an ingenious scientific toy. 

 M. Amagat was so kind as to superintend the construction of one of his instruments 

 for me (it will be a surprise to very many professors of physics in this country to hear 

 that the whole work was executed in his laboratory), and to graduate it by comparison 

 with his well-known nitrogen gauge. My measurements of pressure are therefore only 

 one remove from Amagat's 1000 feet column of mercury. 



The change of temperature produced by compression of water is one of the 

 most formidable difficulties I have encountered. During the compression the 

 contents of the piezometer, as well as the surrounding water, constantly change in 

 temperature ; and the amount of change depends not only on the initial temperature of 

 the water, but also on the rapidity with which the pressure is raised. It was impossible 

 to ascertain exactly what was the true temperature of the water in the piezometer at the 

 instant when the pressure was greatest, and a change of even 0°'l C. involves a displace- 

 ment of the hair index, which is quite easily detected even by comparatively rude 

 measurement. Any very great nicety of measurement was thus obviously superfluous. 

 My readings, therefore, were all made directly by applying to the tube of the 

 piezometer a light but very accurate scale. The zero of this scale was adjusted to the 

 level of the upper surface of the mercury of each piezometer the instant it was removed 

 from the water-vessel, in which it was lifted from the pressure-chamber, and the 

 position of the index was afterwards read at leisure. As the same scale was employed 

 in the calibration of the piezometer tubes, its unit is, of course, of no consequence. 

 The expansibility of water at atmospheric pressure is so small, at least up to 8° C, that 

 no perceptible displacement of the mercury can have been introduced before the zero of 

 the scale was adjusted to it. The effects of the raising of temperature by heating are 

 two : a direct increase of the volume (provided the temperature be above the maximum- 

 density point, and the pressure be kept constant), and a diminution of compressibility 

 (provided the temperature be under the minimum compressibility point). These 

 conspire to diminish the amount of compression produced by a given pressure. At 

 15° C, or so, the first of these is, in the range of my experiments, the more serious of 

 the two, especially in the case of the solutions of common salt. 



