On the Effect of Cold upon the Strength of Iron. 119 
Four bars were then suspended for six hours in the outer air 
in a temperature of — 40° Fah. There was some wind, so it was 
found convenient to test them in the shelter of a snow-house, tem- 
perature — 26, They were carried in one by one on loops of thread, 
and each tested within a minute, with the following result :— 
No. 11 bore 0 broke with 33°3 centimetres fall. 
2 
+B 99 th) th} 
ede 06-5 Care 
These figures appeared to indicate a loss of strength, and were 
the reverse of what I had expected; butit occurred to me that the 
bars when I tested them were undergoing a change of temperature, 
and the power of unequal temperature to produce brittleness is un- 
disputed. Thesame objection applied to experiments made by others, 
since freezing mixtures used to produce the cold are rarely, if ever, 
of the same temperature throughout,andin impact experiments the 
metal had to be removed from the oxygen into a warmer medium 
before it was submitted to the blow. In neglecting to test the 
bars in the open air I had stupidly discarded the one special ad- 
vantage offered to an arctic experimenter. This was the more 
annoying since the number of bars was unavoidably limited ; but 
I decided to test the remainder under conditions that would ex- 
clude any risk of unequal temperature. 
I accordingly waited till the 25rd of January for favourable 
weather, The bars, with loops attached to them to admit of ready 
manipulation, and the guillotine were then exposed for twelve 
hours to a steady temperature of from — 51° to — 54° Fah. At 
the time of the experiment the temperature was — 58° Fah., and 
the utmost care was taken to avoid any transference of warmth 
to the bars either from the lantern or from my hands. 
” ” ” ” ” 29 ” 30 ” 
” ” Hye LO wissen ” ” 
” ” » » 9 286 ” 30 ” 
” ” ska On) n 285 ” 
” ” » 18 , 27° ” ” 
” ” » 9 9 285 ” 295 ” 
The experiments are not numerous enough to warrant gene- 
ralization, and it would be difficult to condense their results into 
any fair statement of averages. But the figures admit of reduc- 
