BETWEEN THE FREEZING AND BOILING-POINTS. 
237 
wire is much more nearly the same for the two flows and more nearly equal to that 
of the water column, the differences being of the order of 1° C. only. 
Sec. 7 .—Preliminary Measurements of the Mechanical Equivalent. 
Our first measurements of the mechanical equivalent in the summer of 1898 were 
made with Thermometer C and Calorimeter B. This had a flow-tube slightly less 
than 2 millims., and, with the exception of the device for eliminating stream-line 
motion, was fitted up in a similar way to the later calorimeters. It is a matter of 
interest to determine the way in which the heat-loss varies with rise of temperature 
for this case. I have summarized the observations which we made at that time to 
determine this, and expressed them here in terms of the same values for the units as 
the later measurements. The results are corrected to the same value of Q, and were 
all obtained approximately at a mean temperature of 30° C. 
Relation of Heat-loss to Rise of Temperature. 
Large flow. 
Small flow. 
Q = ‘54000 gramme per second. 
Q = -27300 gramme per second. 
cl6. 
(EC - 4 • 2 Q, d9)/d6. 
dO. 
(EC-4-2 Q d6)/d0. 
3-0462 
•04445 
2-9717 
•04941 
5•9427 
•04403 
5-8891 
•04904 
8-9131 
•04298 
9-0285 
•04982 
12-2129 
•04070 
11-9785 
•04809 
The readings for the large flow are very consistent, as shown by the jflot in fig. 14. 
For the small flow the variations in the observations are far from satisfactory, but 
they show a similar decrease in the value of the heat-loss with rise of temperature 
as for the large flow. The decided bend in the curves shows that, as the temperature 
of the out-flowing water is decreased, the temperature gradient down the fine-bore 
tube approaches more nearly a straight line (cf. fig. 2, p. 154.) The decrease in the 
heat-loss with increase of temperature points to the more perfect confinement of the 
heated water around the wire in its passage through the tube, which is occasioned by 
its greater difference in density. 
The small flow allows of the more perfect distribution of heat throughout the 
water column in the flow-tube, and the curve approaches a limiting value, as the 
temperature is lowered, much sooner than in the case of the large flow. If we may 
assume the two limiting values of the heat-loss per degree rise in the calorimeter for 
the two flows by extrapolating for a value of dd — 0 in the two cases, and accept 
