April 8, 1898.] 



SCIENCE. 



475 



few millimeters of gas from that of the 

 rest. In this way they arrived at the fol- 

 lowing ' Sixth Law : ' 



"The cooling power of a fluid diminishes 

 in a geometrical progression when its ten- 

 sion itself diminishes in a geometrical pro- 

 gression. If the ratio of this second pro- 

 gression is 2, the ratio of the first is 1.366 

 for air; 1.301 for hydrogen ; 1.431 for car- 

 bonic acid, and 1.41-5 for oleflant gas." 



My own observations show that this law 

 can be approximately trae only in the case 

 of a large balloon, an! at pressures from a 

 few millimeters upward. There is no sug- 

 gestion of it when a small balloon is used, 

 and at small pressures it does not obtain 

 with either large or small balloons. 



It was through misplaced confidence in 

 their Sixth Law that D along and Petit 

 were led to place a value on the rate or 

 velocity of cooling in vacuo, something like 

 a hundred per cent, too high, and as they 

 derived the cooling values of gases by de- 

 ducting the cooling effect of a vacuum from 

 the total cooling observed, all their values 

 for gases are much too low. 



Other experimentalists, also, have studied 

 the transfer of heat by air and other gases 

 at various pressures. Kundt and Warburg 

 (Pogg. Ann., 1S74-5), and Winkelmann 

 (Pogg. Ann., 1875-6), observed that the 

 rate of heat transmission remained sub- 

 stantially constant through a long range of 

 diminishing pressure, and then decreased 

 with further exhaustion. But as they made 

 no measurements of pressure below one 

 millimeter (1316 millionths of atmospheric 

 ■pressure), their results have no quantitative 

 value for low pressures. 



Crookes, in his paper, ' On Heat Conduc- 

 tion in Highly Earefied Air ' (Proc. Roy. 

 Soc.^ 1880), described a similar experiment 

 in which he carried the pressure measure- 

 ments as low as 2M. (two millionths). 

 From the fall in the rate of heat loss which 

 occurred between the pressures of 760 milli- 



meters and 1 millimeter, and 5 M. and 2 M., 

 he concludes: " We maj^ legitimately infer 

 that each additional diminution of a mill- 

 ionth Ivould produce a still greater retard- 

 ation of cooling, so that in such high vacua 

 as exist in planetary space the loss of heat 

 — which in that case would only take place 

 by radiation — would be exceedingly slow." 



In this conclusion Mr. Crookes was, I 

 think, wrong. I find that the curve repre- 

 senting the rate of cooling does not break 

 down materially at pressures as low as a 

 twentieth of a millionth. 



My own investigations on ' The Trans- 

 mission of Eadiant Heat by Gases at Vary- 

 ing Pressures ' form a part of a general 

 study of the properties of high vacua, in 

 which I have long been engaged. 



In the course of my work it became 

 necessary to know how much of the heat 

 communicated by a good radiating body at 

 ordinary temperatures, to a neighboring 

 body at a slighly lower temperature, through 

 an intervening gas, is transmitted by the 

 so-called ether, and how much by the gas ; 

 and whether any of that transmitted by the 

 gas is communicated otherwise than by the 

 process of convection. Also why, and to 

 what extent, do the gases differ from each 

 other in their heat transmitting capacities. 



In the drawings herewith. Fig. 1 is a 

 diagram of the apparatus used in my ex- 

 periments. A is the thermometer whose 

 cooling was observed. It has a very open 

 scale divided into two-tenths degrees C. 

 The zero point is placed a long distance 

 (about 170 millimeters) above the bulb, for 

 obvious reasons. The bulb is cylindrical, 

 about 20 mm. long, and about 7 mm. in 

 diameter, and is coated with lamp-black 

 applied with a very thin alcoholic solution 

 of shellac. After several hours' baking at 

 100 degrees in a good vacuum, this bulb 

 gave constant radiation results. The ther- 

 mometer is suspended by a platinum wire, 

 with its bulb in the center of the large 



