Subject: mea culpa on BU
From: jnkf@math.mit.edu
Date: 07 Dec 2005 23:36:03 -0500

Clarence Wilkerson and Andy Baker both point out that my mistake 
in thinking that the map BU_1 x BSU -> BU pulled c_i (i>1) back to 
1\otimes c_i. By the product formula, it actually pulls c_i back to 
c_1 \otimes c_{i-1} + 1 \otimes c_i.

So the splitting is compatible with the fact that mod p Steenrod 
operations acting on c_i typically involve a c_1 in the cohomology of BU.

Also, my attribution of my error to Mike Hill was also incorrect.

John