1899.] 



cm Measuring Extreme Temperatures. 



99 



The importance of choosing a correct formula is most easily- 

 realised by reference to Fig. 2, which represents the results of 

 extrapolation as applied to deducing the probable temperature of the 

 sun. On the scale of Fig. 2, the dimensions of the experimental 

 range of Fig. 1 are reduced to the thickness of the line at the lower 

 left-hand corner of the diagram. The line at the top represents the 

 intensity of solar radiation, which is taken at 10,000 watts per square 

 centimetre in round numbers. The points at which the various 

 curves meet this line show the corresponding values of the solar 

 temperature. 



Fig. 2. — Temperature of the Sud by Extrapolation. 



The estimates of one million degrees and upwards, which were 

 current in many of the older books on astronomy, were deduced from 

 the law of Newton, and are obviously out of the question. The 

 celebrated formula of Dulong and Petit gives results between 1500° 

 and 2000° C, according to the data assumed, and evidently errs too 

 much in the other direction. At the same time, it must be observed 

 that the recent formula of Weber gives a result which is very little 

 higher. Paschen considered that his results lent support to Weber's 

 formula, and disagreed entirely with Bottomley's. But, according to 

 the writer's reductions, they agree very closely with Bottomley's, and 

 are best represented by the formula ET 57 . The experiments of 

 Petavel agree most nearly with a fifth power law. On the other 

 hand, the experiments of Wilson and Gray, in which the temperature 

 was measured by the expansion of a platinum strip, instead of by the 

 increase of its electrical resistance, appear to be in exact confirmation 

 of the fourth power law of Stefan, and give a much higher result for 

 the solar temperature. The formula of Bosetti is approximately a 



H 2 



