************************************************************************
Differences between the drafts of Dec. 2nd and May 24th.

The numbers quoted in Table II for A_meas was roughly a factor of
two different between the two drafts.  In both drafts,  the equation
defining A_meas was the same, and was supposed to be the raw
asymmetry divided by both the beam polarization and the target polarization.
Unfortunately the numbers we listed for A_meas in the December draft
were not yet divided by the target polarization. We apologize for
this typo.  Additional small differences (MUCH less than an error
bar) also came from recovered runs, slight differences in statistical
weighting, etc.  We should emphasize that there was only a miniscule
differences in the raw asymmetry between December and now.

Another difference came in A_phys.   This was due to two things.
First, in December,  A_phys did not include "nuclear effects",
whereas now it does. This change was implemented because we received
several comments suggesting changes in the way in which we presented
the analysis.  By saving the incorporation of nuclear effects until
the last step, it makes it possible to follow all the corrections
in a fairly unambiguous way in Table II, followed by the paragraph that
describes our GEA calculation.

Finally, there were changes in G_E^n, the worst of which was about
one sigma.  These come roughly from two sources.  Later in December, the
GEA calculations were reanalyzed.  Once complete, the results
included a higher effective neutron polarization (97-98%) as well as a higher
proton polarization (around 5-6%) than what was in the earlier
December draft.  There was also a change in our final results due to a
round-off error that was discovered shortly after the circulation of
the December 2nd draft, which primarily effected the highest Q^2
point.  Both of these changes were in the same direction, and were
"amplified" somewhat in the lowest Q^2 point because our target
polarization was not perfectly perpendicular to the q vector.