VALUE OF THE BRITISH ASSOCIATION UNIT OF RESISTANCE. 
267 
experiments were conducted. Our best thanks are due to Lord Rayleigh for his 
unceasing kindness and his many valuable suggestions which have helped us satis¬ 
factorily to surmount several serious difficulties, and have added greatly to the 
accuracy of the results. 
. Note. 
(Added December 2, 1882.) 
Another source of error should have been mentioned in the previous discussion. 
The external diameter of each layer was measured necessarily before the layers 
external to it were wound on. There will probably be a tendency in each layer to 
compress those below it, and thus to make the mean radius of the coil somewhat less 
than we have assumed. 
In fact, Lord Rayleigh found in his experiments (Phil. Trans., Part I., 1882) that 
the value of the mean radius obtained from measurements made as the coil was 
unwound was less by 1 in 2000 than that obtained from the measurements made 
while winding. 
Several reasons lead us to believe that any effect of the kind would be much less 
with our coils. 
The error is produced probably by the silk being pressed over the top and bottom 
of the layer into the intervals between the wires—the wires being circular in section 
there is necessarily a space between them. 
Now the amount of this free space would be proportional to the area of the cross 
section of the wire, while the amount of silk would depend on the circumference of 
this cross section. 
The diameter of our wire was only about two-tliirds of that used by Lord R ayleigh, 
there would, therefore, be less space in proportion to the amount of silk into which 
the silk could be squeezed, and the error produced would be less. 
Again, if this yielding is going on, it is clear that the average thickness of a layer 
should appear to decrease as we get near the outside surface. A reference to the 
measurements shows that this is not the case. On dividing the whole series of layers 
into sets of five, and taking the average for each set, omitting a layer in which 
paraffin paper had been used, we find the values for the average thickness T28, '123, 
'127, '128, and H27. 
No gradual decrease is here observable. We therefore feel confident that the error 
was much less than in the coils wound by Lord Rayleigh. 
A strong indirect confirmation of the accuracy of the value of the mean radius used 
is afforded by Lord Rayleigh’s recent measurements by Lorenz’ method. 
The same coils were used as in our observations, and Lord Rayleigh has showm 
that an error in the mean radius would affect the two methods by about the same 
2 m 2 
