1302 
where the fact has been attended to that w is small with respect 
to the resistances with which we are concerned. 
There remains therefore a magnetization in positive sense, which 
is smaller as the changes @ in the resistance are smaller. If we 
suppose however 5,4, to be only little smaller than /, so that we 
may put Ha, —h, M proves to be negative of about the same 
value. For A= $ (M2, + Haji) M will be approximately equal 
to zero. In that case the magnetizations of the different conical 
surfaces just neutralize each other. 
hb. We may also suppose that the electromotorie force / decreases 
with equal steps «. Then /# must be substituted for w and « for &. 
Leaving the resistance constant and equal to w we obtain 
ul mn hw? | EE dM 3) mh*w? 
Mou — M ( TT EE ‚sot at == F Ee ’ 
GEN Shy de and so on 
' dE! fe Ee d 
Now: 
ne: E (oy es 
fs 0 (£,--2ne)* 
+ 5 (24!—2) 6’ 5 — : —t.. | m Ben - at 5 
o (Z, — 2ne)* FE op Vi fo» 
If, supposing « to be small, we keep only the first of the terms 
between square brackets, and replace the summation by an inte- 
gration, we finally obtain, after having put again Hs, =h and 
= 
assuming that ¢ is also still small with respect to /5,, 
mfe h® m es. H, h* 
M=—| 3 —— ]= - lS ult 
hw Jobe Eh ETE 
1 
If we may no longer consider ¢ as small with respect to /,, the 
first term between the brackets has still to be multiplied by 
€ . . . 
1— —. In each case M/ can be made arbitrarily small by choosing 
21) 
« small enough. Where however in the expression for M the factor 
EE Nowe ; 
=, occurs. which will be a great number, ¢ must be very small 
L 
with respect to /,, in order to make M small. In the first case 
h . 
we have the factor —, so that there we need not choose w so 
1 
exceedingly small with respect to 2,. 
