On the Viscosity of Ice. 



493 



The ice specimen was formed by freezing water in a cylindrical 

 iron mould, with a conical expansion at one end. To obtain ice as 

 free as possible from included air, I in some cases (but not in all) 

 boiled the water and then froze it. Afterwards it was melted in the 

 mould, boiled, and then allowed again to freeze. In this way nearly 

 all the air was expelled, only a small core of minute bubbles up the 

 axis of the cylinder remaining. In Experiment No. 1, however, these 

 precautions were not taken, and thus the cylinder of ice had linear 

 bubbles of air, radiating in horizontal straight lines from the axis of 

 the cylinder. These were due to expulsion of the air from the water, 

 in which it was dissolved, as the latter froze in concentric cylindrical 

 shells from outside inwards. 



After the ice cylinder had been freed from contact with the mould, 

 by bathing the latter in warm water, it was passed through the 

 conical iron collar (C in figure) which the conical expansion of the 

 ice fitted, and which had been screwed to the upper part of a wooden 

 frame. The other collar (B) was then attached to a moveable plate 

 in the wooden frame, which was capable of motion vertically and 

 horizontally, and, by adjusting screws, both the frame and the collar 

 B could be made accurately horizontal. By a small plummet the ice 

 cylinder w r as made to hang vertically, by placing small pieces of cork 

 between it and the edge of the opening in B. Water was then run 

 into B, and, when frozen, a cylinder of ice was obtained, held above 

 and below in two conical collars. By the above-mentioned adjust- 

 ments it was ensured that the cylinder of ice was perpendicular to 

 the surfaces of the two collars B and C. 



As the weight of the two iron collars was over 4*5 kilos., much 

 care was needed to prevent fracture of the ice, which, unless handled 

 tenderly, broke with the least jar, on being inserted in the machine. 



The lower surface of the upper, and the upper surface of the lower 

 collar had been planed true, and brass rings fitted on them. These 

 rings were also made plane surfaces. They extended so far from the 

 ice specimen that it was possible to use callipers between the surfaces 

 of these brass rings, without interference with the larger ends of the 

 conical collars. The ring on B was provided with four small holes, 

 90° apart on the circumference, and by means of a small plummet 

 the points on the ring of C (which ring was graduated roughly) that 

 were vertically under the four points on the ring above, were deter- 

 mined. By noting these graduations it was rendered certain that the 

 measures would always be taken between the same points in the 

 upper and under rings. Since the ice specimen was inserted in the 

 apparatus with its axis vertical, the lines measured between the 

 points thus determined on the two rings, were, in each case, parallel 

 to the axis of the cylinder. The callipers used measured to the one- 

 fiftieth of a millimetre. 



