380 
DIFFUSE MATTER IN SPACE 
lines cannot be explained in this way because they are absorbed by neutral 
unexcited atoms which require no preparatory stimulation. This is a fatal 
objection to the theory. Moreover, we see no reason why the lines of un 
ionised calcium should not be imprinted on the light as it traverses the 
regions where the atoms are supposed to be unstimulated. 
Another possibility is that calcium and sodium do not exist in great 
quantity in the general cloud, but only in the neighbourhood of the hottest 
stars which eject them by selective radiation pressure or otherwise. After 
ejection the atoms are caught up in the cloud so that they have the 
motion of the cloud and not of the star. This view presents great difficulties 
when we attempt to consider it quantitatively. Remembering that the 
star is moving through the cloud, it cannot very well prepare a screen 
ahead of it. At any rate, the fixed lines should be stronger in receding 
stars than in approaching stars, and if the suggestion were seriously enter 
tained this correlation should be looked for. 
It appears then that we must turn to the alternative theory that the 
fixed lines are produced uniformly by absorption in interstellar space. They 
must accordingly be present in every type of star which is sufficiently 
distant, although it may be impracticable to detect them. Plausible 
reasons can be given why they have hitherto failed to appear. Below 
B 3 it is presumed that H and K begin to be prominent in the spectrum 
of the star proper, and the fixed lines could only be distinguished if the 
star had large velocity. From B 3 to B 8 the velocities are generally very 
small. Stars of lower types are generally not sufficiently remote to give 
the general space-absorption a fair chance, and the increasing multitude 
of lines in the spectrum makes the detection difficult in the lowest types. 
Stars with the necessary requirement of very great distance and large 
velocity can perhaps be found, and it may be that they will give decisive 
evidence for or against the theory; we do not know of any test of this kind 
yet tried*. 
It would seem that a valuable test could be obtained if an attempt 
were made to correlate the intensity of the fixed lines with the distances 
of the stars of types O-B 3. If these stars were grouped according to 
estimated distance a distinct relation should be found; and it is even 
possible that if the test proved satisfactory it would furnish a method of 
determining large stellar distances. Exceptions must however be expected, 
* [The star 66 Eridani, type B 9, mag. 5-2, affords evidence distinctly unfavour 
able to the theory. It is a spectroscopic binary with both spectra visible and the 
calcium lines follow the orbital motion. The relative velocity of the two components 
amounts to 220 km. per sec. so that there would be plenty of room for fixed lines to 
appear between the stellar lines, but none are observed (Frost and Struve, Astrophys. 
Journ. 60, p. 313). I can only suggest that its distance, estimated at 150 parsecs, 
might be insufficient to give the interstellar absorption a chance, but the excuse is 
not very satisfactory to me.]