The position of the entrance of a cave in relation to the body of the cave is an important factor in permitting the cold air to permeate and remain in the cave. In all the caves or gullies I have examined myself, the main mass of ice is well below the level of the entrance, and even if the latter is sheltered against the wind, it is not sheltered against the cold air of winter. This is heavy, and by its own weight sinks well down to the bottom, freezing up in course of time all the moisture that may drip from the roof, or that may come into the cave in the shape of melted snow or cold winter rain. The summer air, which is warm and, therefore, light, can only enter the cave with great difficulty; and, as a rule, before it dislodges the winter air and destroys the ice, another winter's freeze reverses once more the conditions. These principles seem to hold of every known glacière. It is true, that at the Frauenmauer, the floor of the cavern rises somewhat from the entrance; but the highest point of the floor is still below the level of the top of the entrance, so that the cold air can flow over the highest point without difficulty. The same appears to be the case at the Posselthöhle; while at Amarnath in Kashmere, where the floor is said to rise to the back wall, the entrance is about as large as the area of the floor, so that the ice must also be below the level of the top of the entrance. The position or situation of the entrance is important. In almost all cases it has a northerly exposure, and is sheltered against entering winds. If these two conditions do not exist the ice supply surely suffers. Sometimes the entrance is more or less tortuous. In some cases it is protected by a fringe of trees. Still, there is no absolute rule about entrances. The Friedrichsteinerhöhle faces about due south, and at midday in summer, the sun shines all the way down to the ice floor, causing mists to form. In the Kolowratshöhle, the entrance is badly sheltered against the wind and this undoubtedly affects the supply in summer and causes more rapid melting there than in some other cases.[58] [58] See Part III.: Decorah, page 178. Freezing boulder taluses invariably have the ice near the surface, and probably it is never a dozen meters distant from the open air. These taluses are one of the strongest links in the chain of evidence proving the winter's cold theory. The snow and ice on the surface of the taluses and on the surface of the boulders in gullies melts away, while it still lingers underneath the boulders. It seems self-evident that the melting snow water has run to the lowest level and there congealed, and then remained because it was better sheltered than the ice outside.