eA 
150 THE ANTARCTIC MANUAL. 
perature of the barometer is within 1° F. of what we take it to be, there 
is no sense in reading the barometer to units in the third decimal place 
of the inch, and still less so to units in the second decimal place of 
the millimetre. But, it will be said, we have the “attached thermo- 
meter,” and it can easily be read to 7'5° F. Completing the suggested 
syllogism, we have:—Now the temperature of the attached thermo- 
meter is always identical with that of the mercury of the barometer 
to which it is attached; therefore, we know the temperature of 
the mercury of our barometer to 75° F.; and by consequence we 
are justified in reading our barometer to 3 in the fourth decimal 
place of the inch, and to 7-6 in the third decimal place of the milli- 
metre. If the premises are granted, the conclusion is necessary, 
But the second premise is true only in certain conditions which do 
not usually obtain. If the temperature of the room in which the 
barometer is hung is invariable, the premises may be granted and 
the conclusion is valid. If the temperature is not invariable, the 
premises cannot be granted, and the conclusion is false. Sufficient 
attention is not paid to this source of uncertainty in barometric 
readings. After what has been said about the term of cooling of a 
thermometer, the reader will easily see where the fault lies and 
how to remedy it. In so far as the effect of temperature on the 
length of the mercurial column is concerned, the barometer is a 
thermometer with a bulb of the volume and thermal mass of the 
column of mercury. The term of cooling of this mass must be very 
much greater than that of the attached thermometer; so that if at 
any epoch the two chanced to be at exactly the same temperature, 
and the temperature afterwards went through the usual diurnal 
changes, they would never again have the same temperature, except 
for a moment of time when their thermal paths happen to cross, 
The only adequate “attached thermometer” is one which has a 
bulb which is a copy of the barometer, has the same thermal mass 
and the same term of cooling, and is exposed to exactly the same 
condition as the barometer. In important central stations the 
standard barometer should be accompanied by such a thermometer. 
In standardising stations nothing but absolute uniformity of tempera- 
ture should be admitted. The need of correction is the confession 
of imperfection. In exact work no correction should be admitted 
which it is mechanically possible to exclude. 
In Europe and North America, and generally in countries where 
for a part of the year dwellings have to be artificially heated, the 
condition of uniformity of temperature is not secured of itself, and it 
may be troublesome to provide it. The Paris observatory has its 
