BETWEEN THE FREEZING AND BOILING-POINTS. 
197 
marking an attached sheet of paper. The pen was connected to a relay, which was 
excited by an electric current whenever the pendulum of the clock swung through a 
globule of mercury at the middle of each stroke. The stream of water from the 
calorimeter could be made to flow out of either one of two nozzles before it entered 
the tared flask for obtaining its weight. The switch-over device was made from a 
3-way glass tap, and so arranged that when the tap was turned so as to change the 
flow from one tube to another, the time of closing one and opening into the other was 
recorded by the pen on the chronograph sheet. Between the opening and closing, 
the flow was shut off altogether, but as this only amounted to from two to three- 
tenths of a second, the expansion of the rubber tube connections in the water circuit 
prevented the flow of water through the calorimeter from being interrupted, and any 
sudden shock to the conditions was avoided. 
Fig. 7 gives a general plan of the switch¬ 
over device. The handle of the glass tap 
was connected to a long arm, which could be 
moved between two stops, representing the 
position when either nozzle was open. At 
the time of switching over, two marks were 
recorded on the chronograph, the mean of 
which was taken in estimating the interval 
of any flow. These two marks could each be 
estimated to '01 sec. very easily with a 
millimetre scale, and were probably in all 
cases accurate to '02 sec. on 900 seconds. 
The standardization (indirectly to standard 
time) of the clock marking the seconds was 
done at frequent intervals during tire course 
of the present series of experiments by com¬ 
parison with a Frodsham and Keen ship’s 
chronometer. The rate of this chronometer was determined not only by direct 
comparison with the standard clock at the University Observatory, but by repeated 
telephonic communication between an observer at the Observatory and myself in the 
Physics Building. This rate, which was a slight gain, was extraordinarily constant. 
The rate of the clock, a loss, varied considerably between winter and summer, although 
the variation was very consistent and regular. The rates during the corresponding 
months of a year agreed almost to I second a day. 
