142 Dr. C. Chree on the Law 
After the observations were completed, it was noticed 
that through the employment of a sighting-tube intended for 
a larger deflecting magnet, the level of the suspended magnet 
had been in each case slightly too high. A small correction 
"12, answering to the mean error *2 cm. in the level, has been 
applied in the Table to the observed values of P’. 
§ 28. To determine the value of, say, \.—A,, we put 
Ag? Ay" = (Ag—Ay) (Ag+ Aj), 
and as A,—A”, is small, we may for a first approximation take 
Ae +A, ="81/=6'l cm. for AN, Na, and Ne, 
ol gy ay a0), Loo, and 140, 
In this way we find 
TABLE XX, 
| r 
For AN, Na, and Ne. For 60, 138, and 140. | 
cms. | | ems. | 
AA, = "O11 | AA,—A,)//=003 || A,—A,="026 | 2(,—A,)/7=008 | 
Nj—A,="195 —-AA,;—D,)/Z= OSL} AY —A,z="VL2~ | 2(A,—A,)/2= 007 
A,—A,="176 | 2A,—A,\/!=046 || »,—d,="068 | 2(.,—A,)/= "021 
} 
If theory and experiment were perfect, the results for 
r?—/? from the two pairs of distances in Table XIX. should 
be identical, and in Table XX. the value of, say, A3—A, found 
directly should agree with that deduced from the values of 
Ay —A, and A3;—A,. This cannot be claimed, but on the whole 
there is a very satisfactory agreement, considering that a 
complete elimination of the effects of variation of horizontal 
force and temperature could only be hoped from a largely 
increased number of observations. 
The two magnets AN and Na behave practically alike, but 
differ notably from Ne, though to the eye that magnet seems 
an almost exact duplicate of Na. As these three magnets are 
practically equal in length, Table XX. implies that the value 
of p must be about *05 larger in Ne than in the other two. 
In the case of 60, 138, and 140, the agreement in pole- 
distance is closest between 60 and 140, though by different 
makers. The excess, however, of pole-distance apparent in 138 
is partly accounted for by its excess of length, about ‘035 cm., 
over the other two. If weallow for this, we find that the value 
of p in 138 exceeds that in 60 by only about*003. The value 
obtained directly for A,—A3 is probably slightly in excess, 
and the excess of p in 138 over 140 is probably about °01. 
§ 29. Having now proved a close similarity in pole-distance 
