Although it is the endothermic phase transition at 660 km depth that causes the first-order effects of layering, the exothermic, olivine to spinel phase transition at around 400 km depth [ Akaogi et al., 1989; Katsura and Ito, 1989] also has an important dynamical effect in enhancing the flow of material across itself, and has been considered in many studies. Even so, there is disagreement about its net effect on the degree of layering, with some authors [ Zhao et al., 1992; Steinbach and Yuen, 1992] concluding that the overall degree of layering is increased, whereas other authors [ Tackley et al., 1994; Solheim and Peltier, 1994] finding that it decreases the overall degree of layering. This disagreement may be due to differences in geometry or choice of parameters, and warrants further investigation.
Another complexity is that the phase diagram may exhibit a triple point between
-spinel, 
-spinel and perovskite+MgO, above which the Clapeyron slope of the
spinel to perovskite+MgO transition changes to a shallow or positive value. Liu et al.
[1991] and Zhao et al. [1992] show how this acts as a filter on upwelling plumes,
such that hot plumes having core temperatures above the triple point temperature pass
through the phase transition uninhibited, but weaker plumes are deflected. Early laboratory
measurements [ Boehler and Chopelas, 1991; Fei and Saxena, 1990] indicated a
rather low value for this triple point at around 2000 K, which was shown recently by
Liu [1994] to greatly reduce the overall degree of layering induced by the endothermic
phase transition. However, more recent work [ Chopelas et al., 1994] indicates a
significantly higher temperature for the triple point of 2260 K, with a mildly negative
Clapeyron slope above this point. These new results suggest a much more minor role for
the triple point in modulating the dynamics.