Dear authors, I congratulate you to your very interesting work that the referees have attested to be scientifically sound. Thank you also for responding to their requests in detail. I have thoroughly reread the revised manuscript and there is still a (long) list of minor corrections that should be addressed before publication (see further below). There are also four important issues that should be adressed in a further revision process:

P10 L227-228 In the case of CO2, the dilution of oxygen due to 400 ppm of CO2 is significant, and larger than any direct spectral interference.

I don't think you give detailed information (in form of an equation) on how the analyzer obtains its result from the measured quantities. The CRDS methd (and you could recall the fundamental equation) provides direct access to the particle number density of O2 (knowing the spectral data). How is the mixing ratio X_O2 obtained ? In principle you could use p and T, but I think this is not precise enough and some calibration is employed. Please give an equation or a reference where one sees how measured quantities (after correction or not) give O2 mixing ratios. (This would also allow to understand where the dilution effect comes from). The same holds for the conversion into O2/N2: what is the assumption on N2 (or f_N2) made in your algorithm to convert from X_O2 to O2/N2. What are the roles of water and CO2 ? Your data show that this is crucial information that is yet missing. Since your paper is the first description of such a system, the information must be provided.

P21 L500++
Based on the evidence that you provide in this paragraph you should consider writing "The most likely cause for the drift is the optical amplifier ..." or "We believe that the optical amplifier has caused the drift ... instead of "A possible hypothesis for the cause of the drift can be an optical emplifier ..."

P21 L491++
The section describes tests of the systems' water correction function.
You should consider organisation/presentation of this paragraph. If I strongly summarize and exaggerate this paragraph, then the water correction works if there is no water and it does not if there is water in the samples.

Figures/units
Please use a consistent notation to indicate units in axis labels. Most of the time units are given in parantheses, but some figures use brackets (Figs. 2, 5, 8, 9, 10), some use a slash (Fig. 7), some don't use anything at all (Fig. 17), and some even mix brackets and parantheses (Fig. 1).

List of detailed suggestions/corrections

P1 L16-17: Change to : Here we present a detailed description ...
P2 L35-40: Please check grammar, also consider that the choice of the unit (per meg) and the choice of the measurement quantity (O2/N2) are two different things. It would be better first to introduce the quantity O2/N2 and explain why and then the corresponding unit (small relative deviation from a standard).
P2 L40+ Please give a conversion formula that converts between the different units per meg and ppm (could be in the appendix). This would enable to better understand your results (eg. correlation plots in appendix).
P3 L49-53: One gets confused about alphas and ORs in this paragraph. Please define. Use comma after OR. What is alpha ? Has it been defined before ?
P3 L52: give alpha_ocean for the decoupled process (alpha_o = 0)
P3 L59: ... several, mostly custom built techniques ...
P3 L69: monitoring(Note -> monitoring (note
P4 L87: If you capitalize for mentioning the acronym of a DFB laser, please also capitalize the B in feedback
P5 L96-98: A drawing of the optical setup is missing. Please state briefly how the wavelength monitor is integrated in the optics (beam splitter, ...).
P5 L98: It is a little bit strange to read : "the data acquisition system sweeps the laser frequency ...".
P6 L100 (P9 L 198, 202): Strictly speaking, 7878.805547 cm-1 is no frequency, remove "a frequency of"
P10 L221-224: It is not clear what you want say here. The H2O line clearly needs to be considered even if its line strength is only 10 to 20 % of the O2 line.
P10 L217-228: There seems to be little advantage in using AFGL notation. General understanding would greatly benefit from dropping this notation and call isotopologues by their names: water and deuterated water or by their isotope formula ^1H_2^16O etc.
P10 L218: I could not find traces of the heavy water isotopologues in the lower panel of Fig. 5. It seems that the figure has not been updated in the revised version, contrary to what has been indicated in the response to the reviewers. The caption also lacks the update.
P10 L236: AFGL abbreviation -> AFGL code; please see remarks in the beginning.
P11 L255-259: The description is misleading, because line profiles (area = 1) are not multiplied by amplitude, but by the line strength. Please phrase differently. Also what are the weak perturbing peaks ?
P11 L260: I don't understand the phrase "that constrained to be in fixed ratios to be in fixed ratios". Is there a "were" missing ? "ratio" should probably be replaced by "proportion".
P12 L259-260: You might look up the recent HITRAN update.
P12 L263: Please use here and at other instances the notation Filges et al. (2018) and not Filges et al. (Filges et al., 2018) when citing the author in the text.
P12 L252: The water lines ... change , into ; before rather.
P15 L322-324: check passive voice in "there is no advantage to be obtained" and the whole phrase. It does not sound correct.
P15 L346: Giving the absolute value of 5e-9 cm-1 does not help in the argument. 1. there are CRDS systems that measure lower absorptions. 2. You don't give an equivalent value for an absorption that is easily measurable (such as the 16O16O isotopologue).
P15 L333: Is much less precise: It should be sqrt(305/40) ~ 3 times less precise. Is this much less precise ?
P15 L337-339: The phrase "One set of tests aimed at determining the impact of pressure or temperature drifts or of uncobtrolled noise on the concentration measurements" ? can be much simplified to ease reading. What is uncontrolled noise ? This needs to be clarified.
P15 L353: Drift -> drift
P15 L356: insolated ?
P15 L358: regulator -> pressure regulator
P16 L358-360: and reducing the flow ... -> and with an additional orifice to reduce the flow to about 55 sccm.
P16 L369: Use the term "Allan-Werle plot" in combination with Allan variance
P16 L372: Define tau
P16 L377: times scale - time scales
P17 L384: available Oxzilla fuel cell -> available fuel cell
P17 L389: flow was adjusted to. It might be preferrable to use sccm as well.
P17 L391: flow out of the Oxzilla was ... -> flow out of the fuel cell analyzer was
P17 L389++ : The use of different flow units is confusing. Please stick to sccm.
P17 L397: It should be LabView instead of Lab VIEW. Could you specify the reason for using a labview program or is this not interesting ? Then drop this detail.
P17 L398: I know "a priori", but what does "in priori" mean ? A priori wouldn't make no sense here, however.
P19 L447+: "While similar correlation coefficients were observed for both analyzers, different slopes were calculated (Fig. A.1). This is due to the fact that the IRMS measures the O2 to N2 ratio (δ(O2/N2)) in per meg, while the CRDS and the Paramagnetic analyzers provide non-calibrated O2 mixing ratios in units of ppm and per meg, respectively." Is this true ? I cannot come up with the observed slope values when I try to use the definition of per meg and ppm.
P18 L425: fractions -> mole fractions (change here and elsewhere)
P18 L426: change and to or in "very low and very high O2".
P18 L426: standard 6 and 7 -> standards 6 and 7
P18 L427: standard 7 was not measured on the IRMS ... change O2 mixing ratios to singular
P18 L429: between -> measured with
P19 L432: lower in O2 -> lower
P20 L471: accounted -> accounted for
P20 L474: in similar pattern -> following similar patterns
P20 L477: The phrase repeats what has been said before. You could write. Interestingly the drift pattern could be modeled ...
P20 L475: repeated use of drift. Drop the first in the sentence
P22 L506: Based on these plots -> Based on this plot
P22 L506: I think that the essential features are reproduced but I don't think that there is very good agreement. Eg the NDIR shows an exponential response time when the CRDS is not.
P22 L509: measured in, not measured into
P22 L509 & P47 F15: The signals in panels a and b need to be explained. Where do the spikes come from and why does . Without such explanation it is impossible to follow your argument whether the agreement is good or not.
P22 L512: in-built -> built-in
P23 L545: "embedded within", check grammar
P24 L549: a low span -> low span ....
P24 L554+: Check phrase
P25 L555: It might be interesting to know which of the two devices was the first in the series.
P27 L634: no CO2 possible interference from band overlap -> it has no interference from a possible overlapping absorption from a CO2 band.
P28 L659: delete "if it appears"
P29 L677: "However" does not make a sense here.
P29 L680: What is "about a significant decrease" ?
P29 L684: This conclusion is not identical to what is presented in the results section, where CO2 and CO are excluded.
P40 F8 : C -> °C
P40 F8 : upitalic O2 -> upright O2
P41 F9 : mention tau in legend (otherwise it is undefined in the figure caption)
P46 F14: Please increase the size of the labels for CRDS and NDIR in the figure for readability
P46 F14+ later on. SI units for mole fractions are mmol/mol, μmol/mol etc. The atmospheric community often uses permil, ppm etc for the same purpose. It would be better to be coherent and convert the NDIR water data to ppm, if this is the preferred (non-SI) unit of the authors. Using different units for the same quantity is confusing. The offset of the NDIR data is worrysome as it implies negativ values, which are physically impossible.
P46 F14: please shortly mention the reason why the NDIR analyzer always seems to relax into the stable O2 value after a humidity change whereas the CRDS does this only in a few cases.
P47 F15 : Please shortly explain origin of spikes in panels a and b.
P48 F16 : It would probably be interesting to have a histogram plot of the lower panel or a line fit (slope = 0 or different) to these data.
P49 F17 : Figure is not in publishable quality. Axis labels on left hand side are cut. Also adapt label font to y- axis font (bold) and put units in parantheses.
P50 F18 : Time scales of the two time series are different. Please align upper trace on the scale of the lower trace.
P51 FA.1: Change "zooming only to standards 1-5" -> "selecting standards 1-5"
P51 FA.1: Convert all quantities into same units so that deviations from 1:1 correspondances can be spotted easily.
P51 L910: Please change section title into something more specific. It is presently not very informative.

Dear authors,

Thank you for having addressed all different requests in your revised version of your manuscript.

Please find below a list of minor remarks. I also would like to ask you to include an equation that gives your measurand (O2/N2 or p(O2)/p) as a function of measured quantities and calibration gases. This is crucial to evaluate the influence of different parameters on the overall measurement uncertainty and it also avoids misconceptions due to a purely "textual" explication.

Minor comments

L 34 : exponent for year.
L 35 : Suggestion : Write "Atmospheric O2 is commonly expressed in units of per meg due to its small variability with respect to a large background, where ..."
L 42 - 43: I don't think that one can understand the phrase "Note that we convert per meg to parts per million equivalent by multiplying per meg by 0.209500 (the O2 mole fraction of atmospheric air)." Did you eventually mean "Note that we convert delta (O2/N2) values (in per meg) to mole fractions (in ppm) using the conversion factor 0.209500 (the O2 mole fraction of atmospheric air)." ? This is misleading. If I have delta = 0 per meg using the atmospheric ratio as a standard, I should have a mixing ratio of 0.209500 ppm. If I follow your description and multiply 0 by 0.209500, I get 0. It is preferable that you give an equation that unambiguously converts between the (dry) air mixing ratio and the delta values. The dry air mixing ratio clearly involves noble gases, CO2 and other minor components, whereas delta (O2/N2) does not ... I don't see how one can easily convert by a fixed number.
Also, please provide a reference for your value of 0.209500.
L 104: measurement wavelength -> wavelength measurement.
P 10 : As mentioned previously, the AFGL codes 161 and 162 are used exactly twice in the whole document (apart from the definition). Given that a large part of AMT readership has no spectroscopic background it is preferrable to use the more commonly known notation H2O and HDO (or the IUPAC definition) or the words "normal and deuterated water" at these instances. It has to be admitted that the AFGL notation is very handy for all people used to it.
L 151+: Phrase appears twice.
L 257 : I could not find the word "centration" in the dictionary. Do you mean "line centre position", perhaps ?
L 260 : correct the exponent in wavenumber unit
L 305 : Not clear what is meant by "the correlation may be due in part to the fitting procedure itself". You might need to explain a little bit more, even though a reference is given. I suspect (but I am not sure) that you speak about the choice of variables for the fitting function (rather than the procedure or the algorithm). For example, the width and the line area are necessarily correlated, whereas width and absorption depth are likely less correlated.
L 267 : HITRAN gives a value gamma_H2O = 0.0514 cm^-1/atm, which seems to come from semi-emprical calculations : https://arxiv.org/abs/1906.01475 .
Please mention and check origin of that number.
L 311 : "because the O2/N2 and water concentration", add "ratio" after "O2/N2" and "the" before "water concentration"
L 329 : "normalizing absorption amplitudes to line widths does not provide any advantage". It is difficult to understand what is meant here. Could you please clarify further ? Do you mean normalising amplitudes by line widths ?
L 333 : Perhaps you should remove the word "Consequently" here. (As a reply to your comment on a similar and previous remark, please consider that the reader does not necessarily have Fig 1 at hand and in mind when comparing to the 16O2 absorption. Also, the noise level is not evident from that figure 1, as residuals usually reflect systematic effects such as bias in the profile, fringes, etc. some of which might scale with the size of the absorption line).
L 381 : write "tau is the averaging time" and use the greek symbol instead of the word "Tau"

Dear authors,

Thank you for the clarifications and corrections. I would like to ask you some minor corrections.

As it stands, the phrase "Note that we convert per meg to parts per million equivalent by multiplying per meg by
0.209500 (the O2 mole fraction of atmospheric air) (Machta and Hughes, 1970)." is still incorrect or misleading at least. (Please also note that the title of the Machta and Hughes paper in the reference list contains a typographical error.)

First, parts "per million equivalent" seems to be science slang and the phrase should be rewritten as to explain the conversion between the measurement quantities "mole fraction" (commonly expressed in micromole/mole or ppm) and delta (commonly given in per meg units). Just talking about unit conversion could be misleading as the quantities that are converted into each other are completely different (we are not talking about conversion from inch to cm, for example).

Second, the conversion factor is correct if we are concerned with relative changes, ie if delta changes by 1 per meg, then the O2 mole fraction changes by 0.2095 ppm. However, readers not familiar with O2 measurements or the delta notation will understand the sentence as 1 per meg corresponding to 0.2095 ppm and, correspondingly, that the mole fraction of O2 in air of ~ 0.2095 (or 209500 ppm) must correspond to 209500 per meg, which is obviously wrong. Since your article uses both quantities (mole fractions as well as delta values), the definition of quantities and units as well as the conversion formula should be given, preferably based on physical quantities with SI units. Note that AMT is a specialised journal dedicated to measurement techniques. Therefore, the correct definition of measurement quantities and their units is central and cannot be neglected.

Third, it is absolutely not trivial how the CRDS technology can measure delta (or O2/N2), because N2 is not measured (by that technology). The same holds for the mole fraction (O2/air). This is another reason why the measurand (absorption of O2) and its relation to the O2 mole fraction or the delta(O2/N2) should be clearly defined in the manuscript. Please explain the measurement process in more detail (what is measured, what is obtained from calibration and how).

Please understand that all this information must be given before the paper can be published.

With kind regards,
Christof Janssen

Dear authors,

Thank you for your corrections. Unfortunately, you haven't taken into account my remarks on the measurement principle and uncertainties.

Please take into account the following minor corrections.

1. Eq 5. Use a consistent notation for the mole fraction of gases, eg f_{O2}, f_{H2O} etc. ....

2. l. 321 It would be easier to understand if you use "multiplied" instead of "normalized".

3. The use of units still needs clarification. Since you use delta(O2/N2) in your paper, you should give an estimate of the error that is associated with the conversion from ppm to per meg. It is evident that the assumption of the N2 mole fraction being the same in the sample and reference is overly optimistic (l. 42 of your manuscript. As an aside note that it is not clear what you mean by N2 being constant. Partial pressure or mole fraction of N2 ?). There are natural variations of the partial pressures of CO2, Ar, H2O, etc. that require that if you maintain the same total pressure the mole fraction of N2 in sample and reference must differ at the ppm level, even if the amount of N2 in the atmosphere remains constant.
It will be very helpful for the reader if you make the calculation in an appendix.
As I understand it, CRDS can "directly" measure the O2 mole fraction using a calibrated air-like gas mixture (which should not be too dry, otherwise the water correction scheme does not work properly).
Conversion into delta(N2/O2) values, however, requires the knowledge of p(N2)/p(air), but from the measurement you know only p(O2)/p(air). Therefore you must make assumptions on Ar, CO2, H2O and all other substances that contribute to air at the ppm level. In this way you can calculate p(N2) = p(air)-p(O2) - p(all contributors at the ppm level).
What is the error of all these assumptions ? Please discuss/mention in the article. Why do you compare to the standards using delta(O2/N2) instead of f(O2) even if that introduces an additional error ?

Dear authors,

Thank you for your corrections.

I think the quantitative difference between O2 mole fractions and the O2/N2 ratio is clear for the general reader and there is no need for the proposed appendix. What is less clear and what yet seems to be an inconsistency in the article is the following:

The abstract characterises the instrument performance in terms of O2 mole fraction and the uncertainty is expressed correspondingly. However, the article implies that it is rather the O2/N2 ratio that needs to be measured. This can be done easily when the composition of air is known. But if the composition of air is not known exactly, the relative uncertainty of O2/N2 is likely different from that of the O2 mole fraction alone and the uncertainty cannot be determined as easily. Please include a discussion/quantification of the additional factors/contribution to the uncertainty of O2/N2 such that the article can be published.

With best regards,

Dear Authors,

Thank you again for your mail pointing out that the review process is too slow. I hope to be as clear as possible so that any further unnecessary step can be avoided.
Generally speaking, I strongly approve your work and the article. It should certainly be published. However, the current version is inaccurate or erroneous in describing the link between mixing ratios and delta values. This needs to be clarified before publication. There are two points:

1. A simple calculation using a dry three component model atmosphere of O2, N2 and others x = (Ar + ....) with relative abundances of 0.2094, 0.7809 and .0997 (TOHJIMA et al, Geophys Res Lett 30, 1653 (2003)) shows that the O2 mole fraction mf should change with a change dR in the R=O2/N2 ratio as d(mf) = (1+x)/(1 + R + x)^2 * d(R) (assuming that all change in O2/N2 comes from OS and not from N2). The slope in the delta = d(R)/R over d(mf) diagram is thus given by R (1 + R + x)^2 (R(1+x))^-1. Plugging in numbers yields 6.04, which is close but different (why ?) from your value of 5.78 per meg per 1 ppm. The number of 4.8 per meg per 1 ppm on the first page of the manuscript is just wrong. Please correct the introduction and explain the origin of the factor 5.78 or 6.04 per meg per 1 ppm that readers can understand. You should provide details in an appendix. The section that you have added during the last revision refers to the wrong number of 4.8 per meg ppm^-1 and is superfluous at this section in the manuscript. Note that the examples that you have provided in the proposed appendix to manuscript version 7 did not include changes of O2 and contained an error. Your examples were concerned with the addition/dilution of the O2 mixing ratio by gases other than O2 and N2. It is correctly stated that this does not change the O2/N2 ratio, but changes O2 and N2 mole fractions. However in the ongoing calculation you then assume that the mole fraction of N2 remains constant while the mole fraction of O2 gets diluted. This certainly impacts the ratio, but the approach is self-contradictory and leads to the calculation of a meaningless number. Note that dilution cannot be used to determine a delta over d(mf) slope. When you calculate the change of the O2 mole fraction as a function of the dilution dx analogously to the above approach, you get d(mf) = 1/(1 + R + x)^2 * d(x). However, since R does not depend on x, dR/dx=0. So the line traced by dilution by gases other than O2 or N2 should be horizontal (slope of 0 per meg per 1 ppm).

2. The paper lacks a dedicated statement on uncertainties related to the measurement of delta. All of the uncertainty discussion is concerned with mole fraction measurements, but since results need to be reported as delta values, the uncertainty of delta cannot be omitted. Using the above derived transfer coefficient from O2 mole fraction to delta, one can convert a mole fraction uncertainty of 5 ppm into an equivalent value for delta. However, there are additional factors that impact delta (and don't have an effect on the O2 mole fraction), i.e. variability and lack of knowledge of the mole fraction of non-characterized atmospheric compounds. This is why the discussion needs to be extended. Please Mention these additional error sources and give a rough estimate of their contribution so that the reader can judge the degree of agreement with other measurement methods.

Once these two issues are clarified, the paper can be published. Note that these issues have been raised in previous editor remarks, but have not been addressed. I am sorry that they have not been raised at or before the expert review stage, but only came up later in the review process.

Please take the time to carefully address the two above points so that any further revision becomes unnecessary. If the points are not addressed satisfactorily, I will send the revised version to one or more referees, whom I will ask to give a rapid and independent expert opinion on the issue.

With kind regards,

Christof Janssen

Dear Authors,

Thank you very much for your response. I appreciate the level of detail that you dedicate to the discussion of the conversion from O2 mole fraction to delta(O2/N2). I fully agree with your suggestion of modifying the introduction and to move the detailed discussion on the relation between O2 mole fraction and delta(O2/N2) to the Appendix. The first two paragraphs are very comprehensible. You could state before "Following equation 1 in Thojima ..." that no unique conversion factor exists if N2 or the other air components are not known.

Unfortunately, paragraphs 3 and 4 of the analysis and table A1 in the Appendix still lead to confusion and the two paragraphs need to be corrected/rewritten. I cite:

"Note that under the assumption the atmospheric N2 content is constant (i.e. N2sample equals N2reference), we convert relative changes in oxygen given in per meg following equation 1 to oxygen changes in parts per million (equivalent to micromol/mol) by multiplying by the O2 mole fraction (O2reference) expressed as 209500 ppm (the O2 mole fraction of atmospheric air) (Machta and Hughes, 1970). Hence 1 ppm corresponds approximately to 4.8 per meg, or 1 per meg to 1/4.8 (209500/106) ppm.
This is used in our approach since the Picarro ... in the following Table A1"

The above text seems to indicate that in your calculation you consider changes to O2 alone (thus keeping N2, CO2, Ar, ... const). If done correctly, this corresponds to case 3 in your table A1 (ΔO2 only), which indicates that the conversion factor is 6.04 per meg /ppm, but you derive 4.8 per meg /ppm from your calculation ! It is evident that there is an error or an undocumented approximation somewhere in your calculation. This contradiction needs to be resolved.

I might be wrong, but from your arguing, you take the value of 4.77 per meg / ppm as an ad hoc conversion factor somewhere in the range of values that might be observed (between 0 and 6.11 per meg / ppm). If this is true, you should present it in exactly this way. If this is not the case, please give another motivation that does not contradict the results in your Table A1.

Finally you mention the N2 mole fractions of your standards but you don't give them in Table 1. However, this seems to be very important information. With N2, O2, CO2 given, using reasonable Ar values and a measured data for H2O, much of the uncertainty in the conversion from mole fraction to delta should be very small. Please add these data if they are available.

Please revise your manuscript accordingly and
- avoid the term concentration as it is not clear whether you talk about particle number density or mole fraction (which you obtain from the CRDS)
- avoid talking about ppm to per meg conversion if you mean mole fraction to delta conversion.

Minor corrections/suggestions

Following equation 1 in Thojima et al., 2000, the per meg to ppm conversion has a slope of 6.04 if only a change in oxygen is applied as seen in table A1.
->
Following equation 1 in Thojima et al., 2000, the per meg to ppm conversion has a slope of 6.04 if a change only in oxygen is applied as seen in table A1.

The fact that our supplementary plot shows slopes of 5.78 (for the first five standards) or 6.08 (for all standards except ST-7) is due to a mixed influence dilution effects.
->
The fact that our supplementary plot shows slopes of 5.78 (for the first five standards) or 6.08 per meg per ppm (for all standards except ST-7) is due to a mixed influence of dilution effects.

The lower slope of 5.78 documents particularly the influence of the CO2 dilution effect.
->
The lower slope of 5.78 per meg per ppm documents particularly the influence of the CO2 dilution effect.

For example a 10 ppm increase in CO2 lead to an incorrect value of -10 per meg in δO2/N2,est compared to the true δO2/N2,true value.
->
For example a 10 ppm increase in CO2 leads to a bias of -10 per meg in δO2/N2,est compared to the true δO2/N2,true value.

As you can see from table A1 an addition of N2 of 10 ppm leads to a reduced and opposite effect for the difference in δO2/N2 (true – est) because the dilution effect on O2 cannot compensate the change from the increase in nitrogen, therefore it scales with -10 ppm x (oxygen mole fraction/nitrogen mole fraction).
->
As can be seen from table A1, addition of 10 ppm of N2 leads to a reduced and opposite effect for the difference in δO2/N2 (true – est) because the dilution effect on O2 cannot compensate the change from the increase in nitrogen.

In our case this is given – but can certainly be improved – since we are comparing air composition to air standard compositions.
->
In our case this is given – but can certainly be improved – since we are comparing natural air to air standards.

Yet, determinations of the standards that has been used in this study have a range in N2 concentrations of -110 to +110 ppm for the ST-1 to ST-5, whereas ST-6 (+700 ppm) and ST-8 (-6200 ppm) are significantly off compared to our primary standard used for mass spectrometric determination.
->
Yet, determinations of the standards that have been used in this study have a range in N2 concentrations of -110 to +110 ppm for the ST-1 to ST-5, whereas ST-6 (+700 ppm) and ST-8 (-6200 ppm) are significantly off compared to our primary standard used for mass spectrometric determination.

Table A1: ppm to per meg conversion calculations for air-like compositions
->
Table A1: mole fraction to delta conversion for air-like gas compositions according to different scenarios

δO2/N2,true: for instance measured by mass spectrometry; δO2/N2,est: for instance measured by 988 Picarro G-2207.
->
δO2/N2,true: calculated (eg Thojima et al., 2000); δO2/N2,est: use fixed conversion factor of 4.77 per meg / ppm

ΔO2 in line 4 of your Table A1 should be ΔN2

Dear Authors,

Thank you for the revised version of your manuscript which clarifies the raised issues. I therefore congratulate you to your very interesting paper that can be published subject to technical corrections.

To eliminate eventual ambiguities, I propose to delete
- "following equation 1" in the phrase "Note that under the assumption the atmospheric N2 content ..."
- the three phrases from "Since no information about N2 is available ...." to "correspond to the true dO2norm/N2norm." (because all these quantities have not been defined yet and potentially create more confusion than they explain)
- "the following" in the phrase beginning with "Indeed this dilution effects"

You should also replace "this" by "these" in the same phrase and "addition of N2 of 10 ppm" by "addition of 10 ppm N2" at a later instance.

With kind regards
Christof JANSSEN