Thank you for your responses to the review comments. I now invite you to prepare a revuised manuscript based on the revisions you proposed in your responses.

Thank you for your revised manuscript. I agree that you have answered the issues raised by the reviewers, and the paper is essentially ready to be accepted. However I do have a couple of clarifications to ask you for, which make the revision just above the threshold of technical. Additionally I have a few typos (the paper will also need some corrections from a copy editor at Copenicus for the English).

Minor typos
Line 11 and 41 polar regions should be lower case (not Polar Regions)
Line 36, "aerosol blanket"
Line 48. CO2 and CH4 do vary naturally, so this statement is untrue. Maybe "are the two main recognised natural..."
Line 60 "volcanic events"
Line 108, better as (superscript)10Be.
Line 111, these coordinates are already given in line 78.
Line 122 "in the field"
Line 155 "neighbouring"
Line 186 "times"
You call Fig 4 here but I don't remember seeing Fig 3 called, please do so.
Line 196 South Pole (upper case)
Line 270. I don't know what you mean here by "punctual". Please re-word this sentence.


Line 23: "reduce by half". This depends on how many cores are drilled so does not make sense without this information. This also links to issues around line 300 (later). The point is that the standard deviation around the mean should not chnage however many cores one drills. But the estimate of the actual volcanic strength is based on the mean and the standard error of the mean reduces as square root of n. Thus the change from 56% to 41% is probably not very meaningful, but if the geometric SD is 41%, then with 5 cores the SE of the mean (ie the uncertainty in the flux) is 18%. Please discuss with a statistician as I may have misunderstood what you did but anyway the statement in line 23 definitely needs a context that it reduces with 5 cores.

Ariund line 156. A good test whether this outlier approach works is to state how many "troughs" are detected if you use an m-2 sigma approach. Of course you should have a few individual points that fall below 2 sigma, but if (as i suspect) you don't identify any signals in 3 consecutive points, then your outlier detection has worked well.

Fig 7. Please expand the caption, I don't think the reader can easily understand what this figure shows. In the 5 core case, 8 sigma (green hatching) I don't think the reader can tell whether 90% is the probability that the peak occurs in a single core, or the probability that it occurs in all 5 cores? i think it's the former, but then is it seen at 8 sigma in all 5 cores? This just needs explaining better what it is.

Supplement. I appreciate the sentiment to include these figures in response to the reviewer, but other readers will not be able to understand it, unless they read Gfeller. Please add text to explain at least qualitatively what Gfeller's spproach is. Also please call out both S figures from the main text, otherwise there is no reason anyone would ever see them.

Thank you for your work on this paper which i am now pleased to acept. It will need some work on the English in places but I leave that to the copy editor rather than asking for technical corrections.