( 1092 ) 
absorption line produces a local shift of the interference fringes. 
The amount of the shift depends, of course, on the quantity of gas 
traversed. In spectrum 5 e.g. the dispersion anomalies are scarcely 
perceptible in the red; they increase in the main with decreasing 
wave-length (as also the absorbing power of NO, increases in the 
main toward the violet); in spectrum 7, the qnantity of gas being 
greater, we observe very conspicuous anomalies already in the red, 
and when proveeding toward smaller wave-lengths, see them so much 
increase, that bevond 25000 anything like horizontal fringes has 
disappeared. 
At a few places, where more or less isolated lines occur, it is 
clearly visible that the adjacent light shoots out into the next fringes 
like sharp spikes, upward on the red side, downward on the violet 
side of the line. Now, if we suppose the same phenomenon to 
repeat itself near each of the many narrow lines which, crowded 
together, constitute a band or fluting in the spectrum, we must 
expect to find the dark fringes less dark, the bright fringes less 
bright in any region corresponding to a band or fluting. This par- 
ticularity is indeed very conspicuous on the original photographs. 
Where broad dark bands occur in the absorption spectrum, the 
system of fringes shows somewhat vague, contrastless; whereas light 
and dark distinctly alternate in the system, wherever the average 
absorption is small. 
Besides, we observe that in every region crowded with absorption 
lines, the mean index of refraction rises with increasing wave-length, 
and that in regions with few lines it sinks. 
The strips 8 and 9 on the plate show parts of the NO,-spectrum 
with high dispersion. Number 9 is an original-size reproduction‘), 
covering the part of the spectrum which, on 7, is enclosed between 
a and d; of this a part again (lying between 5 and c) is given on a 
three-fold scale in strip 8. 
As in this case, for want of intensity, the time of exposure had 
to be about one hour and a half, and because the arrangement had 
no claim to perfect stability, the details are not so sharp on the photo- 
graph as they showed visualiy. But on direct observation many 
bands were now resolved into fine lines, and there could be no 
doubt that each line clearly influenced the refractive index for 
adjacent waves. Even the reproduction offers the required evidence. 
Indeed, every absorption line of nitrogen peroxide visible on it has 
1) The fringes are there curved in the opposite direction (as compared with the 
other spectra on the plate), owing to a slight alteration in the arrangement of 
the apparatus. 
