viii 

temperature, and that was good enough for the early 
attempts at classification. Sir Norman Lockyer 
always stoutly maintained the existence of the ascend- 
ing and descending series ; but he was almost alone 
among spectroscopists in this. He did not actually 
succeed in separating the ascending and descending 
stars though sometimes he came very near to the 
right criterion. We owe to Russell and Hertzsprung 
the actual separation. They discovered it not by 
spectroscopy, but by measuring the absolute brightness 
of stars ; the greater brightness of the ascending stars, 
due to their large bulk, easily distinguishes them from 
the descending stars, at any rate in the low-temperature 
groups. At the highest temperatures the two series 
merge into one another. 
The disentangling of the two series, and the recogni- 
tion of the true sequence of stellar evolution, is probably 
the most revolutionary and far-reaching of recent 
discoveries in stellar physics. It began to oust the 
older view about 1914, and it is worth noticing that 
the discovery was made from observations coming under 
the province of the older astronomy and not what is 
generally called astrophysics. The data were parallaxes, 
proper motions, double star orbits, etc. The spectro- 
scopists had been misled as to the order of evolution, and 
it was left to the rival branch of astronomy to show the 
way; but they were not to be outdone for long. 
Adams and Kohlschiitter have found an easy spectro- 
scopic method for distinguishing the ascending and 
descending stars. Although our main purpose now 
is to grope in the interior of a star, perhaps we may 
emerge at the surface for a moment to consider what 
is the difference of surface condition of a diffuse 
and condensed star, respectively, which enables the 
spectroscope to distinguish between them. 
SurracE Conpirions. 
The state of the outermost layers of a star can, it 
would seem, be influenced by two factors only, (1) 
the intensity of the stream of radiant energy crossing 
through them and (2) the intensity of gravitational 
attraction holding them to the star. The former is 
measured by the effective temperature, so that we 
have the two variable factors, temperature and gravity. 
The spectrum presumably will vary as the conditions 
governed by these factors vary. We must not expect 
to be able to classify the spectra accurately in a single 
sequence ; they can vary in two directions. The 
ordinary classification depends principally on the 
temperature factor ; we may call this the longitudinal 
sequence. Adams’s new method aims at disentangling 
the transverse sequence corresponding principally to 
the gravity factor. We may say that his method is 
really a way of finding the value of gravity at the 
| 
Supplement to “ Nature,” May 12, 1923 
surface of a star, although it is not yet possible to 
put the value into actual numbers. Clearly, gravity 
will be smaller in the diffuse stage than in the dense 
stage on account of the greater distance from the 
centre to the surface. i! 
The effect of lowering gravity is to make the 
density smaller at corresponding temperature. This 
introduces an important change in the state of the gas, 
namely, ionisation. At moderately high temperatures 
the atoms begin to lose one or more of their most 
loosely attached electrons, a process called ionisation. 
Ionisation is facilitated by low density and prevented 
by high density. The theory of ionisation in stellar 
atmospheres has been chiefly worked out by M. N. 
Saha, who has arrived at many interesting results. 
Here we need only remark that the ionised atoms 
give rise to different spectra, which have long been 
distinguished from the spectra of the neutral atoms. 
The lower density in the atmosphere of diffuse stars 
should strengthen the “enhanced” lines due to 
ionised atoms, compared with the “arc” lines due 
to neutral atoms. The difference in general is not 
very large, but the atoms of certain elements for which 
the conditions are most critical, are specially sensitive 
to the change of density. This is the criterion which 
Adams and Kohlschiitter found empirically, and it 
distinguishes quite easily the ascending and descending 
series. To a limited extent it also distinguishes the 
larger and smaller stars within the same series. 
Although the stars begin to shine on reaching a 
temperature of about 3000° and return to this tem- 
perature at the close of their luminous existence, they 
do not all climb the temperature-ladder to thé same 
height. The more massive stars climb higher than 
the light stars. We can to some extent calculate 
the height to which they will go; but I am afraid 
the figures at present are very uncertain, though there 
is hope of improving them before long. The sun’s 
surface temperature is now about 5900°; I do not 
think that it ever went higher than 6600° ; it had 
not sufficient mass to go beyond. Sirius, nearly 24 
times as massive as the sun, has climbed to IT,000°, 
and at the moment is practically at its maximum, 
having only just turned downward. Still hotter 
stars like Rigel are known, and these must be more 
massive still. At the other end of the scale a star 
of mass less than 1/7 of the sun would not be able to 
reach 3000°, and could scarcely be luminous ; but 
in any case such small masses would be formed very 
seldom, for the reason explained earlier in this lecture. 
It is a well-known fact that hot stars on the average 
are more massive than cool stars ; we see that this 
is accounted for by the smaller stars being weeded’ 
out as the temperature-standard is raised. 
y 

