232 



Mr. J. Joly. 



immediately disturbed. The gain is so rapid during this period that 

 weighing, or observation of the rate of increase, is impossible with 

 the ordinary balance. Between this state of things and that prevailing 

 subsequently, when weighing has become possible, there is, of course, 

 no abrupt transition. But, as observed, it is found simply that 

 weighing (in such a case as I am considering) has become possible, 

 probably in the course of the fourth minute. Let the vibrations of 

 the balance be now observed. The oscillations are perfectly regular 

 for about one minute. If the balance reads tenths of milligrams, a 

 slight preponderance of weight will probably be then observable. 

 Watching the vibrations for 5 minutes from the fourth minute, 

 suppose, and then moving the rider till there is equilibrium, the gain 

 is found to be seven- tenths of a milligram. Observing the balance 

 for another period of 5 minutes, the result is again seven-tenths, and 

 so on. However long observation is carried on, the gain is seven or 

 eight-tenths per 5 minutes. 



Now this is certainly not due to heating of the substance, and it is 

 established by the observations that the seven-tenths of a milligram 

 represents an uniform rate of increment. But from the fourth to the 

 ninth minute observation gave but seven-tenths. It is safe, therefore, 

 to conclude that by th e end of the fourth minute the true condensation 

 due to the calorific capacity of the substance has ceased. 



It remains to consider numerically the importance of this source of 

 error in affecting the degree of accuracy attained by this method of 

 calorimetry. I will illustrate this effect by assuming an extreme case 

 in which the increment is entirely ignored, no correction being made 

 for it except the unconscious one made in effecting the deduction for 

 the carrier, and I will suppose that after weighing has become practi- 

 cable 10 minutes be allowed to elapse, in order to put the temperature 

 of the substance beyond question. I will consider, both in the case 

 of a non-conductor of heat and in the case of a good conductor, the 

 consequent effect on the accuracy of the result. 



(a.) For the former I cannot do better than take the case of the 

 piece of cryolite referred to in (2). It weighs 37 grams. For a 

 range of 90 degrees I get from an experiment on this specimen 

 (" Specific Heats of Minerals," p. 263) that there would be a con- 

 densation of 1'588 grams due to the calorific capacity of the cryolite 

 alone. The increment during the time of observation, according to 

 (2) [ante], will be 074x2 = 1*48 milligrams. Observations on the 

 carrier over a similar interval have, suppose, been made once for all. 

 The deduction due to its thermal capacity then reduces the radiation 

 effect by 0*34 X 2 or 0'68 milligram ; leaving a -f error of 0*8 milligram. 

 This is an error of 0*05 per cent., or 1 in 2000, on the specific heat 

 determined. 



(6.) Let the case be that of a piece of copper having the dimensions 



