M. MELLONI ON THE POLARIZATION OF HEAT. 329 



with the amplitude of the primitive deviation, which we shall call the 

 arc of impulsion ; and we perceive that, the slowness of the motion at 

 the extremity of this arc enabling us to observe it with considerable 

 exactness, it may be afterwards compared with the corresponding 

 steady indication ; which comparison may be easily extended to all 

 points of the circuit if we vary the intensity of the calorific radiation by 

 making the requisite change in the distance between the source and the 

 pile. Moreover, the fixed deviations being given, the corresponding 

 forces may always be determined by experiment*. We are therefore 

 in possession of all the elements necessary for the construction of such 

 a table as may immediately show the ratios of the forces according to 

 the arcs of impulsion. The forces, as we know, represent the tempe- 

 raturesf : thus, by means of our table, the relative intensities of two ca- 



• For the description of the methods, see Bihliotheqiie Universelle, torn. Iv. 

 page 9 ; and Memoires de VJcademie des Sciences, torn. xiv. p. 445 and 446. 



t M. Becqueiel had shown in 1826 {J7m. de Chiinie et de Physique, torn. 

 xxxi. page 371) that the intensities of the thermoelectric currents of copper, 

 platina, and other metals, are proportional to the temperatures through the 

 whole extent of the thermometric scale. Now, the currents which produce the 

 greatest possible deviation in the common thermomultipliers are derived from a 

 heat, which scarcely rises to a few degrees, acting on one of their faces : the 

 proportionality between the forces of magnetic deviation and the temperatures 

 was tlierefore already established by experiment when I was commencing my 

 inquiries into the nature of radiant heat. Hence it is that I have admitted 

 that proportionality, as a known fact, in the preceding jVIemoirs on this subject. 

 Nevertheless, as M. Becquerel had not operated directly on the metals which 

 enter into the composition of the pile, the committee appointed by the Aca- 

 demy of Sciences to examine my experiments on heat, mani.'ested a desire that 

 the proportionality of the forces to the temperatures in the thermomultiplier 

 itself should be placed beyond the reach of doubt by some special experiments. 

 With this view I procured a thermoelectric pile of four very minute elements (bis- 

 muth and antimony) bent like a siphon, in order that each of the two termina- 

 ting faces might be introduced into a separate recipient and different tempe- 

 ratures imparted to them by the contact of heated liquids. The extremities of 

 the two last elements stood out of the vessels into which the ends of the curved 

 electric bundle had been plunged, and communicated with the galvanometer by 

 two copper wires. But as a difference of some degrees was sufKcient to drive 

 the magnetic needles to the extremity of the scale, I placed in the electric 

 circuit a very fine iron wire of several feet in length. The current then be- 

 came so weak that a variation of a centigrade degree of temperature between 

 the two faces produced in the galvanometer no more than a deviation of about 

 one degree. Matters being now in this state, water more or less heated was 

 successively introduced into one of the vessels and thawing ice into the other. 

 The second face was thus kept constantly at zero, while the first took, in suc- 

 cession, the different temperatures of the water, which were determined by a 

 verj' delicate mercurial thermometer. Tlie numbers of the degrees indicated 

 by the thermometer plunged in the hot water were found exactly proportional 

 to the corresponding electric forces or intensities indicated by the deviations of 

 the galvanometer. The experiment was now varied in order to obtain a nearer 

 approach to the circumstances in which the thermomultiplier is commonly cm- 



