1882.] from Wheal Newton. 243 



The plane t is a narrow uneven plane, and gives two images, 

 neither of which seem to be accurately in the zone [m'd']. Its 

 indices, if it be taken to be truly in this zone, will be (5 1 3), a 

 form given by Schroder. This plane requires re — 19 0- 35i', which 

 is 1° too great, a difference which cannot be accounted for by the 

 errors of observation. No nearer approximation can be made with- 

 out introducing high indices. The nearest planes in the zone 

 [ph'e] are (16, 3, 10) or (21, 4, 13). The former plane gives 

 re 18 u, 2i', the difference between which and the observed value is 

 hardly within the possible errors of observation. The latter plane 

 gives re 18 0- 24|', and the only objection is the magnitude of the 

 indices. I have given the angles which these two planes make 

 with in' ; angles which, judging from the close agreement between 

 most of the observed and calculated angles, do not agree satis- 

 factorily with either of the values of m r observed, r lies on the 

 edge of the portion where the crystal has been injured by a tool, 

 and may possibly have suffered a slight strain from the blow. 



Expression for the error due to eccentricity. 



Let G in fig. 3 be the position of the axis, and let the signals 

 be at distances a and b along the lines AC and SC. (These 

 distances are about ten metres for the instruments at Cambridge.) 

 Let A CS = fl ; an angle which can be determined by observation 

 with sufficient accuracy. Suppose the edge to be at N and N~. 

 when the observation is made for each face, and CE the normal to 

 the first face in its first position. Then a, the angle through which 

 the faces or normals are turned, is also the angle NCN X . Let 7; be 

 the difference between this angle and a x that between the two faces. 

 The sign of 7; can be readily determined by considering the nature 

 of the inequality between the angle of incidence and that made 

 by the line of sight with the second face, when it is brought into 

 parallelism with the first. In the case supposed in the diagram it 



is subtractive. Let - — 6, ■ — cf> be the angles of incidence en 



