542 
MR. J. T. BOTTOMLET OR THE 
It will be seen, however, from the method of making the experiments that to fulfil 
this condition was impossible. 
In both of the modes of conducting the experiments the heating arrangement is at 
a considerable distance from the layers which the thermometers A and B represent; 
and the heat supplied travels gradually downward. 
In the arrangement (fig. 1) the rise of the two thermometers A and B goes on con¬ 
tinuously. In the arrangement of fig. 2, where there was only a limited supply of 
heat, namely, that belonging to the layer of hot water poured on the top, both thermo¬ 
meters first gradually rise in temperature and then gradually fall. 
The following, then, was the mode of experimenting and of calculation. The water 
to be experimented on was first carefully stirred from top to bottom to equalize the 
temperature throughout. The heating was applied and the temperatures of all four 
thermometers were read. An hour or so afterwards the regular observations began. 
After that time the thermometer A- was generally found to be commencing to rise 
slowly. Somewhat later the thermometers I and B were found to be rising, and from 
that tune until the lowest thermometer was observed to be rising the reading of the 
thermometers was carried on continuously at intervals of five minutes or of ten 
minutes. The observations are generally carried on for five or six hours and are 
therefore laborious. 
It is unnecessary to put down here the readings taken in the various experiments 
carried out. It is only a small portion of the series of numbers recorded in any one 
experiment that is made use of for calculation. The part that is made use of is that 
which keeps the difference of temperatures of the thermometers A and B as nearly as 
possible constant for a considerable time. In the first mentioned method of heating 
it is that part at which the two thermometers A and B are both rising at nearly the 
same rate. In the second method it is that part of the tune of observation during 
which the two thermometers remain most nearly stationary. During this part of the 
time the increase in temperature of the integrating thermometer is taken. The average 
temperature indicated by A is taken as the temperature of the level which is repre¬ 
sented by it, and the average temperature of the thermometer B is taken as the 
temperature of the level which is represented by it. These are used in the formula 
given above as the temperatures T and T' respectively. 
A specimen of part of a series of thermometer readings, extending over two hours, 
is given in Table I. appended. The heating was carried on by the method of fig. 1. 
The heating was commenced at about 9 A.M., and the observations were carried on till 
4 p.m., when they were discontinued as the thermometer at the bottom of the stratum 
occupied by the integrator began to rise very sensibly. 
Table II. contains results calculated from a number of satisfactory experiments. 
From these and from the consideration of other results, I conclude that the con¬ 
ductivity for heat of water at the temperature at which I have experimented is to be 
reckoned at '0023 square centims. per second. 
