208 
MESSES. EOSCOE AND THOEPE ON THE ABSOEPTION-SPECTEA 
In order to reduce the micrometer-scale readings of the absorption-spectra to wave- 
lengths, a graphical method was adopted. In the first place, the wave-lengths of the 
27 air-lines having been found as described, these numbers were taken as ordinates and 
plotted on curve-paper, the scale-readings of the air-lines being taken as abscissae. The 
points thus obtained fell naturally into six groups or series, and a mean point for each 
group was then found by accurate measurement, and the curve drawn through the six 
mean points. In order to obtain the wave-lengths of the well-defined edge of the dif- 
ferent bands of the absorption-spectra, it was merely necessary to ascertain from the 
curve the value of the ordinate corresponding to each scale-reading as abscissa. The 
positions of the six mean points, each corresponding to a group of air-lines needed for the 
construction of the curve, were ascertained by means of the arrangement seen in Plate 4. 
fig. 1. The wooden rod (A A) has a steel point (C C) fixed at one end, and a graduated 
paper scale at the other. At the end carrying the steel point is a strip of wood (T T P) 
placed at right angles to the rod, which it serves to support, so that the axis (0 C) of 
the steel point is vertical, whilst the scale (S S) lies in a horizontal position, the end P 
sliding on the table when the rod is rotated round the point C C as centre. The curve- 
paper is now adjusted on the table, so that some one division on the scale describes a 
circle through one particular group of spots and also through two other groups, one on 
each side, which circle, if drawn, would pass not far from the mean points of the three 
groups. 
In order now to find the exact position of the mean point of the middle group through 
which the curve is to pass, the distance from each point to the steel centre (C C) of the 
circle is first accurately read off on the divided scale. A point is then chosen and plotted 
near the middle of the group, as a mean of the several distances, such that the sum of 
the shortest distances of the several points of any one group to the curve drawn through 
this mean point is zero. 
This process is repeated for each of the four groups in the middle, the centre of cur- 
vature for an end group being taken to be the same as that for the second group from 
that end. In one case an isolated point occurs, and this is taken as one group and the 
curve drawn through it. 
The curve consists of arcs of circles which pass through the six mean points, with 
radii varying from 64^ inches in length at the upper end to 61^ inches at the lower end. 
Absorption-Spectrum of Bromine. 
This absorption-spectrum was first examined in 1833 by W. H. Miller, but no accurate 
map has hitherto been published. 
The bromine employed in our experiments was a portion of the sample used by one 
of us in the determination of the rate of thermal expansion of this liquid. It was 
obtained by distilling a large quantity of commercial bromine, after dehydration with 
oil of vitriol, and collecting separately the fraction boiling at about 60°, which amounted 
to about two thirds of the whole liquid operated upon. The sample was found to be 
