626 REPORT—1899. 
After passing through a cooler, it was collected and weighed in a tated flask in 
such a manner as to obviate all possible loss by evaporation. 
The range of temperature, d6, was generally from 8° to 10° in the series of 
experiments on the variation of J, but other ranges were tried for the purpose of 
testing the theory of the method and the application of small corrections. The 
thermometers were read to the ten-thousandth part of a degree, and the difference 
was probably in all cases accurate to ‘001°C. This order of accuracy could not 
possibly have been attained with mercury thermometers under the conditions of 
the experiment. 
The external loss of heat, H, was very small and regular, owing to the perfec- 
tion and constancy of the vacuum attainable in the sealed glass jacket. It was 
determined and eliminated by adjusting the electric current so as to secure the 
same rise of temperature, d6, for widely different values of the water-flow. 
The great advantage of the steady-flow method as compared with the more 
common method in which a constant mass of water at a uniform temperature is 
heated in a calorimeter, the temperature of which is changing continuously, is that 
in the steady-flow method there is practically no change of temperature in any 
part of the apparatus during the experiment. There is no correction required for 
the thermal capacity of the calorimeter; the external heat loss is more regular 
and certain, and there is no question of lag of the thermometers. Another 
incidental advantage of great importance is that the steadiness of the conditions 
permits the attainment of the highest degree of accuracy in the instrumental 
readings. 
In work of this nature it is recognised as being of the utmost importance to be 
able to detect and eliminate constant errors by varying the conditions of the 
experiment through as wide a range as possible. In addition to varying the 
electric current, the water-flow, and the range of temperature, it was possible, 
‘with comparatively little trouble, to alter the form and resistance of the central 
conductor, and to change the glass calorimeter for one with a different degree of 
vacuum, or a different bore for the flow tube. Im all six different calorimeters 
were employed, and the agreement of the results on reduction afforded a very 
satisfactory test of the accuracy of the method. 
The general results of the investigation, so far as it has been possible to work 
them out for publication at present, may be gathered from an inspection of fig. 2, 
which includes the results of previous observers plotted on the same scale. ‘The 
curve marked Regnault, 1840, represents the well-known formula of Regnault 
which has been adopted as the basis of much calorimetric work. This formula 
was confessedly approximate, and was deduced from experiments on mixing water 
at high temperatures with water at 15° C. The method could not be expected to 
give any information with regard to the variation of the specific heat at ordinary 
temperatures. The experiments of Jamin and Amaury (J. & A.), 1870, by the 
method of electric heating, gave a very rapid increase of the specific heat at low 
temperatures, but the science of electrical measurement, and the difficulties of the 
electrical method, were not at that date sufticiently appreciated to render the 
results of any value. 
The discovery of the diminution of the specific heat of water with rise of 
temperature from 0° to 30° C. was made by Rowland in his investigation of the 
mechanical equivalent of heat by the method of Joule. His original results, 
reduced to the scale of his own air thermometer, are shown by the dotted curve 
marked Ro. The corresponding values of the specific heat in absolute measure 
are shown by the scale of joules in the right-hand margin. His results have 
recently been reduced to the Paris scale by the comparison of his thermometers 
with a Tonnelot thermometer standardised at the International Bureau, and with 
a platinum thermometer standardised by Grifliths.! The results so reduced are 
indicated by the full curve Rp. The effect of this reduction is to lower the 
temperature at which the specific heat is 4-200 joules from 10° to 7°C., to diminish 
the temperature coefficient, and to lower the point of minimum specific heat to 
about 29° C, 
} Waidner and Mallory, Phil, Wag. Tune 1899, 
