140 MICHELSON— RULING AND PERFORMANCE [April 22, 



true ; and these were straightened by apphcation of an auto-colhmat- 

 ing device which made the deviation from a straight hne less than 

 a second of arc. 



(b) The friction of the grating carriage on the ways was 

 diminished to about one tenth of that due to the weight (which may 

 amount to twenty to forty pounds) by floating on mercury. 



(c) The longitudinal motion of the screw was prevented by 

 allowing its spherically rounded end to rest against an optically 

 plane surface of diamond which could be adjusted normal to the 

 axis of the screw. 



(d) The screw was turned by a worm wheel (instead of pawl 

 and ratchet) which permits a simple and effective correction of the 

 periodic errors of the screw throughout its whole length. 



((?) A correcting device which eliminates periodic errors of 

 higher orders. 



(/) It may be added that the nut which actuates the carriage 

 had bearing surfaces of soft metal (tin) instead of wood, as in 

 preceding machines. It was not found necessary to unclamp the 

 nut in bringing it back to the starting point. 



Finally it may be noted that instead of attempting to eliminate 

 the errors of the screw by long continued grinding — which inevitably 

 leads to a rounding of the threads — it has been the main object to 

 make these errors conveniently small ; but especially to make them 

 constant — for on this constancy depends the possibility of auto- 

 matic correction. 



The accompanying photograph made with a ten-inch grating, 

 6th order (actual ruled surface 9.4 inches by 2.8 inches), used in 

 the Littrow form with an excellent 8-inch lens by Brashear, is given 

 in evidence of its performance. The resolving power as shown by 

 the accompanying scale of Angstrom units is about 450,000. The 

 original negative shows a resolving power of about 600,000. The 

 theoretical value is 660,000. 



Doubtless the possibility of ruling a perfect grating by means of 

 the light waves of a homogeneous source has occurred to many — 

 and indeed this was one of the methods first attempted. 



It may still prove entirely feasible — and is still held in reserve if 



