252 
MESSRS. R. T. GLAZEBROOK AND E. B. SARGANT ON THE 
Part II. 
Profiting by our past experience, the arrangement of the apparatus was modified 
slightly. 
In Part I., as has been explained, one electrode of the coil V dips into the mercury 
cup H, while the other is connected by means of a piece of copper wire with the cup L. 
In Part II. the mercury cups and resistance coils were so placed that the second 
electrode of V dipped directly into the cup L ; the piece of copper wire between the 
two, therefore, was dispensed with. 
Again, in fig. 1 it will be seen that between P and M in the secondary circuit 
there is a variable resistance T, used to adjust the resistance of the circuit until it 
balanced R. 
This variable resistance formed a part of one of the two paths open to the current 
when measuring the deflection, and, as we have said, renders the exact proportion 
into which the current was divided in eacli experiment uncertain to the amount of 
about 1 in G000. 
In fitting up the apparatus for Part II., T was placed between B and N. As before, 
the resistance of the secondary circuit could be adjusted, but that of the primary 
remained unaffected by alterations of T, which in this part formed no portion of it. 
Three sets of rods were used to separate the primary and secondary coils, we shall 
call them A, B, and C, respectively. The rods A were those used in Part I. 
With the rods A three different electromotive forces were used; the batteries 
employed being respectively four ordinary Daniell’s, two ordinary Daniell’s, and five 
Thomson’s tray Daniell’s. 
In position B we had five Thomson’s-Daniell’s, and in position C five Thomson’s- 
Daniell’s and six Thomson’s-Daniell’s. 
As before, the coils were placed in positions I., II., III., and IV., but the order 
of taking the observations was somewhat varied. In Part I. three observations were 
taken in each position without altering the coils; in Part II., however,“after taking one 
observation in position I., one of the coils was reversed so as to bring them into posi¬ 
tion II., and an observation made; the other was then reversed, and so on, and after the 
four measurements had been ta,ken the whole series was repeated. This method 
necessitated rather more handling of the coils than the other; it had, however, the 
advantage that each set of four observations was taken under more nearly similar con¬ 
ditions, while, in consequence of the more frequent setting of the coils, the error due 
to any one chance bad setting was reduced. The time of swing was observed more 
frequently, being taken twice and generally three times for each set of four. The times 
corresponding to the mean throw and deflection are given in the table, being obtained 
by interpolation from those actually observed. 
