picture of ice cubes

non-shattering ice cubes

testing the hypotheses

test 1. ultra towel nest

two trays stacked, bottom tray with towel-nest

Bad cubes come from the bottom tray of the stack. If the Top-Down Hypothesis is correct, these bad cubes could be prevented by causing the bottom tray to freeze top-down. Wrapping the bottom of the bottom tray in a towel (but leaving the top exposed) might accomplish this: the heat loss through the sides and bottom will be slowed.

Typically, the lower shelf of the freezer makes better cubes than the upper shelf, perhaps because the plastic surface of the lower shelf does more to insulate the bottom of the bottom tray than the wires of the upper shelf. The upper shelf, then, is the real test case.

I rubber-banded a towel to the bottom and sides of the bottom tray of a stack. That should keep those little cube-butts warm. The results:

dude freezer water w-temp time f-temp rate shelf stack good
Karl WP tap cool 8:45 3 fu upper top 100%
Karl WP tap cool 8:45 3 fu towel-nest upper bottom 100%

† No significant rate modifier. Of course, this tray was on top of the towel-nest tray, which might make a difference.

Success! The first 100% good cube stack.

This test seems to support the Top-Down Hypothesis, for the reasons stated above.

This test seems to be evidence against the Fast-Freezing Hypothesis, since if anything the bottom tray must, overall, be freezing more slowly than usual.

test 2. ultra towel lid

two trays stacked, top tray with towel-lid

If a towel-nest can lead to good cubes, can a towel-lid create bad cubes? The Top-Down Hypothesis suggests that it can.

I rubber-banded some cling-wrap to the top (and only the top) of a single tray, and then folded a nice cozy towel over the wrap. As warming as a bunch of little knitted cube hats.

Most of these cubes were destined to a life of crime:

dude freezer water w-temp time f-temp rate shelf stack good
Karl WP tap cool 38:30 3 fu towel-lid upper - 20%

More evidence supporting the Top-Down Hypothesis!

This evidence would also support the Fast-Freezing Hypothesis.

further tests

Despite what you’re probably thinking, my patience for insignificant household experiments has its limit. But if I get inspired, I might perform a few more tests:

  1. Try to disprove the Fast-Freezing Hypothesis by producing good cubes freezing very slowly.

  2. Test the strength of cubes formed quickly and slowly.

  3. Make good cubes and bad cubes, and measure (and chart) their temperature gradients (probably by observing opaque core locations).

  4. Clean my trays using vinegar, like Heloise suggests, and see if that makes any difference.


According to the Top-Down Hypothesis, bad cubes form in the bottom tray of a stack because the tray is warmed from above and cooled from below. The evidence supports this hypothesis.

With that in mind, here are a few ideas for avoiding bad cubes.

two trays, one of them with a towel-nest

Obsessive solution: Rubber-band a towel around the bottom tray of any stack. This is silly.

a hand picking an individual ice cube out of the tray

Procedural solution: Make only one tray at a time. If you’re a non-bin cube-picker, this might come naturally: Pick clean the top tray, then fill it and put it on the bottom of the stack. (The new tray might do even better on top, but our experiments seem to show that it does fine on the bottom, as long as the new top tray is already frozen.) Never mind that the way of the picker is the way of the heathen.

For those people and their damned bins, this solution is unsatisfying. We don’t like filling the bin, whatever you may think, and we don’t want to do it twice as often.

a sketch for a tray with a thick plastic bottom

Entrepreneurial solution: Manufacture a specially-designed tray! With thicker plastic walls! Especially for the bottom tray! Then send me lots of $$$$$ when it sells like hotcakes!