156 PROCEEDINGS OP THE AMERICAN ACADEMY 



which the thermal intensities were measured, it was compared with the 

 sun's spectrum by the following method. The pile was made to slide 

 up vertically so as to be replaced by the slit beneath it. This slit 

 could then be moved to different parts of the spectrum, and its posi- 

 tion read off on the same arbitrary scale as the thermopile. In fact, 

 the readings for the two would be identical. Behind this slit was 

 placed the slit of a single prism spectroscope, having a third tube and 

 a photographed scale. As the pile was moved to different parts of 

 the luminous spectrum an image of the slit beneath it was formed on 

 the scale of the small spectroscope and its position could be read off in 

 terms of the arbitrary scale. 



A table was then formed containing in one column the readings of 

 the thermopile scale, and in the second the corresponding readings of 

 the spectroscope scale. Then the spectroscope was turned toward the 

 sun and the positions of the principal Fraunhofer lines read off in 

 terms of its scale. These readings formed a third column, and it 

 was easy, by proper interpolation, to determine the position of these 

 Fraunhofer lines in terms of the thermopile scale. 



It now remained to determine the wave-lengths of these lines. This 

 has already been done with great accuracy by Professor Powell,* and 

 accordingly his values for these lines were made use of. The wave- 

 lengths of the non-luminous portion were determined by means of 

 empirical equations based on observations made in the luminous por 

 tion of the spectrum. Any considerable extension was, of course, 

 accompanied with great error. 



Some experiments with a diffraction grating are contemplated, 

 and, if successful, will greatly help in determining the wave-lengths 

 of the various parts of the curves here given. It must be remem- 

 bered, however, that these experiments are principally of value in 

 their relation to each other, and that the absolute determination 

 of wave-length is therefore of only secondary importance. 



Experiments with Platinum. 



These experiments consisted of ten final and many other prelimi- 

 nary series of measurements of the distribution of heat in the spectrum 

 of a platinum wire when heated to various measured temperatures 

 between a low red and the point of fusion. 



The method of experimenting was as follows. The vulcanite 

 frame, containing the wire with its shunt, was first submerged in water 

 of a known temperature, a faint current sent through it, and readings 



* Pogg. Ann., lxix. 



