Permanence of the Rutherfurd Photographic Plates. 205 



B 



Inasmuch as this correction is added to the readings on both 

 the scale and star, and as the coordinate of the star is obtained 

 bj subtracting one of these readings from the other, it is allow- 

 able to add any quantity not a function of S to the above expres- 

 sion. By so doing we can make the expression always positive. 

 Finally, then, the total correction to be added to all readings of 

 the screw has been taken as : 



B 



— ^— + A; + 1.2 i2 + 0.0035. 



The object of using the quantity : 



+ I.2i2 + 0.0035, 



instead of a simple numerical constant, was to obtain the desired 

 result without making any of the corrections unduly large. A 

 set of tables were computed and the above corrections were taken 

 from them without interpolation. In computing these tables, the 

 screw corrections already determined were interpolated between 

 the limits 9^ and 12^. The screw was not used outside these 

 limits in the present research. 



An example of one of these tables, computed for R= -\- 0.0195 

 is given here. 



.0 



+0.0080 



+0.0053 



+0.0026 



.1 



.0078 





.0050 





.0022 



.2 



.0076 





.0047 





.0020 



•3 



.0072 





.0044 





.0017 



•4 



.0070 





.0042 





.0015 



•5 

 .6 



•7 

 .8 



•9 



.0067 

 .0063 

 .0061 

 .0059 

 .0056 





.0039 

 .0036 

 .0034 

 .0031 

 .0028 





.0013 

 .0010 

 .0008 

 .0006 

 .0003 



3* We take up next the considei'ation of the division errors of 

 the scale. These were determined for the Observatory by the 

 Kaiserliche Normal Aichungs Kommission at Berlin. The method 

 employed for determining the errors of the centimetre lines 

 was that which I have described as Gill's method in the 

 American Journal of Science for May, 1896. This method 



