454 Proceedings of Royal Society of Edinburgh. [sess. 
but with the earlier forms of apparatus this expectation was not 
realised. At least when the current was stopped and the evapora- 
tion ceased the weighing showed irregularities. This may in part 
have been due to the exit for the steam being to the air and at 
the lower edge of the jacket. A current of air would enter when 
the supply of steam ceased. This is no longer possible in the 
present form, in which, however, an independent jacketing supply 
is still provided for. This enters by a wide vertical tube J inch 
in diameter, which delivers a copious supply of steam at a point just 
below and well outside of the lower edge of the inner jacket. 
The supply is regulated so that a small amount of steam issues 
from the jacket to the air, so as to counteract the diffusion of the 
air inwards through the interstices at the top and faces of contact 
of the outer jacket. The rest is condensed on the surfaces of the 
jacket and in a long drain- tube leading from the bottom, the lower 
end of which dips under the surface of the liquid in the collecting 
flask. In this way the entrance of air is avoided. 
The Electrical Arrangements . 
The e.m.f. across the terminals of the coil is limited in the case 
of electrolytes. In the case of water this limit is slightly over 
1J volts, and the voltage used so far has been about 1 volt. Con- 
sequently a large current (about 50 amps.) and a low resistance 
ohm) are required. By the potentiometer it is possible to 
measure these quantities to 1 in 10,000. Compared with the 
low resistance of the coil itself, it is not easy to make the resistance 
of the electrodes negligibly small, but they must be as small as is 
conveniently possible. For there must be considerable uncertainty 
as to the ultimate destination of the heat generated in the part of 
an electrode Avhich is neither in contact with the water inside nor 
with the steam outside. Taking this length as 2 cm. (1 cm. for 
each lead) the resistance for J-inch copper at 100° is *0000143, or 
1 in 1400 of the total. The assumption that half of this heat 
travels to the water cannot involve a greater error than 1 in 2800, 
and the actual error is probably much smaller than this. 
The potential difference across the terminals of the coil is 
measured by a Haider potentiometer reading to T owoo v °lb 
