Does hot water reproducibly freeze before cold water
In plain words
The claim that hot water freezes faster than cold water is old and widely repeated, but careful experiments disagree on whether it is real. The question is whether, under controlled conditions, it happens reliably and for what reason.
Precise statement
Identical samples of water (specified purity, dissolved gas content, volume, container) placed in an environment at $T_{\mathrm{env}} < 273\ \mathrm{K}$ from initial temperatures $T_h > T_c > 273\ \mathrm{K}$. Does the $T_h$ sample reach a defined endpoint (onset of nucleation or complete solidification) earlier than the $T_c$ sample, with statistical significance over many runs, after controlling evaporation, convection, container contact and the stochastic supercooling temperature? An answer is yes with the controlling variable identified, or no with bounds on any effect size.
What would settle it
A preregistered, blinded, multi-laboratory experiment with hundreds of runs per condition measuring freezing-time distributions under controlled nucleation.
Status in the literature
Unverified note
Burridge and Linden (2016) found no faster cooling to 273 K; 2025 preprints attribute observed effects to stochastic ice nucleation (Klimov and Finkelstein, arXiv:2508.05607) or report a repeatable effect when supercooling is suppressed by an ice-nucleating environment (Brownridge, Zinet, Sotta and Ganachaud, arXiv:2503.18820).