To compute rad(Tn(k)) for any ring k , we can treat Tn(k) as a triangular ring with R=k , S=Tn−1(k) , and M=kn−1 (as (R,S) -bimodule). Using fact that rad(T)=(rad(R)0Mrad(S)) and invoking an inductive hypothesis, we see that rad(Tn(k)) consists of n×n upper triangular matrices with diagonal entries from rad(k) .