6G 



hours, and the direction of the motion is not vertical, but only little removed 

 from the horizontal, — when the ice is formed on the ground the thrust is of 

 course upward,— so that in reality the actual pressure at any one point is 

 always very slight. The growth of the icicles of the separately column form is 

 explicable in precisely the same manner. And it seems easy to understand how 

 the ice is formed, when once a beginning is made, but how does the formation 

 commence ? I believe that at first a tolerably uniform but very thin coating of 

 ice forms over the wet surface of the bank. When water, say from a spring, 

 runs over the surface of the ground in a sharp frost, the first-formed layer of 

 ice adheres to the ground, the water flows over it and the ice thickens from it s 

 upper surface. This does not happen in the case before us, because, as I have 

 already said, the water does not flow over the bank from one source alone, but 

 appears in minute quantity over its whole surface. The water, therefore, con- 

 tinues to trickle beneath the ice, and air is drawn in beneath it, and as the ice 

 thickens it is imprisoned as rounded or irregular bubbles, and the intermediate 

 spaces are the tops of our ice columns. But ought we not to find the columnar 

 ice covered by this thin ice pellicle ? As a matter of fact we do find that the 

 openings of the air tubes are narrowed by a thin edge projecting over them, 

 which is part of the pellicle. The rest being much thinner may very easily have 

 disappeared by evaporation. Though I incline to this explanation of tliG first 

 formation of the ice, it follows from the condition of the water supply that the 

 air spaces may take their origin from being opposite the actual minute springlets 

 and therefore freezing does not occur at these points. 



I have seen it suggested that the columns of ice each represents one of 

 the springlets, the water freezing as it appears and the force of the issuing water 

 raising the ice as it is formed. This can hardly be the case, as the water trickles 

 over the outer spaces in a tolerably uniform layer, it would be most unlikely 

 to freeze at those points where it issues from the ground, and the pressure of the 

 water would be quite insufficient to raise the ice, unless the water escaping from 

 each opening were confined by the ice being frozen to the ground around it, 

 which evidently does not occur. 



The surface of the ice is often covered with bits of the soil from the earth 

 beneath, the particles having been frozen in with the first pellicle of ice, and 

 in that form which I have compared to a bundle of icicles on account of each 

 column being separate, each column often has a bit of dirt on the apex, as though 

 .it of earth projecting from the bank had been first frozen, giving the 

 starting point, which after the formation of the ice seems easily understood. 



In some cases I have found the ice structure behind a layer of frozen earth 

 of nearly half an inch in thickness, and usually rather dry and crumbling. The 

 drier earth on the surface soon freezes and takes the place of the theoretical 

 pellicle which appeared necessary when the surface of the ice is nearly free from 

 earth. The columnar ice begins to form at the surface of that layer of . 

 which is wet enough to afford a continuous supply of water. 



