118 
PKOFESSOR HUGH L. CALLENDAR OX 
a spiral of about 1 centim. pitch, closely fitting the tube, was chiefly due to the 
breaking up of the stream-lines. The colour-band rapidly became mixed to a 
uniform tint as in the first experiment. I found, however, from the indications of 
the platinum thermometer wound round the outside of the tube, that the glass was 
considerably heated by the close contact of the spiral. The simpler arrangement of 
the loose stranded conductor was equally effective in mixing the stream-lines, and 
appeared to be free from this defect. 
An excellent illustration of the possible effects of a faulty arrangement of the 
conductor is given by Dr. Barnes in § 6, p. 234. In this case it is possible to calculate 
the actual heat-loss from a knowledge of the correct value of J for the temperature 
of the experiment.* The conditions were purposely chosen to exaggerate the errors 
as much as possible, and it must not be imagined that such large differences could be 
obtained without special pains in the arrangement of the heating conductor. The 
normal heat-loss for this calorimeter at a temperature of 26° with a constant gradient 
of temperature along the flow-tube was approximately ‘070 watt per degree rise. 
Starting from this value, it is possible to calculate the limiting values of the heat-loss 
for either condition of flow, that of the metallic tube or the small concentric con¬ 
ductor, by drawing the curves representing the actual distribution of temperature in 
either case, and making a suitable allowance for the loss of heat from the thermo¬ 
meter. Dr. Barnes has given a pair of imaginary curves for these two cases in 
§ 2, p. 152, but it should be observed that these curves are not drawn to scale, 
being merely intended to illustrate the general nature of the difference, which is 
considerably exaggerated in order to make it clearer. 
Table VII. — Superheating of Straight Conductor in 3 millim. Flow-Tube. 
Number of 
experiment. 
Position of 
conductor. 
Flow, 
Q- 
Watts, 
EC. 
Superheat of 
wire. 
Heat-loss, 
observed. 
Limit, 
calculated. 
(1) 
"1 At side of f 
•600 . 
21-7 
o 
8-0 
•1108 
•1140 
(2) 
J tube t 
•277 
10-4 
4-1 
•0909 
•0920 
(3) 
I In middle of f 
•600 
21-7 
12-8 
•0482 
•0530 
(B 
J tube \ 
•271 
10-7 
7-0 
•0773 
•0610 
The observed values of the superheat and of the heat-loss agree in a general 
manner with those calculated by the theory given in § 32, but there are some 
difficulties. The limit in the last column of lines (1) and (2) is estimated for a 
metallic flow-tube, and it is difficult to see how the heat-loss for a glass flow-tube 
could approach this value so closely when the conductor is in imperfect contact with 
the glass over a small fraction of its surface. The superheat 12°'8 in line (3) agrees 
* In discussing these observations Dr. Barnes takes (EC - 4 - 2 Q dd) in place of the actual heat-loss, 
which unduly exaggerates the difference. The conductor had evidently been annealed , otherwise it would 
have filled the tube instead of resting along the side. 
