230 PROCEEDINGS OP THE AMERICAN ACADEMY. 



the plate were perforated in the centre, and into the latter was welded 

 an open iron tube B, projecting 25 cm. in length and 1-i cm. in diameter. 

 The cylinder was covered, except on the bottom, with a jacket of asbes- 

 tos about 5 cm. thick, and the projecting tube B, which was to serve as a 

 condenser, was wound with spirals of copper wire to increase the cooling 

 surface. Pure mercury was placed in the cylinder and boiled vigorously 

 by means of a number of lamps beneath. The capillary was thus kept 

 at the boiling temperature of mercury under atmospheric pressure. No 

 regard was paid to the variations of temperature arising from changes in 

 barometric pressure, as their effect would evidently be entirely negligible. 



Any desired difference of pressure at the two ends of the capillary 

 was attained by inserting a tube in the ground joint and connecting it 

 with a large air reservoir, R, which was itself connected through the 

 cock ^S" with a suction pump and with an open mercury manometer, M. 

 The gas or vapor entered at the other end of the capillary always under 

 atmospheric pressure. The whole apparatus in the form used for 

 measuring the rate of transpiration of the mercury vapor is shown in 

 Figure 2. 



In making the experiments, the rate of flow of the mercury vapor was 

 first determined in the following manner. While the cylinder was being 

 heated, carbon dioxide was forced through the capillary to prevent the 

 condensation in it of liquid mercury and the formation of its oxide. 

 After the mercury was boiling actively, and its vapor entirely enveloped 

 the capillary, as shown by a mercurial thermometer inserted into the tube 

 B (Fig. 2), it was connected with the suction pump, and mercury vapor 

 drawn through for half an hour. The carefully ground end of a 

 weighed bulb, W, was then inserted in the ground joint, its other end 

 being connected by means of a clamped rubber tube, C, with the air reser- 

 voir, in which the desired reduction of pressure had been produced. At 

 a definite moment the clamp G was opened and the time noted. As the 

 volume of the condensing bulb W was very small compared with the 

 volume of the air reservoir, no sensible change in the pressure was thus 

 produced. The mercury vapor was found to be completely condensed in 

 W about two to three centimeters from the ground joint. It was found 

 that a slight and unavoidable leakage * through the ground joint occurred, 

 and it was therefore necessary to readjust the pressure occasionally. It 



* In tlie case of the mercury experiments no error could arise from this source, 

 as the leakage was inward. In tlie case of those of carhon dioxide and hydrogen, 

 it was proved by bLank experiments that the amounts of carbon dioxide and water 

 which leaked in were less tlian one per cent of the total weiglit. 



