364 Proceedings of the Boyal Society 
while it is cooling from an intense white heat, and that it should 
not do so when cooling from a temperature a little over a dull- 
red heat, is a hypothesis so inconsistent with all known facts as 
to make it desirable to find some explanation more in accordance 
with known principles. Iron is a very bad conductor, and Professor 
Tait has shown (R. S. E. Proceedings, 1873) that the conductivity 
is much worse above than below a dull-red heat. Now, the cool- 
ing of such an iron wire as that used is effected so rapidly that 
the temperature falls through an enormous range of temperature in 
a few seconds. This is effected by convection and radiation from 
the surface. It is quite possible that the internal heat cannot 
be conducted outwards with sufficient rapidity to compensate 
this outer loss. Thus the temperature of the interior of the wire 
is greater than that of the exterior. At very high temperatures 
the rapidity of cooling is enormous. But as the cooling proceeds, 
the deviation from the Newtonian law of cooling is much less. 
Hence the cooling by radiation becomes less, and the heat which 
has been stored up in the interior of the wire has a tendency 
to show itself on the surface. At a dull-red heat the wire becomes 
a better conductor, and this tendency is assisted, so that about 
this stage the temperature throughout the wire is nearly equalised. 
The second experimental fact is explained by this raising of the 
external temperature. The third fact is explained in the same 
way. And it must be noticed, that a difference in temperature 
between the interior and exterior is the only means of explaining 
the rise in temperature of the external air, unless we suppose that, 
while cooling, the wire increases in temperature. And even then 
it would he difficult to understand why the effect is not produced 
by cooling from a temperature a little above a dull-red heat. If 
the wire be massive, or if a poker be used, the cooling is not rapid 
enough to produce the effects ; apparently, because the convection 
currents are not nearly so strong in proportion to the surface 
which has to be cooled. Other causes come into play in this case, 
all tending to prevent the effect from being apparent. The 
explanation I have given shows why the effect is observed only 
when the wire has been originally heated to an intense white 
heat ; for it is only then that a great difference of temperature can 
exist between the interior and exterior. 
