I appreciate the effort the authors have taken to substantially revise and focus the manuscript, including new data on 14CO2 and CO isotope data from an additional site. The evaluation of the Indianapolis dataset is presented in a more adequate way, although I have a doubt whether the error estimate is correct (the standard error of the mean has apparently been used for averaging measurements on clearly different air masses). The uncertainty may be underestimated.
It is nice that Beech Island data are shown, but the straightforward analysis with a Keeling plot model is questionable. The authors already discuss this in some detail, but do not attempt to estimate additional errors. This analysis is in my opinion too qualitative, and it may be useful to compare the Beech Island data to other stations (see below).
I also miss a discussion on the possible role of ozonolysis in producing CO from unsaturated hydrocarbons (e.g. isoprene), in particular the effect that this source would have on the d18O values.
P2, l17: remove “mole fraction of”
P3, l2: It may be good to include some other key references that determined the isotopic composition of sources, rather than only the review by Brenninkmeijer et al.
P8, l11 ff: I suggest replacing carbon-13 by 13C etc.
P8, l14: Mak and Yang are only one of the studies that state this formula, I think it was first explicitly stated in Brenninkmeijer 1993.
P8-9: This paragraph is about the 17O excess of CO. Minor issue: This is NOT observed “particularly in the high northern latitudes” but was also observed at Tenerife. It is a global effect. Further, the authors do not mention or are not aware of the fact that the 17O excess can also be created by ozonolysis of VOC, where the strong 17O excess of O3 is transferred to CO (Röckmann et al., J. Geophys. Res., 103, 1463-1470, 1998). This should be discussed, since it may be particularly relevant for the VOC oxidation source. The effect on the 17O correction for 13C is likely still small, but this source will also affect 18O, which is the target of the present study (see below).
P10, l6: Independent of what?
P10, L22: Is there a reason why you use the symbol I_delta18O here, instead of directly writing delta_source? In the next sentence you write that it is the source signature, so it may not be necessary to introduce a new symbol that is usually not used in the literature.
P12, l24 ff: You mention that you filter unphysical data, and data with less than 5% VOC contribution. In table 2 I see two data points that have about 10% VOC contribution, but for which the signatures are strongly deviating. I understand that these points are included in the evaluation.
P13, l1: The sentence “It is unclear from our data why the XCO2-FF enhancements are occasionally near or below zero.” is a unsatisfactory. Reformulate (at least provide some thoughts) since you have the 14C experts in your author team.
P14, l 8 and 10: Help the reader by writing down the differences (values) between summer and winter for both signatures.
L17: strongly contributes
P15, l3-5: These numbers seem to be the standard error of the mean of your observations. Is it adequate to use the standard error of the mean? It is clear that you do NOT make repeated measurements of the same samples. So it is in my view not adequate to assume that by making more measurements you reduce the error statistically. When I see the large range of values I think that the errors are underestimated.
P15-16: In the discussion of the VOC source signature, I miss the aspect that unsaturated HC can also be oxidized by ozonolysis, where the strong enrichment in both 17O and 18O would be transferred to CO (Röckmann et al., J. Geophys. Res., 103, 1463-1470, 1998). This was originally suggested as a source for the 17O excess of CO (see above), but it would also strongly affect the d18O values since O3 is so enriched in d18O. Does the low value for oxidation of VOC that is found here suggest that ozonolysis of VOC is not a significant source of CO compared to oxidation by OH? Do you have estimates from models on how much of the photochemically produced CO would be produced via ozonolysis?
P16, section 3.3: You apply the Keeling plot approach to the Beech Island data, but I wonder whether this is appropriate. You discuss several aspects, but assess the effect of different processes on the results mainly qualitatively (e.g. would go in a different direction). As you state the assumption of the Keeling plot model is that the bg is constant, but we know that this is not the case since OH strongly affects and processes CO during summer. You conclude from qualitative statements that the Keeling plot approach is still appropriate, but this is not substantiated by quantitative numbers.
An example: The application of the Keeling plot technique conceptually implies that the sample collected on August 8 with 87.9 ppb CO is a “near-background” sample and the sample on July 27 with 179.2 ppb is a contaminated sample and that the additional 90 ppb comes from the local contamination for which you calculate the isotope signature. Including the 87.9 ppb sample has a large effect on the source signature (you would get very different source signatures, outside the reported errors, if you left this one sample out), but does this sample represent the background well?
Can you make some assumptions on the seasonal changes in a conceptual model and assess the effect on the results (similar to conceptual model calculations in the Brenninkmeijer 1999 review, Fig 2)? Or can you try to quantify the effect by comparing to CO isotope measurements at other stations, e.g. Barbados, Montauk Pt, LI or Izana, similar to Mak et al., in their analysis of the Barbados data? Could this provide background values to better assess the total and VOC sources?
P18. ,l4: cite original references. They may also give uncertainties for Table 1.
Table 1: Why are the uncertainties for the OH reaction unknown? Please check original references, they should be given there. Where do the uncertainty estimates on the source signatures come from?
Table 2: Add statement that this table is for the Indianapolis measurements. Can you provide the equation that converts D14C to XCO2 in section 2.4?
Figure 4: Indicate which data were taken for summer and winter periods.
Figure 5: See arguments above on the suitability of the Keeling plot approach.