102 LETTERS TO MARCO xvi 



formed on the bottom of the river and which 

 rises to the top. They advance in proof of 

 this theory that the ice which floats down the 

 river at the beginning of a frost always has 

 weeds and their roots sticking in it, torn up, 

 as they think, from the river-bed. This ice 

 with the weeds in it is no doubt that which 

 gets frozen first at the edges of the river and 

 on the flooded sides, and which, if the river 

 rises afterwards at all, floats up, bringing with 

 it the tufts of weeds and their roots ; in some 

 sense it might be called ground-ice, but it 

 does not come from the deep bottom of the 

 river, as I have often heard stated by farmers 

 and others. This ice is always seen about 

 the second day of the frost, after which, as 

 the river falls a little, no ice is seen for a day 

 or two, the stream running very strongly in 

 its natural bed, not allowing the formation of 

 ice ; but after that, if the frost continues, large 

 sheets of ice begin to form in the eddies, 

 which gradually extend along the banks, 

 leaving but a very small channel in the 



