of Water in Capillary Tubes. 107 



whole field appears dark, and no white line is to be seen down 

 ^he centre of the tube ; whereas, if the water be frozen, the double 

 refraction of the ice depolarizes the light, and the tube appears 

 as if it contained a white substance, which vanishes as the ice 

 thaws. In order to delay the thawing, the small tube should be 

 examined when inside a larger vessel containing a strong solu- 

 tion of common salt. 



The capillary tubes used in my experiments were prepared by 

 drawing out, when melted, the tubes used for alcohol thcrmo- 

 ineters ; and their internal diameters were ascertained by mea- 

 suring them at the broken ends by means of the microscope- 

 niicrometer, because the bore appeared about f of the actual 

 diameter when seen at the side through the curved glass. I 

 made experiments with those whose diameters varied down to 

 ^learly f u\j,jth of an inch ; but when so small as that, the depo- 

 larizing action of the ice is so weak, that it is impossible to he 

 guite sui'e whether they contain water or ice. Those, however, 

 from aOT^h to j^oth of an inch in internal diameter are in every 

 respect most satisfactory ; for there is no difficulty whatever i^ 

 deciding this point with polarized light in the manner just de- 

 scribed. "When water is contained in such tubes, we can readily 

 reduce the temperature to below —15° C. without using any 

 pare to keep them quiet ; and even when we shake them very 

 violently, it will not freeze. The temperature may even be re- 

 duced to —16° C. (3° F.) without taking any precaution to keep 

 it still; but at —17° C. the water freezes at once, even when 

 Vept perfectly quiet. It will thus be perceived that there is a 

 difference of about 11° C. (20° F.) between the temperature at 

 which water freezes in tubes of about ^th of an inch in diameter 

 and in those less than ^no*'^' which is so great a difference that 

 there is no chance of its being an error of observation. 



So far as I was able to ascertain, there was no very decided 

 difference in the temperature in the case of tubes varying from 

 j^ijth to y^oth of an inch in diameter, nor in the case of tubes 

 from :|;th to :f ^th ; but in a tube of about xnn^^ ^^^^ water froze 

 at about —13° C, but not at —11°, which is clearly a tempera- 

 ture intermediate between that required for those of -^^th. and 

 ^i(jth. It must not, however, be supposed that water will not 

 freeze in minute tubes at a temperature very little lower than 

 0° C, if the water be in contact with ice. Wben a tube of con- 

 siderable diameter in one part is ever so fine in another, the 

 contact of ice causes the water to freeze right down into the ca- 

 pillary part of the tube at a temperature very little lower than 

 0° C; and ice th:;-.vs as usual at O'' C. when in tubes in which 

 water will not freeze until thp tcuipcraturc has been reduced to 

 -16° C. 



