NATURAL PHILOSOPHY. 141 



"I have verified this method by determining the capacities of 

 iron and of copper, which are the most difficult to obtain exactly, 

 because they are very small. I found 0.098, 0.093. M. Regnault 

 obtained the numbers 0.113, 0.095, which are a little larger ; but 

 he operated with a higher temperature. 



"II. Of Oases and Vapors. The advantages of this method 

 are especially apparent when treating of aeriform fluids. A gas- 

 eous current passes through a glass tube to the middle of a cork 

 of badly conducting material ; a thermometer there measures its 

 temperature. It immediately enters a second tube through the 

 folds of a metal spiral or a bundle of twisted wires traversed by 

 electricity, that is to say, through a focus; it becomes heated 

 and meets a second thermometer, which measures its increase of 

 temperature. Before emerging, the gas is led round the first tube, 

 to prevent any loss by radiation and conductibility ; and when the 

 temperature has become stationary, we may say that all the heat 

 of the focus, which is known, is taken by the gas, the tempera- 

 ture of which is increased by a measured quantity ; hence the 

 specific heat can be deduced. 



" There are two advantages in this method. The first is, that 

 the greatest cause of error which Delaroche and Berard, and after- 

 wards M. Regnault, met with, is suppressed. In their experi- 

 ments the gas reached 100 in a calorimeter at 10 ; and the great- 

 est difficulty was felt in appreciating the heat which passes by 

 conductibility from the hot tube to the cold calorimeter. In my 

 method the gas reaches, at the ordinary temperature, say 10 ; it 

 passes from the spiral at about 20 ; the difference is 10 ; it Avas 

 90 before ; the present error is at most one-ninth of the former. 



"Here is the second improvement. The whole of my appara- 

 tus is the size.of a finger ; it is of thin glass ; it might be of mica, 

 even of goldbeater's skin ; it weighs no more than a litre of gas, 

 and expends no more heat in reaching the final temperature. Ten 

 litres of gas are sufficient to make one measurement. Thus the 

 difficulties that had for a long time to be overcome, in order to 

 obtain a uniform current, disappear, ordinary gasometers suffice, 

 and the method is applicable even to vapors. A first determina- 

 tion gave the number 0.242 for air instead of 0.237, which M. 

 Regnault found. 



" Thermometers, even, may be dispensed with, and the tem- 

 perature measured by the increase of resistance in the wires. It 

 is known that a resistance r at zero becomes r (1 -f- at) at t de- 

 grees. That bein<* the case, let 2 equal bundles of wires be placed 

 one after another in a tube ; then, having decomposed the total 

 circuit into 2 equal derived circuits, let us make each of them 

 pass, first, through 1 of the 2 bundles of wires, then into a differ- 

 ential galvanometer ; the latter remains at zero. But if a current 

 of gas at t degrees be sent through this tube, it will pass at t -\- 9 in 

 the first spiral, at t-\-29 in the second ; they take a difference of 

 temperature 0, a different resistance, and the galvanometer is 

 deflected. It is reduced to zero on introducing, by means of a 

 special rheostat, a platinum wire into one of the circuits. The 



