﻿M. C. Marignac on the Specific Heats of some Liquids. 135 



done, this knowledge is of little importance) of being perfectly 

 sure that it has always the same temperature. 



To remove this difficulty, I use as source of heat a mercurial 

 thermometer with a large reservoir. As it is important that the 

 heat be transmitted as rapidly as possible, I have had thermo- 

 meters constructed the reservoir of which is formed of a tube of 

 9 millims. internal diameter, and about 50 centims. long, spirally 

 coiled, so as to admit of being entirely immersed and agitated in 

 a cylindrical vessel containing from 150 to 250 cubic centimetres 

 of liquid. The value in water of these thermometers varied from 

 15 to 21 grammes. The transmission of the heat was suffici- 

 ently rapid for the maximum to be reached in 30 seconds. The 

 stems of these thermometers were graduated in divisions of about 

 a millimetre, of equal capacity, and corresponding to about T \y 

 of a degree. 



Before the experiment, the thermometer is heated, in a stove, 

 to a few degrees above the temperature at which it is intended 

 to be immersed in the liquid. It is then taken out and gradually 

 approached to the vessel, and the immersion takes place the 

 moment the mercury arrives at the point determined. I do not 

 think that the error which one may commit in operating thus 

 will amount to a quarter of a division (^q of a degree) ; and as 

 the thermometer was immersed at a temperature about 40° 

 above that of the liquid of which the heating was to be mea- 

 sured, it is evident that the possible error from this source could 

 not exceed Yeoo- 



The second and more serious objection is the inexactness of 

 all the methods proposed for correcting the errors produced by 

 the radiation of the vessel which serves as calorimeter. But 

 there is a very simple means of rendering correction needless. 

 Instead of comparing, as has generally been done, the variations 

 of temperature produced by the same source of heat in arbitrary 

 weights of water and the liquid whose specific heat we wish to 

 measure, it is sufficient to reverse the problem, and determine 

 the relative weights of water and of the liquid which undergo 

 the same alteration of temperature by the addition of the same 

 quantity of heat. In this way, and especially if the experiments 

 are made immediately after one another, in the same atmospheric 

 conditions, the influence of all exterior causes acts in precisely 

 the same manner in each case, and consequently involves no 

 perturbation. 



Besides, it will be understood that here absolute identity of 

 temperature is not requisite. Provided that, in the two experi- 

 ments compared, the temperatures differ by only a few hun- 

 dredths of a degree (which it is very easy to ensure after a single 

 preliminary experiment), all correction may be dispensed with. 



