CELESTIAL SPECTROSCOPY. 81 
perature to which hydrogen must be raised before it can show its char- 
acteristic emission and absorption, we shall probably be right inattribu- 
ting the relative feebleness or absence of the other lines, not to the 
paucity of the metallic vapors, but rather to their being so hot relatively 
to the substances behind them as to show feebly, if at all, by reversion. 
Such a state of things would more probably be found, it seems to me, 
in conditions anterior to the solar stage. A considerable cooling of the 
sun would probably give rise to banded spectra due to compounds, or 
to more complex molecules, which might form near the condensing 
points of the vapors. 
The sun and stars are generally regarded as consisting of glowing 
vapors surrounded by a photosphere where condensation is taking 
place, the temperature of the photospheric layer from which the greater 
part of the radiation comes being constantly renewed from the hotter 
matter within. 
At the surface the convection currents would be strong, producing a 
considerable commotion, by which the different gases would be mixed 
and tot allowed to retain the inequality of proportions at different 
levels due to their vapor densities. 
Now the conditions of the radiating photosphere and those of the 
gases above it, on which the character of the spectrum of a star depends, 
will be determined, not alone by temperature, but also by the force of 
gravity in these regions; this force will be fixed by the star’s mass and 
its stage of condensation, and will become greater as the star continues 
to condense. 
In the case of the sun the force of gravity has already become so 
great at the surface that the decrease of the density of the gases must 
be extremely rapid, passing in the space of a few miles from atmos- 
pherie pressure to a density infinitesimally small; consequently the 
temperature-gradient at the surface, if determined solely by expansion, 
must be extremely rapid. The gases here however are exposed to the 
fierce radiation of the sun, and unless wholly transparent would take 
up heat, especially if any solid or liquid particles were present from 
condensation or convection currents. 
From these causes, within a very small extent of space at the surface 
of the sun, all bodies with which we are acquainted should fall to a con- 
dition in which the extremely tenuous gas could no longer give a vis- 
ible spectrum. The insignificance of the angle subtended by this space 
as seen from the earth should cause the boundary of the solar atmos- 
phere to appear defined. If the boundary which we see be that of the 
sun proper, the matter above it will have to be regarded as in an essen- 
tially dynamical condition—an assemblage, so to speak, of gaseous pro- 
jectiles, for the most part falling back upon the sun after a greater or 
less range of flight. But in any case it is within a space of’ relatively 
small extent in the sun, and probably in the other solar stars, that the 
H, Mis, 334, pt. 1——6 
