of Edinburgh, Session 1879 - 80 . 
431 
with sufficient prism dispersion just to separate the various bright 
lines of the source from one another. 
My former assistant, Professor D. H. Marshall, made for me, in 
1870, a series of careful measurements of the change of plane of 
polarisation of the lines C and F of hydrogen by this method, using 
a vacuum tube with a narrow bore, no slit, and a prism of small 
angle. It was found to give fair but not excellent results. Although 
no greater thickness of quartz was employed than the plates supplied 
along with Duboscq’s saccharimeter, the planes of polarisation of C 
and F were separated by the thickest of them upwards of 130° ; but 
the determination of the exact point of extinction is not easy. In 
measuring with practically homogeneous light, like that of a spirit- 
lamp with chloride of sodium or of lithium, the prismatic dispersion 
was, of course, not required. The great merit of the rotatory polari- 
sation process consists in the fact that there is scarcely any additional 
loss of light incurred by using a foot or two of quartz instead of a 
few millimetres, and thus in proportion increasing the amount of 
rotatory displacement ; while the thicker the quartz the less is the 
inevitable percentage error of observation. Also the position of 
each bright line is determined in terms of a standard quartz-rotation, 
and needs no comparison spectrum. It remains to be seen whether, 
on trial, it may be found possible to have a great length of quartz 
cut with sufficient accuracy, and whether the bright lines are narrow 
enough for this mode of observation. I have ordered a 6-inch 
cylinder of quartz, and hope soon to have observations made with 
it. Meanwhile it seems likely that this combination of polarising 
and analysing prisms, with a quartz plate, and a small direct vision 
spectroscope (with very wide slit), may be well adapted for measure- 
ments of position of the bright lines in the spectra of auroras, comets, 
and nebulae, where it is not easy to employ either a comparison 
spectrum or a wire micrometer. 
