The question of the variation of the specific heat of water is so fundamental in calorimetry, and the results of different observers and different methods are still so discordant, that no apology is needed for the publication of fresh experimental evidence. The continuous electric method, which I carried out in conjunction with Prof. Barnes, was specially designed to avoid the main sources of error of the older methods in which mercury thermometers and open calorimeters were employed. In this method. the rise of temperature of a steady current of water, heated by a steady electric current in its passage through a fine tube hermetically scaled in a vaccumjacket, was observed with a differential pair of platinum thermometers. Errors due to lag, or to uncertainty of water-equivalent, or to evaporation or heat-loss in transference, were thus eliminated, and a higher order of accuracy was secured in the temperature measurements. The results of the continuous electric method in the case of water showed a variation of specific heat amounting to less than one half of 1 per cent. between 10° and 80°C., with a minimum at 37.6°C., followed by a very slow and steady rise. The mean value from 0° to 100°C. agreed to 1 in 2000 with the experiments of Reynolds and Moorby by the mechanical method, and the values from 5° to 35° C. agreed to a similar order of accuracy with the experiments of Rowland. But the value at 80°C. was 1 per cent. lower than that found by Lüdin's (Zürich, 1895), employing the ordinary method of mixture with an open calorimeter and mercury thermometers. Lüdin's results for the variation over the range 30° to 100°C. agreed more closely with the continuous electric method than those of any previous observers; but showed a minimum at 25°C., and a maximum at 87°C., which could not be reconciled with the experiments of Reynolds and Moorby on the mean specific heat from 0° to 100°C., or with the most probable reduction of Regnault's experiments between 110° and 190°C.
On a Mechanical Method of Solving Quadratic and Cubic Equations, whether the Roots be real or impossible
