126 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1919. 
4. The spectrum of the spirals is practically the same as that given 
by a star cluster, showing a continuous spectrum broken by absorp- 
tion lines. A few spirals show bright-line spectra in addition. 
5. The space-velocities of the various classes of celestial objects are 
summarized in the following short table : 
THE FACTOR OF SPACE VELOCITY. 
1. The diffuse nebulae. 
Velocities low. 
2. The stars. 
Velocities vary with spectral type. 
Class B stars : Average speeds 8 miles per second. 
Class A stars : Average speeds 14 miles per second. 
Class F stars: Average speeds 18 miles per second. 
Class G stars : Average speeds 19 miles per second. 
Class K stars : Average speeds 21 miles per second. 
Class M stars : Average speeds 21 miles per second. 
3. The star clusters. 
Velocities about 100 miles per second. 
4. The planetary nebulae. 
Average speeds 48 miles per second. 
5. The spiral nebulae. 
Average speeds 480 miles per second. 
The peculiar variation of the space velocity of the stars with 
spectral type may ultimately prove to be a function of relative mass. 
The radial velocities of but few spirals have been determined to date ; 
future work may change the value given, but it seems certain that it 
will remain very high. 
It will be seen at once that, with regard to this important criterion 
of space velocity, the spiral nebulae are very distinctly in a class 
apart. It seems impossible to place them at any point in a coherent 
scheme of stellar evolution. We can not bridge the gap involved in 
postulating bodies of such enormous space velocities either as a point 
of stellar origin, or as a final evolution product. 
On the older theory that the spirals are a part of our own galaxy, 
it is impossible to harmonize certain features of the data thus far 
presented. If this theory is true, their grouping near the galactic 
poles, inasmuch as all evidence points to a flattened or disk form for 
our galaxy, would indicate that they are relatively close to us. In 
that event, we should inevitably have detected in this class of objects 
proper motions of the same order of magnitude as those found for the 
stars at corresponding distances. Such proper motions are the more 
to be expected in view of the fact that the average space velocity of 
the spirals is about 30 times that of the stars. I have repeated all the 
