718 Dr. H. A. Houstouu on the Relative 



various parts of the spectrum, because the latter of course 

 was not a normal one. 



The lamp A was compared with a Hefner lamp at its 

 normal voltage by means of a spectrophotometer, and the 

 final results are expressed in the second column of the follow- 

 ing table. This column gives ratio of lamp A to Hefner 

 lamp, and we see that lamp A is stronger at the blue end. 

 The distribution of energy in the spectrum of the Hefner 

 lamp has been determined by K. Angstrom*, and he has 

 represented his results by the formula 



_ 7-8.5 



\- b e~ A , 



where A, is measured in //,. This function is given in the 

 third column of the table. On multiplying the second and 

 third columns we get the original distribution of energy in 

 the spectrum of lamp A. To allow for the different absorp- 

 tion of light in the spectroscope the slit and eyepiece of the 

 latter were removed, and a beam of light which passed 

 through the spectroscope and an additional lens and ground 

 glass was compared by the spectrophotometer with another 

 beam from the same source, which was twice reflected by 

 mirrors, and passed through a similar ground glass. The 

 results are shown in the fourth column. The figures are 

 proportional to the percentage transmitted, and they show 

 that the blue end of the spectrum is absorbed to a greater 

 degree than the red end. The figures in the fourth column 

 are of course not quite beyond question, as the one beam 

 went through an additional lens instead of the eyepiece, and 

 the other beam was probably slightly affected by the two 

 mirrors, but there was no other way of measuring this 

 quantity, and the absorption almost all takes place in the 

 prism. It should be noticed that it is only the relative 

 values, not the absolute values of the figures in the second, 

 third, and fourth columns that have any meaning. 



To allow for the spectrum being prismatic we have to 

 make the " slit-width " correction, L e., multiply the distri- 

 bution of energy by -j- , which gives the inclination of the 



tangent to the calibration curve of the spectroscope. The 



spectroscope was one of the earlier ones not furnished with 



a wave-length scale, and s was the reading of the prism 



drum in hundredths of a revolution. The fifth column 



dX 

 gives y-? and the sixth column gives the product of the 



* Plivs. Rev. xvii. p. 312 (1903). 



