176 O. N. Rood on a Secondary Spectrum. 
will always be half the size of the actual physical spectrum 
which it represents. It is evident that the construction is inde- 
pendent of the relative position of the lines representing the 
two primary spectra, demanding only that they should be repre- 
sented by straight parallel lines. 
It is not actually necessary to go through the labor of execut- 
ing the projections on paper, as a far more accurate result can 
otherwise be obtained. t on inspecting the construction, fig. 
1, or one with dissimilar primary spectra, it at once becomes 
evident that the distance of any two lines apart in the secondary 
spectrum (on the line AB), will be equal to one-half the differ- 
ence between the corresponding distances in the primary con- 
v1 
stituents, or ee cm We 
and hence in the actual or physical secondary spectrum we 
shall have w=—l, 
positive, those to the left negative. 
Eaxperiments.—Three principal cases arise: 1st, where the op- 
posing spectra are of equal or nearly equal lengths ; 2d, where 
rismatic spectrum is the longer of the two. I give below for 
these cases the measurement made on the two primary § 
and on the actual physical secondary spectrum, adding for each 
of these cases the results obtained from the construction, which, 
for a reason hereafter to be mentioned, are generally somewhat 
more correct than those obtained with the telescope. 
ist Cask. 
Spectrum from funt glass prism of 60°. 
Ka Ha Na Th Hp. gt oy Hg 
Pao t 0.1.48 1 Ser bt [00 1 (4s 1 eo 
Spectrum of 1st order, grating No. 2. 
t 
Me ny He a a a 
[4°75 | 2 | 6-45 | 645 | 71 | o-75 | 1°85 | 125 | 
