246 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



trum. A dividing engine with a ^ mra. screw was used. This screw, 

 wlieu used to measure lines on a six-foot grating possesses for a run of 

 20 cm. an accuracy better than one one-hundredth of an Angstrom unit. 

 The plate was observed by means of a stationary microscope fitted with 

 a four-inch objective and micrometer eye-piece. The value of one cen- 

 timeter of plate length in terms of Angstrom units was determined by 

 measuremeuts taken between two given sun lines, located by Rowland's 

 map. In order to get a photograph of the sun spectrum and the mag- 

 nesium spectrum on the same plate, a shutter was placed before the slit 

 and so adjusted that the upper or lower half could be exposed at plea- 

 sure. By this means two narrow spectra were obtained, one above the 

 other. The method is preferable, when working with the first spectrum, 

 to the usual mode of protecting the plate itself by a swinging screen. 



When it became necessary to test the nature of the li^ht 

 producing the lines at 1873 /x and 924/1. in the magnesium 

 spectrum, the following arrangement was adopted : The 

 quartz prism of an angle of five degrees was placed some 30 

 cm. distant from the slit, but not on the straight line joining 

 slit and grating. Thus when the source was in position A, 

 the light did not fall upon the prism ; but when it was in 

 position B, the light passed through the prism. In the first 

 case, the upper half of the slit was closed by the shutter : 

 in the second, the lower half was closed. The result was, 

 of course, to produce two spectra, one above the other ; the 

 one due to light which had passed through the prism was 

 thus shifted toward the red. The general appearance may 

 be seen from Plate No. I. 



It now becomes necessary to consider more in detail the 

 behavior of the light of these ultra-violet spectra when 

 passed through the quartz prism. The results for magne- 

 sium may be seen from the following table. The lettering 

 of the groups on the plans of the spectra is, of course, 

 arbitrary. The chief or real groups are marked with an 

 unsubscripted letter. The reproductions are marked with 

 letters bearing a subscript. The jjlates are from drawings 

 made from the original photographs. Beginning with mag- 

 nesium Plate II., six principal groups of lines are to be 

 noted ; they extend from the pair of lines marked Ai, — 

 wave length 1966 /x, — to the strong group C4, — wave 

 Besides these strong lines, there are six fainter groups 



B 



FiGUEK 2. 



length 924/1. 



