EFFECT OF PRESSURE UPON ARC SPECTRA. 
149 
Hale and Kent # have remarked, for the spark discharge in compressed gases, that 
when a bright line is beginning to show signs of reversal, or when bright and dark 
lines occur in pairs, the observed pressure shifts are irregular, probably because the 
overlapping lines prevent settings from being made on their true centres.” The 
phenomenon is, however, in the present research presented by lines of Set A which 
are symmetrically reversed, and these should not be affected by overlapping. For 
unreversed lines belonging to Group III. (which reverses unsymmetrically) the 
existence of a faint absorption line on the edge of an emission line would imitate a 
very large displacement of the bright line, as Hale and Kent suggest, but these 
have been searched for carefully, and no displacements are included in this paper in 
which there is any suspicion of error due to such a disturbing cause. Plate 6, fig. 2, 
shows two lines, fl and f 2, belonging to Group III. ; the former is reversed, but the 
latter presents no trace whatever of any absorption which might affect the accuracy 
of the readings. The evidence points to the phenomenon being due to pressure and 
not to errors of observation. 
Table VI. 
(The Displacements are in Thousandths of an Angstrom Unit.) 
Mean values of 
Atmospheres. 
Plates. 
Number of 
reversals. 
displacements of Groups 
I. 
II. 
III. 
25 | 
D19„ 
431 
102 
215 
203 
D18„, 21 6 , 22, 
15/ 
45 
120 
212 
20 | 
D8 0 , 9 fl , 16 ff 
431 
54 
194 
490 
D27, 
24/ 
52 
116 
280 
15 { 
D19« 
261 
52 
94 
310 
D24 6 , 25 b 
18/ 
43 
95 
171 
10 { 
D3R 
261 
27 
74 
166 
7 ft 
19/ 
23 
63 
118 
Ratios of displacement, 
Group 
I. 
1*2 
1 
1-2 
1 
II. 
1-7 
1 
1-2 
1 
III. 
l = 2f 
1 
1-8 
1 
1-8 
1 
1-4 
1 
t See next section, p. 151. 
The evidence that favours a connection between the tendency to reverse and the 
abnormal displacements is comprised in Table VI., in which the displacements of the 
lines shown by a photograph are compared with the number of reversals on that 
photograph. In those cases in which two or more photographs show the same 
number of reversals at the same pressure, the mean values of the displacements have 
been taken. The table shows that, in all cases in which there is a discrepancy 
* Hale and Kent, ‘Publications of the Yerkes Observatory,’ Vol. HI., Pt. II., 1907, 
