IN AQUEOUS solution, AND THE EXISTENCE OF CO^IPLEX IONS. 
12 
found that for salts of simple type, the difference in transport numlDer of the same 
salt, as determined in different ex]:)eriments, is exceedingly small: thus six experi¬ 
ments with 2 N, KCl gave 0‘492, 0'490, 0'494, 0'488, 0'488, and 0'489, aiid two 
experiments with N, KBr gave 0‘473 and 0'474. 
For salts of less simple type than these, only two duplicate measurements have l)een 
made; these are, for magnesium sulphate 0’5N, 0'693 and 0’694, and for a 0'2 N 
solution of the same salt 0'649 and 0‘646, 
For the measurement of the current, a galvanometer with the resistance of about 
100 ohms was employed; it was far too sensitive for the amount of current used, 
and was therefore always connected with a shunt. Provided the total deflection 
of the needle amounted to at least 70 scale divisions, readings correct to 1 part 
in 200 could he made ; to ensure accuracy in the measurements, therefore, it was 
necessary that, whatever the strength of the current, the deflection should be not 
less than about 50 scale divisions. 
Since the current varied greatly in different experiments, a series of shunts was 
prepared by the use of whicli widely varying values could l)e given to the scale 
divisions. The apparatus is shown in fig. 4, wliere G is the galvanometer and S tlie 
Fig. 4. 
B 
box of coils, one or all of which can l)e used as a shunt. In the fiofure the three 
coils between 2 and 5 are connected with the Ijinding screws of tlie galvanometer, 
and Ijy moving the connecting wire c and the wire d to different mercury cups, 
a large number of different values can loe given to the shunt. The value in amperes 
of each scale division was determined Ijy direct measurement wlien the various coils 
were connected with the galvanometer, and by this means it was possilde, however 
the amount of current varied, to ensure in all cases a deflection of from 80 to 100 
