210 G. H. KNIBBS. 



smaller, or that the error found is a systematic and not an acci- 

 dental error of the proper motions, as later on he demonstrates to 

 be the case. With a view to deciding this question, the 905 stars 

 of greater proper motion were divided into two series; the division 

 and results being as follows : — 



&£ R.A. D. dp 



Bradley's 747 268-°l -0-°8 + 0-"0013 Epoch 



Southern 158 271*0 +2-3 - -0135 1810-0 



A further division was made, with the following result : — 



ft* R.A. D. dp 



Bradley's, north of + 23*2 decl. 309 269-5 + 1 °2 + 0*0058 

 „ + 23-°2to -23-°2 decl. 416 265-7 -1-5 + -0122 



Southern stars 158 271-0 +2-3 - 0-135 



Epoch 1810-0 

 These deductions for the direction of the solar motion are practically 

 identical, or at least shew no systematic difference. 



Continuing the analysis, the surface of the sphere was divided 

 into equal areas by hour-circle quadrants, and as before by parallels 

 of declination (0°, ll-°3, 23«°2, 36-°2, 51 -°9 and 79-°6, referred to in 

 the 1897 paper). It was shewn that the solution by the method 

 of least squares led to the values 



R.A. = 269-°8, D.= +16-°5, dp-0-"0381 

 dp denoting a correction to the precession-constant, that is, it led 

 to a result approximating to the position of the apex obtained 

 with the usual assumptions ; and, in reference to the precession- 

 constant, one in agreement with the recent investigations of various 

 authors who have followed Airy's method. The strict solution 

 gave however 



• R. A. = 270-°6, D. = + 0-°l, dp = - 0-"0028 

 a solution substantially identical with that previously obtained 

 from the 905 stars. Kobold argued that the comparison of these 

 results significantly shews that we have not to do with mere 

 accidental errors, but with systematically occurring differences of 

 motion, thus with the uniformly acting motus parallacticus. 



