of Edinburgh, Session 1885 - 86 . 
633 
A third reason of the too high readings is that some of the air heated 
on the outside of the case is driven by air currents past the annular 
openings and enters the central tube. The too high readings given 
by the fan apparatus would be principally due to the heat conducted 
through the walls of the air passage and radiated to the thermometer, 
and very little of it to heat radiated by outside objects entering by 
the air passage. 
In the attempts to devise arrangements to correct these errors, 
I have been guided very much by two conclusions arrived at 
in a previous part of this investigation. The one conclusion was, 
that if we wish a body to acquire the temperature of the air 
it must be infinitely small, and in practice it should be as small 
as possible; the smaller it is the nearer will its temperature be 
to that of the air. The other conclusion was, that all radiant 
heat must be prevented from falling on the bulb of the thermometer, 
because the radiant heat is absorbed, not only at the surface, but 
also internally, by the glass walls of the bulb, and the passing air 
carries away heat only from the outside of the bulb, which explains 
the reason why, in practice, it was not found possible to check the 
heating effect of a very small amount of radiation by any velocity of 
air current that could be made to pass over the bulb. 
This investigation thus naturally divides itself into two parts. 
The first, how best to check the heat absorbed by the outside of the 
case, and prevent it getting to the thermometer ; the second, how 
best to check the entrance of radiant heat at the opening by which 
the air enters. 
Taking these two points in their order, we shall confine for the 
present our attention to the question how best to prevent the 
thermometer being heated by the hot case by which it is surrounded. 
In the screen shown in fig. 3 of the previous paper, the transference 
of heat was checked by the non-conducting concentric tubes and 
air passages. The cooling effect of the air in this arrangement 
cannot, however, be perfect, on account of the length of the surface 
over which it flows in an even stream ; only the lower edges, where 
the air first touches these tubes, will be cooled to near the tempera- 
ture of the air, the upper parts being warmer and warmer the 
further they are away from the entrance. 
Guided by the principle already referred to, it appeared that the 
