378 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



the width of one square, the wave-lengths between tlie D lines, and for 

 a distance of two squares to each side oi" them, were lost by absorption. 

 By means of step by step measurements, taken at different temperatures, 

 it was possible to get a very good idea of the dispersion curve within this 

 very narrow range of the spectrum. The appearance of the spectrum 

 on the reticulated background at two different temperatures is shown in 

 Figure 3, the dispersion at Dj being somewhat greater than at Dj. Obser- 

 vations were taken of the dis- 

 persion within a quarter of a 

 square of Do, that is, within a 

 distance of the absorption line 

 equal to j\ of the distance 

 between the U lines. The 

 deviation at this point was 

 calculated to be about eighty 

 times the deviation at the he- 

 lium line. If we number the 

 wave-lengths at the lines of the 

 mesh between Dg and D2 1 to 

 9, tlie observations can be re- 

 corded as follows. The light 

 very close to D2 (5 of a square 

 distant) was seen to be dis- 

 tinctly pulled down through a 

 distance of seven squares with a vapor density sufficient to deviate the 

 spectrum at the first line {= 5888.3) through a distance of one square. 

 The temperature was then raised a trifle. The light vei-y close to D^ 

 now disappeared, partly as a result of absorption and partly through being 

 deviated out of the field, when the spectrum at the first line was deviated 

 through three squares, that at the second line one square. The temper- 

 ature was now raised until the spectrum at the ninth line was deviated 

 through one square. The deviation at the second was now five, that at 

 the third three squares, and that at the fifth, two squares. 



These results for the different wave-lengths between D3 and D2 are 

 given in the following table : 



