io 



with / = O at the beginning gives 



^=^--(i-rr'j 



1 dM 



r dt \ 



r 

 and for small values of - and t, as long as M has not reached 

 » L 



zero, with sutlipient approximation 



1 dM 



L dt 



so that, if M reaches while / is still small, 



L 



will be the tinal value of the current, 



In our experiment the constants were //„=:= 400, J/„ = l,26xl<)% 

 L = \0\ so that / could rise to 0,126 C. G. S. or 1.26») Amps. 

 The current caji thei'efore reach the threshold-value 0.8 Amp. even 

 with a tield of rather more than half the strength assumeil in the 

 calculation (cf. one of the expei-iments in § 4). From 'the moment 



r 



at which this value is reached ordiiuxry lesistance appears and 



will be no longer small: the further increase of i above the threshold 



value Id follows a ditferent law from below ij). 



For an accurate calculation of the jirocess above iu, it woidd 



be necessar)' to take into account the complicated law of increase 



of the resistance ^vith the current beyond io- For our pur|»ose it 



is sufficiently accurate to assume, that when iu is exceeded I)y a 



small amount, the resistance becomes suddejdy r' of the order of 



magnitude above the vanishing [)oint. 



dM 

 In that case, — — remaining the same as before, the current will 

 at 



I dM 

 be able to rise by a small amount I — in=z-- — , which will soon 

 •^ r dt 



dM 

 be reached, Avill then become constant and. onJ/and — becomin": 



dt "" 



zero, disappear again in a short time. In view of the value of .1/ 



and r' we may, if Al does not change very rapidly, disregard i — i^^ 



unless we intend an explanation of all the details of the experiment. 



We therefore come to the conclusion, that, J/„ being sufticienlly 



1) The more accurate data given liere dider somewhat Irom lliuse in the Dutcli 

 text. 



