Question about commercial implementation of mscales

This message is going primarily to owners of the commercial product,
however
anyone is welcome to comment on it.
mscales implements the multiple scales perturbation method, and returns
results as e-lines.  Following is a result from the demo, which is
solving the
Van Der Pol equation.

This is e7,
[at(x1,[%scale0 = t, %scale1 = e * t])
 = ((3 * ms0 * ms1^2 * sin(3 * t) - ms0^3 * sin(3 * t) - ms1^3 * cos(3 *
t)
 + 3 * ms0^2 * ms1 * cos(3 * t) + 32 * ms3 * sin(t) + 32 * ms2
 * cos(t))/32),
at(x0,[%scale0 = t, %scale1 = e * t]) = ms1 * sin(t) + ms0 * cos(t)].

This is e1,
[x = e * x1 + x0]

At this point in the demo, the secular equations are solved for ms0 and
ms1 by switching to
polar coodinates, and the solution is substituted into e7.  ms2 and ms3
are set to 0, and
the result is subsituted back into e1, which will fail, as e7 uses
AT(X1,...) instead of
x1 in the solution.  Does anyone have an idea why the writers of mscales
would do things
this way, that is using a AT?
As it stands now, the demo seems broken.
Dan Stanger