310 Prof. R. W. Wood on the Dispersion of Sodium 
to have a normal spectrum, and the spectroscope was replaced 
by a collimator and plane grating. The light from the 
erating entered a camera made of a long wooden box, 
provided with an achromatic telescope-lens of 2 metres 
focus. 
The spectrum, which was received on a plate of ground 
glass, was about 25 ems. long and 1 mm. wide. Its position 
was accurately recorded on the ground glass by means of a 
narrow pencil mark ruled along its centre, the ground surface 
having been sponged over with oil to increase its transparency. 
The burners of the furnace were now lighted and turned 
down very low so that the tips of the flames fell a little short 
of the tube. As soon as the temperature of the tube became 
constant, the curved branches of the dispersed spectrum were 
outlined on the glass with the pencil. The burners were then 
turned up a little and the deviations of the oppositely curved 
branches increased, their new positions being recorded in the 
same manner. A number of records were made in this way, 
and the tube was then aliowed to cool off. The spectrum 
straightened, and was found to occupy the same position on 
the plate as before, showing that no displacement of the ap- 
paratus had occurred. A cadmium spark and sodium flame 
were now placed before the slit of the grating spectroscope, 
and the position of the bright lines recorded on the plate. 
The deviations which different regions of the spectrum had 
experienced were then carefully measured. The object of 
making a number of records was to show whether the rela- 
tive deviations depended on the temperature of the tube and 
density of the vapour. No indications of any such irregu- 
larity were found. The values found in this way agreed 
admirably with the results obtained with the interferometer. 
It moreover furnished data regarding the dispersion in the 
blue and violet region, which the interferometer failed to. 
do with any accuracy for reasons which I have given. 
Photographs of the dispersed spectrum are reproduced on 
fig. 8, Plate XII. They are neither retouched nor enlarged, 
and were made with sunlight; the channeiled absorption 
spectra in the red and green appear on the plate. 
Numerical Results. 
The results obtained by the three methods are given in 
the second column of the table in the section devoted to the 
refraction and dispersion of sodium vapour of great density, 
which follows. For the sake of uniformity they have all 
been reduced to the same scale. A shift of 100 helium (Ds) 
