=o 
: 
id ‘ 4 a 
sf 
208 RICHARD THRELFALL. 
spring balances as before—measured pieces of brass being inserted 
between the pole faces. In a series of obseryations the induction 
was kept constant by varying the magnetomotive force. The 
observations were taken just as in the previous case. A little care 
is necessary in defining what is meant by the total induction. If 
the bars are long and thin then of course the solenoidal condition 
is fulfilled pretty closely, and there is no ambiguity, but with large 
air or brass gaps, say amounting to two diameters of the bars, the 
lines begin to leave the iron just in front of the middle point of each 
bar (at all events when the bars are about fifty diameters long). 
The “total induction” therefore has no very exact meaning with 
respect to the iron unless it be specified where it is to be measured. 
At the time the experiments were made I did not (as I now 
consider), sufficiently attend to this point, though I used a testing 
coil of about four times the diameter of the bars and kept this coil 
just to one side of the gap when the latter was large. It is prob- 
able, therefore, that I have considerably over estimated the trac- 
tions with the larger air gaps, for the induction must have been 
greater than I took it to be. I have decided not to reinvestigate 
this point, for the curve is of use in giving approximate ideas of 
traction only ; and no one, after looking at it would design a 
mechanism with air gaps so long as those which are probably 
inaccurate. I have made a little allowance for this (most un- 
scientifically of course), in drawing the curve. In fact my suspic- 
ions were first aroused by examining the part of the curve corres- 
ponding to the larger air gaps. 
It will be seen that I examined a good many cases and the 
results show that when the non-magnetic field is of sensible 
dimersions, the differences in the permeability of the samples 
examined do not lead to any very abnormal results. The curve 
is drawn by reduction for a bar one centimeter in diameter, and 
the air gaps which must be expressed in diameters appear there- 
fore in centimeters. The ordinates give the values of the tractions 
at corresponding points in terms of the calculated tractions when 
the surfaces are in contact. One set of observations refers to 
