Internal referee report 1 (A. Caliandro)It is important to describe the shape and the size of the ROI. In section 3.2 you write that sources for the xml model are extracted in a region of 10deg around HESS J1857. I guess you used the same ROI for the data selection. If it is the case (that you used 10deg ROI for the data selection) I suggest to use for the xml model a larger extraction radius (for example 15deg). In this way you avoid possible issues at the borders of your region, and the diffuse emission will be better constrained. Plausibly, this will lead to a TS value for your source slightly higher. To evaluate the systematics due to the Galactic diffuse emission, you chnge the normalization by +-6%. This is fine, but 6% refers to the uncertainty of the model we used with P6V3. The new model you are using to analyze P7V6 data most probably and hopefully is better than the previous one. So the systematics you evaluated are over-estimated. YOu ´can have a feeling on how large is the over-estimation comparing the systematics due to the diffuse with those due to IRFS. Sincerely, I do not know whether a study equivalent to those performed for W51C and W49 is already done for the new Galactic diffuse model. Since this paper is a CatII, I do not think that this kind of work should be included in it. But in the text should be clarify what I wrote above. Anyway, this is a general problem that we should discuss within the collaboration, involving the diffuse group.
#------------------------------------------------------------------------------------------------#
I will comment section by section here below.
-- Introduction. I found quite confusing the third paragraph of the introduction (L36) In particular the period: In such sources, TeV radiation can be ex- plained by Inverse Compton (IC) scattering of an old population of leptons produced earlier in the pulsar’s life on ambient photon ?lds (CMB, IR, ...) or by ? decay from the interaction of accelerated hadrons with nuclei of the interstellar medium.
What do you mean exactly with 'an old population of leptons'? Then, in the first paragraph you say that 'The dissipation of the rotational energy of a pulsar leads to the creation of a relativistic wind made of electron/positron pairs.'. So why you expect pi0 decay in PWN? Maybe a more detailed explanation, and some reference will improve.
-- Timing analysis This is a request mainly for the radio people. It is possible to evaluate the significance of the last frequency derivative in the ephemeris? In other words, how much the fourth derivative improve the fit respect to a fit with just three derivatives?
-- Spatial and spectral analysis For the description of the diffuse models you could also reference the 2FGL paper.
The description of pointlike is quite poor. Further also gtlike is used in binned mode here (I guess). I would avoid to enter in the details of the pointlike description in tis paper. I suggest to just state that pointlike is optimized for the search of the extension of a source, while it is not the case for gtlike. I would cite also the paper in preparation about search for spatially extended sources, in addition to Kerr 2011. -- Shape and position of the source Since you finisc the previous chapter talking about the refitting of W44, to avoid confusion I would rename this one as ' Shape and position of the HESS J1857 counterpart' Since you cite Fig1 (b) as first, I would exchange the positions of the maps, so that the TSmap for E>10GeV is in panel (a). The TSmap for E>10GeV is not a residual map, as you write in the caption. Otherwise we should see no source, but just statistical fluctuations. The ellipse errors on the localization of the source should be specified. you state that the localization is consistent with the position found in HESS. What about the consistency with the position of the radio pulsar? -- Spectral analysis I already commented on the systematic uncertainties. The analysis concerning the evaluation of the pulsar DC emission (described in the last paragraph) is a little bit tricky. I ask here the same question than above. Is the radio position of the pulsar compatible with the pointlike localization? If so, you can not disentangle the pulsar DC emission from the PWN emission.
Authors response to referee 1*"It is important to describe the shape and the size of the ROI.* *In section 3.2 you write that sources for the xml model are extracted in a region of 10deg around HESS J1857. I guess you used the same ROI for the data selection. If it is the case (that you used 10deg ROI for the data selection) I suggest to use for the xml model a larger extraction radius (for example 15deg). In this way you avoid possible issues at the borders of your region, and the diffuse emission will be better constrained. Plausibly, this will lead to a TS value for your source slightly higher."* This is a typo, we used a 15° extraction region to build the model. We fixed it in the text. " *I found quite confusing the third paragraph of the introduction (L36)* *In particular the period:* *In such sources, TeV radiation can be explained by Inverse Compton (IC) scattering of an old population of leptons produced earlier in the pulsar’s life on ambient photon ?lds (CMB, IR, ...) or by ?decay from the interaction of accelerated hadrons with nuclei of the interstellar medium.* *What do you mean exactly with 'an old population of leptons'?* *Then, in the first paragraph you say that 'The dissipation of the rotational energy of a pulsar leads to the creation of a relativistic wind made of electron/positron pairs.'. ** **So why you expect pi0 decay in PWN? Maybe a more detailed explanation, and some reference will improve." We removed these part of the text in the new version to be able to fit in the 4-page limit of A&A letters. *-- Timing analysis* *This is a request mainly for the radio people. It is possible to evaluate the significance of the last frequency derivative in the ephemeris?* *In other words, how much the fourth derivative improve the fit respect to a fit with just three derivatives?*
By including a fourth derivative the RMS of the timing residuals decreases in a bout 30%.
It goes from 1551.106 us to 1081.297 us, which is equivalent to a decrease from 19 to 13 milliperiods.
The reduced chi-square of the fit also decreases significantly, changing from 22 to 5. Even though they have no physical meaning, the use of higher order derivatives helps to flat the residuals,
allowing a more accurate folding of the gamma-ray photons on the rotational period. *The ellipse errors on the localization of the source should be specified. you state that the localization is consistent with the position found in HESS. What about the consistency with the position of the radio pulsar?* We added the average error on the position. *Is the radio position of the pulsar compatible with the pointlike localization? If so, you can not disentangle the pulsar DC emission from the PWN emission.* The radio position of the pulsar is consistent with those of the GeV source. Nevertheless, it is clear that the hard spectrum that we obtain at very high energy does not ressemble the one of a pulsar. Then, to prevent any contamination of the upper limit on the pulsar by the PWN, we added the PWN best fit parameters in our source model. This way, we can be sure that the derived upper limit is robust (as was previously done for HESS J1825). internal referee report 2 (name)Authors response to referee 2 | 
=Task completed
= To do
