1869.] On some Becent Spectroscopic Researches. 217 



equal to about seven prisms was needed to detect the amount of 

 motion wliicli was to he expected. 



The circumstance that the writer possessed two fine compound 

 prisms by Hofmann, of Paris, and some other considerations, led to 

 the construction of an apparatus of the form represented in the 

 woodcut on the opposite page, a is an adjustable slit, & an 

 achromatic collimatiug lens, of 4' 5 inches focal length; c repre- 

 sents the small telescope with which the spectrum is viewed. The 

 train of prisms consists of two compound prisms, giving direct 

 vision, d and e, and three simple prisms, /, g, h. The compound 

 prism marked e is permanently connected with the telescope c, 

 with which it moves. The compound prism d is so fixed that 

 it can be removed at pleasure when the dispersive power of the 

 instrument is reduced from about six and a half prisms of 60° 

 to about four and a half prisms of 60°. It was with this 

 instrument that the observations were made; but an apparatus 

 in many respects superior has since been constructed for the 

 writer. 



A difficulty of some importance, and which had been expected, 

 had now to be overcome. It was necessary to contrive a method 

 by which the light from a substance to be compared with a star 

 could be made to pass through the prisms in such a manner that 

 the spectrum of comparison should be with certainty in absolute 

 coincidence with the spectrum of the star. When the light to be 

 compared is reflected into the instrument in the usual manner, by 

 means of a small prism placed before the slit, a very small alteration 

 of the position of the source of the light will cause the two spectra 

 to shift upon each other to an amount much larger than the small 

 difierences of position which were to be sought for. In the former 

 researches of the writer this source of possible error was constantly 

 kept in view, and guarded against by a frequent comparison of the 

 lines of sodium reflected into the instrument with those from a flame 

 containing sodium placed before the object-glass. When the lines 

 of sodium, as seen in both spectra, were truly coincident, the appa- 

 ratus was considered to be in perfect adjustment. This method, 

 however, was not found to be trustworthy for the more delicate 

 investigation now in hand, since there existed the risk of a slight 

 accidental displacement of the spark, or of the mirror by which it 

 was reflected into the instrument, which would produce an error 

 much more serious with the larger dispersive power now in use. 



After the trial of some other methods, an arrangement was 

 adopted which was found to be perfectly trustworthy. Two j^ieces 

 of thin glass, silvered by floating them in a silvering solution, were 

 fixed at an angle of 45° in front of the slit, a small space of about 

 Vo- inch being left between them for the passage of the pencils of 

 light from the object-glass. The invariability of position of the 



