588 HENKY A. EOWLAND 



made by the incident and diffracted rays, and is independent of the 

 number of lines per unit of length ruled on the grating. If this num- 

 ber is too small, however, the different order of the spectra will be too 

 much mixed up with each other for easy vision. A convenient number 

 is 15,000 to 20,000 lines to the inch, or from 6000 to 8000 to the 

 centimetre. The defining power is defined as the ratio of the wave- 

 length to the distance apart of the two spectral lines which can be just 

 seen separate in the instrument. Thus the sodium or D lines have 

 wave-lengths which differ from each other by -597 ftp, and their aver- 

 age wave-length is 589-3 pp. A spectroscope to divide them would 

 thus require a defining power of 988. The most powerful gratings have 

 defining powers from 100,000 to 200,000. Lord Eayleigh's formula for 

 the defining power is 



D = Nn. 



When D is the defining power, N is the order of the spectrum, and n 

 is the total number of lines ruled on the grating. As the defining 

 power increases with N, and since we can observe in a higher order as 

 the number of lines ruled in a unit of length decreases, it is best to 

 express the defining power in terms of the width of the grating, w. In 

 this case we have for the maximum defining power D' = 20,000 w for 

 small gratings, or D' = 15,000 w for extra fine large gratings, w being 

 the width of the gratings in centimetres. It is seldom that very large 

 gratings are perfect enough to have a defining power of more than 

 10,000 w, owing to imperfection of surface or ruling. The relative 

 brightness of the different orders of spectra depend upon the shape of 

 the groove as ruled by the diamond. No two gratings are ever alike 

 in this respect, but exhibit an infinite variety of distributions of bright- 

 ness. Copies of glass gratings can be made by photography, contact 

 prints being taken on collodiochloride of silver or other dry plates. 

 Eeflecting gratings can be copied by pouring collodion or gelatine over 

 the grating and stripping off the films thus formed. The latter warps, 

 however, and destroys the definition to a great extent. The grating 

 always produces a brighter spectrum in the violet than a prism. In 

 the green the reflecting speculum metal grating may be brighter than 

 a prism spectroscope of five prisms, and for higher dispersion surpasses 

 the prism spectroscope both in definition and brightness in all portions 

 of the spectrum. 



To produce the pure spectrum from flat gratings, two telescopes are 

 generally used, as in Fig. 1. 



