Strong evidence, available from the experiments of Nakaya and his co-workers and bolstered by an analogue experiment, supports the argument that type of snow-crystal growth is determined principally by the excess of the ambient vapor density over that at equilibrium with the ice crystal at its own temperature. Changes in crystal habit occur when this vapor-density excess is sufficient to overcome the inhibitions, first to edge growth and secondly to corner growth, that must exist if the edges and corners have higher surface vapor-densities than the flat faces of the crystal. Equilibrium vapor density for liquid water exceeds that for ice by the order of 10 per cent; slight supersaturation, such as can exist in water cloud, will increase this excess considerably. Ventilation has been neglected in analyzing the Nakaya experiments. Its effect was probably slight, since the ventilation was due to natural convection. The effect should be much greater in the atmosphere, where the ice crystal is falling... Abstract Strong evidence, available from the experiments of Nakaya and his co-workers and bolstered by an analogue experiment, supports the argument that type of snow-crystal growth is determined principally by the excess of the ambient vapor density over that at equilibrium with the ice crystal at its own temperature. Changes in crystal habit occur when this vapor-density excess is sufficient to overcome the inhibitions, first to edge growth and secondly to corner growth, that must exist if the edges and corners have higher surface vapor-densities than the flat faces of the crystal. Equilibrium vapor density for liquid water exceeds that for ice by the order of 10 per cent; slight supersaturation, such as can exist in water cloud, will increase this excess considerably. Ventilation has been neglected in analyzing the Nakaya experiments. Its effect was probably slight, since the ventilation was due to natural convection. The effect should be much greater in the atmosphere, where the ice crystal is falling...