There was a poster* at the conference yesterday that discussed the history of modeling experiments in geology. It had a couple of pictures of the classic cornstarch experiment for modelling the development of polygonal structures. This article provides a good example.

Order and disorder in columnar joints (pdf)

Abstract. – Columnar joints are three-dimensional fracture networks that form in cooling basalt and several other media. The network organizes itself into ordered, mostly hexagonal columns. The same pattern can be observed on a smaller scale in desiccating starch. We show how surface boundary conditions in the desiccation of starch effect the formation of columnar joints. Under constant drying power conditions, we find a power law dependence of columnar cross-sectional area with depth, while under constant drying rate conditions this coarsening is eventually halted. Discontinuous transitions in pattern scale can be observed under constant external conditions, which may prompt a reinterpretation of similar transitions found in basalt.
Starch patterns are statistically similar to those found in basalt, suggesting that mature patterns contain inherent residual disorder, but are statistically scale invariant.

So it is basically a desiccation experiment. The resulting cracks start out as a large network, then the interiors begin to develop a smaller scale pattern. Interestingly enough, the first stage is more orthagonal while the second stage is clearly polygonal.

Try it yourself! Next time I'm at the market, I'm gonna get a couple of pounds and play around.

I suggest some other variations. Add a very fine-grained silica sand. Thicker layers. Different layers. Try repeated wetting. Freeze the sample. Evaporate the water under a vacuum.

What will all this tell us? Not much. But it's fun!

*In this case, a poster session involves hanging your research on the wall as a poster, and then you stand around for four hours and talk to anyone who is interested in your work. It's alot of fun, and I prefer it to listen to oral presentations. I've presented my work a couple of time like this and always find it rewarding.