The Mpemba effect — the observation that hot water can freeze faster than cold water under certain conditions — has been debated for centuries. What we are observing is not the Mpemba effect. What we are observing is water freezing from the inside out.

The first documented case occurred during routine quality testing of a municipal water batch destined for the Laceworks residential supply. The batch — 200 liters, standard treatment, standard chemistry — was placed in cold storage for crystallization analysis. When the technician checked the sample sixteen hours later, the exterior was liquid at 2 degrees Celsius. The interior was a solid block of ice at -4 degrees. This is not how freezing works. Water freezes from the surface inward as heat dissipates from the exterior. This sample had frozen at its core while its shell remained liquid. The ice was structurally normal. The liquid was chemically normal. There was no barrier, no insulation, no membrane separating the two phases.

Since that first observation, we have documented 340 inversion events across 12,000 batches tested over two years. Approximately 2.8% of batches invert. The batches are chemically identical — same source, same treatment, same storage conditions. We have tested for trace contaminants, isotope ratios, dissolved gas content, mineral variation, microbial presence, and electromagnetic exposure history. Every variable we can measure is identical between inverting and non-inverting batches. No predictive model achieves accuracy better than chance.

The inversions are becoming more frequent. The 2.8% rate is an average. In the first six months, the rate was 1.1%. In the most recent quarter, it is 4.6%. The trend is linear and shows no sign of plateauing. Lab Director Ingrid Mutombo-Svensson has requested that this data not be included in the public water quality report. I am including it in this internal report because someone should be paying attention to the fact that our water is forgetting how to freeze.