240 POOR 



among themselves, while a similar series obtained by another 

 observer will show large discrepancies among the individual 

 observations. It was, therefore, necessary to assign weights to 

 the observers, and an approximate idea of their weights was 

 obtained as follows. Each observation, both in a and o, was 

 compared with the ephemeris and the differences in the sense, 

 observed minus computed, found. Then upon the supposition 

 that the weights were all equal, corrections to the ephemeris were 

 found for numerous dates. These corrections were apphed to 

 the observed differences for each observation, and thus were 

 obtained for the series of observations of each individual ob- 

 server a series of residuals which closely represented the actual 

 errors of observation. From this senes of residuals the weight 

 was obtained by the formula 



ni^{n — I) 



where n is the number of observations and v, v^ the residuals. 

 In applying the formula, the value of ;//, the mean error of unit 

 weight was assumed as 



For a, ;w^zizoM5 For (5, 7n=^±i\" .$ 



The weights as thus determined were adhered to closelv in 

 assigning the weights to the various observations, the nearest 

 tenth being taken in most cases. 



As the comet differed radically in appearance and in ease of 

 observation in the various appearances, the observations of each 

 appearance were treated separately, and the weights assigned 

 as in the table. The initials in parentheses after the name of an 

 observatory represent the various observers. 



