Joty—On the Conservation of Mass. 390 
In the experiments I am describing, the value of m would be one-sixth of 
Heydweiller’s result, and obtained over the interval of 30 seconds; and 
Heydweiller’s result indicated a weight-loss of 0:08 milligrammes in 200 grammes. 
The total mass is now 160 grammes. The value of V—v becomes, therefore, 
0:00008 x 3 x 10° 
_— 3 . : 
6 x 80 x 2x 160 4 x 10° ems. per second; 
and the force causing such acceleration would be 0°640 dynes. It would appear, 
therefore, that the sensibility of the apparatus (0:04 dynes) is such as to show the 
one-sixteenth part of the force calculated from the equation and Heydweiller’s 
results. In short, the method is sensitive to about 6 per cent. of the calculated 
amounts. This result will serve as typical of the dimensions of the experiments 
presently to be given in detail. Owing to small changes made in the quantities 
employed and mass of containing vessels, &c., the sensibility will vary a little 
about this amount ; but calculation in each case would serve no purpose. 
It remains to consider if the experiment is not open to errors of a magnitude 
comparable with the small effects being dealt with. 
The most obvious source of error will reside in temperature changes accom- 
panying the reactions progressing in the closed vessels. Forces arising from this 
source will be irregular or indeterminate in their effects; that is, if the beam is 
symmetrically placed in the containing case, it will be impossible to predict the 
direction in which draughts arising from escaping heat will urge the beam. Ifthe 
vessels are nearer one side of the case than another, there would probably be 
effects resembling attraction towards the nearer wall. The only way, of course, to 
deal with such source of error is to seek to eliminate it altogether for an imterval 
of time sufficient to permit of the observation of any force arising from the re- 
actions. So far as the experiments on copper sulphate are concerned, the jacket of 
about one centimetre thickness of dry cork-dust surrounding the vessels appears to 
sufficiently accomplish this object. The fall in temperature attending the solution 
was, in fact, found to be only a little above one degree Centigrade; and that this 
small rise could produce an effect across the jacket in the space of two or three 
minutes is not to be anticipated. However, direct experiments, to be given later, 
appear to negative any error from this source arising during the first three 
minutes. 
Another and more instantaneous effect will arise out of a change of temperature 
of the air around the vessel. Its contraction or expansion, as the case may be, 
will lead to a small inflow or outflow of air at the mouth of the cylinder. When 
it is considered, however, that only the small volume of air in the interstices of 
the insulator is concerned, and that only a part of this changes its volume by a 
little more than one-third per cent., and further, that this flow is distributed over 
G 2 
