OF ARTS AND SCIENCES. 247 



length vibrates upon a shaft set exactly in a line with the precision 

 screw. At one end a magnet, fitted to the curvature of the head of the 

 screw, is attached by eight pivots in such a way as to give parallel mo- 

 tion with respect to the arm. The outer portion of the head of the screw 

 consists of a rim of soft iron, which operates as an armature to draw 

 the magnet to it when the circuit is completed. The other end of the 

 arm works between two stops, one of which is adjustable. The action, 

 then, is this : the circuit being completed, the magnet becomes firmly 

 attached to the head of the screw, and by the movement of the arm 

 from one stop to the other, it is carried over a given space. The cir- 

 cuit being broken, the arm during the reverse movement carries the 

 magnet with it without disturbing the precision screw. In order to 

 guard against every possibility of disturbance, a second magnet holds 

 the head of the screw in place while the first one is moving back to 

 prepare for the next increment of motion. By varying the distance 

 between the stops, any desired motion whatever, within certain limits, 

 can be given to the screw. From repeated experiments, it is found 

 .that about twenty movements of the arm for x^oo^ ^^ ^ revolution 

 of the screw-head can be made without varying more than one from 

 this number. 



If now the lower stop is replaced by a wheel made to revolve simul- 

 taneously with the head of the screw, and if to the periphery of this 

 wheel a curvature is given corresponding to the known errors of the 

 screw, it is obvious that the screw can be made to correct its own 

 errors. Thus, if at any point in its revolution the screw gives too 

 small intervals, the periphery of the wheel must be filed away enough 

 to increase the space ruled by the amount of the error. I am indebted 

 to Professor Joseph Wiiilock, the late Director of Harvard College 

 Observatory, for the suggestion of this elegant method of overcoming 

 the residual errors of the screw. 



How are Robert's Lines of Varying Degrees of Fineness ruled on 



Glass ? 



First of all, the evidence seems quite clear that they are ruled with 

 a diamond having a knife-ediie. In all of the cases which I have 

 examined the lines start in with a curvature, which is maintained 

 throughout the whole extent of the band. I have been able to 

 produce this result only by setting the cutting edge of the diamond 

 slightly inclined to the direction of the line ruled, and this inclination 

 seems to give a decided improvement to the character of the lines. 



