Again, we thank the reviewer for the valuable comments.  In this
revision we think we have modified the text to include the comments
the reviewer has requested.

> This manuscript should be published after the authors have made 
> changes and after they have considered some improvements to the
> manuscript.  The work is a valuable contribution to neutrino 
> physics measuring the Q2 distribution of quasi-elastic events 
> on nuclei.

> I am disappointed that the authors did not take many of the 
> prior suggestions into account to improve the manuscript.  They 
> commented extensively on my previous report, which was helpful 
> for me to understand the paper.  But much of this commentary 
> belongs in the manuscript so that the journal readers will also 
> understand the paper.  For a full length Phys Rev article it is 
> always better to err on the side of completeness.

> The following remarks are to be considered in addition to 
> remarks already made in the made in the first report.


> I suggest the title "Measurement of the quasi-elastic axial 
> vector mass in neutrino interactions on oxygen".  Section IIC 
> and the comparison to previous observations makes it quite clear 
> that the authors consider the oxygen mostly irrelevant.  There 
> is little discussion of the effects of Al, C and H on the result.

Changed.

We would respectfully characterize the reviewer's last statement 
as follows:

Among the effects of Al, C, and H(D) for which a model exists and 
can be quantified, none has a significant effect on the fits to the 
data presented here.  This is discussed in several locations (though 
not addressing Al in particular) in II-C, the use of Proton 
rescattering and R_nonQE, and 2tk->1tk migrations, and many of the 
subsubsections within V-C.  The known effects of the nucleus are 
dwarfed by the two large experimental errors.

And we had already made a stronger statement in the conclusion, 
in response to the reviewer's first set of comments,
and have added one sentence to reinforce it.  The relevant comparison
is not how we have treated the Aluminum in our detector.  Rather, 
this result is different from deuterium;  if this is due to the 
effects of the nucleus (rather than errors associated with the K2K 
experimental setup), then there is currently no explanation for it.

Unfortunately, the detector uncertainties are large, and we can 
only state that our best fit MA is higher than deuterium, but 
consistent at the two-sigma level.


> Could the authors please describe figures with more than one 
> plot on them in a more descriptive manner than "top" and "bottom".  
> In figure 6, for example, the dashed curve is sometimes on top and 
> sometimes not.  The caption uses "top" and "bottom" and is 
> ambiguous.  This caption problem was raised in the first 
> referee report.  Please do not use "top" and "bottom" to describe 
> different traces on the same plot.  Use "solid" and "dashed" etc.

Fixed.

> The first sentence of section IVA should have the words 
> "assumed to be" added to the definition of the muon candidate.  
> This assumption introduces some Q2 bias, particularly in the 
> non quasi-elastic sample.

"muon candidate" and "assumed to be the muon" seem to say the same
thing, though perhaps with different emphasis.  The text now
follows the reviewer's suggestion.  The reviewer's comment caused
us to recognize an ambiguity in how we stated the related detail
in Section III-C.  It more clearly indicates that roughly 1% of 
the muon candidates will not actually be a muon.


> Please put the list of Q2 values in the paper.  In the response 
> to the first report it is stated "We can easily supply such a list."
> But it is not in the paper.  Such a table will be a valuable part 
> of the publication record.

Table is added (now presented as Tab. IV), with the necessary
explanations and captions.

> In the discussion of systematic uncertainty, one might expect some, 
> such as the MRD errors to be asymmetrical.  But only symmetric 
> errors are used.

A good caution.  

The muon momentum scale error is effectively symmetric, because it
is a combination of correction and then the error on the corrected
result.  If we had not done the correction we describe, and instead
use the nominal momentum scale, then we would have quoted a much 
larger error on the result, we would certainly have given it as an
asymmetric error.

The other largest error, collectively coming from the uncertainty
in the Enu spectrum, is also effectively symmetric.

Among the smaller errors, a better model for the nonQE background
shape would likely produce a non-symmetric error, but when it is
folded in with the really large errors, it would not make a
difference, so we have chosen to present the result in the more
simple form.

> The rather odd appearance of figures 8 and 9, in which the fit to 
> all of the data lies at the edge of the fits to smaller subsamples 
> needs an explanation in the text.

We have added more discussion of this to the text.

The q2min cut is naturally this way because the data systematically 
prefer a higher value for MA as the high-statistics, low Q2 data is 
not included in the fit.  (A correction to our previous reply:
This trend is true regardless of how the other fit parameters 
are or are not constrained.  This comment was meant to apply
to the energy region fits)

The energy region fits is affected by the correlations.  The main one
that causes the fit values to all be "low" is that the flux in
the energy bin in question remains a free parameter while the
others are fixed.  If that parameter is also fixed to its best
fit value, then the fit value for the middle point rises slightly.


>
>
>


We have also added one sentence at the very end, based on a 
conversation one of us had with a colleague in June who participated
on some of the earlier bubble chamber experiments.  We agree with 
his philosophy on this point, and think this reviewer does also.
Note the middle sentence inserted in the final lines of the paper.

  We do not assume that neutrino interactions on oxygen should be the
  same as for deuterium.  In this sense we have measured the effective
  $M_A$ for oxygen.  However, we find only a small effect on the shape 
  of the $Q^2$ distribution for $Q^2 > 0.2$ (GeV/c)$^2$ from known
  nuclear effects.