sn921 -- unstable to a combination of stretch and date of max.  Varies between
  s = 0.86 and s = 1.28 depending on initial conditions for the two.

sn94102 -- unstable.  The fit that rob has (s=1.04) is probably the better 
 one, but this is still pretty worrisome.

sn94191 -- stable . This is not a bad looking fit.

sn94264 -- stable.

sn94281 -- stable, but error bars misestimated with default values.

sn94361 -- stable, but really ugly.  There's a point that misses by 2.5 
  around max.  Error bar on s a bit smaller than Rob's.

sn94351 -- no B reference.  I did two sets of fits.  One was fixed,
 the other let B float.  Both are pretty solid, but give different
 magnitudes and stretches.  Rob's fit corresponds to the fixed one.

sn9568 -- solid

sn9569 -- solid

sn9570 -- solid

sn9579 -- solid

sn95103 -- solid

sn9104 -- solid

sn95110 -- solid

sn95116 -- mostly solid.  Values don't move around, but the
 error bars change a bit as you modify the initial values.

sn95126 -- mostly solid, but the error bars are a bit wierd.  Strangely, 
  when the fit is initialized close to it's final value the error bars
  go to zero for MINOS.

sn962 -- solid

sn963 -- solid

sn969 -- solid

sn9617 -- solid

sn9621 -- Mildly unstable . Pretty ugly.  The values move around with the
  initial conditions, but generally less than the quoted errors.  But
  it's still a troubling one.

sn9624 -- solid

sn9626 -- solid

sn970 -- solid. one of the VR guys.  The R fit isn't so great, unsurprisingly
 (although the dual fit is stable) so I have checked the results of a V only 
 fit (IV) which is stable (V).  I then fixed the stretch to that value
 and did a dual fit (VI).  This didn't make the fit look better, but
 also didn't really affect any of the values.

sn971 -- solid, but the MINOS error bars are screwy.

sn972 -- solid . Not great looking.

sn974 -- solid

sn975 -- solid, but no V ref.  I-III both fixed (pretty solid,
  slight change in stretch), IV-V V float (solid).

sn976 -- solid, albeit ugly.  B/R only.

sn9710 -- solid. VR RI.  Has V final ref and R pre-sn ref, so I left it fixed.

sn9733 -- solid

sn9738 -- solid

sn9739 -- partially solid. no V ref.  I-III are both fixed, IV and V float V.
  Fixed is pretty solid, but not entirely, and float is solid.  Floating is
  probably better, but this is not one of the SNe which Rob claims should
  float.

sn9742 -- solid. no V ref.  I-III are both fixed, IV and V float V.  No V band
  data near peak, so color is suspect.  The two kinds of fit are solid,
  but different.

sn9748 -- solid. no V ref.  I-III are both fixed, IV and V float V.  Both are
  pretty solid, although the errors move around a little as initial 
  conditions are varied..  This is one of the float offset ones listed by Rob.

sn9760 -- pretty solid. no V ref.  What, did we not bother with V refs 
  for anything on this run?  As usual, I-III fixed, IV and V float V.  
  Fixed are pretty solid, with the errors changing slightly.  Float V 
  also has slightly shifty errors.  This is one of the ones Rob lists as 
  needing a floating offset.

sn9765 -- mostly solid, although errors are fairly variable depending on
  initial conditions.

sn9779 -- not stable.  no final refs at all, but pre-refs.  See plot III --
  different initial conditions change the stretch by more than the error
  bar, and the chisquare is pretty much the same as fits I,II

sn9781 -- pretty solid.  Very noisy data.  The error bars change around a
  bit based on the initial conditions.

sn9784 -- solid. UR BI, U final ref, B pre-SN ref.

sn9785 -- solid, but should have floating V.  B final ref, 
   V not quite pre-ref.  I-III fixed, IV and V float V.
   In the fixed fit B misses high.  Fit III gives different results, 
   but the chisquare is considerably worse than fit I/II, so it is not 
   worrisome.  Floating the V really changes the results, 
   and dramatically lowers the chisquare. The float fit seems stable.

sn9794 -- Unstable.  Fits I/II and III have the same chisquare but very
  different fit values.  Fit III corresponds to Rob's fit.