128 ROYAL SOCIETY OF CANADA 
The first theory that suggests itself on examining the facts is 
that this phenomenon has its origin in some structural alteration in 
the crystal itself. However, there is a simpler explanation, which 
appears to account for the majority of the observed facts. 
The specimens of ice used in these determinations are divisible 
into two classes,— natural and artificial. Artificial ice may be 
obtained in many ways, but for the purposes of the present paper it 
will be sufficient to deal with three main classes. If a vessel be filled 
with water and immersed in a mixture whose temperature is below 
the freezing point, ice will form around the inside walls of the vessel— 
let this be called the interior formation. If the vessel be filled with 
the freezing mixture and placed in water, ice will form around the 
outside of the vessel — this will be referred to as the eaterior formation. 
Lastly, if a long tube be filled with water, and the freezing mixture 
applied to the end, by gradually moving this forward as the column 
of water freezes, a column of ice will be formed without any tendency 
to fracture the walls of the tube. 
The ice formed inside Plücker and Geissler’s dilatometer affords 
an example of the interior formation; the ice-mantle in a Bunsen ice 
calorimeter is an exterior formation; while an instance of the column 
formation, which was first devised by Boyle, is seen in the overflow 
dilatometer used by Bunsen in his density determinations. 

In fig. 1 is seen a section of a vessel in which the first formation 
is going on, the freezing mixture being applied from the outside. 
Before the freezing commences, the outside surface of the vessel 
is approximately at the temperature of the freezing mixture, while 
the inside surface is very nearly at the temperature of the contained 
water, these being the conditions necessary to allow the heat to be 
conducted through the walls of the vessel. Since the average tem- 
perature of the vessel is below zero, it follows that it is slightly smaller 
than it would be if it were at a uniform temperature of zero. Thus 
the surface on which the first layer of ice forms will tend to enlarge 
when it assumes the zero temperature. After the first layer of ice 
is formed, its average temperature will also fall below zero, since the 
heat is conducted through the ice as well as through the walls of the 
