400 PROCEEDINGS OP THE AMERICAN ACADEMY 



errors of this scale, and a determination of the values of its divisions. 

 The results show that the maximum error in one division is 0.003 

 scale div., a quantity entirely insensible in the present investigation, 

 since at best it is possible to estimate to hundredths of a division. 

 The actual value of one division is jg^jjth of an inch = 0.053 of a 

 millimetre. 



During the progress of the work two ends were held in view : First, 

 to obtain data for comparing the appearance of the group under differ- 

 ent conditions of the atmosphere, of dispersive power, etc. ; and sec- 

 ondly, to measure the positions, and thus determine the wave-lengths 

 of all lines seen with certainty within the assumed limits. For the 



o 



first purpose mentioned, an enlarged copy of Angstrom's chart was 

 prepared, and below this were roughly sketched, on the various days 

 of observation, the lines which were seen, — the grating and power 

 being noted in the margin. All lines which could be measured were 

 afterwards accurately entered upon the chart, and those which were 

 not seen a second or third time were entered by estimation. In this 

 way I obtained twenty-three observations of the group, between 

 November, 1879, and May, 1880, using eight diffraction gratings and 

 two prisms. Four of the gratings were ruled by Rutherfurd, of 

 17,296, 8,640, 13,321, and 7,021 lines per inch respectively; three 

 by Professor Rogers, of 2,540, 2,540, and 5,080 lines per inch; 

 one by Brunner, of 2,540 lines per inch. Grating 17,296 is on 

 speculum metal, the ruled surface measuring 4.3 centim. in length 

 by 4.8 centim. in width. Observations were usually made upon 

 the spectrum of the second order. Dispersive power = 1860. 

 Grating 13,321, silvered glass. Grating 7,021, on speculum metal 

 covered with glass. The three gratings of 2,540 lines were ruled 

 upon glass prepared by Brunner Freres, and were used for trans- 

 mitted spectra. Grating 5,080 was ruled upon ordinary '' commercial 

 crystal " glass, and seems far superior to several gratings ruled upoa 

 better quality of glass, and showing less error in the spacing and cut 

 of the lines. This grating was also used for transmitted spectra, A 

 comparison of the three 2,540 gratings seems to show that those ruled 

 by Professor Rogers are fully equal, if not superior, to Brunner's. 

 In the preparation of the chart given in Plate II., Rutherfurd's grat- 

 ing of 8,640 lines was used. This is ruled upon speculum metal with 

 the following memorandum upon its edge: — "Aug. 2, 1877, 8,640 per 

 in,, 17,640 spaces. Man'f by D, C. Chapman with Mr. Rutherfurd's 

 Engine. 175, 2 Ave., N. Y." The length of the lines is 4.3 ceiitira., 

 and the width of the ruled surface 5.2 centim. As far as working 



