468 KEPOET — 1882. 



which holds the object to be divided, and in this case the first and last-mentioned 

 errors conspire to produce a periodic irregularity in the motion of the platform 

 recurring at each revolution of the screw. 



If, however, the nut itself be left free to take any position which it likes, subject 

 only to the condition of not turning round with the screw, the efi'ect of these two 

 errors is to cause the axis of the nut to describe a cone about the axis of the screzv, 

 and the forward motion of the apex of this cone is free from the periodic error in 

 question. 



If, then, the nut be held in gymbals, which is in effect to hold it by a point in 

 its axis, and the motion of the nut be communicated to the platform through the 

 (n-mballing, the error depending on the pitch of the screw does not appear in the 

 motion of the platform. 



This plan has been carried out in the ruling machine. Fig. 2 shows a per- 

 spective sketch of the screw, nut, gymbals, rings, and pushing piece which connects 

 the nut with the sliding platform. 



The pushing piece has a spherical projection at E which fits into a spherical 

 cavity in the end of the platform. 



It will be seen that if it were not for the arm F, the nut together with the push- 

 ing piece would be capable of turning round with the screw. This arm, whicli can 



pivot freely about the axis yy', rests on 

 Pjj, 2 a guide G, parallel to the axis of the 



screw, and is continually held at right 

 angles to the screw by means of a pair 

 of jaws, which are carried along by a 

 screw A', of the same pitch as, and turn- 

 ing simultaneously with, A. 



If the screw were perfect, the guide 

 on whicli the ends of the arm F rest 

 ought evidently to be a straight line, but 

 if the errors of the pitch are known, the 

 guide can be so shaped as to make the 

 advance of the nut equal to what it 

 would be if the screw were perfect and 

 the guide straight. 



If the nut is more than half as long again as the pitch of the leading screw 

 of the lathe, errors of the third kind will not matter. 



It was needless to state in detail the means by which the errors of the screw were 

 measured, but they were made with considerable care, and he did not think any 

 variation in pitch exceeding iuo~o ^^^^ could have escaped detection. 



The screw itself had an error of about ^^^^ inch in its length, that is to say, at 

 one part the threads were ^-■'5^ inch distant irom the position which they should 

 have had, had the pitch been uniform throughout. 



II. The platform on which the grating rests while being ruled is a plate of 

 cast iron which rests on a plane, P P P, at three points or rather small surfaces, and 

 is touched by a straight edge, S S, at two. This five-point contact ensures motion 

 in a straight line. It will be observed that owing to the method employed to con- 

 nect the nut and platform, it is not necessary that the direction of motion of the 

 latter should be more than approximately parallel to the axis of the screw. 



Through the platform pass three verticallj' adjustable studs, bi,b^,b^, on which 

 the grating rests, and from one side of the platform one, and from an adjacent side 

 two, cast iron ears, X,, 'K^, project, against which the sides of the grating are pressed 

 by screws from the sides opposite. 



III. The tool carriage, c, fig. 1, which is a very rigid girder of cast iron, is 

 guided much in the same Avay as the platform. 



The tool-holder itself, m, is supported on bearings 7c, k', which permit the 

 tool to be raised or lowered on the gratings. The tool, t, has besides independent 

 vertical and azimuthal adjustments in the tool-holder. 



IV. In order to rule lines at equal intervals the screw has to be turned through 

 a constant angle between the ruling of each line. 



\ 



