H. A. Rowland—Studies on Magnetic Distribution. 458 
TABLE III. Bar °39 inch diameter. 0 at end of bar. 
Q’e. e Q. Q 
L b- | Caleu- | POF °F) Ob. | Calen- | BO of 
served. 1 & served. | lated. : 
0 0° 0- 0 
it 19°7 16°2 —4°5 19°7 16°2 —4°5 
2 16°3 15:3 —1°0 36°0 30°5 —5°5 
3 16°0 15°5 — °5 52°0 46°0 —6'0 
4 15°8 15°9 ae | 67°8 61°8 —6°0 
5 16°5 16°3 — 2 84°3 18-1 —6°3 
6 17:0 16°9 — ‘1 101°3 5-0 —6°3 
vi 17°6 176 0 118°9 112°6 6°3 
8 18°4 18-4 0 iora 130°9 —6°4 
9 19°2 19°4 + +2 156°5 150°3 —6'2 
10 20°3 20°5 + 2 176°8 170% 61 
ll 21°8 21% — ‘1 198°6 192°2 —6'4 
12 22°8 23°1 + 3 221°4 215°3 —6'1 
13 24°8 24°7 — '] 246°2 239°9 —6'3 
14 26°8 26°5 — 3 273°0 26674 6°6 
15 28°38 28°4 — 5 301°8 294°6 Ee ey) 
16 31°8 30°5 —13 333°6 325°1 8°5 
Q’, = T-6(10-°RL 4 19—O87L). Q’ — BQ(10°L— 19—H7L), 
errors of Q' are all positive between 0 and 8 inches; and this 
has always been found to be the case at this part of the bar in 
all my experiments 
The explanation of this is very simple. In obtaining the 
formule we assumed that the magnetic permeability of the bar 
# was a constant quantity; but it has been shown by Dr. 
Stoletow and myself, independently of each other, that  in- 
creases as the magnetism of the bar increases when the latter is 
not great. Hence between 0 and 8 inches the resistance of the 
bar R is greater than at succeeding points, and hence a less 
number of lines of induction pass down the bar from 8 towards 
0 than would be given by the formula which has been adapted 
to the average value of R at from 9 to 14inches. In Table 
IL. this same fact shows itself towards the last of the table, 
and would probably be more prominent had the table been 
carried her. However, in this table all things have com- 
bined to satisfy the formula with great accuracy. 
2. 
-— 10 
Distribution at end of bar. 
In Table IIL we come across a fact of an entirely different 
nature from the above. Fig. 2 is the plot of this table, and 
gives the values of Q’« at different parts of the 
