greatest difficulty that the lines could be distinguished at all. Finally another 

 observer was requested to determine the cftect, if any, upon tlie relative bright- 

 ness of tlie spectra on dilYerent sides of the collimator, produced by varying the 

 inclination of the grating to the collimator and the results agreed precisely with 

 the foregoing hypothesis. 



There is another defect in this grating which seems to me to be likely to 

 exist in all gratings of a similar space and magnitude. It is the faintuessof the 

 illumination at the two extreme ends of the spectrum, especially at the violet end. 

 In the American -Tiiurnal of Science for November 1880., Professor Young, in 

 describing the excellent ipialities of a similar grating, makes the same comjilaint 

 in regard to the violet end of the spectrum. I am inclined to the opinion that 

 the cause of this is to be looked for in the minor periodic variations in the third 

 factor of the expression for intensity of illumination, referred to above. Besides 

 the great maxima of this fimction which determine the location of the spectra of 

 the various ordei^s, there are numerous others of extremely small magnitude 

 compared with the first. They are distributed throughout the whole range 

 within which the great maxima are visible and, as they contain the spectral 

 components of white light, they will tend, notwithstanding their extreme faint- 

 ness, to produce a general illumination of the whole field, so as to destroy, to 

 some extent, the blackness of the background against which the ordinary 

 spectra are seen. Fraunhofer called these "spectra of the second class" and 

 Angstrom refere to them in his memoir in a discussion of the relative merits of 

 gratings in which a given number of lines is made to cover a great or a small 

 space. — The grating space being constant the number of these minor spectra in 

 a given space will 1« proportional to the number of lines. But as the brightness 

 of the great spectrum at anv point varies a.s the square of the number of lines it 

 would seem that the one could never overtake the other. When we consider, 

 however, that the minor spectra contain all of the components of white light it 

 seems probable that where, as in the extreme ends of the ordinary or great 

 spectra, the light is, at best, exccH3dingly faint, that of the minor .«pectra may be 

 sufficient to nearly overpower it and to render observations difficult in these 

 regions. If this be correct it must be admitted that, while in general there is 

 great advantage in the use of gratings of great fineness, for the examination of 

 the extreme regions of tiie sixjctrum a smaller nmubcr of lines covering the same 

 space would yield better re.'iults. In the present instance it will he observed that 

 neither the A nor the H lines were included in the measurements as they could 

 not 1« observed with sufficient satisfaction to make a mea.surement desirable. 



MKTII0D8 OF OUSKUVATION. 



In making tlie measurements necessary to the calculation of the wave-length 

 of any line, the essentia] jiarts of the apparatus, aside from the graduated circle, 

 are the slit with collimatiug len.s, the grating and the telescope. In general 

 there will be three ditTerent arrangements |)0S8ible. 



