326 M. MELLONI ON THE POLARIZATION OF HEAT. 
then to indicate a very marked difference in this respect between the 
laws of polarization of heat and those of light. 
Are calorific rays really susceptible of polarization? Are they all 
equally and completely so? Such are the questions which I propose 
to consider in this memoir, endeavouring at the same time to account 
for the contradictions (more or less obvious) exhibited by the results 
at which the different observers just quoted have arrived. 
The instrument invariably employed by me in these researches is an 
excellent thermomultiplier constructed by M. Gourgon. In order to 
give the reader an idea of its great sensibility, it will be sufficient to 
state that the natural heat of the hand placed near one of the extre- 
mities of the tubes with which the pile is furnished, will impel the 
index to its maximum of deviation when the temperature of the atmo- 
sphere is below 15°. The pile, which has its two terminating faces 
perfectly symmetrical, consists of thirty pairs (bismuth and antimony,) 
formed into a bundle measuring eight lines in the diameter of its trans- 
verse section and ten lines in length*: the tubes or cylindrical appen- 
dages which envelop its two faces are nearly of the same breadth as 
the pile, but three times as long. The astatic system of the galvano- 
meter, which consists of two needles very powerfully magnetized, mea- 
suring O™™"-47 in diameter and 53™™ in length, makes but two 
oscillations a minute. If, however, after the communication with the 
pile has been interrupted, the system is turned aside 35 or 40 degrees 
* The symmetry, or rather the equality of the two opposite sides of the pile is 
condition indispensably necessary, in order to render the observations inde- 
pendent of the slight changes of temperature that may take place in the sur- 
rounding air during the experiments. In fact, if the bars of bismuth and anti- 
mony were stronger, or their solderings less extended, on either of the two 
sides than on the other, the heating or cooling of the air would no longer be 
communicated (by contact) with equal promptitude to them both, and the ex- 
tremities which presented the least mass in proportion to the extent of the sol- 
dering would be heated or cooled more rapidly than the opposite ends. This 
circumstance would produce a current which, by its intervention, would disturb 
the calorific effect of the rays that are received by the anterior face of the pile 
Hence it is obvious that with piles having their opposite faces unequal, exac. 
measures of the calorific radiations are attainable only in that case in which the 
temperature of the atmosphere undergoes no sensible variation. If it varies, 
the results will be less accurate in proportion to the greater rapidity of the va- 
riation and the greater length of time required to make the experiments. The 
piles [a rayons et a biseau,] described by Nobili in the 57th volume of the 
Bibliothéque Universelle have not their opposite faces symmetrical; they are 
therefore not free from the defect just mentioned. The author himself admits 
it in the 8th page of the same volume, where, after having given the different 
rates of calorific transmission obtained in a series of bodies by means of his pile, 
he adds, that the conditions of temperature which could affect the results, and 
consequently change them by their variations, are the following: 1st, the ter - 
perature of the source ; 2nd, that of the bodies themselves, and particularly th . 
of the surrounding air. 
