THE OUTSIDE OF A STAR 
349 
(5) He + . A4686. Still increasing in the hottest type O -— 
P e = 5-1.10- 6 , T = 30,000, 
M.10- 4 , T = 35,000. 
(6) He. A5876, 4471, 4026. Maximum in type B 2— 
P e = 5-9.10~ 6 , T = 14,000, 
P e =l-2.10- 4 , T= 16,000, 
P e = 4-0.10 -4 , T = 17,000. 
In types M to A where the temperature scale is fairly well known the 
results are used to determine P e in the reversing layer. Having discovered 
in this way the general order of magnitude of P e , we can use the results 
of examples (5) and (6) to extend the temperature scale to types B and O. 
243. Many other examples will be found in the papers quoted, and the 
general body of evidence is that ordinarily P e is about 10~ 4 atmospheres. 
This then may be taken to be the average pressure in the reversing layer. 
(Of course, the pressure cannot be just the same through the whole series 
of stars, but as mentioned in § 233 the range is moderately small.) But 
certain lines give much lower pressures (e.g. Ca + in example (4)) and are 
evidently formed at high levels where the pressure is of the order 10 -7 to 
10~ 8 atmospheres. 
Fowler and Milne remark that it is the principal lines of the elements 
that give the exceptionally low pressures. Their view is that, since atoms 
in a state to absorb a principal line are about 10 4 times as abundant as 
those absorbing subordinate lines, a comparatively thin layer suffices to 
produce full absorption; hence in the wave-length of the principal line 
we can see only a little way into the star, and the high level and low 
pressure are accounted for. We do not think this explanation can be correct. 
If the material is considered to be so superabundant that it prevents us 
seeing down to the ordinary reversing layer, abundance ceases to be a 
matter of primary concern. So long as we have enough to form a practically 
opaque screen in front of the region from which nearly all the photospheric 
radiation comes, i.e. in front of the reversing layer, variations of abundance 
will scarcely affect the appearance of the line. Further, if Fowler and 
Milne’s view were true the spectral types at which the line just appears 
and disappears would still be determined by the reversing layer conditions, 
and not by the high level pressure which determines the maximum. 
It seems more likely that the clue to this deviation is given by Milne’s 
work on the chromosphere (§ 252). The elements which show the deviation 
are subject to strong selective radiation pressure—which is associated 
with the occurrence of strong principal lines in the intense part of the 
general spectrum.