4 ROYAL SOCIETY OF CANADA 



mixture, and with rapid cooling by pouring liquid air over the lower 

 lulb; and also slow and rapid cooling while the water was slightly 

 agitated by tapping. The instrument has been used after standing 

 for long periods in air at the ordinary temperature, and I have stopped 

 the process at all stages for a fev/ minutes and then' reimmersed the 

 bulb. The only result has been the ordinary flat cake, though some- 

 times a few needles have been seen projecting downwards from it. But 

 though a little water has appeared above the cake when it was loosened 

 from the sides, it has on the whole remained at the surface. 



I have no explanation to offer, and in spite of the witness of mem- 

 bers of the class and my own vivid recollection, now find some difficulty 

 in believing that the ice really sank to the bottom. It is so difficult to 

 see how the ice could be more dense than the supercooled water that I 

 should be inclined rather to look for a clue in some effect of surface 

 tension or the capillary attraction of the water rising through the spaces 

 between the compacted mass of needles. 



