18 Studies on Solution. 



0.0742- 0.0206 



J. 



Correcting for expansion upon dissolving, using data from Landolt 

 and Bornstein, we have 



A _ 0.0742- 0.0200 

 J. yu 



In this case the correction amounts to 1 per cent. Furthermore, the 

 example just cited is one in which this correction is at its maximum. 

 In the cases for solutions which are more dilute and for wave-lengths 

 where the water absorption is smaller, this correction is much less. 



In measuring absorption bands which are narrow in comparison 

 with the range of wave-lengths passing through the second slit of the 

 spectrograph, a correction for the finite width of the slit must be made. 

 All of the bands studied in this investigation were so broad as to make 

 such a correction negligible. 



It should be noted that the spectrograph and radiomicrometer of this 

 investigation are useful for a detailed quantitative study of band 

 structure. At no time in the present work has the full resolving 

 power of the instrument been called upon. Readings could be taken 

 at wave-length intervals of 4juju without fear of measuring overlapping 

 regions of the spectrum. 



THE ABSORPTION COEFFICIENT OF THE SOLVENTS. 



WATER. 



The water used throughout this investigation was the same as that 

 used in the work on conductivity carried on in this laboratory. The 

 water was dust-free and had a mean specific conductivity of 1.8X10" 1 

 reciprocal ohms. In view of the fact that the values of a for water are 

 used in the calculations of A for all the water solutions, the absorption 

 curve of water was repeated 6 times, and the recorded values are thus 

 each a mean of 12 separate measurements. 



The absorption curve for water in this region of the spectrum has 

 been drawn by one other observer, Aschkinass. 1 His curve is also 

 plotted on figure 3 for the sake of comparison. The lack of agreement 

 in the location of the position of the bands at 979jUM and at l,190ju/z is 

 probably due to the fact that Aschkinass used a quartz-prism spectro- 

 graph. The determination of wave-lengths in this region of the 

 spectrum is more uncertain in the case of the prism than the grating 

 spectrograph. The values of a for the maximum of the sharp band 

 at 979/xju given by Aschkinass are lower and those for the minimum 

 at 1 ,070///i higher than the corresponding values recorded in the present 



iWiedem. Ann., 55, 401 (1895). 



