When the ice had lain a short time, cracks appeared on the surface exposed to the sun, and spread like a network from the edges towards the centre of the surface. At first there was no regularity in the connection of these lines, and the several meshes were of very different sizes. After a time, the larger meshes split up into smaller, and the system of network was found to penetrate below the surface, the cracks deepening into furrows, which descended perpendicularly from the surface, and divided the ice into long thin rhomboidal pillars. The surface-end of some of these pillars was strongly marked with right lines parallel to one of the sides of the mesh, and it was found that there was a tendency in the ice to split down planes through these lines and parallel to the corresponding side-plane. Parallel to the original surface of the mass of ice, the pillars broke off evenly. The side-planes had a rounded, wrinkled appearance; and their mutual inclinations--as far as could be determined--were from 105° to 115°, and from 66° to 75°. When these ice-pillars were examined by means of polarised light, they were found to possess a feeble double-refracting power. The writer of the article in Poggendorff suggests a question which he was not sure how to answer:--Is this appearance in correspondence with the original formation of the ice, or does it only appear under slow thaw? It is worthy of remark, that from the 1st to the 11th of February the thermometer was never higher than 22°·8 F., and during that time fell as low as 21° below zero, i.e. 43° below the freezing point. Professor Tyndall has informed me that in the winters of 1849, 1850, 1851, he found the banks of a river in Germany loaded with massive layers of drift-ice, in a state of thaw, and was struck by the fact that every layer displayed the prismatic structure described above, the axes of the prisms being at right angles to the surfaces of freezing. It may be, he adds, that this structure is in the first place determined by the act of freezing, but it does not develop itself until the ice thaws.