For this investigation of coma, we use a plano-convex lens as described by
the following input file:

2
1.0
1.5  -5.0  150.0 1.0
1.00  10.0  1.0e20 1.0

The on-axis focus is given by 

par_ray cxpl 0.0 0.0 0.0 0.1
Rays intersect at (x,y) = (298.333295, 0.000000)

Now, putting the screen at 298.333295, we find the following focus
characteristics as a function of beam offset:

offset	focal point	+ screen pos.	- screen pos.
1.0	298.329465	0.000013	-0.000013
5.0	298.236595	0.001612	-0.001612
10.0	297.946009	0.012936	-0.012936
20.0	296.778059	0.104538	-0.104538

From this we clearly see the spherical aberration kicking in, but no coma,
as the screen position offsets are perfectly symmetric.

Now let's try an angular offset of 0.05:

par_ray cxpl 0.0 0.0 0.05 0.1 298.333295
Rays intersect at (x,y) = (296.989172, 14.932565)

Setting the screen to 296.989172, and comparing the screen positions to the
nominal 14.932565.  We get the following:

offset	x-inter		y-inter		+screen pos.	-screen pos.
1.0	296.985354 	14.932296	-0.000088	-0.000063
5.0	296.892773	14.925763	-0.003517	-0.000288
10.0	296.603090	14.905306	-0.020601	 0.005307
20.0	295.438723	14.822831	-0.135867	 0.073513

The + and - screen positions are relative to the nominal value.  We can see
by the migration of the x-intercept of the two rays that we have spherical
aberration in addition ot the (corrected-for) field curvature.  But we also
have coma: we see this as a shift in the average position on the screen (toward
the negative).  This is apparrent from the beginning (small offsets), but
is eventually somewhat compensated by spherical aberration.  But the
average of the two rays on the screen defnitely walks downward.

Now we go to an angular offset of 0.10:

par_ray cxpl 0.0 0.0 0.1 0.1 
Rays intersect at (x,y) = (293.013084, 29.466168)

Setting the screen to 293.013084, and comparing screen positions to
29.466168, we get:

offset	x-inter		y-inter		+screen pos.	-screen pos.
1.0	293.009302	29.465637	-0.000159	-0.000133
5.0	292.917589	29.452765	-0.005315	-0.002072
10.0	292.630614	29.412456	-0.027855	-0.001841
20.0	291.477002	29.249923	-0.165683	 0.044582

Now the effect is more exaggerated: the negative comatic aberration
persists longer, and the spherical aberration does less to smear out the
effect.  The effect, in both offset cases, evolves dramatically as a
function of offset (double the offset and see a huge increase in the
effect).

On the whole, correcting for field curvature, this lens still displays
spherical aberration (focus shifts inward for larger offsets, and screen
sees resulting blur).  But we also definitely see coma.  The offset on the
screen is systematically negative, and rapidly increases its negativity as a
function of offset.  This is the signature of coma.

To evaluate the degree of blur, we compare the offsets to the focal length
of approximately 300 units.  Using the +/- 20 unit displacement (for an
f/7.5 lens), we see that on-axis blur is at the 0.1 scale, or 0.0003
radians, so that we could resolve 3 cm at 100 m.  By the time we get to an
angle 5.7 degrees off-axis, the outer rays have walked 0.06 units off the
small-offset focus, almost doubling the blur effect from spherical
aberration (on-axis blur).