Emma's paper on X-ray profiles, comments on section 4 D. Cohen 7 December 2011 There is a lot of good information and cogent analysis in the results (\S4) and discussion and conclusions (\S5) sections of the current (v6c) version of the paper; and in the associated figures and tables. I do think, however, that we need to significantly overhaul both sections. In order to think carefully about the optimal structure, it might be useful to outline what we currently have in v6c; so I'll do that on a paragraph-by-paragraph basis. And I'll put my own comments in square brackets. Current structure of section 4: \S4 Results P1: Lines are well fit by the OC model (with only norm, taustar, Ro as free parameters) except for delta Ori, which seems to have a slight red-shift. The fits are good once that's accounted for. [delta Ori is a close binary system; Harvin et al. 2002 shows radial velocities for the primary that range from +100 to -100 km/s. The redshift is likely due to this orbital motion. But leading with this in the results section seems maybe not appropriate. It could easily be incorporated into the discussion in the middle of \S3.2 where the RV of xi Per is discussed. But, noting that our fits are generally good seems like a reasonable thing to include at the beginning of \S4.] P2: taustar results -- wavelength dependence is evident in most stars (though not epsilon and zeta Ori). [Looking at Fig. 3, I'm not sure I even agree with this empirical statement. Especially after we censor the lines at 15 and 19 A in epsilon Ori, due to resonance scattering -- which will already have been discussed in \S3. I wonder if we should include chi-square values in Table 2, for our mass-loss fitting, just to be able to say good-vs-bad fits; and then just comment on how "evident" the wavelength trend is. E.g. I really don't see a trend in HD 150136 in Fig. 3.] Next, the "mass-loss rate results" are discussed (and Tab. 2 and Fig. 3 called out). The values of Mdot are summarized and categorized with respect to the order of magnitude of their Mdot reduction (9 Sgr and xi Per are reasonable; the others less so). Pretty much every star is mentioned here by name. [In addition to 9 Sgr and xi Per, eps Ori and zeta Ori are reasonable -- at least with respect to theory (Vink); Those two Ori stars have late spectral subtypes, so the theoretical values are pretty low. Plus, there's the resonance scattering in epsilon Ori. I'm pretty confident that those four stars are the ones we want to claim are "normal" with believable X-ray profile based mass-loss rates.] P3: Various sources of error in the taustar and the Mdot determinations are discussed. P4: Ro results are discussed (and Tab. 2 is cited again, and Fig. 4 called out). P5: V_inf results are briefly discussed (and Tab. 3 and Fig. 5 are called out). [OK, I like how there are only a few paragraphs and each does a specific job. We need to think about how to modify specific paragraphs in light of the new ways, or at least nuances, we're using to interpret the data. And then we need to think about how to arrange these paragraphs.] [Off the top of my head, I'd say, here's how I'd outline a new section 4: P1: The profile fits to the lines are generally good. However, at the only modest resolution of the Chandra gratings, an OC01 profile with taustar ~ 0 is essentially equivalent to a Gaussian, with Ro controlling the width for a fixed vinf. So, just because the profile fits are good, doesn't mean that the EWS mechanism (and OC01 profiles) explain the physics of the X-ray production. Reasonable values of taustar and Ro and a wavelength trend of taustar that is at least consistent with that expected from the atomic opacity of the wind is the minimum requirement to conclude that the EWS mechanism is operating in these stars and to interpret the ensemble taustar values as a mass-loss rate measurement. P1.5 (?) For some of the stars in the sample, the data quality is simply too poor to draw any conclusions. For this reason, we exclude HD 206267, 15 Mon, and tau CMa from further discussion. (We could show parts of those spectra -- compare with Fig. 1, perhaps -- to show just how bad a Chandra grating spectrum has to be before it's useless for this kind of work; and/or we can refer to the overall Chandra counts listed in Tab. 1 and note that the stars we're excluding had only two or three lines that could be analyzed. Further note that HD 93250 is a special case, with only three measurable lines, but a lot more continuum counts than the other stars. But that's because its plasma is so hot; it's bremsstrahlung dominated. And that's because it's a CWS system -- with a close binary companion recently discovered via interferometry (Sana 2011) -- see next section.) P2: We summarize the taustar and Ro values for each star in Figs 2 and 3, respectively. But I don't think we need to summarize the actual values or reductions in the text, as Emma currently does in her P2. (For epsilon Ori, we should gray-out the two resonance scattering lines, and show mass-loss rate fits that both include and exclude those two lines.) I think we can just briefly mention uncertainties -- on taustar and Ro (statistical from the fitting); on Mdot (scatter in the taustar values); and systematic (on taustar, mostly due to assumptions like vinf or beta, but also on Mdot, mostly due to the wind opacity model). I don't think the uncertainties discussion needs its own paragraph. Should we also perhaps briefly discuss weird lines -- like Ne IX complexes that might be effected by blending with iron lines? Or Si XIII in zeta Ori which has a crazily small upper limit on taustar [we - Jon, Alex, Maurice and I - talked about excluding that line from the fit or calculating Mdot both with and without that line complex; Maurice thought that the lines in that complex looked weirdly peaked.] [We will have to refit all the Mdots using our newer opacity model; I've already done it for the solar abundance opacity model, but I haven't remade the plots for Fig. 3. Should I do this or should Emma?] [Should the higher, "expected" taustar trends shown in each panel of Fig. 3 actually use the Vink theoretical mass-loss rates rather than the "traditional" unclumped H-alpha rates? I think so. It's a bit tricky, but I think we should show the Mdot determinations, graphically, in Fig. 3, as we currently do, but not discuss the Mdots themselves until section 4, the "discussion" - especially since we're going to claim that half of them are essentially meaningless. I guess Table 2 should still be called out for the first time in Sec. 4, though; just discussed in Sec. 5.] [I'm also having second thoughts about even fitting the CNO opacity model at all, since there's only one line complex for which it matters, and the new CNO mass-loss rates will be even more similar to the solar ones, given the even greater similarity of our two opacity models, now.] P3: Vinf determinations and Fig. 5 and Table 3. [OK, so my suggestions make for a quite short section 4, overall, deferring more of the Mdot results to sec. 5 then Emma did.]