432 Sir J. F. W. Herschel on a Machine 



of the horizontal axis; then make very slowly one complete revo- 

 lution of that axis, noting carefully at its commencement, and end, 

 and at every 30°, the reading off of the straight edge IJ, where it is 

 intersected by the fiducial line of the vertical scale, (which should 

 always be allowed to attain perfect rest free from lateral oscilla- 

 tions). If the point of intersection be found not to have varied 

 on the straight edge, it is evident that the coincidence of B and C 

 must have been perfect; but if otherwise, its extreme variations 

 will mark out the diameter of the small circle which B continues 

 to describe about C. This must be destroyed by shifting the place of 

 B on the slider, and if needed, by altering the place of the slider 

 itself on the axis (which may possibly have been originally erro- 

 neous, so as not to allow of the line described by B in its groove, 

 passing through C at all). As soon as this is done, and the in- 

 variability of the above-mentioned intersection ascertained, a fine 

 line must be drawn directly across the slider and its groove at 

 each end, and thus the zero points of the divisions both of the 

 slider and its verniers at each end are secured. The divisions should 

 be carried along the whole length of the cross piece and along both 

 edges of the slider, on one forwards, and on the other backwards, 

 by which means the machine is equally adapted for positive and 

 negative values of e. If the cross-piece be long enough, the ma- 

 chine will of course serve for values of e equal to or greater than 

 unity as well as less. 



(12.) If we would read off e on the horizontal straight edge, we 

 must first set the index hand to 0, and then to 180°— the differ- 

 ence of the readings is equal to 2e, being the diameter of the 

 circle described by B transferred to the straight edge by perpen- 

 diculars to the horizon. 



(13.) The only adjustment of any degree of delicacy is that of 



