barus.1 PORCELAIN AIR THERMOMETRY. 187 
following Deville and Troost, a compensator is used to correct the stem 
error, this apparatus may temporarily replace the insulating tubulure of 
the thermoelement. 
A iinal remark must be added with reference to silicification 1 of plati- 
num. If the wires of the thermoelement be in contact with the surface 
of the air thermometer, and the temperature high enough to make the 
glaze appreciably viscous, the platinum is invariably attacked and de- 
stroyed as far^as it is in contact with the glaze, and it is found em- 
bedded in it after cooling. The apparent effect is fusion, for the plati- 
num thus silicified fuses at a temperature below that of the glaze. 
Quite aside therefore from the change of constants silicified platinum 
is unavailable for high-temperature measurement, because of its fusi- 
bility. Care must therefore be taken to avoid unnecessary contact of 
the thermo electric wires with the glazed bulb, which is fortunately feas- 
ible, because the operator is able to see into the furnace as far as the 
air thermometer through the hollow axle at the side of the muffle on 
which the thermo-couple is inserted. 
One great advantage enjoyed in using the present furnace is this, 
that even in case of extremely high temperatures it is possible to ap- 
proach very near it without discomfort. The temperature of the exterior 
Surface does not exceed 300°, and may be even further reduced by ap- 
propriate jacketing of asbestus. Hence all manipulations near the fur- 
nace can be conducted with facility. In the case of the large furnaces, 
described in Chapter I, the radiation from the furnace is intense, and 
all close approach to it is difficult. 
A sketch of a revolving muffle for the comparison of two air ther- 
mometers, containing either two distinct gases, or the same gas at dif- 
ferent temperatures, is given in Fig. 3G«, where B B is the body of the 
furnace, H, G, the two burners, CD the revolving muffle, and and N 
the two air densities to be compared. The revolving muffle is ellip- 
soidal in shape, and the appurtenances are the same as those described 
above. The stems of the air thermometers project at m and m, and 
capillary tubes connect each with the respective manometer. In the 
comparison of air thermometers both the expansion of the porcelain and 
the stem error compensate each other, and the data are thus as accurate 
as the environments are identical. 
Finally, I desire to make a few remarks on the complete disposition 
of apparatus, as given in the frontispiece. The lower part ot the figure 
lr riie tendency of platinum to combine with silicon, forming very fusible products, 
has long been known. Tait, endeavoring to use a bath of fused silicates tor constant 
temperature, found the platinum wires, wliicb he submerged therein, disintegrated. 
More recently, Colson (C. R., xciii, 1881, p. 1074; C. R., xciv, 1882, p. 2(5); Violle 
(C. R., xciv, p. 28) ; Pcniolet (C. R., xciv, 1882, p. 99), have given especial prominence 
to this and similar phenomena. In these eases there seems to he a diffusion of the 
solids into each other, and the melting point of silicified platinum may he reduced to 
That of glass. I have often found platinum wire fused on or even embedded in the 
glaze of the air-thermometer bulb. 
(8-11) 
