SECT. 7] THE PHYSICS OF SEA-ICE 829 



A number of general reports have appeared on sea-ice, several of which are 

 listed among the references. The most famous is the account of Malmgren (1927) 

 of his work on the Maud expedition. His results were quoted extensively by 

 Sverdrup et al. (1942) in The Oceans. 



2. Mechanical Properties 



The properties of density, ultimate strengths, elastic parameters and plasti- 

 city may be considered to be dependent on the temperature of the ice, its 

 salinity, or brine content, and its crystal structure. These three variables are not, 

 of course, truly independent of each other, both the last two being dependent 

 on the freezing rate. An ice cover is an integrated record of the meteorological 

 and oceanographic history of at least the preceding part of the winter. 



The density of sea-ice is variable within wide limits. Malmgren (1927) and 

 Laktionov (1931) report specific gravities between 0.857 and 0.92, the low 

 values referring to old surface ice from which considerable drainage had 

 occurred. Pounder and Stalinski (1960) found an average of 0.943 for Arctic 

 sea-ice in Barrow Strait. 



A . Ultimate Strengths 



Because of their practical importance, the ultimate strengths of sea-ice in 

 tension, compression and shear have been measured by many observers under 

 a variety of conditions. In addition to the independent variables listed above, 

 ice is so anisotropic that the direction of stress application is very important. 

 Failure to record all these variables limits the value of most of the experimental 

 work in the literature. The many methods used to measure tensile strength 

 can be divided into two main classes : small sample tests and in situ beam tests. 

 The former class includes flexural tests on small, simply supported beams and 

 ring tensile tests in which a hollow cylinder is fractured by applying a compres- 

 sive load across a diametral plane. Butkovich (1958) gives a summary of these 

 methods. Small scale tests are relatively simple and quick to perform and thus 

 can be done on enough samples to apply statistical methods. This is an advan- 

 tage with a material so inhomogeneous as sea-ice. The samples can also be 

 brought to a uniform temperature before testing. 



In situ tests on cantilever beams and freely supported ones are laborious and 

 time consuming but the results are more directly applicable to practical con- 

 siderations on the bearing strength of the ice cover. These large scale tests 

 almost invariably yield smaller values for the tensile and shear strengths than 

 those obtained in small scale tests. The explanation usually offered is that 

 failure will normally start at some imperfection in the ice structure. These 

 imperfections are expected to be distributed statistically throughout the 

 volume, so that a larger sample is more likely to contain a weakness than a 

 smaller one. 



