The manuscript reports methodological contributions in three parts:

1. collection and purification of CH₄ out of air, down to ambient concentrations, out of volumes up to 1100 L.
2. Measurement of δ¹³C, δD, ∆¹³CH₃D, and ∆¹²CH₂D₂ by high resolution isotope ratio mass spectrometry on the Thermo MAT253 Ultra
3. Validation, standardization, long-term repeatability, and calibration of a temperature scale with internal measurements.

Then, it contributes to the experimental literature with the measurement of clumped isotopologues in CH₄ in atmospheric air, and a simple box model interpretation of these results and their implications.

In all these categories, the manuscript represents a very valuable contribution to the literature. In particular the thoroughness of the methodological description will be valuable to many future investigators in the field. And the measurement of atmospheric samples reported here, represents an early contribution to what will be an important global dataset, measured by various techniques and labs, of the atmospheric CH₄ clumped anomaly, putting source and sink estimates to use to constrain budgets of this important greenhouse gas. In light of its thoroughness and significance, this manuscript should be accepted, with some minor revisions as outlined below.

Concrete comments:

(1) The measurement of atmospheric samples that forms the culmination of this manuscript should be highlighted more specifically in the abstract and the introduction.

(2) The authors should ensure that the methods part of the manuscript is written in a way suitable for a scientific publication, as opposed to an internal protocol. The level of detail is very welcome, but care should be taken, for instance, that terms such as silica gel are written consistently (and not as Si-gel) and common abbreviations like LN2 are defined at their first mention.

(3) Not all readers are necessarily conversant in the expression of gas amounts as mL at STP, so the assumptions of this nomenclature should be introduced, and perhaps a conversion to molar units given.

Line-by-line comments

97: the use of ‘-‘ in this sentence, where it could also be mistaken for a negative sign, is confusing, as is the use of both ‘and’ and ‘&’ to divide parts of the sentence. Consider rephrasing.

123: The ion currents of the rare isotopologues are indeed the relevant parameter for what precision is achievable, but so early in the manuscript would benefit from some more explanation of what controls these numbers.

208: it is not clear to me from this sentence where this additional aperture is, and how it compares to the exit slit at detector H4 mentioned in line 225

228: avoid use of ‘etc’ as far as possible—if there are any other important contaminants, it is better to be specific.

248: Incorporating these measurement times in Table 1 could be useful to the reader

350: In Fig. 2, it is not clear to me what darker and lighter connecting segments signify.

406: the phrase ‘larger sample size of the bulk air’ is ambiguous to me

410: what are the other consequences of running GC columns at 40°C instead of 50°C? Why could this temperature not be used for all samples?

438: are there no issues with ice clogging the first glass trap when water is frozen out of air at -70°C?

440: has the need for two RDTs been quantfied, or is this just a factor of safety?

451: is PS4 the gauge labeled ‘P4’ in the drawing?

515: “dotted”

556: clarify that the heated gas is a subsample of AP613, which is  at present ambiguous

591: why would high count rates on a CDD lead to divergence from the expected error estimate?

655: in Fig 7 it is ambiguous whether AP613 or AP613 that has been equilibrated at UU was measured on the UMD Panorama.

This paper presents the development of 1) a measurement method for methane isotopologues including isotopologues bearing 2 heavy isotopes using a high-resolution mass spectrometer and its validation, 2) the collection and purification of CH4 from samples at a range of concentrations. The methods are then applied to perform measurements of CH4 in atmospheric air. Both the technical developments, which are very well detailed, and the natural sample measurements are valuable contributions to a growing field. I have a number of comments and suggestions, but they are altogether of limited impact on the paper, and I recommend acceptance of the manuscript with minor revisions.

Comments:

The atmospheric CH4 measurements should be mentioned in the abstract! It is one of the very first study to do so, and this should be given more visibility. This is made all the more important by the contrast with the expected clumped values from modelling, which, if confirmed as not a local effect (see next comment), will challenge assumptions about the understanding of the methane budget and motivate further investigation.

The results on atmospheric CH4 deserve more discussion. The bulk isotope values measured for the three samples of Utrecht air cover a 2 permil range in d13C and a 30 permil range in dD. The range in ?13CDH3 is also larger than the analytical uncertainty (but with no correlation to the bulk values). This should be commented on or discussed, especially since the values are compared to the modelled average atmospheric values. Are those ranges meaningful or not? Are there local sources or differing wind patterns that could cause such variations in Utrecht air CH4? This can potentially inform sampling strategies.

There is no mention of possible tailing from the adduct 13CH5 on the measurement position chosen for 12CD2H2, despite the very close proximity. Previous studies using an Ultra, albeit in HR, not HR+ mode, have had to include an ion correction. If the problem is absent in HR+ mode for your instrument, it would be useful to state and show it explicitly. Fig 3(d) shows the detail of the relevant peak shape, but it is not evident by eye that the contribution from the adduct tailing is negligible.

Line by line minor comments/typos/etc:

l45 missing () around Li et al.

l56-59 no order to the references either chronological or alphabetical? this is present in other parts of the manuscript (e.g., l73-74, l78). Usually, the publication's guide style will suggest one or the other.

l66-67 it would be good to include a supporting citation

l63-68 presents three groups for sources, but the figure 1a that supports the following discussion of the sources contains 4 groups (which are more for accumulations or seeps of methane from the crust than for the atmosphere).

l96-97 the sentence is a bit tautological: clumping anomaly is a measure of clumping of heavy isotopes. It could be better for the non-familiar reader to replace "clumping" in the second part. I would suggest signature or value rather than anomaly in general. Deviation from the stochastic may be better (especially for methane where we see positive and negative clumping commonly).

l112 m/z is incorrect here as you are not speaking about the ions but the molecules.

l123: this is very important, but the number of counts is not illustrative for someone not very familiar with the issue, especially if not compared to the count rate on the major beam. Consider giving the abundance ratios of the isotopologues instead?

l125 and in the rest of the manuscript: mL is given for the required size of the sample. If it is mL STP, this should be written explicitly (at least the first time in the manuscript).

l137: the sources for figure 1b are not referenced here or in the caption and encompass more than the ones at l136.

l141: Yeung 2016 GCA paper could be referenced as well here

l142 repetition of l136-137c

l153 missing "a" between require and fewer

l181-183: the long form explanation of "stochastic" would be more useful in the introduction where it was used on its own.

l208: the aperture trims the beam in the Y-dimension while the slits do that in the X-dimension. The sentence may lead the reader to think that the aperture is located close to the other slits which is not correct.

l223: resolution and mass resolving power are used interchangeably. It could be useful to give a reminder of how the mass resolving power is calculated

Table 1: measurement times for each configuration would be useful in this table too.

l251: what motivates measuring alternatively and not each in one block? flexibility for differences in sample sizes, countering potential drift, or some other reason?

l278: I do not understand the sentence. The adjustment is checked on the signal intensity, not the pressure, so should the tolerance not refer to the signal?

l315: the procedure described is for the ?-Al2O3 in Eldridge, they used Ni on silica-alumina rather than Pt for the high temperature equilibration. Do you mean that the Pt catalyst was also activated with the same procedure?

l366: it's the measurement or the determination that takes 3mL, not the composition(s)

Fig2: mention in the caption that acronyms are in the main text?

l383: avoid LN2 or define it first. It is a technical rather than scientific term (which would be N2, l).

l389 should be Å rather than A°

l395-396: apparent contradiction for the elution times of N2, I'm guessing 22 and 35 are the start and end of its elution?

l409: strictly speaking most samples will have Kr, but it is only for atmospheric samples that its amount is of the same magnitude as CH4's.

l410: any reason not to use 40C for all runs? the elution time for CH4 seems to be just 5 minutes later from figure 9.

l436: would the system be able to detect a CH4 breakthrough or are there reasons to think this is unlikely to occur?

l450: "is" should be "has been"

l451: PS4 should be P4

l476: for the average reader it would be useful to define or reference what is meant by zero air.

l501: ""1s standard error" do you mean that when you write 1s it is one standard error (rather than one standard deviation?). If so, it should read 1s (standard error). Same later in the manuscript.

l515 doted should be dotted

l556: AP613 is described as a source of gas here, rather than as a gas before in the manuscript. Is it because here you mean that it was used as the source to produce the equilibrated aliquots? This seems the case from l568, please rephrase.

l590-591: I think most readers are going to be confused about why higher count rates lead to more deviation from the shot noise limit, it would be better to develop here. Eldridge et al 2019 used data filtering because the noise was not gaussian, did you need to use anything similar? if not, do you have hypotheses other than not being perfectly on the flat top for the extra noise?

l592-593: but the calculation of the ? values requires a really precise dD, which is the main motivation for the long measurement of 12DH3/12CH4 here compared to conventional.

l620-624: the structure is a bit confusing at the first read, please rephrase. Ultimately it is also just a statement that your 0-enrichment gives you the expected result on average.

Fig7: the legend is a bit confusing; it would be better to remove "heating experiment" here I think

l695-696: it would be useful to know which parts of the system were likely to cause large offsets

l715: some repetition of Section 2's text around this line.

l742: it is the (temperature-induced) signal to noise ratio that is critical here rather than just the measurement precision.

l769: depending on the font used it can be hard to differentiate - and —, consider using from X to Y instead of —?

Fig 11 I have not found a complete table of the sample compositions plotted in fig11. It would be good to at least have them in supplementary material.