136 CARNKGIE INSTITUTION 



problem relates to the distribution in tlie heavens of stars of various 

 spectral types. The extensive photographic surveys of the Harvard 

 Obsen-atory have provided the most abundant material for this 

 study. The method of placing prisms over the telescope objective, 

 employed by Fraunhofer and Secchi in their visual work, is here 

 applied photographically. Thus the spectra of several hundred 

 stars are obtained on a single plate. The entire sky has been pho- 

 tographed in this way. An examination of these plates has led to 

 the discovery of a large number of objects of interest, among them 

 six novse, seven spectroscopic binaries, 75 stars of the fifth type, 

 nearly 200 variable stars, besides 500 whicli occur in clusters. A 

 detailed study has also been made of the spectra of the brighter stars 

 north and south, several hundred lines being sometimes discussed in 

 the spectrum of a single star. 



Stella>- Spectroscopes. — A brief consideration of the steps by which 

 the instruments emplo3^ed in stellar spectroscopy have been ejn- 

 ployed should be instructive in its bearing on further progress. 

 The earlier spectroscopes gave small dispersion, and in most cases 

 were merely the ordinary sm.all instruments of the laboratory at- 

 tached to telescopes. When photography was first applied by Hug- 

 gins it became apparent that special devices would be required to 

 keep the star on the slit throughout the exposure. Hence his use 

 of polished slit jaws, on which the reiiected stellar image could be 

 watched with an auxiliary telescope. The attempt to measure ra- 

 dial velocities made higher dispersion necessarj- and resulted in 

 Vogel's specially designed spectrograj/ii, in v^^hich freedom from 

 flexure was an essential feature. It soon appeared, however, that 

 if the prisms varied appreciably in temperature during the exposure, 

 displacements of the lines would result. Hence Deslandres intro- 

 duced apparatus for maintaining the prism box at a constant tem- 

 perature, now universally employed. The success of the Mills 

 Bpectrograph was due to no radical change in design or method, but 

 rather to the use of a larger, stifTer, and more powerful instrument, 

 improved in both optical and mechanical details, and employed under 

 favorable conditions with a telescope of great light-gathering power. 

 Frost's design for the Bruce spectrograph is a further step in the 

 same direction. Each of the attachments for producing the com- 

 parison spectrum recently adopted at Mt. Hamilton and Potsdam 

 embodies a distinct element of advance. In the new Chile spectro- 

 graph Campbell has introduced a radical modification of the mechan- 

 ical design, for the purpose of reducing flexure. The optical parts 



