Hi, I'm sending you a ps file comparing two different bswf for the p1/2
 shell and p3/2 shells.
 We comment first the p1/2 results.

 Here we compare the results obtained with the nlsh bswf, same as used in
 our PRL and oyur publication,
 with  other, nl-low, I've recently  obtained with a phenomenological
 lagrangian  that I've fine tuned by hand to reproduce the cross-sections
 at low-pm of Leuschner as closely as possible (and I really got I nice
 fit).

 As you see,  with only 4 points in the low-pm
 region it is hard to decide which wave function fits better the data in
 this region (have a look at the linear plots!). The chi2 is worse for the
 nl-low result, but this is mostly due  to the contribution from the
 points in the high pm region. I  believe that the chi2 for the
 nlsh wave fucntion is just 'too good'. With only 8 points in total, it is 
 likely that they all happen to be near the curve. This low chi2 
 makes the error in the sp. factors deduced from nlsh results to be
 very small (too small, I believe).

 The important thing is that the sp. factors I get with the nl-low bswf is
 0.64(5) with cc1 and 0.66(7) with cc2, to be compared to 0.69(4) and
 0.73(2) for the case of the nlsh wave function.

 Two things:

   1) The error is larger for the sp. factors deduced with NL-low.
 I expect that the nl-low will fit better the results than the nl-sh
 in general, so having 20 points instead of 8 will reduce the error
 in the sp-factor deduced from the nl-low bswf, and increase the error
 of the sp-factor from the nlsh ones. 
     The values of both sets of sp. factors are compatible within
 the error, but certainly different. I trust better the values
 obtained with the NL-low bswf.

  2) The Rtl and Atl are slightly better reproduced by the new set.
 With only 4 points in every case, I won't dare to say that this is a
 definite reason to believe that nl-low is more adequate, but certainly
 it claims for more experimental points over the whole pm-range!.
---------------------------------------------------------------------------
 Figure for the p3/2 shell: I used here the nl-low wave
 function  but here I compare the effect of the doublet, including in the
 calculation
 a 100% of it, 50% or not at all. There is also an effect in the
 sp. factors, that go from 0.59(8) with 100% of the doublet to 0.63(9)
 with 0% of the doublet. The values with nlsh wave function would be
 a 15% higher, approximately,  compatible within error bars with the 
 ones from nl-low.

 I'd like to have more experimental points also in this case, to confirm
 the problem we have in Rtl in between 50 to 200 MeV, and also to reduce
 the uncertainty in the sp. factor. It could be that with more points,
 the chi2 will not decrease much, so the sp. factor obtained with this
 wave function for this shell will still have a large uncertainty. This
 would mean than the wave function that fits Leuschner data does not 
 work fine for JLab, and that would mean there is something missing in
 the treatment of the reaction: MEC or a Q2 dependent effect of
 correlations or whatever... So far I'd say there is a clue that the
 agreement of theory and experiment is not so good for this shell as for
 the p1/2 one, but with only 4 points is hard to see.

 Finally one comment, the sp. factors I obtained fitting to your data
 are compatible with the ones I get fitting to Leuschner or Chinitz within
 error bars, what is not very significant at the moment as the error
 bar is quite big in one of the cases.

 Let me know if you want to include any of these curves in the proposal.
 Regards
 Jose