Lord Rayleigh's Investigations in Optics. 55 



tion. Such a result may be anticipated when with full aper- 

 ture the top and bottom of the spectrum appear better defined 

 than the central parts. The principal cause of error is pro- 

 bably a deviation of the extreme rays from horizontality in 

 their passage through the prisms, due either to imperfect 

 action of the collimator or to the curved faces of preceding 

 prisms. 



It is not easy to decide whether the highest resolving-power 

 is more likely to be obtained by gratings or by prisms. Up 

 to a certain point the resolving-power of gratings is principally 

 a question of the accuracy with which the lines can be ruled. 

 If the deviations of the lines from their proper places do 

 not exceed (say) one fourth of the interval between consecu- 

 tive lines, the definition in the first spectrum will not be 

 materially injured. To obtain corresponding results in the 

 2nd, 3rd, 4th, . . . spectrum, the errors must not exceed |, 

 1*2, y 1 ^, . . . respectively of the same interval. Every effort 

 should also be made to rule as great a number of lines as pos- 

 sible, even if it be necessary for this purpose to reduce their 

 length. 



I have lately had the opportunity of experimenting with 

 two Rutherfurd gratings, one on glass, with nearly 12,000 

 lines (a present from Mr. Rutherfurd himself), the other on 

 speculum -metal, with nearly 14,000 lines (kindly lent me by 

 Mr. Spottiswoode). The lines are at the rate of 17,296 to the 

 inch. Both these gratings give admirable results in the 

 second spectrum, where, according to (14) § 4, the resolving- 

 power in the orange exceeds that obtainable from 25 centims. 

 of extra-dense flint. In the third spectrum the gain of re- 

 solving-power is still apparent, but illumination is rather 

 deficient. 



It is much to be wished that spectroscopists in possession 

 of powerful instruments would compare their actual perform- 

 ances with those of which they are theoretically capable. A 

 carefully arranged succession of tests of gradually increasing 

 difficulty, like those applied to telescopes, would be of espe- 

 cial value. In my own observations I have usually attended 

 principally to the definition of the fine lines bordering b 2 and 

 to the double line b z . 



For experiments on dark heat, to which in some respects 

 gratings are well adapted, resolving-power is secondary to 

 illumination. In order to avoid confusion of spectra, it would 

 be well to eliminate the second spectrum altogether, which 

 could be effected by ruling equally strong lines at alternate 

 intervals in the ratio of one to three. 



Terling- Place, Witliam, 

 Nov. 8, 1879. 



