EFFECT OF PRESSURE UPON ARC SPECTRA. 
153 
undoubtedly causes a decrease in the amount of metallic vapour surrounding the arc, 
there must be a large number of atoms capable of absorbing certain vibrations even 
at 100 atmospheres, since at that pressure broad reversals are still apparent for the 
lines cO, c/0, d2, &c. 
It is interesting to note that it is at the pressure at which the maximum number 
of reversals occurs (20 to 25 atmospheres) that the change occurs in the amount of 
condensation upon the window of the pressure cylinder, mentioned on p. 131. This 
suggests that the size of the chamber possibly has some influence in determining the 
pressure at which the maximum number of reversals is found. 
The reversals are of two types, (a) symmetrical, and (/3) unsymmetrical. Examples 
of (a) are 63, 64, cO, c/0, c/2, Plate 4, and el, el, Plate 8. Example of (/ 3 ) is e3, 
Plate 8. 
In the measurement of ail reversed lines the centre of the reversed portion has 
always been taken as its true position. For symmetrical lines this is probably strictly 
correct, and for the accuracy of the measurements it is fortunate that only lines of 
this type are found with broad reversals. It may not be accurate for type (3, but any 
error introduced is not large, because the total width of a reversal is only a small 
percentage of the displacement associated with lines of this type. 
It not infrequently happens that the same line occurs on two different plates taken 
at the same pressure with different amounts of reversal (see page 149). Anderson* 
has discussed the possible effect of the duration of the exposure on the widths of 
reversed lines, but the plates taken during the progress of this research were all 
sufficiently exposed for this cause to be negligible, and the determining factors are 
the amount and the temperature of the vapour in front of the arc, which are liable to 
variations on account of the fluctuations in the current which accrue from the 
continual striking of the arc, and from its comparatively short duration. 
On this account the photographs show an integration of the effects of the 
temporary distributions of the vapour round the arc, and in view of this fact the 
labour of measuring the widths of the reversals has not been undertaken, especially 
as the widths are different on different plates at the same pressure, but inspection of 
the plates brings forward the fact that in general those Ivies ivhich reverse similarly 
are also displaced by equal amounts under pressure. For example, the lines 64, c/0, 
c/2, el are reversed symmetrically and have approximately the same widths of reversal, 
and their wings are of the same nature ; it will be seen, too, that their displacement 
curves also agree. Lines 61 , cl, c/1, eO also reverse similarly and equally, and they 
are displaced to the same extent (Diagram VIII., p. 151). The converse is not 
necessarily true. 
The reversals shoiv no tendency to undergo change of type as the pressure is 
increased; those lines that at low pressures are unsymmetrically reversed are, if 
reversed at all at higher pressures, still of type /3. Those of type a preserve their 
* Anderson, 1 Astrophysical Journal,’ XXIV., 238, 1906. 
VOL. CCVIII.-A. 
X 
