NO. 8 WATER-VAPOR TRANSPARENCY — FOWLE 2'J 



The points connected to form the third trace a'h'c'd'e' were ob- 

 tained by the insertion of a plane-parallel quartz plate one-half a 

 centimeter thick, between the energy source and the slit. Otherwise 

 the conditions were for curve a'h'c'd'e' the same as for the curve 

 abcdc. The significance of this third curve will be presently 

 considered. 



Taking into account the widths of slit and bolometer these curves 

 were first corrected to represent as well as possible " pure " spectrum 

 conditions as described under the work relating to the shorter wave- 

 lengths. See also Appendix I. 



The next step was the reduction of all the branches of the two 

 upper curves to the same scale. Provision was made to accomplish 

 this by arranging the observations so that in the last three branches 

 ordinates were obtained at the same abscissae for two successive 

 slit conditions. However, when the ratio of ordinates under the 

 wide slit to those under the narrow slit was taken, this ratio was 

 found to progressively increase as the wave-length increased. The 

 cause of this change, as stated in the description of the slit, remained 

 obscure for some time until it occurred to Dr. Abbot that it was 

 probably due to the spreading of the beam by diffraction when the 

 narrower slit was used to produce the lower branch of the curve. 

 As already stated and explained (1. c.) it occurred only when the 

 beam passed through the long tube containing the water vapor and 

 not when employing the path in the spectroscope alone. In the 

 former case, the beam jvist filled the collimator mirror and dififraction 

 caused it to spread ofif the mirror. In the latter case the beam con- 

 siderably more than filled the mirror and the diffraction produced 

 no harmful result. 



This disturbance, therefore, rendered difficult the determination 

 of diaphragm values for the reducing of the various parts of the 

 curve to the same scale. With the spectroscope alone no trouble 

 was found, for the slit ratios as measured mechanically and by means 

 of the deflections agreed closely and the energy curve over the whole 

 range of wave-lengths could be reduced to a uniform scale. ' It was 

 thought best on the whole to use the same slit ratios for the tube work 

 that were used with the spectroscope alone. To avoid error in the 

 tube work as far as possible the overlapping parts of curves made 

 with the smaller slit were rejected. It might, however, be feared that 

 some error from diffraction affects the tube work even with wide 

 elits. This matter will be treated later, and such errors will be 

 shown to be probably negligible. 



