186 MEASUREMENT OF HIGH TEMPERATURES. [bull. 54. 
the gas may be shut off altogether, and the furnace closed by shutting 
the flue holes with a fireclay plug or with asbestus. If all influx of air 
is cut off the furnace (below 500°) cools very slowly. During these 
stages of cooling it is not undesirable to continue a rotation of the muf- 
fle at a rate sufficiently slow to prevent currents of air from entering 
the polar parts and being hurled off centrifugally at the equatorial parts 
of the muffle. Here it is well to mention that the speed of rotation 
usually employed in the present experiments was about fifty revolutions 
l>er minute. Possibly this rate is too great, particularly at very high 
temperatures, at which the muffle becomes iucipiently viscous, and 
much slower rates are therefore preferable. Protected from direct flame 
"by the muffle, the fragile porcelain bulbs are heated with great regu- 
larity, and the liability of the thermometer to breakage is therefore nil. 
In order to determine the constant of the air thermometer (zero read- 
ing) it is sufficient to place a mercury normal thermometer in contact 
with the bulb (or better in the central tube of the re-entrant bulb) in 
the cold furnace before heating. The same operation must be repeated 
in the cold furnace after heating. The difference of zero readings thus 
obtained is a test of the validity of measurement, inasmuch as it shows 
whether during the course of the comparisons the bulb has remained of 
fixed normal volume and whether air or gas has leaked into or out of 
the air thermometer. The mercury thermometer is inserted and held in 
position by the same spring-clamp which, during calibration, holds the 
thermo-element. In the form of bulb, shown in Fig. 32, to which most 
of the p resent remarks apply, the mercury bulb is pushed quite into 
the re-entrant tube and an identity of environment of mercury and air 
thermometers is thus secured. Of course it is more nearly accurate, 
though less convenient, to submerge the bulb in water of known tem- 
perature in order to get the zero or fixed point. All this will be discussed 
below. 
During the heating of the furnace to white heat, the hollow axles, 
which necessarily lie in the circumambient sheet of flame, are heated 
more easily and kept at a greater intensity of red heat than the body oil 
the muffle. This is the objectionable feature, due to the absence of the* 
third rotation referred to on page 182. But this error, thus introduced, 
is confined to a small part of the stem ; it does not extend as far as the] 
muffle, in the interior of which, moreover, the hot air is churned around! 
the bulb by the rotation; its harmful effect is fully obviated by the use] 
of the re-entrant bulb, in which heat is communicated to the therino- 
.electric poiut junction through the surface of the air-thermometer bulb; 
it is entirely absent during the cooling of the furnace, conditions under 
which most of the experiments are made. Finally, the furnace has been 
so made that the zone of variable temperature, which surrounds the stem 
of the thermometer, is as narrow as possible when measured along the 
length of the stem. Indeed, the correction to be applied for the part of 
the porcelain stem, which has varying temperature, is almost nil. If, 
(840) 
