148 PROCEEDINGS OF THE AMERICAN ACADEMY 



as Kellaud assumed. By use of a suitable arrangement, it would be 

 very easy to use such ranges of temperature as should make the 

 experiment decisive. If the points were unequally heated by accident, 

 the wax curve would not be symmetrical with respect to a line per- 

 pendicular to that joining the two heated points. There could be no 

 trouble with lack of homogeneity of the plate, as preliminary experi- 

 ments show that from a single heated point the curve is perfectly 

 symmetrical and distinct, with a probable error iu finding the line 

 of demarcation practically insensible. I am indebted to Professor 

 Gibbs for the suggestion that the New Hampshire Infusorial Earth 

 might be advantageously used as an almost perfect non-conductor of 

 heat. 



Whatever might be the result of a series of experiments like those 

 referred to above, it is evident, from the differential equation, that 

 Fourier's solution cannot be a correct one. If the uniform tempera- 

 ture of the plate were taken as the zero of the scale, the temperature 

 of any point in the plate due to the two heated points would be tho 

 sum of the temperatures that would be given to the same point if each 

 heated point acted alone. On this hypothesis, it would be very easy 

 to find a point in the plate so situated that, when the plate had 

 reached its final state, the point would have a temperature nearly 

 double that which either of the heated points have, which is mani- 

 festly absurd. It does not appear that writers upon this subject have 

 noticed this fact. 



If experiment shows, as it probably will, that the propagation of 

 heat in the interior of a solid is determined by the expression 



— c^JM, 



dx 



it will not be hard to determine J{v) experimentally. The infusorial 

 earth, it is believed, will prevent any sensible loss of heat from con- 

 duction or radiation, so that the flow of heat in the plate will be de- 

 termined solely by the law of internal flux. Let a plate of any metal 

 other than copper be heated at a single point, and after it has reached 

 its final state let the temperatures of the different points of the plate 

 be determined relatively by means of the thermoelectric currents 

 obtained by touching different points of the plate by the copper ter- 

 minals of a Thomson's Short-Coil Galvanometer. These terminals 

 are to be held by wooden pincers, and pushed through the light infuso- 

 rial earth so as to explore the points of any line radiating from the 

 point where the heat is applied. 



