the Galvanic Current through Iron. 223 



cimals of the measuring resistance. In this way we can get 

 with great exactness two more decimal-places ; hence I made 

 the experiments as follows: — 



(1) w approximately determined; 



(2) Si Sq . . . 5 W _! s n s n -i • • • s 2 % observed, and from them a 

 system of simultaneous values of $| ... s n derived ; 



(3) a x + b 1} « 2 + o 2 . . . a n 4- b n observed ; 



(4) Experiment 2 repeated inversely— 



s n s n - 1 • • • s 2 S l S 2 ' • ' S n- 1 s n > 



This again gives a system of simultaneous deflection-values. 

 Finally the mean of these two was taken afresh. It then re- 

 presents a system simultaneous with the determinations 3, and 

 can by means of them be exactly reduced to a system of ad- 

 ditive or subtractive resistances. 



Here the current remained closed, on the average, only a 

 few seconds ; so that a thermal influence could hardly assert 

 itself (compare below). On the other hand, the extra currents 

 in thick wires made the procedure impossible ; hence I was 

 obliged to confine myself to thin ones. 



6. Hard iron wire / 10 . Z = 24300, d = 021. 1-3 Daniel! 

 elements. (Preliminary experiment.) 



2^=103*9 deflections (to the left*): 



8l =Q% * 2 = 2-3, s 3 = 4:'7, s 2 = 2% s^l-2. 



Therefore the simultaneous values are 



$1=0-6, 5 2 = 2-6, §3=4-7, 



Now there were found 



^ + ^ = 4-7, tf 2 + 6 2 =8-6, « 3 + 5 3 =12-2. 



Therefore 



^=103-913, w 2 = 103-930, w B = 103*939, 



therefore in the mean 



w 2 — W 



1=S 12 = 0-00017, 



in fact much less than at p. 



In the following Tables of the results of the exact experi- 

 ments, the index in the first column states the number of the 

 Daniell elements; the second column gives the sum (a + b) of 

 the deflections, left and right, for a unit more or less in the 

 last decimal in the approximate statement of w ; s and / are 

 the mean values of the deflections found before and after the 



* "To the left" denotes constantly "to the side corresponding to a too 

 little measuring resistance." 



