20 



ROYAL SOCIETY OF CANADA 



Applyiiif^ this expression to the observations on Feb. 13th and 15th 

 (supra), there is given : 



The agreement of AT and log. A for the two days is good, wlien it is 

 considered how roughly the distance S vvas measured. The reading at 

 5 feet on Feb. l.^th indicates, as would be expected, too low a tem- 

 perature. Assuming, for this depth, log. A and ^from the depth of 8" 

 and 2 feet, log. S comes 1*894 instead of 1-778. It must be remembered 

 again, however, that this is only a rough genei-ality. 



If it may be assumed that the depth to which the sun's rays would 

 be no longer indicated l»y the thermometer would be Avhen (t—t^) = 0, 



then 



log. S = log. A, 

 and hence on 



Feb. 13th S = U feet, 



" 15th 5 = 15 feet, 

 a very good agreement, considering the extent of the extrapolation. 



Of course, this must not be taken as the total depth of penetration, 

 for had the thermometer bulb been black it would have detected the 

 rays to a much greater depth, but it may be considered the least limit. 

 Conversely it was shown by observations at night that under a clear skj'^ 

 the thei-mometer bulb at a depth of about one foot showed the eti'ect of 

 radiation' from the water. The temj^erature indicated b}' the bulb fell, 

 when left undisturbed, nearly -001° C. in 15 minutes. 



Summing up the evidence in favour of radiation as the cause of 

 anchor ice foi-mation, there is : 



a. Effect measured on the bulb of the thermometer. 



b. Growth of anchor ice much more readily on dark objects. 



c. Ab.sence of anchor ice under a cover, such as surface ice or a 

 bridge, which could act as a check to radiation. 



d. Immediate disintegration of anchor ice under a bright sun. 



e. Non-formation on cloudy nights. 



Sec. 111. U.N THE Fkee/ing Point of Watek. 



It had been hoped that before this experiments could have been made 



to determine the dependence of the absolute temperature of a freezing- 



• L. c, p. 25. " 



