( 558 ) 

 TABLE VI. 



tlie iinal aim in inquiries of this kind to come to a rational expres- 

 sion for the frequency of barometric deviations as a function of the 

 distance of centres of depressions and of their average deptli and 

 extent but, even if we assume the most simple relations between 

 pressure and distance of the centrum, we must expect to find rather 

 complicated, exponential expressions, which can be treated -only by 

 expansion into series. It is, therefore, desirable to summarize the 

 characteristics of the frequency curve in an empirical formula of 

 the form : 



c-U-\t^ (.4 -f Bx + C'.i;' 4- i»(.-' + Ex') (3) 



The constants of this formula can be easily determined and, if we 

 succeed in establishing a rational expression, there will probably be 

 no difficulty in indicating their meteorological meaning. 



The frequency curve, positive and negative deviations being taken 

 together (Table 1), is then represented by the expression : 



Z=2e-H-'^'{A-^Cx' + Ex'), (4) 



Avhich represents a symmetrical curve, and the formula for the 

 differences of Table III becomes -. 



Y = 2e-Jl'-^' {Bx 4- Dx') f5) 



If, as in our case, tlie dc\iations x are departures from the arith- 

 metical mean, 



