I've always disliked the explanations I've seen for the Mpemba effect. They seem to be based on poorly defined terms, odd assumptions, and uncontrolled variables. The (poorly) linked 2010 paper by Brownridge is the first description that I'm inclined to trust: https://www.binghamton.edu/physics/docs/Preprint%20and%20Sup...
Impressively, he describes a controlled duplicable setup in which he's able to reliably freeze (using an exact definition) identical quantities of water in identical containers under identical conditions such that the hot water freezes before the cold water. Incredibly, he was sometimes even able to observe this when starting with 100C water vs 0C water!
As you'd guess, there is a little bit of a trick to it, but it's one that not nearly as obnoxious as some of the earlier explanations: no evaporation resulting in a smaller volume or remaining water, no spontaneous ice spikes with greater surface area, and no differential melting of the thick layer of frost on the freezer shelf required.
Rather than giving a spoiler, I'll let the paper describe the exact technique he used. OK, a hint from the paper to encourage you to actually read it: cold water will indeed always reach 0C before identically handled hot water, but contrary to popular belief reaching 0C is merely a requirement for freezing rather than a guarantee.
I think it's because there is more gas in cold water. This results in ice that has more bubbles in it and conducts heat more poorly, and prevents the latent heat from being removed from the yet unfrozen parts of the water.
Impressively, he describes a controlled duplicable setup in which he's able to reliably freeze (using an exact definition) identical quantities of water in identical containers under identical conditions such that the hot water freezes before the cold water. Incredibly, he was sometimes even able to observe this when starting with 100C water vs 0C water!
As you'd guess, there is a little bit of a trick to it, but it's one that not nearly as obnoxious as some of the earlier explanations: no evaporation resulting in a smaller volume or remaining water, no spontaneous ice spikes with greater surface area, and no differential melting of the thick layer of frost on the freezer shelf required.
Rather than giving a spoiler, I'll let the paper describe the exact technique he used. OK, a hint from the paper to encourage you to actually read it: cold water will indeed always reach 0C before identically handled hot water, but contrary to popular belief reaching 0C is merely a requirement for freezing rather than a guarantee.