774 Proceedings of the Royal Society 
furnish me with the means of accurately measuring high temper¬ 
atures (by a process described in previous notes of this series). As 
this hope may possibly not he realised, I can as yet make only 
rough approximations to an estimation of the temperatures of these 
neutral points. 
So far as I am aware, the phenomenon discovered by Cum¬ 
in ing and analysed by Thomson has hitherto been described 
thus : When the temperature of the cold junction is below the 
neutral point, the gradual raising of the temperature of the 
other produces a current which increases in intensity till 
the neutral point is reached, thenceforth diminishes; vanishes 
when one junction is about as much above the noutral point 
as the other is below it, and is reversed with gradually in¬ 
creasing intensity as the hot junction is farther heated. To 
discover how my recent observation affects this statement, I first 
simply heated one junction of a circuit of iron and (hard) platinum 
gradually to whiteness, by means of a blowpipe, and observed the 
indications of a galvanometer—both during the heating and during 
the subsequent cooling when the flame was withdrawn. The heat¬ 
ing could obviously not be effected at all so uniformly as the 
cooling; but, making allowance for this, the effects occurred in 
the opposite order, and very nearly at the same points of the scale 
in the descent and in the ascent. [I have noticed a gradual dis¬ 
placement of the neutral points wdren the junction was heated and 
cooled several times in rapid succession ; hut as my galvanometer, 
though it comes very quickly to rest, is not quite a dead-beat 
instrument, I shall not farther advert to this point till I have made 
experiments with an instrument of this more perfect kind, which 
is now being constructed for me.] The observed effect of heating, 
then, was a rise from zero to 110 scale divisions when the higher 
temperature was that of the first neutral point, then descent to 95 
at a second neutral point, then ascent to a third, descent to a 
fourth, neither of which could be at all accurately observed, and 
finally ascent until the junction was fused. 
With an alloy of 15 per cent, iridium and 85 per cent, platinum, 
the galvanometer rose to 53‘5 at a neutral point, then fell to — 50 
at a second, then rose to a third at — 39-5, and thence fell, but I 
could not observe a possible fourth neutral point on account of the 
