BAROMETRIC PRESSURES ON THE GREAT LAKES 



109 



the N determinations, then the probable error of one determination may be 

 computed from the formula 



in which Si' 2 stands for the sum of the squares of the residuals in the group. 



TABLE No. 27. 

 [Mean observed and mean corrected elevation =579+ feet.] 



The probable error of a single observed elevation at each of the five 

 stations was computed in turn from this formula applied to each group of 

 N residuals as shown in the tables Nos. 19 to 23. It will be noted that N 

 was usually 5, but was sometimes 6, for the group at the end of a month, and 

 sometimes less than 5 where observations were missing or were rejected. 

 The mean of these many values of the probable error of a single determina- 

 tion, one for each group, was taken as the probable error of a single deter- 

 mination for that station, of the kind under consideration, observed or 

 corrected. 



The probable errors so computed are shown in table No. 28. 



When one considers the probable errors for an observed elevation as 

 shown for each of the five gage stations in table No. 28, the question naturally 

 arises, Why should the observations at the different stations on any lake be 

 given equal weight? Two considerations led to the decision to assign equal 

 weights to the observed elevations in this investigation. First, that has 

 been the usual practice, so far as the investigator knows, in connection with 

 other studies on the Great Lakes. It did not seem desirable to depart from 

 past practice except for clearly good reasons. Second, it was clearly 



