PHYSICAL PROPERTIES OF WATER, ETC. 23 



of the mercury in the reservoir, and yet not so narrow as to prevent free motion of the 

 mercury. 



" Important Remark. — During the successive operations the large piston should 

 always, by means of the regulating pump, be kept at such a height that the rod cc shall 

 not come in contact with the wall of the opening FF, and not high enough to make the 

 wide lower part of the small piston come against the piece M (this, of course, when the 

 smaller of the two upper pistons is used : — that whose lower part is thickened). 



" There are two pistons pp for this manometer. The ratio of the section of the 

 larger to that of PP is 1/61-838, and the reading per atmosphere is 12-290 mm. 



"For the smaller, the ratio of the sections is 1/277 75, and the reading per 

 atmosphere is 2'736 mm. 



" The former serves for the measurement of lower pressures, up to the point at 

 which the oil passes visibly round the large piston. For higher pressures the latter 

 must be used. 



"'The treacle must be changed from time to time ; first, because, after a while, some 

 of it passes the small piston ; second, because it gradually dissolves in the glycerine, and 

 at last becomes hardened round the small piston, so as to make the friction too great. 

 The small piston and its cylinder should occasionally be cleaned with the greatest 

 care, and anointed with neats-foot oil." 



In all my later experiments I have used exclusively the smaller of the two small 

 pistons. The scale which I fitted to the manometer tube was a long strip of French 

 plotting paper. It had shrunk slightly, so that 752'5 divisions corresponded to 

 750 mm. Neglecting the difference in the values of gravity at Lyons and at Edin- 

 burgh, the number of scale divisions per atmosphere is 2'73G x 752'5/750 ; and its 

 logarithm, i.e. the Gauge Log. above spoken of, is '43856. 



V. Compressibility of Glass. 



Buchanan's process, already referred to, consists simply in measuring the fractional 

 change of length of a glass rod exposed to hydrostatic pressure, and trebling the linear 

 compressibility thus determined. The only difficulty it presents is that of directly 

 measuring the length of the rod while it is under pressure. I employed a couple of 

 reading microscopes, with screw-travelling adjustment, fixed to the ends of a massive 

 block of well-seasoned wood. This block was placed over the tube containing the glass 

 rod, but quite independently,— the two distinct parts of the apparatus being supported 

 separately on the asphalt floor of a large cellar. No tremors were perceptible except 

 when carriages passed rapidly along the wooden pavement of the street, and even then 

 they were not of much consequence. 



