Richard:
If you're used to looking at transverse ray aberration plots its very
easy to see the difference:  coma has a second-order dependence on
aperture, while oblique spherical has a third order dependence on
aperture.  Its much easier to understand it visually than verbally,
unfortunately.  A Newtonian telescope is a classic example of an
optical system with coma, and a Topogon is a classic example of an
optical system with oblique spherical aberration.  Both improve when
you stop down, but oblique spherical will disappear faster because it
has a higher-order aperture dependence.  Looking at spot diagrams can
get confusing because vignetting in the tangential direction can cause
oblique spherical aberration to look a little bit like coma.

Unlike the Newtonian telescope mentioned above, photographic objectives
rarely suffer coma at their optimum magnification.  Thats because coma
is an easy aberration to correct.  An exception is what I call "axial
coma" that can occur due to tilted and/or decentered lens elements.  If
you change magnification from nominal then you will see coma,
especially in longer focal length lenses.  Most macro lenses for small
format cameras use at least two independently moving lens groups to
correct coma over a very wide range of magnifications.

Oblique spherical aberration, on the other hand, is a very difficult
aberration to correct, and it is the limiting aberration in many types
of photographic objectives.  One way it can be reduced is by the use of
aspheric surface(s) located in an area where the marginal ray height is
fairly large.  Unfortunately, such surfaces are difficult to make
because they almost always have tight tolerances.  The famous 58mm
Noct-Nikkor lens used an aspheric surface not to correct coma, but
rather to correct sagittal oblique spherical aberration.

Brian