of Action between Magnets. 143 
between AN and Na on the one hand, and between 60, 138, 
and 140 on the other, we occupy a favourable position for 
determining an approximate value for p by the method of § 21. 
The necessary comparison between calculated and observed 
quantities is made in Table XXI. In the case of 60, 138, 
and 140, I have assumed a mean value 6°40 for / and com- 
bined all the observations. To the present degree of 
approximation, that should give a satisfactory mean result. 
TABLE X XI, 
| Calculated values of 
Observed 
Magnets. | p. | mean value of 
| P(A,X’). | P+ )Q | P’(A,N’). pe 
| | ae 2.) i. | 
| 18 & 24 | 22:5 & 30) 18 & 24 | 22:5.&30) 18 & 24 22:5 &30 
ems, cms. ems. ems, ems. | cms. 
= oe | = 
and “86 10°68 3°38 219 | 1406 | 12:87 13°88 | 13°30 
AN "84 10°19 3°08 2:00 13-27 | 12°19 
28 fase Vi lichs 3°70 240 | 1488 | 13:58 
eee | 23 | 157 | Lor | sso | B24 | 
a | re | 72 |r | 929 | 868) 985 | 901 
68 | 793 | 188 | 122 | 981 | 915 | | 
By interpolation we deduce for the value of p: 
| 
| From observations ...... | at 18 & 24 ems. at 22°5&30ems. Mean. 
| : | 
Mean for AN and Na......... | "855 | "872 i £86 
Mean for 60, 138, and 140... “862 | ‘874 ee Si 
The results can be readily checked by reference to the fact 
that when X and 2! are equal 
PA, ’) = —V = — pp? P?/4. 
§ 30. The excess of these values for p over Bérgen’s 
limiting value *815 may arise possibly from some instrumental 
peculiarity, or some defect in the theory; but itis difficult to 
imagine that any such cause can account for the apparent 
difference between Ne and its companion magnets. To 
obtain, however, a further check, I got Mr. Baker to carry 
out a series of observations in which AN, Na, and Ne were 
deflected in turn by the same collimator-magnet KCl in the 
Kew magnetometer at the three distances 22°5, 30, and 40 cms. 
