March and April are when we find out what the winter did. Containers come in from farms, municipal yards and construction sites with a split in the bottle and a story attached, and for years we accepted them, granulated the bottles, rebottled the cages and thought no more about it.
Last spring Marisol started writing down the circumstances. Forty containers, forty short conversations with the people selling them. The pattern is tighter than we expected.
The numbers
| Factor | Count | Share |
|---|---|---|
| Fill level 20–80% at freeze | 33 | 83% |
| Fill level above 80% | 4 | 10% |
| Fill level below 20% | 3 | 8% |
| Stored outdoors, uncovered | 36 | 90% |
| Split at mid-height | 31 | 78% |
| Split at base radius | 9 | 22% |
| Contents: water or rainwater | 27 | 68% |
| Contents: dilute solution | 13 | 33% |
| Insulated or heated in any way | 2 | 5% |
Why partly full is worse than full
This is the finding that surprises people, and it reverses the intuition almost everybody arrives with. If freezing is the problem, surely less water means less problem?
No, because of where the ice forms and where it can go. A full container freezes inward from all surfaces roughly evenly, and the expanding ice migrates toward the only free volume available — upward, toward the fill opening, which is the designed relief path. It often bulges the top and survives.
A half-full container has an enormous horizontal surface exposed to cold air. It forms a thick slab across the top first, then continues freezing downward and outward. The slab is constrained laterally by the sidewalls and it has nowhere to go but through them. That is why 78% of our splits were at mid-height, right around where the liquid line had been.
How long a freeze it actually takes
More than most people expect. A 275-gallon container of water has serious thermal mass. In our own yard, with a logger in a test container, sustained ambient below 28 °F took between 36 and 60 hours to develop ice thick enough to threaten the wall, depending on wind exposure.
That is genuinely useful, because southeastern Pennsylvania rarely delivers a 60-hour hard freeze without warning. A drain-down on a Thursday afternoon when the forecast turns is almost always enough, and it costs nothing.
What the two survivors did
Two of the forty had some protection: one fitted jacket and one band heater without a jacket. Both still split, which is worth stating plainly rather than quietly omitting.
The jacket alone slows heat loss but adds no heat — it buys time through a cold snap, not through a week. The heater without a jacket fought a losing battle against an uninsulated 275-gallon surface area and, according to the owner, ran continuously for a fortnight before tripping. The combination is what works: a jacket makes a modest heater sufficient, and a heater makes the jacket more than a delay tactic.
The cheap answer, again
Drain it. For anything not needed until spring — site water, stormwater capture, irrigation supply, wash-down stock — draining below 10% with the valve open and the lid vented is free, reliable and cannot fail. Twenty-seven of our forty were holding plain water that nobody needed in January.
The containers were not the problem. The calendar was.