RADIATION IN ABSOLUTE MEASURE. 
605 
the absolute positions of the curves, but most of these depend on some differences of 
circumstances. The only case which I cannot explain is that of the silvered globe 
in air, November 6, I do not understand why the loss of heat was so small in an 
experiment which bears every sign of being thoroughly good. I have no note (such 
as would certainly have been made) of any change of galvanometer zero (indicating 
possible alteration of magnetic field) ; nor was there alteration of the apparatus of 
any kind.* Extreme dryness of the air is a possible explanation. 
A feature, which seems to me remarkable, about the curves representing these late 
experiments is that none of them show any sign of emissivity, as it were, accelerated 
with increase of temperature, that is to say, having an increasing rate of increase of 
emissivity with increasing temperature, but rather the reverse. The curves are all » 
slightly concave towards the axis of temperatures. It is, however, well ascertained, 
particularly from the experiments of Schleiermacher and myself, that, at all events, 
in the case of wires losing heat there is a very rapid acceleration or increase of rate 
of increase of radiation with increase of temperature. This is most easily seen by 
looking at the curves of my former paper, ‘ Phil. Trans.,’ 1887, A, Plate 25. Looking 
at the matter in this light also, it will be seen from those curves that the curvature 
(proportionate acceleration) is greatly masked when more than ^ millim. or ^ millim. 
of air pressure is present, and that the lower the air pressure the more striking is the 
convexity towards the axis of temperatures in that series of curves. 
Referring now particularly to the tables accompanying this paper. Table I. has 
been already explained (pp. 591, 592, 593). Table II. shows the results of four nearly 
successive days of experimenting in air and in vacuum on a globe wdth two different 
coatings of lamp-black. In the first pair the coating was much more dense than in 
the second pair. In the latter case the globe was passed quickly a few times through 
the gas flame of my blowpipe table. In this way a coating of the finest texture was 
obtained; the colour of the copper globe was distinguishable through it. I had a 
reason for wishing to keep the coating very thin in connection with experiments on 
incandescent wires which I am making. 
Table III. shows the results of similar experiments on a silvered globe. 
Both these sets of experiments were made with the apparatus of fig. 1 ; and I 
believe that in neither was the interior of the enclosure sooted in the way whicli 
was afterwards done. 
Tables IV. and V. show the results of experiments on the sooted globe in air and 
in vacuum, and on the silvered globe in air at very high temperatures. These experi¬ 
ments were made in apparatus of fig. 2, Plate 17, and were most satisfactory in 
every way. 
I come lastly to Table VI. It was in the hope of finding time (three months at 
least) for repeating and extending these experiments that I have delayed the publi- 
* To imagine a cause for unexpected excessive loss would be easy; but it is not easy to account for 
the globe retaining its heat. 
