section--especially if a barrel
or trough is to be used--in order to have the resistance with the
plates at a safe distance apart, small enough to give 60 amperes with
40 volts.

Let us try a 10 per cent. solution of salt. Suppose the maximum
current this will carry is 1/4 ampere per square inch, which will give a
cross section of the solution of at least 60 / 1/4 = 240 square inches.
Now, the specific resistance per inch cube (i.e., the resistance
between two opposite surfaces of a cube whose side measures 1 inch) of
the 10 per cent. solution of salt used in test No. 3 was 2.12 ohms.
The drop, CR, will be 2.12 x 1/4 = 0.53 volt per inch length of solution
between electrodes. Hence, the electrodes will have to be 40/0.53 = 75
inches apart. This would require about three barrels connected in
series. This was taken merely as an illustration, because its specific
resistance was known when the current density was 1/4 ampere per square
inch. This solution, however, will carry safely 1/3 ampere per square
inch, but I used the previous figure, since I did not know its
specific resistance for this current density, because its specific
resistance will be lower for a larger current density on account of
the higher temperature which it will have, for the resistance of a
solution decreases as its temperature increases.

To reduce this length would require a solution of higher specific
resistance, that is, a solution containing less than 10 per cent. of
salt, and an increase in the cross section, since the maximum carrying
capacity also diminishes as the percentage of salt diminishes. Only
approximate calculations are useful because variations in temperature,
amount of salt actually in solution and the rate at which heat can be
radiated, all combine to give results which may vary widely from those
calculated.

As a matter of fact, it is seldom necessary or advisable to use a
solution containing over 2 or 3 per cent. of salt. The best way to
add salt to a liquid rheostat is to make a strong solution in a
separate vessel and add as much of this solution as is needed. This
avoids the annoying increase in conductivity of the solution which
happens when the salt itself is added and is gradually dissolved.

Liquid rheostats are ever so much more satisfactory for alternating
than for direct current testing. The electrodes and solution are
practically free from decomposition, and a given cross section seems
to be able to carry a large