In a recent paper Alscher and Grabert claim to prove that a dominant (tree-
level) stationary phase approximation to the Wiener regularized continuum s
u(2) coherent-state path integral for the quantum propagator P := (z(F)\ T
exp(-i integral(0)(tau) H ds)\ z(1)) becomes exact, provided H is a linear
combination of the su(2) generators with arbitrary time-dependent coefficie
nts. I find the derivation of this in fact obvious result unduly complicate
d and somewhat obscure. The authors start from a classical spin action inco
nsistent with necessary boundary conditions and therefore are forced to inv
oke a nontrivial regularization of the action to render the latter meaningf
ul. Alternatively, when a su(2) symplectic potential consistent with the bo
undary conditions is employed, no regularization is required to obtain the
leading quasiclassical asymptotics.