11 0 
PROFESSOR HUGH L. CALLENDAR ON 
20 to 30 centims. of water at 20° C., for different arrangements of the conductor, and 
different conditions of flow, being greater when the flow was non-linear. Since 1° C. 
corresponds to a fall of 42,700 centims. under gravity, the rise of temperature due to 
the friction would be less than a thousandth of a degree. This is a quantity which 
ought not to he neglected in working to a ten-thousandth of a degree, but the effect 
was practically eliminated by the method of observing the difference of temperature, 
which was expressly intended to eliminate small residual sources of error of this 
character. For each value of the flow, the difference of temperature was observed 
“ cold” before turning on the electric current, and the “ cold reading” was subtracted 
from the difference observed with the current passing. We are therefore concerned 
only with the change in the head due to diminution of viscosity with rise of tem¬ 
perature when the current is turned on. I found by calculation from the known 
variation of the viscosity, and also verified by direct observation in each case, that 
this amounted to only 10 per cent, of the head for a rise of 8° C. at 20 C. This 
would be equivalent to 2 or 3 centims. fall, or less than a ten-thousandth of a degree, 
a quantity which might safely be neglected. The correction would be much smaller 
at higher temperatures, owing to the great diminution in the viscosity. 
In the final apparatus, as employed by Dr. Barnes, the difference of head would 
be somewhat greater owing to the rubber spirals on the copper sleeves, and the 
rubber cord on the central conductor. The question was raised at the Dover meeting 
of the British Association, and I wrote to Dr. Barnes asking him to measure the 
head under the actual conditions of experiment, but the apparatus happened to be 
dismounted at the time. We may safely conclude, however, that the difference 
of head could not have exceeded 1 metre of water, in which case the correction 
would be less than 1 in 40,000 at 20° Cl, and might be fairly neglected. This 
correction corresponds with that for the heat generated by stirring in the Griffiths 
and Schuster, methods of calorimetry. It amounted in Griffiths’ apparatus to 
about 10 per cent, of the heat-supply, but was apparently negligible in Schuster’s 
experiments. 
(30.) Radial Distribution of Temperature in the Fine Flow- Tube. 
We assume in the elementary theory of the experiment that the temperature 
is uniform across the section of the flow-tube at any point. It is important to 
consider how far and under what conditions this is true. Given the rate of external 
heat-loss at any point it is easy to calculate the difference of temperature between the 
inside and outside surfaces of the glass tube, but the distribution of temperature in 
the liquid can only be calculated if we assume the flow to be in straight lines 
parallel to the axis of the tube, and the conductor to be circular in section and 
concentric with the tube. Even in this simple case the solution cannot be made 
complete, owing to the variations of viscosity and conductivity with temperature ; 
