Dear Alan,

Thank you for your comments. Here come the replies.

Bye
Peter

> Hello Bernhard, Peter,
> Congratulations on putting together this paper; sorry
> for the delay. I've gone through the May 3 draft up to Section 5 and, to save time, will send some comments now and some later.
>
> (a) There are a number of English corrections, usually using past tense when present tense is needed (the rule is:
> either tense is OK, but you must use one tense consistently
> throughout a single paragraph), or singular when plural
> is needed, or the wrong preposition. There are a number of
> sentences that, while not grammatically wrong, are awkward for an English speaker (I don't know if you worry about these). Anyway, I'll send these corrections in a separate email.  
>
Thanks a lot, they will be most welcome.

> (b) I have two physics concerns:
> The fit quality in Fig 2a does not look especially good, i.e., the MC does not describe the peak at 1.58 GeV/c2. What
> are those events (i.e., from the truth table)? What is the chi2/dof of the (presumably binned) fit? Do you include a systematic error to account for uncertainty in the background shape? (you should) The discussion of systematic errors on p11 is a bit terse; more explanation here would strengthen the paper.
>  
>
Please find enclosed a plot which shows the residuals (difference in the measured and fitted number of entries in units of sigma) for Fig 2a. From the plot we deduce that the observed 'peak' at 1.58 is most probably a fluctuation. This is supported by the distribution for individual targets (Fig 8 in the note ,http://www-hera-b.desy.de/subgroup/physics/herab/analysis/mb2003/open_charm_articles/note_061017.pdf).  On the other hand, a slight discrepancy between the ansatz and the real background form cannot be excluded. The corresponding impact on the signal yield is small, and has been included in the systematic error (under 'event counting'), and its evaluation is described in the note. Note also that the ccbar background disapears above 1.75GeV (see Fig 7 of the note).
Chi2 of the fit is 92.4/84 dof, prob = 25%.
You are probably right about the description of the sys error determination. We shall expand that paragraph.

> There is "background" under the phi peak in Ds->phi pi coming
> from Ds->K*K, Ds->KKpi that seems unaccounted for. You can use the MC to make this (small) correction. In fact the effect
> may be negligible, but it seems we should at least mention it.
> Note: this is different from the D*->D0pi mode, because in that case background under the D0 peak will not peak under the D* mass
> (there are no non-resonant D* decays).
>  
>
The effect is indeed very small, and has been accounted for. From MC we determine that  3% of selected events are Ds decays other than phi pi+. Note that this is much smaller than the statistical (30%) and systematic (20%) errors. We will think about how to formulate this.

Bye
Peter