ON THE MECHANICAL EQUIVALENT OP HEAT. 335 
and in the few trials in which this became a sensible quantity it was to be added as 
a correction. 
The Loss of Heat by the Leakage of Water from the Automatic Cock. 
36. This was the product of {iVc), the weight of water which escaped multiplied by 
the total rise of temperature. Since the water passing the cock was on its way to the 
high temperature thermometer, where any such water was caught it was put into the 
tank, and so required no correction. This leakage was very small, at most 2 oz. in a 
trial, but as there mmst be some evaporation as the water escaped through the hot 
gland, which, though small, might be of some importance on account of the latent 
heat of evaporation, it was desirable in some way to enclose this stuffing-box in an 
indiarubber bag closing on the spindle, so that the vapour could not escape, and this 
was eventually accomplished very effectively and neatly by Mr. Foster, in a way 
which did not interfere at all with the free action of the cock (Art. 14, Part II.). 
The result of this, besides preventing any subsequent loss of water, in this way, . 
was to show that any error that had previously existed from evaporation was 
inappreciable. 
The Loss of Water at the Sivitch. 
37. Apart from evaporation which would result from the exposure to the air and 
in passing the air-gap into the switch, there was no loss, as the water descended 
almost tangentially on to the surface of the tube on the switch which received it, 
the switch itself being a vertical knife-edge extension of this surface, which passed 
through the vertically descending water at starting and stopping; and further, to 
prevent any minute drops of water going astray from the bursting of an occasional 
bubble in passing, a sheet brass hood was placed round the descending pipe directly 
the trial started. 
The outside of the weighing tank is completely exposed to observation, and is 
perfectly tight. The valve in the bottom, being a 4-inch leather-faced screw-valve 
on a brass seat, is also tight, but for satisfaction it was arranged to place a clean tin 
dish under the valve before starting a trial, and only to remove it after the water 
was weighed, so that there should be absolutely no loss of water from any of these 
causes. 
That there must be some loss of water by evaporation to the air as long as the 
temperature of the water, after leaving the condenser, was above that of the dew¬ 
point of the surrounding air, was certain. By using sufficient cooling water it would 
be possible to bring the temperature down to that of the dew-point; but it was 
found that this could not be done under all circum.stances without a larger condenser, 
for which room was wanting, and, as long as the water lost by evaporation was the 
