660 PROFESSOR THOMSON ON THE ELECTRO-DYNAMIC QUALITIES OF METALS. 
with red lead so as to hold mercury, which was poured into each hole. The copper 
slabs projected outside to distances of about an inch, and each bore a bundle of 100 
No. 18 copper wires soldered to it, two of which, connected with diagonally opposite 
copper slabs, served as battery electrodes, while the other two were clamped to the 
ends of the conductor to be tested. The four slabs have only to be connected by 
two conducting arcs parallel to one pair or to the otlier of the sides of the square, to 
send the current one way or the other through the conductor; which was done by 
means of two heavy brass castings, as shown in the diagram (fig. 1). This com- 
mutator has been used in a considerable variety of experiments, and has been found 
very convenient. It gives the means of reversing almost instantaneously a very 
powerful current, without the necessity of bending any of the electrodes or deranging 
any part of the apparatus, and the conductors involved in it are so strong that it 
occasions very little resistance. 
28. As the supposed dififerential effect had appeared not to be increased after the 
first five minutes of the flow of the current in either direction, shorter periods of 
various lengths were tried, and more frequent observations of the thermometers were 
made, for the purpose of discovering the gradual variation of the temperature in 
the conductor, towards its final distribution as affected by the current. Four more 
large iron cells were added to the battery, which made it consist in all of eight 
cells, arranged as a single galvanic element, exposing 20 square feet of iron to 10^ 
square feet of zinc surface. As the strength of current thus produced would be 
nearly double of that given in the previous experiment, any true effect of the kind 
sought would be augmented in the same ratio ; and might be expected, both on this 
account and because of the improved system of observation, to become much more 
decided. These expectations, however, were not borne out by the results. The irre- 
gularities certainly became much diminished, but with these the differential effect on 
the thermometers, following the reversals of the current, either quite disappeared, or 
became very much less considerable than that which had been observed in the first 
experiment, and which I afterwards was led to attribute to some derangement in the 
position of the conductor occasioned by shifting the heavy clamps and stiff electrodes 
from between its two ends, causing the thermometer bulbs to alter a little in their 
positions in the hollows. 
29. Many experiments, both on the copper and iron conductor, were made, from 
October 1852 to March 1853, and the results of the observations (on each of the 
two principal thermometers either every half-minute, or every quarter-minute, during 
an experiment of about two hours) carefully reduced; with much labour at first 
when arbitrary scale thermometers were employed, but afterwards with far greater 
ease when centigrade thermometers, constructed for this investigation at the Kew 
Establishment, were received and brought into use. 
In the months of September and October 1853 the investigation was taken up 
again. The thermometric observations which had been made in the previously com- 
