picture of ice cubes

non-shattering ice cubes


David, Jess, and I don’t really know anything about crystal formation, but here’s what we do know:

Thoughts on growing a perfect crystal:

Thoughts on growing a strong crystal:

h0t nak3d ice pix!!!!

Before we go any further, let’s have some more gratuitous pictures of ice cubes. Have a gander at these beauties.

blurry view of inside freezer

The first dim view of the world as seen by a baby cube emerging from its watery mother.

adult cube view of inside freezer

Later, when the maturing cube’s sight organs have shed their nascent film, it can see clearly as far as the freezer door.

a hand reaching inside a freezer

One day, the hand will reach into the freezer.

two cubes on a blue plate, one broken

Two cubes, one shattered by forces beyond its control, contemplate fleeting time and the sunset from the vantage of a blue plate.

picture of ice spikes (stalagmites) in a tray

An ice spike.

You can read more about ice spikes here or here. I grew one ice spike with tap water and one with filtered water.

Also notice how the tops of these cubes have protruded from the tray; apparently the top froze first, and then the expansion of the lower part of the cube (as it froze) raised the top part out of the tray.


Some of the ultimate variables affecting our crystal formation are these:

There are a number of proximate variables which we can more directly affect:

i heart proximate variables!

The idea is that the proximate variables can be expressed in terms of the ultimate variables: For instance, the type of tray might ultimately be changing the number and location of nucleation sites (because of scratches or material irregularities), the local temperature gradient (because of the heat conduction of the material of the tray and the extent to which it allows evaporation), and the overall freezing rate (because of similar factors).

After a few experiments, David, Jess, and I chose a set of variables which we felt allowed for a nice range of possibilities without consuming too much time or effort. The variables are described as they are used, on the next page. No doubt we would get more sophisticated hypotheses if we had experimented using more (or better) variables. But it’s good enough for rock-and-roll, and we got pretty good results using the variables we used.