OF ARTS AND SCIENCES. 205 



the outer wooden case below the shelf is provided with wooden doors, 

 wliich should be shut in front of the metal box before approaching the 

 balance. When the thermometer hanfrinsr with its bulb inside the box 

 indicates the same temperature as one hanging from the balance case 

 outside within one tenth of a centigrade degree, the conditions are 

 most favorable for accurate weighing. 



In order to maintain the atmosphere within the box in a constant 

 hygrometric condition, we placed in it four large open dishes of sul- 

 phuric acid, two on the bottom of the box and two on a shelf near the 

 top, as described in our previous paper. The acid was first boiled with 

 a fraction of a gram of ammonic sulphate, to remove the last traces 

 of nitrous fumes, and renewed as occasion required. 



We have found that we can most accurately estimate slight differ- 

 ences of weight with a balance thus loaded by observing the amplitude 

 of the first swing of the pointer when the pans are relieved. With 

 the balances of Becker, after the beam has been set on its bearings by 

 the usual cam motion, the pans are still supported by two delicate arms, 

 which are held by a weighted lever against the under side of the pans, 

 but which can be pressed down by pushing against the button shown 

 iu the drawing on the front of the balance case. Such a mechanical 

 arrangement is peculiarly adapted to this method of weighing, as it 

 relieves the pans suddenly and under the same conditions, so that the 

 amplitude of the subsequent swing is an accurate measure of the differ- 

 ence of weight acting at the two ends of the balance beam ; and in 

 this case the inertia of the heavy load renders the effects surprisingly 

 uniform. We give these details, not because they involve any new 

 principles, but because the results have been reached after a long se- 

 ries of experiments, and our experience may save others much loss of 

 time. With the apparatus arranged as we have described it, a dif- 

 ference of weight amounting only to one tenth of a milligram was 

 distinctly indicated, although the load exceeded half a kilogram on each 

 pan, and the compensation was made so perfect that the equilibrium 

 was not perceptibly altered by any ordinary changes of atmospheric 

 conditions. 



In order to reach such compensation, a counterpoise is first selected 

 which on rough measurement appears to be nearly of the same volume 

 as the balloon, and at the same time a little lighter in weight. The 

 apparatus is then mounted with this counterpoise as described, the stop- 

 cocks of the balloon being closed. A series of observations are then 

 made of the variations of weight arising from changes of atmospheric 

 pressure and temperature, as carefully measured as possible. From 



