1) N(Ds) in Fig.1b
-------------------
(a) Explain why the random background under the Ds is smaller
in the RS sample, compared to the WS-sample.

(b) The fit does not describe the shoulder around 2010 MeV.
Explain the cause of the shoulder at 2010 MeV.

(c) The shape of the Lambdac->pKpi, with misidentified p->K,
is obtained from MC. Since the PID performance is known to
be different in data and MC, this leads to wrong results.
It is therefore necessary to use real data to obtain the shape 
of the misreconstructed Lambdac->pKpi.


2) Ds1 in Fig.3
----------------
The entire paper hinges on the assumption that peak at 2390 
MeV is uniquely caused by the Ds1(2536). One has to proof
that this peak cannot be caused by another 
projection/reflection/resonance than the Ds1(2536). 

At least it would be required to show with MC that the 
m(D0K+) spectrum is consistent with the presence of the 
decay  Ds1(2536)->D*0K+.


3) N(Bs) uncertainty
---------------------
(a) The text states that the procedure yields N(Ds)=2195 +/- 60 RS Dfb
events. The note says 2245 RS Dfb events. What is the correct value?

(b) The average detection efficiency is quoted as (1.09 +/- 0.03)%.
This is obtained from MC. Show how much the efficiency changes if the
charged track multiplicity from data is used;
and use the difference as systematic uncertainty.

(c) In the calculation of the efficiency, the relative 
BR(Bs->Ds*)/BR(Bs->Ds) is assumed to be 2.34 +/- 0.01.
Where is the source of this ratio? The assumption that the Ds*/Ds ratio
is the same as in B- or B0 decays is not justified, since the PDG reads:
B0 (PDG):  BR(B0->D-*munu)/BR(B0->D-munu)= 5.01/2.17 = 2.31 +/- 0.06
B+ (PDG):  BR(B+->D0*munu)/BR(B+->D0munu)= 5.68/2.23 = 2.55 +/- 0.006
In addition the Bs predictions from theory disagree to the measured B0/+
counterparts:
Bs (1003.5576): BR(Bs->Ds*munu)/BR(Bs->Dsmunu)= 7.09/2.85 = 2.49 +/- 0.3

(d) What is the uncertainty on the (Bs->Ds) efficiency if 2.49 +/- 0.3 is 
assumed?

(e) The "overall uncertainty on the Bs->Xmunu yield is 6.6%" is based
on MC. This is insufficient, as one needs to take into account the 
additional systematic uncertainty from differences between data and MC.


4) Comparison of data and (private) MC
---------------------------------------
Data-MC comparisons for the distributions relevant for the analysis are 
required.

(a) The comparison of track multiplicity for data and MC is needed, since
a different track multiplity will affect the (Bs->Ds) detection efficiency of 
(1.09 +/- 0.03)% as estimated from MC.

(b) The PID performance needs to be compared between data and MC, and 
must be shown.
A different PID performance will affect both the (Bs->Ds) efficiency,
and the number and shape of misreconstructed Lambda_c, mistaken as Ds.

(c) The pt spectrum of the Bs is unknown in the data.
The pt and IP distribution of the (Ds-mu) pairs need to be compared between
data and MC, shown, and an uncertainty must be attributed to it.


5) The Title
------------
The title of the paper does not cover the entire content of the paper.
The final result of the paper includes the BR, and this should be mentioned
in the title. 
Also, a decay with a BR of ~0.3 % is not really worth of being called "rare"
(in Conclusions).