1874.] ^^ ^ [Chase. 



representing the mean of several years' observations, it is not strange 

 that great care should be needful in order to determine the approximate 

 character of the normal flexures. 



My previous discussions having shown that the lunar rain-curves at a 

 given station vary somewhat at diiferent seasons of the year, I first com- 

 puted the normal curves at "Husband's" for each month of the year in- 

 dependently, and then "smoothed" the curves by taking the fourth suc- 

 cessive means between the daily normals of successive months. This 

 second series of normals, although insufficient for any conclusive infer- 

 ences based on comparisons between consecutive months, should furnish 

 approximate evidence of the normal changes, as wel] as means for making 

 eighteen entirely independent comparisons between curves with intervals 

 of five or six months. The normals for these two series of curves are 

 given in Tables III. and IV. If there were no other than an accidental 

 connection between the several curves, the chances of agreement or dis- 

 agreement between the normal excesses or deficiencies of rainfall, in each 

 independent comparison, would be equal, there being a probability of 15 

 days agreement and 15 days disagreement. The actual accordances and 

 discordances and the ratios indicative of a vera causa in lunar action, are 

 given below, 1 being the ratio of probable accidental agreement. 



TABLE II. 



Normal Proportions of Rainfall at " Husband'' s" on Lunar Days of each Calendar 

 Month, for Independent Comparisons. 



