Intelligence and Miscellaneous Articles. 157 



T=2000°, it is found that the excess t—d, instead of being J° con- 

 formably to observation, ought to be several hundreds of degrees. 

 Therefore the law of Dulong and Petit, so accurate for tempera- 

 tures from zero up to 300°, ceases to be so when these limits are 

 passed. 



If, by reasoning analogous to that which we have now criticised, 

 we kept the formula 



a t —a 9 =oca T , 



merely modifying the value of the coefficient a which would be de- 

 duced from my experiment, 1*0028 would be found for this value 

 instead of 1-0077. Next, applying the formula to the excess t — 

 which I obtained at the summit of Mont Blanc, we should arrive 

 at the number of 3330° for the temperature of the sun. But the 

 reasoning is not legitimate ; the law of Dulong and Petit being in- 

 accurate from 300° to 2000°, it cannot be admitted to be applicable 

 above 2000°. And since the coefficient a (which, according to 

 the law, should be constant) diminishes from 1*0077 to 1*0028 when 

 we pass from 300° to 2000°, it is probable that it would take a still 

 lower value at temperatures exceeding 2000° ; and this would con- 

 duct us to a higher number than 3330° for the temperature of the 

 sun. 



But that is only a probability, not a certainty ; and, in brief, I 

 confine myself to saying I do not think we can actually in this way 

 succeed in approximately measuring the temperature of the sun. 

 My impression is, that it is considerably above the highest tempe- 

 ratures we can attain by means of combustion, which are estimated 

 at about 3000°; but does it exceed them by some hundreds, or by 

 some thousands of degrees ? That is a question to which I should 

 not like to venture a reply. 



[:. In regard to this, permit me ,to speak again of some trials, made 

 partly in jouy laboratory at the Ecole JSTormale, which, imperfect as 

 they are, show once more the great comparative intensity of the 

 solar radiation. If a source of light (a gas-jet, for instance) be 

 viewed through one or several plates of blue cobalt glass, and if the 

 thickness of glass is suitable, the flame appears of a purple tint, 

 from the cobalt permitting the extreme red rays, as well as the blue 

 and violet rays, to pass through, and intercepting the radiations of 

 medium refrangibility. If through the same thickness of glass we 

 observe a luminous source at a higher temperature, and conse- 

 quently richer in highly refrangible rays, it no longer appears 

 purple, but blue ; the thickness of cobalt glass must be increased 

 in order again to obtain the purple tint : in fact, this does not much 

 alter the proportion of red rays transmitted, while the blue rays 

 are sensibly weakened. There is therefore a relation between the 

 thickness of cobalt glass which produces the purple tint and the 

 temperature of the source, at least if we have to do with white light 

 emanating from an incandescent solid or liquid body. "With some 

 improvements, one might base on this principle the construction of 

 a sort of pyrometer which would perhaps be useful in certain cases. 



