If we liquify a metal and then leave it to cool, some times it remelts on its own, without any input of heat energy. This effect was seen in Co-B alloy. We call it Flash effect.
Years ago we observed a similar phenomenon (even though quite different) in a martensitic transformation (solid-solid first order phase transformation): By slow cooling, a single crystal sample in beta phase (high temperature phase) transformed slowly to plates of martensite (low temperature phase). The plates had some pinning effects, at a given point, a new plate grew a considerable amount, and this produced other martensite plates to shrink. Later, by further cooling, all martensite plates grew (some grew slightly before lowering further the temperature). The interpretation was that temperature was not strictly uniform, and the apparition of a part of martensite implied the release of latent heat, which altered locally the temperature. The temperature wave propagates, and if some plates with low hysteresis (hysteresis is somewhat related to deffects) are in the neighbours, they will shrink as a result of the appearance of the new plate (the shrinking is a "reverse" transformation)
Might I discuss more about these effects, I'd like also to know more, and try to understand better, please send me an e-mail: