Proceedings of the Royal Society 
f)54 
Beetz’ Table. 
Ewing and MacGregor’s Table. 
Observed. 
Calculated. 
Difference. 
Observed. 
Calculated. 
Difference. 
2387 
2315 
+ 72 
1876 
1883 
- 7 
2864 
2864 
0 
2264 
2264 
0 
3417 
3408 
+ 9 
2828 
2828 
0 
3921 
3992 
-71 
2969 
2997 
-28 
4450 
4487 
-37 
3145 
3166 
-21 
4502 
4502 
0 
3264 
3298 
-34 
4528 
4545 
-17 
3344 
3344 
0 
4594 
4615 
-21 
3367 
3379 
-12 
4638 
4621 
+ 17 
4641 
4630 
+ 11 
4626 
4638 
-12 
4628 
4641 
-13 
4632 
4649 
-17 
4640 
4651 
-11 
4632 
4645 
-13 
A comparison of these tables shews that, within the limits of 
common observation, Professor Beetz’ remark will apply as well 
to the agreement of his own numbers as of ours. That the 
agreement for very dilute solutions is not so good we have already 
satisfactorily explained.* Professor Beetz did not attempt to 
measure their conductivity, because he found that when there was 
less than a certain percentage of salt in his solutions, his method 
was no longer proof against the effects of polarisation. 
I have already said that the only way of testing our method is 
the comparison of the results which it furnishes with those of a 
known perfect method. But the comparison to which I refer is 
not such an one as Professor Beetz has proposed and executed. 
Its first condition must be the possession of a standard method ; 
its second the elimination of all unnecessary variables. It must 
be such as to allow r suspicion of the cause of differences in results 
to rest only upon the method itself. If there be x sources of 
error, it cannot be fastened upon any one. The solutions examined 
must be the same, the vessels containing them the same, the 
standards of resistance the same. The only pieces of apparatus 
which may , are those which must be changed. The fulfilment of 
the second condition is easy. The first is generally regarded as 
fulfilled by the use of Professor Beetz’ method. Kohlrausch and 
* Page G4 of our paper. 
