319 



leave a pretty large margin of uncertainty. According to the formula 

 given in § 2 the probable errors of a determination from (5) and 

 (6) for de Bilt with /^ — 0.957 and 0.97(i resp. are 1.62 and 

 1.21 m.m.; they prove however, as well as equ. (4) that local 

 intiuences play an unimportant part. 



In the same manner as from (4), it appears from (5) and (B) tiiat 

 the influence of the eastern stations Mülhausen, Dresden and Sylt 

 is considerably greater than that of the western stations: Valencia, 

 He d'Aix and Lerwick. 



For the partial c.c. between Helder and Valencia we have found 

 0.470 (Table IV) whereas for that between de Bilt and Valencia, as 

 deduced from (6), we find 0.583, an agreement which can be 

 considered faii'ly satisfactoi'y if we take into account that the data 

 used in computing these values are totally different. 



As mentioned in § 3, for the first series general normal values 

 have been used, given in the "Berichte" so that it is possible tiiat 

 in this case the sum of (he deviations for each station is not exactly 

 equal to zero which, of course, would influence the value of the c.c. 



H is, however, more probable that the cause of this disagreement 

 must be ascribed to an insnflficiency of the number of observations 

 used in ^ 5 and § 6, because the values of k found in the first 

 investigation (§ 4) are all greater than those derived from the groups 

 treated in § 5 and § 0, from which a generally smaller value of 

 the c.c. would follow. Owing to the mutual dependence a number 

 of 240 daily observations cannot be considered as equivalent to 126 

 tenday means and it is a general law in statistical investigations 

 that the computed relations show a tendency to give smaller limiting 

 values as the data increase in number. 



7. Finally the question may be put, what will the condition 

 equation become when the two groups of three suriounding stations 

 are taken together so that the deviation of atmospheric pressure in 

 the central point is determined by 6 circumjacent stations within 

 angular distance of about 60°. 



The numeration of the stations then becomes : 



The c.c. computed in § 5 and § 6 and all products can be used 



