PHYSICAL PROPERTIES OF WATER, ETC. 11 



diminishes as the pressuTe is increased, quickly at first, afterwards more and more 

 slowly. 1 This was contested by Orsted, who found no change of compressibility up to 

 70 atmospheres. Many other apparently authoritative statements have since been 

 made to the same effect. Unfortunately Perkins' estimates of pressure are very 

 inaccurate, so that no numerical data of any value can be obtained from his 

 paper. 



Colladon 2 is sometimes referred to as an authority on the compression of liquids. 

 But, referring to Canton, he states that there is no difference in the compressibility of 

 water at 0° C. and at 10° C. His words are : " Nous avons trouve' que l'eau a la menie 

 compressibilite a 0° et a + 10°. Nous avons ddja fait observer les causes d'erreur qui 

 ont du alterer les resultats des experiences de Canton." There can be no doubt whatever 

 that there is a difference of 6 per cent. , which is what Canton gives ! 



In Regnault's experiments 3 pressure was applied alternately to the outside and 

 to the inside of the piezometer, and then simultaneously to both. From the first 

 Appendix to my Report on the Pressure-Errors, &c, it will be seen that the three 

 measurements of changed content thus obtained are not independent, the third giving 

 the algebraic sum of the first two ; so that, unless we had an absolutely incompressible 

 liquid to deal with, we could not employ them to determine the elastic constants of the 

 piezometer. For the compression of the liquid contents is added to the quantity 

 measured, in the second and third of the experiments. Thus Regnault had to fall back 

 on the measurement of Young's modulus, in order to obtain an additional datum. In 

 place of this, Jamin afterwards suggested the measurement of the change of external 

 volume of the piezometer ; and this process was carried out by Amaury and Descamps. 

 But there are great objections to the employment of external, or internal, pressure 

 alone in such very delicate inquiries. For, unless the bulbs be truly spherical, or 

 cylindrical, and the walls of perfectly uniform thickness and of perfectly uniform 

 material, the theoretical conditions will not be fulfilled : — and the errors may easily be 

 of the same order as is the quantity to be measured. 



Finding that he could not obtain good results with glass vessels, Regnault 



employed spherical shells of brass and of copper. With these he obtained, for the 



compressibility of water, the value 



0-000048 per atm. 



for pressures from one to ten atmospheres. The temperature, unfortunately, is not 

 specially stated. 



Grassi, 4 working with Regnault's apparatus, made a number of determinations of 

 compressibility of different liquids, all for small ranges of pressure. 



1 The carefully drawn plate which illustrates his paper is one of the very best early examples of the use of the 

 graphic method. Phil. Trans., vol. cvi. p. 541, 1826. 



2 Mem. Inst. Savans hrang., torn. v. p. 296, 1838. 3 Mint. Acad. Sci. Parts, torn. xxi. pp. 1 et seq., 1847. 

 4 Ann. de Chimie, ser. 3, torn. xxxi. p. 437, 1851. 



