494 



Dr. J. F. Main. 



I used the rings because I have not been able to devise means, as 

 yet, for so firmly fixing objects in the ice that there would be no 

 danger of displacement by their own weight, or during the measure- 

 ment by callipers. The results obtained are therefore affected by 

 errors, if any, which may be due to action at the collars, such as 

 slipping through, owing to distortion of the conical enlargements. 

 This effect is but slight at temperatures below the freezing point ; 

 though near and above freezing point the collars, when the ice is 

 under tensile stress, will not hold it, but, in the course of a few hours, 

 by distortion of its conical enlargements, it slips through them. At 

 lower temperatures, to act as' a check on the measurements taken 

 between the rings on the collars, and to determine if any appreciable 

 effect was due at those temperatures to slipping through the collars, 

 I gummed two pieces of paper on the ice specimen near the top and 

 bottom. On each of these a small pencil mark was made, and by 

 means of a rule, T got a measure of the distance between the points. 

 By repeating this measure on a subsequent day, a rough value was 

 obtained of the extension between the marked points. The distance 

 could be estimated to about the quarter of a millimetre, and showed 

 that, with low temperatures, nearly all the extension observed was due 

 to the stretching of the piece, and not to a shearing action of the ice 

 in the collars. 



In the tabular results of three experiments which follow, the first 

 column gives the date ; the second the hour, reckoned from Berne 

 mean midnight, at which the observations were taken ; the third 

 column gives the temperature in degrees centigrade, read by an 

 attached Kew-verified thermometer, graduated to tenths of a degree; 

 the fourth column gives in millimetres the mean of the distances of 

 the points on the upper from those on the lower ring. These dis- 

 tances, four in number, were taken at opposite extremities of two 

 perpendicular diameters of the rings. The next column furnishes the 

 mean extension in millimetres in the interval that has elapsed since 

 the last observation. The sixth column gives, in hundredths of a 

 millimetre, the hourly mean extension deduced from the preceding 

 column, and the interval which has elapsed between the two observa- 

 tions. The two next columns give, in kilogrammes per square 

 centimetre, the mean and the maximum stress respectively. By the 

 rapid evaporation from the ice cylinder, even at low temperatures, 

 the amount of this stress increased from day to day, with the same 

 load at the further end of the compound lever. Since the evaporation 

 was different at different parts of the ice cylinder, owing probably to 

 variations in texture, and still more to the effect of the proximity of 

 the collars to the ice above and below, which protected the ice near 

 them from evaporating so quickly as in other parts, the diameter of 

 the cylinder at different heights varied. Thus the mean stress, as 



