NO. 8 WATER-VAPOR TRANSPARENCY FOWLE 29 



sently to be described. Additional confirmation is given by the agree- 

 ment in form from 9 to 13.5 /x of curves taken here with a range of 

 from 0.003 to 0.028 cm. ppt. H2O. All the energy curves taken 

 through the spectroscope alone (which involved a range of between 

 0.003 and 0.012 cm. ppt. H„0) were reduced and carefully compared. 

 Upon reducing to the same scale between 9 and 10 /x they showed a 

 remarkable agreement even over the whole range from 9 to 13.5 fx 

 indicating no change in transmission of energy within this range 

 of wave-lengths when the ppt. H^O increased from 0.003 to 0.012 cm. 

 A yet later set of observations taken on a cold winter day through 

 the long tube with the total ppt. H.O equal to 0.028 cm. also agreed 

 over this range. It would be very improbable that the percentage 

 absorption would be uniform over the range from 9 to 13.5 /* unless 

 the absorption over this whole range is zero. Any departure from 

 such a uniform absorption would have tended to cause the curves 

 to diverge at the longer wave-lengths. 



In accordance with the well-grounded assumption just discussed 

 all curves were made to agree in intensity between the wave-lengths 

 9 and 10 jx. 



DIFFRACTION EFFECT 



It has been stated that with the energy curves made with the beam 

 passing through the spectroscope alone no diffraction effects were 

 noted in determining the diaphragm or slit ratios. It was also 

 stated that the curve just referred to taken with the minimum amount 

 of aqueous vapor (0.028 cm. ppt. HgO) in the path through the great 

 tube coincided in form from 9 to 13.5 n. with those taken through the 

 spectroscope alone. This justifies the inference that the width of the 

 slit (10.69 mm.) used from 9 to 13.5 /a in the "tube" observations 

 was sufficient to avoid diffraction losses. Beyond this wave-length 

 the energy in the tube curves is practically all depleted by the carbon- 

 dioxide band central at 14.75 /* but which produces an appreciable 

 effect, according to Rubens, all the way from 12 to 16.5 fx. Its effect 

 is most important from 14 to 15.5 /x. 



STRAY LIGHT 



Before proceeding with the discussion of the transparency of 

 aqueous vapor attention must be directed to the most troublesome 

 source of error of all. The intensity of energy observed at any 

 point in a spectrum will be due in part to the true energy of the 

 wave-length considered and in part to energy scattered there from 

 other regions of the spectrum. Naturally a portion of energy proper 



