gave scatterplots illustrating large splits in three large trees, and observed that a certain new model (``beta = -1") seems a better fit than the usual models. Miklos Csuros (C.S. Dept, Yale) has made 5 new scatterplots corresponding to other large published trees. These data reinforce the observation above.
Here is Bibliographic info. One "new" tree is based on same underlying data as a tree in the paper.