430 Mr. Arrott on some new 
In order to prove this, and also to show that the effect is 
not dependent on any particular state of the metals, a porous 
vessel was filled with a mixture of strong solutions of proto- 
and per-sulphates of iron; this vessel was then placed in 
another filled with the same mixture, a platina plate was 
plunged into each vessel, and the circuit completed by a deli- 
cate galvanometer; not the slightest effect was produced; the 
plates were then put in communication with the poles of a 
common battery, and the amount of current which passed 
measured by means of a voltameter. No gas was evolved in 
the solutions, npr was any iron reduced, but the persulphate 
increased in quantity at one side, and the protosulphate at 
the other. When 80 measures of gas were collected, the bat- 
tery was removed, and the plates being connected by the gal- 
vanometer as at the beginning of the experiment, a powerful 
current was produced in a direction the reverse of that of the 
battery; the effect was the same whether the same or fresh 
plates were used, and if they were simply washed with water, 
they might be moved from one vessel to the other, without 
producing the slightest effect on the current, provided their 
connexion with the galvanometer was also changed. The 
arrangement was again connected with the battery, in such a 
manner that the current should move in the solutions in the 
opposite direction to the former battery current, 80 measures 
of gas were again collected, and the battery being removed, 
not the slightest current was observed on connecting the 
plates with the galvanometer, but every thing was in the same 
state as at the commencement of the experiment. The power 
of the solution to produce the current gradually dimi- 
nishes, but is not entirely destroyed until the second or return 
current becomes equal in amount to the first. The quantity 
of the return current cannot be measured correctly, without 
using the battery to enable it to pass quickly; because, from 
the extreme slowness of the action towards the end of the 
experiment, the solutions unavoidably mix, and thus cause 
great error in the amount. A similar experiment was tried 
with nitric acid, containing a large quantity of the lower oxides 
of nitrogen with precisely similar results. 
Where the circuit is entirely metallic, but not homoge- 
neous, we have also a return current; for the heat produced 
in a thermoelectric arrangement causes a current the reverse 
of that which produces it; but we cannot in this case deter- 
mine the amount, from the impossibility of retaining the heat 
produced, and of preventing it from extending to those parts 
which ought to remain coid; the same may be the case also 
in a homogeneous circuit, but from the extremely low inten- 
