168 EEPORT — 1892. 



number 13"5956, wliich expresses its relative density (specific gravity). 

 The pressure being given by the equation 



jp=hgd, 



we must choose for rj some normal value. Now, there can be no hesitation 

 on this point Meteorologists, metrologists, and geodesists have adopted 

 the normal value at 45°, and at sea-level ; physicists have begun to adopt 

 this unity, and any exceptional cases can only be attributed to pre- 

 conceived ideas. The normal value of g is 



cm 



Hence fffrm . cm' 



P 



L8 



980-65 



sec" 



= ;» X 980-65 X 13-5956 f-^"^ . ^""'l. 



Lsec" cm^J 



Let^=10«; then /i = 75-005 cm. 



Starting from this result I propose to adopt as the imit of pressure 

 the pressure exerted by a column of mercury of 76 cm. at 0° under the 

 normal conditions of gravity, and to call this unit the barie. This 

 decision would in no wise prevent one from keeping as secoudary units 

 the pressures exercised by colnmns of mercury 1 mm., 1 cm., or 1 m. in 

 heiglat, which bear a simple ratio to the unit pressui'e.' 



I do not think that any measurements have been made in which it 

 would be of use to take into account the compressibility for a pressure of 

 one barie ; but as this correction varies as the square of the height, 

 the limit beyond which it is no longer negligible is 2 or 3 metres of 

 mercury, a correction must be applied in certain experiments on this 

 account. 



The Temperature Scale. 



The thermometric system is an entirely arbitrary one, and is subject 

 to only one condition, namely, that of offering a simple relation with 

 precise experimental data. The adoption of a normal unit of pressure in 

 nowise renders it necessary to make use of this unit for the determina- 

 tion of the higher fixed point of the system of temperatures. The 

 adoption of a thermometric system (that is to say of a fundamental 

 interval and of its subdivision) by all physicists alike would evidently be 

 extremely useful, and it is doubtless to avoid breaking suddenly with 

 popular customs that the Centigrade system has not yet come into 

 universal use. 



Altogether different is the question of a thermometric scale, that is 

 to say of a function of some natural phenomenon which represents 

 the temperature when it is made to satisfy two equations of condition. 

 Carnot's principle, as Lord Kelvin has shown, allows us to define a scale 

 of temperatures independently of the thermal properties of any given 

 body. The most accurate experiments may lead to the belief that, 

 ■within the limits of temperature measured as yet, the hydrogen thermo- 

 meter furnishes us with a scale of temperatures which is practically 

 equivalent to the thermodynamic scale ; and it is for this reason that 

 this standard scale has been adopted by the Comite International des 

 Poids et Mesures. I would propose, therefore, that all physicists should 

 be recommended to reduce all exact measui'ements to the Centigrade scale 



' For the unit of pressure to be completely defined it is necessary to state 

 whether account shall be taken of the compressibility of mercury and of its vapour 

 pressure, of which the value at 0° is as yet not very well determined. 



