On the Method of Condensation in Calorimetry. 365 



all three act, or whether the first two are not practically efficacious of 

 themselves, I think that theory is here in accord with experiment in 

 showing that on this count no appreciable inaccuracy is to be appre- 

 hended. 



There are a few other considerations. From the facts that there is 

 a fall of temperature across the film, the gradient descending nearly 

 to the temperature of the substance on the one side, and that through- 

 out the period of condensation the film continually gravitates down- 

 wards as it receives accretion from without, it might be expected that 

 water which had not yet risen to steam temperature might in some 

 cases reach the bucket. This occurs, but need obviously be no source 

 of error. If experiments be made in glass calorimeters, the formation 

 of a dew on the underside of the bucket will often be noticed ; to this 

 extent it does no harm. But in the case of sudden and copious 

 precipitation, especially in the case of smooth substances, where water 

 runs off more freely at points close to the surface of the body, the 

 precipitation on the outside of the bucket may collect in such quantity 

 that drops fall off. I have occasionally noticed this, and it should be 

 borne in mind ; but with the flat form of bucket used, and the safe- 

 guard afforded by the supporting claw, there is little danger of this 

 occurring up to rapid precipitations not much exceeding a couple of 

 grams. Over that amount a second catch water, which may be of 

 small dimensions, should be provided beneath the claw. 



The substance hanging in the steam is safe from radiation errors. 

 The phenomena attending accretion of weight terminate in the con- 

 ditions which prevail in the case of a liquid in presence of a saturated 

 atmosphere of its own vapour, the whole maintained at constant 

 temperature and pressure, when the relative weights of liquid and 

 vapour remain constant statistically, and absolutely so far as we can 

 measure. If radiation from the hot substance to the cooler walls of 

 the calorimeter occurred, constancy of weight would not be attained. 

 Jt is certain that the least radiation across the steam-jacket would be 

 revealed in continued slow accretion of weight. But there is nothing 

 so striking in experiments effected by this method as the steadily 

 maintained equilibrium of the balance once the substance has attained 

 the temperature of the steam, the pointer vibrating with perfect 

 equality so long, apparently, as we chose to prolong the experiment. 

 If the steam supply be cut off for a moment there is loss of weight, 

 and on restoration of steam there is finally on the whole gain of 

 weight, for radiation, as well as evaporation, occurs in the absence of 

 steam. If the pressure be appreciably increased by throttling the 

 exit pipe and partially closing the boiler for a couple of seconds, the 

 balance reveals increase of weight ; but so long as the pressure is 

 neither increased nor diminished equilibrium is absolutely preserved. 



It is obvious that disturbing effect due to water-dust or mechanically 



