520 HENRY A. KOWLAND 



part and using the electric arc. With sensitive plates, the time can be 

 diminished to one-fifth of this. 



For eye observations, a very low power eye-piece of 1 or 2 in. focus 

 is best. This, with a focus of 21 ft. 6 in. is equivalent to a plane grat- 

 ing with a telescope of a power of 100 or 200. 



In measuring the spectra, an ordinary dividing engine with errors 

 not greater than 10*00 inch can be used, going over the measurements 

 twice with the plate reversed between the separate series. The plates 

 are on so very large a scale that the microscope must have a very low 

 power. The one I use has a 1 inch objective and a 2 inch eye-piece. 

 The measured part of the plate is about a foot long, the plates being 

 19 in. long. 



All the spectrum photographs taken at different times coincide per- 

 fectly, and this can be used for such problems as the determination of 

 the atmospheric lines. For this purpose, negatives at high and low 

 sun are compared by scraping the emulsion off from half the plates and 

 clamping them together with the edges of the spectra in coincidence. 

 The two spectra coincide exactly line for line except where the atmo- 

 spheric lines occur. 



This method is specially valuable for picking out impurities in metal- 

 lic spectra, using some standard impurity in all the substances to give 

 a set of fiducial lines; or better, obtaining the coincidence of all the 

 metals with some one metal, such as iron. Making the iron spectrum 

 coincide on the two plates, the other spectra can be compared. This is 

 specially possible because the focus of a properly set up concave grating 

 need not be altered in years of use, for, when necessary, it can be ad- 

 justed at the slit, keeping the distance of the grating from the slit con- 

 stant. 



The spectrum of the carbon poles is generally too complicated for 

 use with anything except the more pronounced lines of metals, there 

 being, at a rough guess, 10,000 lines in its spectrum. However, in pho- 

 tographing metallic spectra but few of these show on the plate, as they 

 are mostly faint. The spark discharge gives very nebulous lines for 

 the metals. 



Most gratings are ruled bright in the higher orders, but this is more 

 or less difficult, as most diamond points give the first spectrum the 

 brightest. Indeed, it is very easy to obtain ruling which is immensely 

 bright in the first spectrum. Such gratings might be used for gaseous 

 spectra. Short focus gratings of 5 ft. radius of curvature, very bright 

 in the first order, require only a fraction of a second exposure for the 

 solar spectrum and the spectrum of a gas can be obtained in less than 

 an hour. 



