Last night it rained, but the skies cleared before morning, letting the air temperature drop to about 2 degrees C. The weather, it seemed, would be perfect for lacy white frost patterns on cars. But it didn't quite turn out that way. A little bit before sunrise, I went out on my rounds. I took a few pictures of ice on parked cars and then headed for the rice fields. Neither tub had ice, so I walked to the abandoned car on the other side of the roadway. Nothing of note there either, but in those few minutes a blade of ice had started across one tub. Soon there were several. With a stick, I went under one blade, lifted it up, and noticed the thin serrated fingers that had been growing into the water, hidden under the surface. The disturbance of the water surface caused the growing mesh of ice lines to drift, unattached to the tub walls. I don't see this in puddles. In the puddles, the ice is always anchored to the shore, presumably because the ice starts from the shore, where there are more things to nucleate from.

In the tub I saw a curvy piece and fished it out. Laid out on the rusted lid of a nearby barrel, the ice looked strange, like a fossil of some prehistoric creature.

The critter is about 3-4 inches head to tail and paper thin. As the crystal grew from liquid water, called 'melt' by material scientists, it is said to be "melt-grown". I call such crystals "puddle crystals". Anyway, soon ice was starting on the other tub. Out in the middle of an open patch of water, I saw a small hexagon. With a stick I tried lifting it out.  I  found that the central hexagon had six large branches that had been hiding just under the surface. But the crystal slipped off my stick and cut under the water's surface - never to be seen again. Needing something better than a stick, I found a piece of a plastic lid that had been left in a nearby ditch and tore off a piece. Then next time I saw a small, isolated hexagon, I lifted it out and put it on the barrel lid.  See the photo below.

Though the central hexagon that I could see on the water was less than a half inch across, the full crystal was about one and a half inches across. Curious how the water would cover up all six branches symmetrically...  Anyway, despite the large width of our 'puddle star', it too is paper thin.  And being so flat and featureless on both sides, it evenly reflects the bluish skylight. The serrated pattern on the branches make this crystal a dendrite, like the dendritic snow crystal. But a puddle crystal grows much faster than a snow crystal because the melt (i.e., liquid water) is more densely packed with water molecules than the air from which snow grows. Also, when a snow crystal branch grows, it removes water vapor from the nearby air. This depletion of vapor is most severe near the base of the branches, and thus snow-crystal branches hardly widen - they mainly grow at the tip and at the tips of the side-shoots (called side-branches). But the melt is never depleted of water molecules, so the branches of puddle crystals keep widening until they merge. Another difference between puddle crystals and snow crystals is that puddle crystals viewed broadwise have a curving outline. The snow crystal usually has a polygonal outline, that is, made from straight lines.

- JN