900 



SGIUNCE 



[N. S. Vol. XXXIV. No. 887 



The diffraction grating possesses so many 

 advantages in simplicity and convenience 

 of manipulation that it is even now used in 

 preference to these modern instruments, 

 save for such refinements as require an 

 exceptionally high resolving power. But 

 has the resolving power of the grating been 

 pushed to the limit? We have seen that 

 this depends on the number of rulings ; and 

 it is certainly possible to increase this 

 number. But the theoretical value is only 

 reached if the rulings are very accurately 

 spaced; for instance, the resolving power 

 of the Rowland grating is only one third 

 of its theoretical value. This is a direct 

 consequence of inaccuracies in the spacing 

 of the lines. If a grating could be con- 

 structed of say 250,000 lines with exact 

 spacing, the resolving power would be 

 equal to that of the most powerful echelon. 

 The problem of the construction of such 

 gratings has occupied my attention for 

 some years ; and while it has met with some 

 formidable difficulties, it has had a fair 

 measure of success and gives promise of 

 still better results in the near future. 



The essential organ in all ruling engines 

 in actual use is the screw which moves the 

 optical surface to be ruled through equal 

 places of the order of a five hundredth to 

 one thousandth of a millimeter at each 

 stroke ; and the principal difficulty in the 

 construction of the machine is to make the 

 screw and its mounting so accurate that 

 the errors are small compared with a 

 thousandth of a millimeter. 



This is accomplished by a long and tedi- 

 ous process of grinding and testing which 

 is the more difficult the longer the screw. 

 A screw long enough to rule a 2-inch grat- 

 ing could be prepared in a few weeks. 

 Rowland's screw, which rules 6-inch grat- 

 ings, required two years or more — and a 

 screw which is to rule a grating 15 inches 

 wide should be expected to take a much 



longer time, and in fact, some ten years 

 have been thus occupied.* 



I may be permitted to state a few of the 

 difficulties encountered in this work — ^some 

 of which would doubtless have been dimin- 

 ished if my predecessors in the field had 

 been more communicative. 



First, is the exasperating slowness of the 

 process of grinding and testing the screw. 

 This can not be hurried, either by grinding 

 at greater speed, or by using any but the 

 very finest grade of grinding material. 

 The former would cause unequal expan- 

 sions of the screw by heating; and the lat- 

 ter would soon wear down the threads till 

 nothing would be left of the original form. 



Secondly, in ruling a large grating, 

 which may take eight to ten days, the rul- 

 ing diamond (which must be selected and 

 mounted with great eare) has to trace a 

 furrow several miles long in a surface as 

 hard as steel — and often breaks down 

 when the grating is half finished. The 

 work can not be continued with a new dia- 

 mond and must be rejected and a new 

 grating begun. 



Thirdly, the slightest yielding or lost 

 motion in any of the parts — screw, nut, 

 carriage or grating, or of the mechanism 

 for moving the ruling diamond — is at once 

 evidenced by a corresponding defect in the 

 grating. When after weeks or sometimes 

 months of preparation all seems in readi- 

 ness to begin ruling, the diamond point 

 gives way and as much time may have to 

 be spent in trying out a new diamond. 



When the accumulation of difficulties 

 has seemed insurmountable, a perfect 

 grating is produced, the problem is con- 

 sidered solved, and the event celebrated 



* A method of ruling gratings accurately, which 

 is independent of any mechanical device, is now 

 in process of trial, in which the spacing is regu- 

 lated by direct comparison with the light-waves 

 from some homogeneous source such as the red 

 radiations of cadmium. 



