140 



In the 2nd paper (p. 27) thèse curves were called pro- 

 portional curves resp. of the first and of the second kind. 



A distinction between the two kinds simply by the aid 

 of the given frequency curves is impossible. Whether it 

 will ever be possible to obtain the data necessary for 

 such a distinction I do not know. For the présent expo- 

 sition it may at ail events be suffîcient to treat only the 

 first kind of curve, referring those who might be interested 

 in the curves of the second kind to the 2nd paper (p. 27). 



As for any one cause the reaction becomes À fold, the 



total reaction, that is the ordinate , ifthereaction-curve, 



z 



becomes À fold. Therefore: The ordinates of the reaction 

 curves of proportional frequency curves are proportional ^), 

 If the quantities corresponding to the À fold causes are 

 distinguished by the suffix À, we get for the numerical 

 expression of this property: 



(.\ 1 1 -, 1 , 



(a) — ; : —r = a or z a = --.z . 



ZA Z A 



I found this critérium of proportionality satisfîed, with 

 surprising approximation, in the case of the summer 

 and winter barometerheights at den Helder, the data 

 for which I owe to the courtesy of my friend Dr. v. d. Stok. 



The observed frequencies, as well as the values of the 

 quantities z and z' computed from them, will be found 

 in tab. 10. 



The last column shows the proportions 



It is true that thèse still show small irregularities, but 

 they are not greater than might hâve been expected 



^) This holds for proportional curves of both the l^t and the 2°d 

 kind. In the fîrst the proportion is as 1 : a in the second as 1 : 1/ a 

 (sec 2°d paper). 



