the clay-soiled line representing the day tune during ^* *^\«* 

 occur This appeared to be proved by removing a portion, leaving the bank 

 xposed, and by finding the exposed space next day covered by a fresh layer of 

 ce a little over an inch in thinness, and the surrounding ice *«*«jed£ 

 nearly the same amount, and presenting a clayey line at near y the san>« 

 Lm the bank. It also proved what was perhaps sufficiently plain before, that 

 the ice grew from the bank. 



But why does it assume this peculiar structure? In the first place, on 

 removing any of the ice it is seen that the bank is wet, so ^tly Jhjwto 

 STt it ecnLs some attention to detect that there is «■ f^*»% ™ 

 its surface ; and, secondly, -that the water is supplied by the bank £ el £* 

 it does not flow over it from the top, but flows out of it in a multitude f tiny 

 pLlets. This trickling water will not freeze at the bottom of the air columns 

 S ah°isavery bad conductor of heat, but where the base of ^each ice column 

 uches the bank, the ice being a better conductor, will more effectua^y abs a 

 the heat from the trickling water at the same time that it offers it .suitable 

 mlrial on which to crystaUise, and the base of each column wdl ttaa .grow 

 by addition from beneath. This is strictly parallel to the case of the ribs of 

 ice that form beneath the continuous surface layer at the margin of ponds and 

 Xers when the water is falling. The ice that is left by the water of course 

 ceases to increase in thickness, but as it conducts the heat from the ice m 

 contact with the water, the ice thickens more rapidly at he line f™*™*' 

 the water than elsewhere, and when the waterfalls stall lower tins thicken d 

 portion often remains in contact with it when the water has .receded ^ from .the 

 L at a further distance from the bauk, and if the water fall gradually enough 

 afterwards freezing may continue along this line until a very thick : „b . « 

 thicker it may be, than the solid ice, as the ice on either side of it hav mg no 

 water o freeze conducts the heat from the rib of ice instead. These ribs are 

 Iftn very numerous and obvious in mere puddles and ditches, ut the ice 

 being ml rigidly supported across a small area "^* "^^ £ 

 question of their formation, the capillary attraction between the water and ice 

 being an important element in determining their position. 



It may be objected here that the ice either does, or with the first slight 

 addition will touch the clay bank so closely that there can be no more room 

 for a further layer. The force manifested during freezing is, we know, immense, 

 and hea^ as tL whole sheet of ice is, this force is doubtless sufficient to raise 

 the whole mass, though it is one solid piece, and apparently as immoveable as 

 amass of iron. This consideration led me to expect to find the labyrinthine 

 J Li the ice imprinted on the wet clay of the bank, but nothing of the sort 

 is to be seen. There is however, I think, a fallacy involved in expecting it. 

 Thou^ I whole force required in raising the mass of ice, say one inch rom 

 Le bank, is very great, the freezing at the base of each ice column has only to 

 raige tha particular column, and at the rate say of only one inch in twelve 



