Notes on the comments raised by the reviewer:


* page 2, row 17 from the bottom: BLS1s - ok

* page 3, row 11: ... ion species, which are... - ok

* page 3, row 5 from the bottom: ...ions in their unexcited configuration... (remove comma after ions) - ok

* page 3, row 4 from the bottom: ...plasma electron temperature... - ok

* page 5, last but one row: ...holding, the inferred... - ok

* page 7, first row: ...process, which must account for the several contributions that overlap... - ok

* page 7, row 7 from the bottom: you used E(B-V), but Cardelli et al. themselves showed in their paper that it is better to normalize the extinction laws with A(V). 

Answer: The role of E(B-V) as a proxy to the calibration of the actual extinction curve changes with wavelength. For this reason, though the original measurements of extinction were performed as color excess estimates, we compared the derived extinction coefficients with the A(V) provided by Schlegel et al. (1998). We thereby calculated our extinction curves in order to match the appropriate value of A(V). The mis-leading description has been corrected in the text.

* page 8, row 6: ...the techniques that are required... - ok

* page 8, row 10: ...contributions that are most... - ok

* page 8, row 25: The calibration... - ok

* page 8, row 31: generally, it is better to rebin with 20-30 counts per bin to use the chi square test. You will have also lower statistical fluctuations (see page 9, first row).

Answer: The 15 counts per bin acts as our "not-worse-than-that" hard limit,
mostly affecting the high energy tail of fainter spectra or the lower
response of specific instruments. The low energy channels (which are
concerned with the discussion in page 9) are generally characterized
by the maximal values of counts. Though higher binning values can
smear out the noise fluctuations, the resultant energy resolution would
be reduced, leading to little or no improvement in the distinction of
different models for this part of the spectrum. Clearly, channels with
low counts have smaller weights on the global fit.

* page 10, Table 2 and Section 4: the emission lines in the region of the X-ray
spectrum between 6 and 7 keV (rest frame) are many. There is not only the neutral iron Kalpha (6.38 keV), but also the Kalpha of the ionized iron (at different levels: 6.7 keV for FeXXV, 6.97 keV for FeXXVI,...), the Kbeta (7.1 keV). In my opinion, it is not correct to say that the energy of the Kalpha is changing. The energy of an atomic transition does not change with the ionization degree, while what is changing is the type of ion or the type of transition (7.1 keV is generally the Kbeta). Please have a look at Ross et al., 1999, MNRAS, 306, 461.

Answer: With the achievable spectral resolution, the Fe line system cannot be
resolved. As a consequence, we have an emission blend, originated by
K-shell processes, whose central energy increases as soon as the
contributions of highly ionized species become more and more
relevant. In this sense "the energy of the iron emission line shifts
up in energy from 6.4 through 6.7 to 6.97 keV as iron is ionized"
(Ross et al. 1999). We clarify this point changing the text
accordingly and pointing out that, in most cases the dominant feature
is Kalpha, with at least one suspect case of significant weight from Kbeta.

* please revise the bibliography according to the Section 2.3 of the PoS authors manual. - ok