1. P345 CLAIM A, "irreducible affine scheme" should be "affine space". Also, in its proof, I made an incorrect statement that (UX-X) is linear.
    To save the proof, use the fact that (StabUX2)X1 - X1 is linear, which follows from the choice which made in (6.2) with an additional hypothesis that eigenvalues of unmarked Jk are small enough (in a possible range). Then utilize Spaltenstein's affine paving argument for type A and the fact that (StabUX2) acts freely on F(gB) when we replace X with X2 (as in the second paragraph).
  2. I cannot verify P347 Corollary 6.9. To save the later argument, we restrict our assertion only to the preimage of union of open dense AND codimension-one orbits. Its proof is gone as follows:
    Open dense/codimension one orbits are open dense/codimension one along the fiber. Hence the collasping map gives a surjection of tangents (as in De Concini-Lusztig-Procesi 2.2, which concerns only dense open parts). (This remark might be deleted if I was wiser than expected.)

I thank Anthony Henderson for a discussion lead to the discovery of some of the errors at Sydney in the beggining of November 2010.

Last modified on 28 November 2010 by Sorry. I cannot type it here.