NO. 8 WATER-VAPOR TRANSPARENCY FOWLE 21 



CORRECTION FOR SLIT AND BOLOMETER WIDTHS 

 When a spectrum is formed with a sHt of finite width (the sHts 

 here in use were often necessarily wide), the energy at any point 

 in the spectrum includes an appreciable range of wave-lengths 

 depending upon the angular width of the slit. This apparent so- 

 called impurity of the spectrum is further increased by the finite 

 width of the bolometer. Indeed it is easily seen that the range of 

 deviations observed by the bolometer at any point of the prismatic 

 spectrum is equal to 2 (a+b) where a and b are the slit and bolometer 

 widths expressed in angular values subtended in the spectrum. In 

 appendix I will be found the derivation of a formula for partially 

 correcting for this impurity of the spectrum. This formula was 

 applied in every case where the resulting correction would be of 

 importance. Such a formula can only partially correct for errors 

 in the readings of the maxima and minima in a spectrum. It of 

 course cannot reproduce from a nearly continuous record, such as 

 is shown in the upper part of figure 6, a purely line absorption 

 spectrum such as is produced by water vapor. Such a formula is 

 best applicable to a continuous spectrum such as would be given, for 

 example, by a black body. 



Having corrected the curve in the upper part of figure 6 for the 

 widths of slit and bolometer, it becomes as shown at a' in the lower 

 part of the figure. Here the ordinates of the longer wave-length 

 portion (5 to g fi) are magnified 10 times relatively to those of the 

 shorter wave-length section. Curve b, on a uniform scale with a', 

 records the energy passing through the air, carbon dioxide and water 

 vapor in the 117 meters additional path of the large tube. Curve c 

 on a similar scale is the black-body curve corresponding to a tem- 

 perature of 2,200° K. 



In order to obtain the amount of energy absorbed by the aqueous 

 vapor it would be highly desirable to observe the energy first through 

 the 117 meter path in vapor, then through the same path free of 

 vapor. This was practically impossible. As a substitute for the 

 latter condition, the energy was recorded with the lamp turned so 

 as to observe its energy when passing through the spectroscope 

 alone. With such a process the deflections at any wave-length could 

 not be directly compared but had to be first reduced to the same scale. 

 This was done by making the two sets of curves coincide near devia- 

 tions — 5' and 4-5' in spectrum regions where the many experiments 

 of this observatory indicate no appreciable absorption by atmos- 



