CONTINUOUS ELECTRIC CALORIMETRY. 
113 
satisfactorily be eliminated in the plate-method, even by the employment of a guard- 
ring, are easily avoided in the tube-method by making the tube small in proportion 
to its length, (c) The error of the plate-method due to direct radiation through the 
liquid, which is quite important with a thin transparent stratum, is completely 
eliminated, since all the heat which is lost by the inner surface of the tube must be 
absorbed by the liquid itself. 
Graetz, (• Wied. Ann./ vol. 18, p. 79) has applied a non-electrical steady-flow 
method to the determination of the conductivity of a liquid, which bears a close 
superficial resemblance to that above described, but in reality differs from it in several 
fundamental points. In his method a stream of liquid at a temperature between 
30° and ICbO. flows through a thin capillary tube, 10 centims. long, and 0'6 millim. 
bore, immersed in a water-bath at a temperature of 7° C. to 10° C. The mean 
outflow temperature is observed, and is also calculated in terms of the conductivity on 
the assumption that the flow is linear. The temperature of the external surface of 
the tube is assumed to be the same as that of the water-bath. This is the most 
obvious defect of his method, as the assumption could not be even approximately 
true unless the current of liquid through the tube were extremely slow. 
Unfortunately in that case the difference of temperature between the outflow 
and the bath, on which the measurement depends, tends to vanish ; it is also more 
difficult to obtain the true mean temperature of a small stream owing to defective 
mixing in the outflow tube, and accidental sources of error due to end-effects are 
exaggerated. These defects might be avoided in various ways. The true mean 
temperature of the tube itself might be determined by making the tube very thin, 
and observing its expansion, or preferably its electrical resistance. Or the tempera¬ 
ture of the outside surface might be indefinitely approximated to that of the bath by 
making tbe tube very thick, and supplying a vigorous circulation around it. The 
true mean temperature of the outflow might also be determined for small flows by 
adopting a spiral circulation similar to that employed in the present investigation. 
There would remain, however, a most essential point of distinction between the two 
methods. 
In the electrical method, heat is continuously supplied by the current at a nearly 
constant rate as the liquid flows along the tube. The observation depends on the 
limiting difference of temperature at the end of a long tube when a steady radial 
distribution has been reached. The advantage of this is that the solution is 
independent of the initial or variable state, the calculation of the effect of which is 
much less certain on account of the steep gradients and excessive differences of 
temperature involved, which tend to produce disturbances in the flow. In Gkaetz’ 
method the initial differences, amounting to 20° or 30° C r , were much larger than in 
the electrical method, and the result entirely depends on the correctness cf the 
assumptions made in the solution of the initial state. The final or limiting state in 
his method is one of uniform temperature, and cannot be utilized at all. 
VOL. CXCIX.-A. Q 
