106 
PROFESSOR HUGH L. CALLENDAR OX 
of high conductivity, on the outside of which a rubber spiral was wound to fit the 
outflow-tube as closely as possible. The accuracy of fit was found to be much more 
important in the case of water than in the case of mercury. The reason of this is 
that, the thermal conductivity of water being 10 or 15 times less than that of 
mercury, the accurate averaging of the outflow temperature is more dependent on 
the uniformity of the spiral circulation and the complete elimination of asymmetric 
stream-lines. 
In order to obtain a perfect fit for the sleeves with their spiral screws, it was 
necessary that the bore of the outflow-tube should be as nearly uniform as possible, 
and accurately straight. It was most essential that there should be no constriction 
at the points of junction E and F with the vacuum-jacket, and that the external 
portions of the tubes AE, FD should not be of smaller bore than the portions inside 
the vacuum-jacket, though it would not matter much if they were a little larger. 
These details of the design, which determined the choice of the dimensions of the 
tubes, had all been carefully worked out before the ordering of the first six 
calorimeters with vacuum-jackets in October, 1896, and the importance of the 
straightness and uniformity of the tubes was clearly explained in the specification. 
There was some difficulty in making the calorimeters accurately to specification, and 
when they arrived about three months later, it was found that this particular detail 
had been somewhat overlooked. It was consequently a difficult matter, even with a 
soft rubber spiral, to secure sufficient perfection of fit, and the accuracy of many of 
the earlier experiments was seriously impaired. 
The effect of an imperfect fit was to permit part of the heated stream to escape 
directly past the thermometer, so that the temperature indicated by the outflow 
thermometer was lower than the true mean of the flow. This defect was less 
apparent with a large flow or a large rise of temperature, either of which conditions 
tend to promote mixing of the liquid and the attainment of a proportionately greater 
uniformity of temperature. A good illustration of this is afforded by the results 
quoted by Dr. Barnes in Section 7 of his paper, p. 237, which were obtained with one 
of the first three calorimeters in which the bore was undoubtedly defective. The 
apparent diminution of the heat-loss per degree rise with increase in the rise of 
temperature is probably due in part to the more perfect mixing of the stream caused 
by the greater differences of temperature, which promote instability of flow and 
increase the formation of eddies. I may add that I have repeatedly observed the 
same effect in my experiments on steam by a similar method. With steam it is 
much more difficult than with water to secure a true average of the outflow 
temperature. Any imperfection of fit or circulation immediately produces the 
observed effect in an exaggerated form. 
The greater irregularity of the results (Barnes, Section 7) for the small flows is 
probably in part accidental, but is also characteristic of imperfect fitting of the 
rubber spiral in the outflow-tube, which is obviously more detrimental in the case of 
