Accurate measurements by electrical means of temperatures at various levels in shelf ice, glaciers or ice caps are of themselves insufficient; they must be accompanied by data on densities at the same levels and on incoming and outgoing radiation, as well as other things. Depths of measurements must be known accurately, and changes in them due to changes in surface level must be recorded. Levels chosen should be , 1, 2, 4, 8, 16, 32 and 64 m., and readings (if repeated to ensure accuracy) need be taken only once daily down to 4 m. and weekly or so at lower levels. Measurements should be free of external influences such as heat from buildings or deep drifts. Only when all other necessary data are available and all precautions observed will it be possible to explain the thermal structure of the ice sheet as shown by the sub-surface measurements of temperature.

Snow and ice from their mutations and from the part they play in nature as precipitation, agents of erosion and modifiers of climate come within the range severally of physics, meteorology, geology, physical geography, oceanography and climatology. The behaviour of ice crystals in glaciers has close connection with crystallography and metallurgy.

It may be of interest at this stage to recall the papers and discussions at meetings since the commencement of the activities of the Association for the Study of Snow and Ice until the war put a temporary end to them.

Ice streams composing a compound valley glacier may be juxtaposed, inset or superimposed, and Demorest’s concepts of ice flowage provide a plausible explanation for these relations. If valley floors are accordant, juxtaposition results because the waxing force of gravity and obstructed gravity flow in the tributary glacier overcomes the waning force of obstructed extrusion flow from the trunk glacier enabling the tributary to force its way into a juxtaposed position with the trunk glacier in the main valley. If discordance in valley floors is less than the thickness of the trunk glacier, an insert position develops by the same mechanism. If the discordance is greater than the thickness of the trunk glacier, superimposition results. Locally, superimposed or inset relations may be established by exceptionally rapid advances, even with accordant valley floors, but any superimposed ice stream shortly becomes inset by sinking into the underlying ice, at least in temperate glaciers.