3^ 



SCIENCE. 



kinds of forces are acting simultaneously upon the same 

 glacier, and while huge icy mountains are at intervals of 

 centuries rising from their dense, watery bed, other and 

 smaller ones are more frequently dropping from its sea- 

 ward face, for those formed by dropping are far smaller 

 than those which rise into the sea, as the following 

 diagram will serve to show. Although about seven- 

 eighths of an iceberg is submerged, it must not be 

 inferred that, when its height has been determined, seven 

 times that height is its depth below the sea level. If of 

 a tabular shape, this proportion becomes more nearly 

 correct ; but if of a pyramidal or conoidal cross section, 

 which is far oftener the case, the lineal proportions of 

 height to depth approach each other more closely, while 

 the volumes, necessary to hydrostatic equilibrium, remain 

 invariable. Their great height, as compared with their 

 breadth shows that these lineal proportions do not obtain 

 beneath the sea level, or the mass.it homogeneous could not 

 be in a state of stable equilibrium, and would topple over, 



which sometimes happens when the conditions of equili- 

 brium are disturbed by the unsymetrical decrease of its 

 different faces. 



The height of bergs, estimated or measured by various 

 Arctic voyagers, varies greatly. During the warm 

 months of summer, when they are most frequently en- 

 countered by navigators, they are often surrounded by a 

 hazy mist, due to the condensation of the surrounding 

 moisture by their chilly faces, and the effect is to make 

 them appear much higher than they really are, and to 

 render estimates of their height particularly unreliable. 



As about seven-eighths of an iceberg is under water, 

 the curious spectacle, which has often been seen in Polar 

 latitudes, of these monsters ploughing their way against 

 a rapid current, loaded with heavy pack-ice, and in the 

 very teeth of a strong gale of wind, can be readily under- 

 stood on the theory that the surface current is shallow, 

 and the drifting colossus is only obeying the mandates of 

 a deeper and more powerful agent. 



ON HEAT CONDUCTION IN HIGHLY RAREFIED 

 AIR* 



By William Crookes, F.R.S. 



The transfer of heat across air of different densities 

 has been examined by various experimentalists, the gen- 

 eral result being that heat conduction is almost inde- 

 pendent of pressure. Winkelmann (Pogg. Ann., 1875, 

 76) measured the velocity of cooling of a thermometer 

 in a vessel filled with the gas to be examined. The diffi- 

 culty of these experiments lies in the circumstance that 

 the cooling is caused not only by the conduction of the 

 gas which sui rounds the cooling body, but that also the 

 cui rents of the gas and, above all, radiation play an im- 

 portant part. Winkelmann eliminated the action of cur- 

 rents by altering the pressure of the gas between 760 

 and 1 millim. (with decreasing pressure the action of gas 

 currents becomes less); and he obtained data for elimi- 

 nating the action of radiation by varying the dimensions 

 of the outer vessel. He found that, whereas a lowering 

 of the pressure from 76010 91.4 millims. there was a 

 change of only 1.4 per cent, in the yalue for the velocity 

 of cooling, on further diminution of the pressure to 4.7 

 millims. there was a further decrease of 11 percent., and 

 this decrease continued when the pressure was further 

 lowered to 1.92 millim. 



About the same time Kundt and Warburg {Pogg. Ann., 

 1874. 5) carried out similar experiments, increasing the 

 exhaustion to much higher points, but without givir.g 

 measurements of the pressure below 1 millim. They en- 

 closed a thermometer in a glass bulb connected with a 

 mercury pump, and heated it to a higher temperature 

 than the highest point at which observations were to be 

 taken ; then left it to itself, and noted the time it took to 

 fall through a certain number of degrees. They found 

 that between 10 millims. and 1 millim. the time of cool- 

 ing from 60" to 20° was independent of the pressure : on 



* Abstract of a Paper read before the Royal Society, Dec. 16, 1880. 



the contrary, at 150 millims. pressure the rate was one 

 and a half times as great as at 750 millims. Many pre- 

 cautions were taken to secure accuracy, but no measure 

 ments of higher exhaustions being given the results lack 

 quantitative value. 



It appears, therefore, that a thermometer cools slower 

 in a so-called vacuum than in air of atmospheric pres- 

 sure. In dense air convection currents have a consider- 

 able share in the action, but the law of cooling in vacua 

 so high that we may neglect convection has not to my 

 knowledge been determined. Some years ago Professor 

 Stokes suggested to me to examine this point, but find- 

 ing that Kundt and Warburg were working in the same 

 direction it was not thought worth going over the same 

 ground, and the experiments were only tried up to a cer- 

 tain point, and then set aside. The data which these 

 experiments would, have given are now required for the 

 discussion of some results on the viscosity of gases, 

 which I hope to lay before the Society in the course of a 

 few weeks ; I have therefore completed them so as to 

 embody the results in the form of a short paper. 



An accurate thermometer with pretty open scale was 

 enclosed in a \]/ 2 inch glass globe, the bulb of the ther- 

 mometer being in the centre, and the stem being enclosed 

 in the tube leading from the glass globe to the pump. 



Experiments were tried in two ways : — 



I. The glass globe (at the various exhaustions) was 

 immersed in nearly boiling water, and when the tempe- 

 rature was stationary it was taken out, wiped dry, and 

 allowed to cool in the air, the number of seconds occu- 

 pied for each sink of 5 being noted. 



II. The globe was first brought to a uniform tempera- 

 ture in a vessel of water at 25 , and was then suddenly 

 plunged into a large vessel of water at 65 . The bulk 

 of hot water was such that the temperature remained 

 sensibly the same during the continuance of each experi- 

 ment. The number.of seconds required for the thermo- 

 meter to rise from 25° to 50 was registered as in the 

 first case. 



