102 THE ANTARCTIC MANUAL. 
These melting and condensing temperatures are intended to give 
fixed points of reference on the thermometer below the ordinary 
freezing- and above the ordinary boiling-point, and they can be 
chosen to suit the work in hand. Thus, if work is being done where 
temperatures about 108° or 109° C. are to be measured with the 
accuracy which is demanded of a thermometer on which the Centi- 
grade degree occupies a length of one or perhaps two centimetres, it 
is not sufficient to verify the ordinary boiling-point, because the part 
of the scale of the thermometer used may be ten or fifteen centi- 
metres distant from it, and to verify the scale by careful calibration 
entails great labour. It is, however, very simple to expose the 
thermometer in a vessel of the form about to be described to the 
action of steam condensing in a mixture of common salt and brine, 
which gives a perfectly fixed temperature at a given barometric 
pressure. And these conditions can be reproduced at any future 
time, and the readings of the thermometer in the immediate neigh- 
bourhood of this fixed temperature can thus be verified as accurately 
as if they had been as near to the ordinary boiling-point of distilled 
water. 
For temperatures above the boiling-point of water it is convenient 
to use “boiling mixtures,” where the dry salt is put in a suitable 
vessel holding a thermometer, and the steam is blown through it 
until the bulb of the thermometer is immersed in a boiling mixture 
of brine and solid salt, and the stem is immersed in its steam. 
One of the most convenient salts for this purpose is chloride of 
sodium, on account of its almost uniform solubility at different tem- 
peratures. At ordinary atmospheric pressure it raises the condensing 
point of steam by 8°°4 C., or 15°-1 Fahr. If a weighed quantity of 
salt has been used and the apparatus has been also weighed, then 
by continuing to blow in steam after all the salt has been dissolved, 
and weighing the apparatus when the temperature has fallen to 
certain definite degrees, a series is obtained of the temperatures at 
which steam condenses in solutions containing definite amounts of 
salt. This has been done for a number of salts. It is a much 
more accurate way of determining the boiling-point of a solution 
than by boiling it over a lamp flame. This holds generally. To 
determine the botling-point of a liquid it should be boiled by tts 
own steam. 
The apparatus used is shown with the boiling flask and lamp in 
Fig. 2. The steam tube is U-shaped; the one leg has a large body, 
CB, 15 cm. long and 4cm. wide. This is continued upwards in 
the tube A C, which is 15 cm. long and 12 mm. wide. The exit 
