( 555 ) 



liere the reverse ()I)t;iiiis, tlie frequency of barometric lieiglits showing 

 a deficit for small and a surplus for moderate deviations. 



In an earlier paper (this volume p. 314) I have shown that, in 

 taking together series with different factors of steadiness, each series 

 occurring with equal subfrequency, we must expect to find too great 

 a number of small deviations. 



From this follows the apparently somewhat i)aradoxical conclusion, 

 tiiat a sum of frequency numbers as those of barometric deviations, 

 ail showing negative differences for small deviations, may, when 

 taken together, lead to a resiilting curve in whicii these differences 

 have vanished or even turned positi\e. 



This conclusion is of some importance liecause an investigation 

 into the frequency of barometric heigiits, in which the different 

 months are not treated separately, may lead to normal curves (the 

 skewness being left out of account) whereas in fact no normal curve 

 exists and appears only as an artificial consequence of the combi- 

 nation of incomparable frequency numbers. 



The exceptional behaviour of the months of February and .July might 

 then be explained by assuming that the different series of f)aromelric 

 curves corresponding with different winds (barometric windrose) are 

 more differentiated in these two months than in the other ones. 



A second remark is that frequency numbers as given in Table I 

 cannot be accepted as a measure for the variability of the atmo- 

 spheric pressure in the course of a montli, at least not if we adhere 

 to the conception of this variability as generally admitted. 



On the one hand we have here to do witli the superposition of 

 two kinds of variability, J*' the secular variability as shown by the 

 variability from year to year of monthly means and 2"'^ the vari- 

 ability from day to day, which might be called the interior variability 

 for the month in question ; it is the latter definition which corresponds 

 with the usual conception. 



On the other hand, daily means or obser\ations taken at fixed 

 hours are by no means to be regarded as being independent of 

 each other. 



The questions, thei'efore, arise : how can we separate the two kinds 

 of variability, and to what degree are daily mean values of baro- 

 metric observations to be taken as dependent upon each other in 

 the different months. 



For a knowledge of the climate of a place the latter question is 

 of importance ; it might also be formulated thus : what is the average 

 duration of a V)aromefi'ic disturbance, a question whicli can hardly 

 be answered by means of direct investigation, 



