On Crookess Layers at Atmospheric Tensions. 57 



was shown by Professor Barrett, at the Brighton meeting of the 

 British Association, with the improvement of adding soap to 

 the water, an addition which seems essential to the full success 

 of the experiment. A copper ball, some six cm. in diameter, 

 furnished with a staple by which it can be lifted, was brought to 

 a bright red heat, and while glowing was lowered into a large 

 beaker of soapy water. As the ball approaches the cold surface 

 of the water heat passes from the ball to the water by conduction 

 or penetration* as well as by radiation • accordingly the interven- 

 ing air becomes intensely polarzied, and the Crookes's stress that 

 accompanies the polarization makes a hollow in the surface of the 

 water. Let the ball be lowered till it is half submerged, the de- 

 pression in the water is now nearly hemispherical, but not quite 

 so, since the interposed layer of polarized gas will be thinnest at 

 the bottom, where, to withstand the pressure of the water, it must 

 exert most force. The stresses at any point of this polarized 

 layer consist of a constant stress P nearly equal to the tension of 

 the open atmosphere, acting equally in all directions, along with 

 a variable Crookes's stress p, acting for the most part nearly in 

 the direction of a radius of the ball ; the most marked deviation 

 from this direction being close to the horizontal surface of the 

 water, where the action of the upper hemisphere of the ball gives 

 an inclined direction to the Crookes's stresses, and helps to round 

 off the surface of the water. The amount of the Crookes's 

 pressure acting on the water will vary with the depth, being such 

 at each point that it gives a component equal and opposite to 

 the resultant of the pressure of the water at that depth, and of 

 the surface tensions round the point. Whenever the Crookes's 

 force is not quite in the direction of this resultant, there will be 

 a free tangential component, and this must produce surface cur- 

 rents in the water. These, however, cannot be observed in the 

 present experiment because they are of small amount, and too 

 much mixed up with convection currents arising from the heat 

 that reaches the water by radiation and diffusion. 



When the ball is lowered until it is quite submerged it will be 

 surrounded on all sides by its envelope of polarized air, thinnest at 



* The heat which diffuses across a layer of gas passes under what are known as the 

 laws of conduction if the number of gaseous molecules present is sufficiently large. If 

 fewer molecules are present the heat passes under other laws, which ma3 r be distinguished 

 from the laws of conduction by calling them the laws of penetration. 



