252 CARNEGIE INSTITUTION OF WASHINGTON. 



produce a deflection of 0.2 mm. and the deflection can be measured with 

 certainty to 0.02 mm. A Leeds and Northrup Coblentz-Thompson galvan- 

 ometer was secured during the year, and after some modifications has been 

 used within an evacuated tube. 



REGISTERING MICROPHOTOMETER. 



Mr. Pettit and Mr. Nicholson have constructed a thermopile of six junctions 

 to replace the photoelectric cell formerly used in the registering microphoto- 

 meter, and have investigated its behavior under a variety of conditions. 



Since the thermopile can not be compensated for variations in the source of 

 illumination, the constancy of this source is of vital importance. Various 

 lamps were tested, using both alternating and direct current from the gen- 

 erators. Fluctuations of 5 per cent in the radiation over periods of from 0.5 

 to 2 seconds were found with both currents. A storage battery and straight- 

 filament lamp were adopted finally. For some of the denser plates artificial 

 illumination proved to be unsatisfactory, and for these sunlight has been 

 employed with such success that it has been used for a large part of the 

 work. 



The accuracy with which the machine registers the density gradient of a 

 plate depends in large measure on the width of the slit in front of the thermo- 

 couple. If this is made too small, however, the effects of the clustering of the 

 silver grains and defects in the film produce false depressions and elevations in 

 the curves. This diflticulty has been overcome by placing two parallel plane 

 mirrors in the microscope barrel, which increase the angular aperture of the 

 beam brought to a focus on the thermopile and thus integrate the light from 

 over a greater range on the plate. This arrangement also produces greater 

 galvanometer deflections. The first slit above the plate has been removed 

 and a projected image from a collimator substituted. Through the addition of 

 a multiplying gear, installed in series with the original gear, ratios as high as 

 500 may be obtained between the microscope stage and registering apparatus. 



If the machine is to be operated at a fair speed, a galvanometer of short 

 period should be used in order to avoid distortion due to lag; but, on account of 

 vibrations in the building, it has been found impracticable so far to use a 

 period of less than 4 seconds. It will be necessary to employ some form of the 

 Julius suspension before the short-period galvanometer can be used. 



About 50 plates have been taken with the machine for various purposes. A 

 comparison of the density curves of spectral lines obtained with the photo- 

 electric cell and with the thermo-couple shows that considerable improvement 

 has been made. The applicability of the machine to the measurement of 

 radial velocities was tested with a spectrogram of Capella taken with a disper- 

 sion of 3 prisms. The curve for a region 8 mm. long near X4200 was made on 

 the machine with a similar curve for the comparison spectrum. The measure- 

 ment of these curves with a millimeter scale gave 70.3 km., while the velocity 

 from the spectrogram measured in the usual way was 69.7 km. Curves have 

 also been obtained of the iron line X6173 in the spectrum of sun-spots to test 

 the reality of the apparent displacement of the p-component, and some tests 

 have been made of the use of the machine as a surface photometer for measur- 

 ing the density of star images and the distribution of density in extended 

 objects, such as the solar corona and nebulae. 



