108 
PROFESSOR HUGH L. CALLENDAR ON 
prevent excessive superheating by the current. It was found that the dimensions 
above given satisfied the required conditions very fairly, and although there were a 
few minor details which could not he satisfactorily settled until the complete apparatus 
had been fitted up, it was not found necessary to make any changes in the essential 
features of the design. The three calorimeters ordered two years later by Dr. Barnes 
were of the same design in all important points, except that two of them had 3 millims. 
bore flow-tubes ; but these were found to be less suitable, and were employed in very 
few of the tests. 
(28.) Improvements in the Design of the Calorimeter. 
Although Dr. Barnes was naturally unwilling to introduce any radical changes in 
the original pattern, which had proved to be capable of giving very good results, there 
can be no doubt that it was capable of improvement, and I had in fact already noted 
several points in which alterations were desirable. 
The importance of uniformity of bore in the flow-tube, and particularly in the out¬ 
flow-tube, has already been referred to (§ 26). This was remedied in the later 
apparatus. Two of the side tubes were also removed, and the other two bent parallel 
to the flow-tube, to facilitate insertion in the cylindrical form of jacket. These 
modifications were of small importance, and were made in the first calorimeters after 
the apparatus was received. A more important improvement in the same direction 
would be to have the tube for exhausting the vacuum-jacket inserted at F, fig. 6, in a 
direction parallel to the flow-tube, which would permit the water-jacket to be made 
much smaller, thus securing a more vigorous and uniform circulation. 
It would be most important for future work to endeavour to reduce (1) the risk of 
error from conduction at the outflow end of the calorimeter at high temperatures ; 
(2) the correction for the heat-loss, which amounted to 4 per cent, at the higher 
points of the range. 
The conduction error might be reduced by including a greater length of the outflow- 
tube in the vacuum-jacket. This would he much more effective than lagging the 
exposed parts with flannel, since the flannel lagging is exposed to the temperature ot 
the laboratory, which is much lower than that of the jacket. Besides, the lagging is 
apt to become damp, and takes a long time to reach a steady state. The length of 
outflow-tube inside the vacuum-jacket might easily be increased to 25 centims. instead 
of 12, with 5 centims. outside. This would greatly diminish the possible uncertainty 
of the conduction loss. 
The heat-loss depends chiefly on the extent of surface of the flow-tube and ther¬ 
mometer bulb. The thermometer bulb could not be made much smaller, but the 
greater part of the loss arises from the flow-tube, which might be reduced to about 
half the external diameter, although the bore coidd not conveniently be made less than 
1-5 millims. This would also be an advantage as diminishing both the heat capacity 
