NO. 19 ENERGY SPECTRA OF STARS — ABBOT AND ALDRICH 5 



dismount the radiometer suspension and pack it into a little wooden 

 box without any chance of breaking the invisibly fine quartz fiber, 

 or the delicate suspension. In this box he carried it in his suitcase to 

 Mount Wilson, and within a half hour after arrival at the observing 

 room hung it, entirely unharmed, within the radiometer case, pre- 

 viously leveled. Two powerful magnets reduced the time of single 

 swing in air at radiometer pressure (about 0.2 mm. Hg.) from 

 about 2 minutes when free, to about 10 seconds under magnetic con- 

 trol. With two other magnets coarse and fine adjustments in azimuth 

 could be made, so that the spot of light for reading purposes could 

 be brought readily to any part of the reading scale at 5 meters 

 distance. The light spot was furnished by one short length of the 

 filament of a 200-watt Mazda lamp, situated about 2.5 meters above 

 the radiometer. Two pinhole diaphragms 2 meters apart limited the 

 beam to about 6 millimeters diameter where it fell upon the quartz 

 plate through which it entered the radiometer. The spot was further 

 largely shorn away there by other obstructions. Still it is feared that 

 too much light entered the radiometer in this beam. Variations of 

 voltage of the lamp may have produced radiometrically minute vibra- 

 tions of the suspension. The ordinary range of these vibrations on 

 the scale at 5 meters was less than i millimeter, and sometimes hardly 

 observable at all, yet it is hoped to reduce this range in future 

 experiments. 



The positions of the light spot on the scale, as the spectrum was 

 swung from one vane to the other of the radiometer by pulling a 

 cord, were read on the special device described and shown in figure 4 

 of Mount Wilson Observatory Contribution No. 380, of 1929. The 

 places of the light spot were observed on a divided circle of 100 divi- 

 sions, one complete revolution of which corresponded to a movement 

 by a screw of ^ inch, approximately 3 mm. So the positions were 

 recorded to 0.03 millimeter. Maximum deflections in the spectra of 

 stars ranged from about 100 to about 300 divisions. It was found in 

 reducing the observations that the average individual deviation from 

 the mean in a set of readings was 42 divisions. Generally four swings 

 from one vane to the other were made at each wave length. Hence 

 the average deviation of the mean of such a set of readings would be of 



— = 21 divisions. This is from 20 down to 7 percent of the deflec- 

 tion at maximum in the spectra of the stars observed. 



This percentage accidental error is obviously far too great to give 

 satisfactory spectral energy curves. In future work we hope to re- 

 duce it: I, by increasing the radiometer effect by substituting hydro- 



