BETWEEN THE FREEZING AND BOILING-POINTS. 
155 
temperature gradient will then be represented by (2) fig. 2. The water 
flowing through the interior of the tube will receive less and less heat from 
the layers along the sides as the velocity of flow increases. To attain the 
given temperature 0 ] on the outflow-thermometer, the temperature gradient 
from B to A will rise rapidly, and suddenly decrease as the water is mixed 
around the bulb. The total heatdoss from the water will then increase 
with increasing flow in proportion to the slope of the temperature-gradient 
from B to A. This condition would be perfectly fulfilled by replacing the 
fine glass tube by one of metal through which the electric heating current 
could be made to flow. Less perfectly it is fulfilled when the heating wire 
lies along the sides of the tubes, and supplies heat to the layers of water 
nearest the walls of the tube. 
In both these cases the heat-loss would not be independent of, hut would depend 
on, some function of the flow. To ensure a perfectly uniform temperature, equal to 
that of the flow-tube, throughout any section of the water column, it is necessary to 
produce thorough mixing at all points, and avoid the formation of stream-lines. If 
this is fulfilled, we can be safe in assuming the temperature gradient at least 
approximately linear from B to A in both the above cases. Also that the total 
quantity of heat lost per second by radiation from the water in its passage through 
the length of tube included in the vacuum-jacket is the same, quite independent of the 
velocity of flow. 
An experimental study of the two cases above given, where the water flowing in 
parallel stream-lines receives heat from a platinum wire, which may be moved from 
the centre to the sides of a 3-millim. bore flow-tube, will be given in Section 7. 
Conduction .—The heat-loss by conduction from the ends of the calorimeter will 
evidently be very small where a bad thermal conductor, such as water, is used. 
Where metal wires are introduced to convey the electric current to the central 
heating wire, the conduction of heat from the water by the wires assumes a much 
more serious character, more especially when the calorimeter and jacket are maintained 
at a temperature very different to that of the surrounding air. At the inflow end 
the effect, when the calorimeter is at a higher temperature than that of the air, is to 
lower the temperature 0 O of the inflow water by a small amount. The effect at the 
outflow end is similar, but smaller, on account of the direction of flow. It is evident 
that this can always be measured and eliminated for any given flow by recording the 
temperatures of the inflow and outflow thermometers before the electric* heating 
current is turned on. The only conduction effect that these “ cold ” readings will 
not take account of is the conduction from the outflow-tube due to the rise of 
temperature (0 1 — 6 0 ). This must be separately measured, in other ways. It can be 
estimated and its maximum effect obtained, for any given difference in temperature, 
by surrounding the outflow-tube beyond the water-jacket by a water circulation, the 
temperature of which can he changed at will. If it is made as small as possible by 
x 2 
