I assume the bricks are still there? Ice crystals need a nucleation site to form around - a speck of dirt or protrubrance from a rough surface. The surface of your bricks would be ideal.My hypothesis is that ice crystals form on the rough surfaces of your bricks, then detatch and float to the surface. Here they stick to the underside of the ice layer already formed... I imagine it would look like snowfall, but going up instead of down. The images are aggregated ice crystals stuck to the bottom of the ice sheet. Of course, all the crystals from the vertical faces of the bricks would stay roughly in plane as they rose, explaining the dense white outlines. If you get bored, you could work out the relative gradients of the bricks by measuring the intensity of each face in the image.Christopher Binny

Re the previous comment: Hmm..ice crystals ? surely not - the pond would have to be freezing from the bottom or supercooled, both pretty unlikely.I think that the image is composed of small bubbles generated on the surface of the bricks, perhaps by microorganisms or air held in the structure of the brick. These floated upwards in the still water and arrived at the underside of the ice and were frozen into position. They didn't coalesce because the surface ice was still progressively forming. This idea fits with a detailed examination of the image. The lighter outlines are because there is vertical brick surface to generate bubbles which arrive in the same place under the ice. The images vary in 'brightness' depending on whether they are partially buried in mud which reduces the surface area available to generate bubbles. Where one brick is lying on top of another there is no image of the edge of the lower brick that is underneath the upper brick.I assume that the two smaller images at the top are from smaller bricks - if they are the same size as the other bricks, it makes things very tricky to explain.Mike Adams

The following answer has been selected and edited by New Scientist staffPond water contains a certain amount of dissolved gas, including oxygen. Because of the physical properties of water, the colder it is the less gas per unit volume it can hold. Water is at its densest at a temperature of about 4 °C.As the water temperature in the pond drops, cooled by the colder air above, the surface water sinks to the bottom by convection. Once the whole body is at 4 °C this convection stops because further cooling makes the surface water less dense. The surface starts to freeze and the coldest water begins to release its dissolved gas. Some of this would bubble upwards, but much more would diffuse down, still remaining in solution, until eventually the water surrounding the bricks becomes supersaturated.The rough surface of the bricks, particularly around the edges and corners, provides nucleation sites for dissolved gases. Gas molecules collect preferentially around the edges of the bricks, eventually producing bubbles. As these reach a critical size they break away and float straight upwards in the still water. Because there is a layer of ice on the surface, the bubbles become trapped and frozen into it. As the ice layer thickens and bubbles continue to rise from the brick, the 3D shape develops. The rate of bubbling was probably very slow, as was the rate of freezing, so the very detailed effect was able to form.David Jackson, Liverpool, UK

Could there be magnification of the effect if the bricks do not lose heat quite as quickly as the water?

What an amazing effect! Next year, toss in a crucifix or a bust of Princess Di and you could make a fortune on e-bay.