330 Scientific Proceedings, Royal Dublin Society. 



water in B slowly warming the air in the tube above CC, and thus 

 creating a gradual increase of pressure on the surface CC' '. The 

 pressure of the vapour of the liquid also adds its effect to that 

 of the expanded air, and sufficient time must be allowed to elapse 

 for the vapour to attain its maximum pressure. This error may 

 be neutralized by taking three pairs of readings : first, CD' and 

 the temperature f, then CD and its temperature t ; then, again, 

 CD' and its temperature t" ; if the readings are made at a uniform 

 rate, the mean of the first and last readings of CD' and of t' and 

 t" may be taken as the correct readings of the warm liquid and its 

 temperature. The source of this error may further be lessened 

 by making the volume of air above CC as small as possible ; 

 hence the advantage of raising, rather than depressing, the level 

 CC, and the object of the constriction of the glass tube forming 

 the cross-piece. Capillarity, it is true, causes the liquid surface to 

 be more or less curved, but by using U tubes of sufficient bore, 

 and reading the same part of the meniscus in each case, the error 

 from capillarity may be rendered insensible. 



An actual experiment with petroleum is given as an illustra- 

 tion of the working of the apparatus. 



Absolute Dilatation of Petroleum. 

 Mean of first and last reading of li = 139*4 mm. 



f = 46°-0 C. 

 Beading of h = 134*2 mm. 

 t = 7°'5 0. 

 li - h = 5*2 mm. 

 f - t = 38°-5 C. 



Whence the coefficient of absolute dilatation between 

 7° and 38° C. is 



0-000992 for 1° C, 



which, allowing for the low range, does not sensibly differ from 

 the coefficient given in Clarke's Tables, 



The same apparatus may obviously be used for determining 



