Thomas H. Huxley once wrote the famous line:

The great tragedy of Science: the slaying of a beautiful hypothesis by an ugly fact.

Great man, catchy phrase, but perhaps a bit overdramatic. To me, the slaying of a “hypothesis” (i.e., pet theory or just idea, really) can be the beautiful thing. It means that one can do a lot of damage with just a simple observation. I like it, even when I'm shooting down my own theory. Here's an example:

Some time ago in my experiments, I saw the following ice growth sequence:


What you see there is an extremely small, thin ice crystal growing from the tip of a glass capillary into air. (Size-wise, the glass capillary is about 5 micro-meters in diameter, or about 1/10th the thickness of the hair on your head.) I saw it happen several times. As the crystal grew, it developed the prism facets that generally define the hexagonal crystal shape. Other people had seen such rounded growth before, generally within a few degrees of zero (C), though in all cases, the crystal had been extremely thin. You can also see this thin, rounded (non-facetted) form in some hoar-frost formations.

The mystery here is why the disk grows without the prism facets for awhile. I never saw this with thicker crystals, so I formed a little theory. The theory involves the source of the water molecules to the curved region of crystal: some come from the vapor in the air and some wander over from the flat, non-growing crystal faces on front and back. In the 1960s, people had tried to measure this “wandering distance”, but never determined a consistent value. My theory predicted that once the crystal thickness exceeded about twice this distance, the curved edge would transition to flat, giving rise to the hexagon.

But then I looked closely at this case:


That sequence shows the side and front view. The crystal doesn't discernibly thicken when the flat prism facets appear. Zing! That theory shot down.

On to the next pet theory. So maybe the key factor is the diameter of the disk: One might argue that the curvature of the crystal surface must be below a certain value, which means a large enough diameter is needed, for the flat facets to appear. Or, the size of the resulting prism faces must be larger than that needed to have several surface steps (which help ensure flatness).

But then I look at this case:


In that case, I see both smaller and larger prism faces forming at the same time (same thing can be seen in the previous image). So, I guess the curvature or size of the resulting face is not the main factor. Zing!

Some researchers had observed slight bending of the prism facet above -2.0 C in equilibrium. They postulated a “roughening temperature” at -2.0 C. This might explain the rounded disc edge, but wait! 1) This disc edge becomes facet as it grows, so it is not merely temperature, and 2) these crystals are below -2.0 C.
Zing again!

Well, there are always “impurities” to blame! Crystal growers are fond of blaming some trace, active chemical, or “impurity” for inexplicable results, so we could theorize that the above show the effect of surface impurities. As the crystal grows, the area over which the impurities distribute increases, thus diluting their concentration, and thus reducing their effect. Yes! This could explain these results.

But, wait, what about this:


The above shows two sequences (same capillary) in which some prism facets have formed, but some remain round. In the right-side case, one corner even rounds as it grows. Hmm, not a likely result of impurities. Zing!

So, I am down to one last theory. I do not yet have the data to shoot it down. And I'm not telling you, or I'd ruin the fun. We just need more data.

All in all, I think Huxley needs a little tweaking to serve my view:

The great beauty of science: the slaying of a pet idea by a simple observation.

-JN