42 



A FEW NOTES ON THE USE OF GKATINGS 



[Johns Hopkins University Circulars, No. 73, pp. 73, 74, 1889] 



The ghosts are very weak in most of my gratings. They are scarcely 

 visible in the lower orders of spectra, hut increase in intensity as com- 

 pared with the principal line as the square of the order of the spectrum. 

 Hence, to avoid them, obtain magnification by increasing the focal dis- 

 tances instead of going to the higher orders. The distances from the 

 principal line in my gratings are the same as the distances of the spectra 

 from the image of the slit when using a grating of 20 lines to the inch. 

 They are always symmetrical on the two sides, and about -^ inch for 

 the violet and inch for the red in a grating of 21 ft. 6 in. radius in all 

 orders of spectra. When the given line has the proper exposure on the 

 photographic plate, the ghosts will not show, but over-exposure brings 

 them out faintly in the third spectrum of a 20,000 grating or the 6th of 

 a 10,000 one. They never cause any trouble, as they are easily recog- 

 nized and never appear in the solar spectrum. In some cases the higher 

 orders of ghosts are quite as apparent as those of the first order. 



The gratings with 10,000 lines to the inch often have better definition 

 than those of 20,000, as they take half the time to rule, and they are 

 quite as good for eye observation. They can also be used for photo- 

 graphing the spectrum by absorbing the overlying spectra, but there 

 are very few materials which let through the ultra violet and absorb the 

 longer wave-lengths. The 10,000 gratings have the advantage, how- 

 ever, in the measurement of wave-lengths by the overlapping spectra, 

 although this method is unnecessary since the completion of my map of 

 the spectrum. By far the best is to use a 20,000 grating and observe 

 down to the D line by photography, using erythrosin plates from the F 

 line down to D. Below D, cyanine plates can be used, although the time 

 of exposure is from 10 to 60 minutes with a narrow slit. The solar 

 spectrum extends to wave-lengths 3000, and the map has been contin- 

 ued to this point. Beyond this, the coincidence with the solar spectrum 

 cannot be used, but those of the 1st and 2d or 2d and 3d spectra can be. 



Some complaints have been made to me that one of my gratings has 

 no spectrum beyond 3400. even of the electric arc. I have never found 

 this the case, as the one I use gives w. 1. 2200, readily with 30 minutes 

 exposure on slow plates, requiring 5 minutes for the most sensitive 



