of Edinburgh , Session 1871-72. 775 
fusion of the iron. As shown on the plate, the first of these occurs 
at about 240° C. of a mercurial thermometer. 
With another alloy supposed to be of the same metals, but of 
which I do not yet know the composition, also made into a junction 
with iron, the behaviour was nearly the same, but the readings at 
the successive neutral points were 28, - 137, - 132. The tempera¬ 
ture of the first is about 200° 0. by mercurial thermometer. 
An iron-palladium circuit showed no neutral points within the 
great range of temperatures mentioned above ; though it showed 
a remarkable peculiarity which must be more closely studied, as it 
appears to point to the cause of the above effects in a property of 
iron. It was therefore employed to give (very roughly) an indica¬ 
tion of the actual temperatures in these experiments. But as for 
this purpose it is necessary to measure the simultaneous indica¬ 
tions of two circuits whose hot and whose cold junctions are respec¬ 
tively at the same temperatures, I was obliged to employ a steadier 
source of heat than the naked flame. I therefore immersed the hot 
junctions in an iron crucible containing borax glass, subsequently 
exchanged for a mixture of fused carbonate of soda and carbonate 
of potash; but, to my surprise, the former of these substances at a 
red heat disintegrated both the platinum and the alloy, and thus 
broke both circuits without sensibly acting on the iron, while the 
mixture (evidently by the powerful currents discovered by Andrews, 
Phil. Mag. 1837) interfered greatly with the indications of the 
thermo-electric circuit, as will be seen by the dotted curve in the 
plate. [I may remark here that the deviations of this curve from 
its form when these currents are prevented are quite easily observed 
and plotted by the process next to be mentioned, so that the study 
of the Andrews’ effect may be carried out with great accuracy by my 
method.] Finally, determining to dispense altogether with fused 
salts, which conduct too well besides acting on the metals, I simply 
suspended a red-hot bombshell, vent downwards, in such a way that 
the hot junction was near its centre. This arrangement worked 
admirably, until a white heat w T as required, for this melted the 
shell. In its place a wrought iron tube (an inch in bore, four 
inches long, half an inch thick, and closed at the upper end) has 
been substituted and answers excellently. It does not cool too fast 
for accurate reading at the higher temperatures, and by elevating 
