TRANSACTIONS OF SECTION A. 627 
The experiments of Bartoli and Stracciati (1891) were made by the method of 
mixtures, and are expressed in terms of a thermal unit at 15°C. The correspond- 
ing curve, marked B. & S., bears a general resemblance to that of Rowland, but 
shows a minimum at 20° OC. The errors and limitations of this method are well 
known, and it is difficult to suppose it capable of an order of accuracy higher than 
one part in a thousand, or to resist the impression that this excessive lowering of 
the minimum point is due to some constant error inherent in the method, whick 
cannot be eradicated by mere repetition of similar experiments. 
The experiments of Griffiths (G) over the range 15° to 25° were made ky 
observing the rate of rise of temperature of a mass of water in a calorimeter 
heated by an electric current. His work threw a flood of light on the difficulties 
of electrical calorimetry as usually practised, and explained the failure of previous 
observers to secure satisfactory results by this method. Over the range of his 
experiments he found approximately the same rate of diminution of the specific 
heat as that given by the experiments of Rowland when reduced to the same 
scale of temperature. 
The curve marked CB in the figure, extending from 4° to 60°, represents the 
Fie, 2.—Variation of Specific Heat of Water. 
a ed 
J. & A, Jamin and Amaury, 1870. ; G. Griffiths, 1893, ‘ KC 
2, & 8, Bartoli and Stracciati, 1891. A cauaadnna Gadmaad seein Clay Cell 
Ro. Rowland, 1880 ; tp. reduced to Paris scale, C.B. Callendar and Barnes, Mieke or i 
results so far obtained in the present investigation. The points indicated by small 
circles represent samples of single observations with different calorimeters, and are 
inserted to give an idea of the order of agreement attainable by this metho?. 
The order of accuracy diminishes as the temperature rises, owing to the greater 
difficulty of satisfactorily regulating the temperature of the water-jacket at the 
higher points. It is hoped, however, by a slight modification of the heating and 
regulating apparatus, to secure good results at temperatures as high as 90° C. 
on spp water-jacket, and to obtain an observation at 100°C. with a steam- 
jacket. 
According to the authors’ experiments, the curve representing the variation of 
the specific heat is much flatter than that given by Rowland, and has a minimum 
at 40°C. instead of 29° C. The experiments of Rowland did not extend sufficiently 
heyond the minimum point to afford a really satisfactory determination. The 
$§2 
