DETERMINATION OF DEFLECTION OF LIGHT BY THE SUN'S GRAVITATIONAL FIELD. 327 



but having regard to possible systematic error, probably a quarter weight would more 

 nearly represent the true value. 



This determination is ignored in the subsequent discussion. 



36. It is easy to calculate the effects of any errors in the adopted scale, orientation, 

 &c., on the final result (deflection at the limb). We give some illustrations. 



An error in the adopted scale of y of 10 units (in the fifth place of decimals) would 

 lead to an error 0"-68 in the result from either plate. Thus the probable error 2-1 

 in the determination of e' gives a probable error 0"-14 in the final result ; or, if we 

 adopted the largest (rather discordant) value found for e' instead of the mean, we should 

 reduce the result by 0"-52. 



An error of 10 units in the orientation gives an error in the result of 0"-45 for plate X, 

 and 0"-22 for Plate W. It is therefore of less importance, and further it is not likely 

 to be systematic. 



Errors in the measurement of x only affect the result through the orientation. For 

 Plate X, a probable error of 0"-20 in the x-measures would give an error 4-0 in 

 the orientation, leading to an error 0"-18 in the result ; whereas an error of the same 

 magnitude in the y measures gives directly an error 0"-35 in the result. For Plate W, 

 the probable error of 0"-20 in x gives an error 3-5 in the orientation and 0"-08 in 

 the result, compared with 0" 38 for similar inaccuracy in y. It is particularly fortunate 

 that the x-measures are so unimportant for Plate W, because, as already mentioned, the 

 images trailed on that plate. 



Finally, it will be remembered that in order not to commit ourselves to the Einstein 

 hypothesis prematurely we neglected the correction |E, in determining the orientation. 

 This will make a difference of 0"-029 in the results from Plate W and 0"-092 from 

 Plate X. The effect is that the deduced deflection needs to be decreased, and the mean 

 correction 0"-06 should be applied to the mean result obtained, or rather, to make 

 the adopted deflection for x consistent with the deduced value from y, the correction 

 needed is 0"-04. 



Discussion of the Results. 



37. The four determinations from the two eclipse plates are 



X-G . . . . l"-94 

 X-H . . . . l"-44 

 W--D . . . . l"-55 



W I .... l"-67 

 giving a mean of 



l"-65. 



They evidently agree with EINSTEIN'S predicted value l"-75. 



