322 C. G. KNOTT AND J. G. MACGREGOR ON THE 
junction. It was then lowered until the junction occupied a position about the | 
centre of the tube, and about an inch above the open end. The junction 
speedily acquired the temperature of the region inside the tube or hollow 
cylinder, and gradually cooled with it. The cold junctions were immersed in 
distilled water contained in small beakers, which were all set in one large vessel, 
through which a constant stream of cold water flowed from a cistern. In this — 
way, notwithstanding the proximity of the heated cylinder, the temperature of 
the cold junctions, as determined by a mercury thermometer, was kept nearly 
constant during a whole series of observations. In many cases no change was 
perceptible, and the variation never exceeded one-fifth of a degree centigrade. 
To guard against the possibility of galvanic or of electrolytic action, the portions 
of the wires of the cold junctions which were in the water were carefully 
covered with a thin coating of a non-conducting varnish. 
To determine the temperature of the hot junction, the thermo-electric pair 
composed of the alloys M and N was used. It was for this reason solely that 
both of these wires were bound to the wire under investigation. They were 
peculiarly fitted for measuring temperature, since the variation with tempera- 
ture of the electromotive force of a thermo-electric circuit formed of wires of 
these alloys had been thoroughly investigated by Professor Tair and _ his 
students. The special peculiarity of this variation established in Professor 
TaiT’s paper above cited, and fully corroborated by our own results given below, 
is that, through considerable ranges of temperature, the electromotive force of 
the M-N circuit may be taken as directly proportional to the difference of tem- 
perature of the hot and cold junctions. The current intensities were measured 
by the defiections produced on one of THomson’s “ dead beat” mirror galvano- 
meters. The deflections were so small that they might be regarded as propor- 
tional to the currents producing them. The galvanometer terminals and the 
free extremities of the wires, which formed the multiple junction, were so 
connected to the mercury pool commutator, that any one of the possible thermo- 
electric Junctions could be thrown into circuit with the galvanometer at will. 
It was not necessary to take measurements of the electro-motive force of all 
the possible pairs. Ordinarily the pair composed of either M or N and the 
wire under investigation, and that formed of M and N, were joined in circuit 
with the galvanometer alternately and in rapid succession at equal intervals 
of time. The circuit containing the latter pair may be distinguished as the 
thermometric circuit; and the mean of any two consecutive readings of the 
thermometric circuit, made in this way during the cooling of the hot junction, 
gave the measure of the electromotive force which corresponded to the tem- 
perature of the hot junction at the instant at which the reading for the other 
circuit was taken. In one or two cases the wire to be investigated was taken 
in thermo-electric connection with a wire of known thermo-electric relations 

