WOOD. — NITROSO-DIMETHYL-ANILINE. 59 



numerical values from the photographs, the distances between the spectra 

 were measured with a dividing engine at tlie principal cadmium lines. 

 From these distances the actual angular deviations were calculated, 

 making proper allowance for the fact that the plate stood at an angle, — 

 i. e., tlie focal length of the lens of the spectrograph was very different 

 in the different regions of the spectrum. The angle of the prism was 

 calculated from the deviations in the yellow and green, for which region 

 tiie refractive index had been already determined with considerable 

 accuracy. In spite of the rather crude method, the results obtained 

 with the different prisms agreed surprisingly well. 



It will be seen from the photographs, some of which are reproduced 

 in Figure 1, Plate 3, that the deviated and undeviated spectra come 

 together at the ultra-violet edge of the principal absorption band, indicat- 

 ing a refractive index not much above unity. 



The prism angles in these experiments varied from 20' to 2°. 



It was found possible to get a fairly accurate determination of the re- 

 fractive index for wave-length 48 from these photographs, the mean of 

 several determinations giving 2.28 as the value. The values found on 

 the ultra-violet side of the band are given in the form of a curve in 

 Plate 2. This portion of the curve is especially interesting, as the 

 effect of the ultra-violet absorption bands is most strikingly shown. 

 The weak band at wave-length 25 causes a dip in the curve, which 

 has a maximum at 285, after which it again rises rapidly as it ap- 

 proaches the heavier absorption band situated not far from 20. 



On this same Figure I have given the dispersion in the visible spec- 

 trum, together with the value n = 1.85 for wave-length, A = 43, the 

 centre of the absorption band, which was determined from the constants 

 of elliptical polarization. The trend of the curve within the absorption 

 band I have sketched in freehand, as well as the general form and posi- 

 tion of the absorption bands. 



I have already shown that the reflecting power of the nitroso for the 

 wave-lengths immediately adjoining the absorption band on the ultra- 

 violet side, is exceedingly small, which is not surprising when we con- 

 sider that the refractive index for this region is not very different from 

 air. If now we consider a quartz-nitroso surface, instead of an air-nitroso 

 surface, we should expect fairly strong reflection at this point, the 

 nitroso acting as the rarer medium, while further down the spectrum, 

 at about wave-length 29, where the indices of the quartz and nitroso are 

 very nearly the same, we should expect scarcely any reflection at all. 



This was found to be the case. An acute prism of quartz was 



