feus.] VISCOSITY OF GASES. 305 
calculated. The tubes a and b are again made of semicircular section 
with their flat faces juxtaposed. 
Eeturning from this digression to 
the typical form, Fig. 49, it appears 
that the practical dimensions given 
to the instrument need not exceed 
those of an ordinary mercurial ther- 
mometer. The capillary tube may 
easily be wound in a closed helicoidal 
form with an external diameter of ~ cc _. .,. „ . ,.„ 
Fig. 55. Disposition of apparatus for differ- 
not more than 0.6 cm , and a length not eutia i measurement. 
greater than 2.5 cm . Such a spiral can therefore be made to lie wholly 
within the central tube of my reentrant porcelain air thermometer, Fig. 
33, and the comparison between transpiration pyrometer and air ther- 
mometer may then be effected with nicety in the same way as has been 
explained above, page 180. It is merely necessary to replace the ther- 
mo-couple by the transpiration pyrometer in the diagram of the re- 
volving muffle, to carry out the present comparison in detail. 
Again, it is curious to note that by selecting capillary tube of small 
external diameter and keeping the terminals a and b insulated by plates 
of mica and the turns of platinum capillary tube apart (Fig. 49) the 
platinum spiral may be heated to any degree of redness by the current 
of a dynamo electric machine. To obtain extreme degrees of tempera- 
ture it is^of course necessary to surround the tube cc with some non- 
conducting substance, for instance layers of carded asbestos. If the 
free space surrounding the axis of the helix of platinum wire be large 
enough to admit the insulator of a thermo-couple snugly, a comparison 
between the indication of the transpiration pyrometer and of the ther- 
mo-couple may be made at once, the degree of high temperature being 
regulated by the electric current which circulates through the helix of 
the platinum capillary tube. This form of calibration apparatus ap- 
pears to my mind to be at once the simplest and the most elegant as 
well as the most efficient form yet devised. Perhaps I may add that if 
the resistance of a hot platinum capillary tabe be measured the thermal 
variation of resistance is obtainable for long ranges of temperature. 
Supposing, therefore, that for a given (diatomic) gas the law of sixth 
roots characterizes the thermal variations of mean free path, it appears 
finally that the present method is especially well adapted for the high 
temperature study of dissociation phenomena in gases ; for dissociation 
here means the degree of discrepancy between the law holding under 
the given condition of dissociation and the normal law of sixth roots. 
In the case where gases permeate platinum at high temperature (the 
case of hydrogen and of many hydrocarbon gases) it is necessary to 
inclose the transpiration pyrometer in a glazed porcelain tube which 
fits snugly around it. This tube is closed at the hot end, and filled with 
the hydrogen or other gas in question at a pressure equal to the mean 
Bull. 54 20 (959) 
