Dear all,

I collected comments from Marcel, Niels and Jacco

General remarks:

* It would be nice to reflect on the physical meaning of the observed values: 
Is the double ration expected to be close to 1? 
Does one learn anything w.r.t. factorization? 
Are other numbers in agreement with CKM factors?

* In the analysis the charm-hadron masses are constrained to their nominal value. 
For the background under the peak this might lead to biases to the observed mass. 
Has this been checked?

* In lines 64-66 an arbitration is is performed. 
Why is such arbitration done? 
A specific mass phase space is treated different than the rest. 
Why not assign the event to both decays, which is what would have been done if 
the D and Ds analyses would have been performed in separate analyses?

The motivation for the analysis could be stronger, especially as
we aim to submit to PRL. In particular we would suggest the following:
- line 9 
  [Add in the introduction a statement on testing factorization, for example:]
  "... to study beauty-baryon properties. In particular, a comparison of
   b-meson and b-baryon branching fractions can be applied to test factorization 
   in these decays. Furthermore ..." .

- line 111 
  [Refer back to this here, perhaps?]
  "The observed ratio is approximately |Vcd/Vcs|2 as expected, 
  if non-factorizeble effects are small."

- line 175 
  [Also here?]
  "... measurements have been made for the decays Lb->LcD(s),
  consistent with expectations from B0 decays, assuming
  small non-factorizeble corrections."


Specific comments:

- line 5: 
How is the sentence " .. many decays at the percent level [1]" 
an example of the previous sentence?

- line 48:
Why mention TMVA?

- line 49:
(5 + 23x3) = 74 variables are used in the BDT training? Why so many? 
Do you have enough events in the upper side band to prevent overtraining?

- line 100 and other:
Usually we omit the equation numbering for results, which in your case
are all equations, apart from (1), as you have such a wealth of results!
Be aware of it, but in discussion of the results the numbering does have 
advantages.

- line 107:
cancelation -> cancellation

- line 111: 
Refer back to this here, and add the decay constants, because
|Vcd/Vcs|^2*(fD/fDs)^2 = 0.054 * (205/258)^2 = 0.034 : 
"The observed ratio is approximately |Vcd/Vcs|^2 x (fD/fDs)^2 = 0.034 as
expected, if non-factorizable effects are small."

- figure 2: 
Inconsistent label of Bs0(bar) -> D- Ds+ in the plot compared to the text 
Bs0 -> D+ Ds- in caption. 
Same inconsistency in label for B0 -> D- Ds+.

- figure 3: 
- The axis label is too complicated.
  Please replace the y-axis label to something less cryptic,
  like 'Yield ratio', or 'N(LcDs)/N(DDs)'. 
  You don't need to add the epsilons, as N is nowhere defined as raw yield.
- Replace "The curve shows the efficiency-corrected yield
  ratio of ... observed by LHCb [24] fit to the data" by
  "The same pT dependence is observed in the efficiency-corrected yield
  ratio of ... as shown by the curve that fits the data of that measurement 
  of LHCb [24]".

- line 154: 
Has it been proven that the momentum scale is the dominant uncertainty 
in absolute mass and not e.g. the opening angle systematics in the Velo 
(e.g. length scale of the Velo)?
Typically, for high mass low multiplicity decays (large opening angle) 
the p-scale dominates and for lower mass and high multiplicity decays 
(small opening angle) the Velo systematics dominate.

- lines 198-200: 
Is the most recent acknowledgment sentence used for the Yandex sentence?

- line 221: 
Ref. [11] is published:
JINST 9 (2014) P01002 [arXiv:1311.3893 [physics.ins-det]].

-line 243, 253:
Refs [22],[26] Collaboration -> collaboration.

- line 248:
Ref. [24] LHCb-PAPER-2013-0XX -> LHCb-PAPER-2014-004.
Title: "Measurement of the \pt and $\eta$ dependences of \Lb production, 
        and the \LbLcPi branching fraction"