While the paper looks good on the whole, there were certain 'major' comments from both referees that I do not see as being adequately addressed. I would like to see these tackled properly before the paper goes to publication.

Firstly, regarding referee #1's comment concerning anthropogenic influence, I note that the authors have rebutted this point, however I see that as a sufficiently important issue that this should be addressed in the manuscript itself. I would ask that the authors amend the text to reflect this.

Second, referee #2's points 1 and 2 about the metrics being reported has not been addressed, which they regard as being very important (see their comments). However, on reading the rebuttal and the paper referenced by the referee, it seems to me that the authors may have misunderstood what the referee was asking for. The metrics in question are defined in the paper Kulmala et al. (2012) doi:10.1038/nprot.2012.091, where 'GR' is defined as 'growth rate' and 'CS' as 'condensation sink'. Furthermore, the term 'formation rate' is applied to the rate of particle number formation at a defined size, not the growth rate referred to in section 3.2 and in the rebuttal.

In light of this clarification, can the authors generate these metrics and comparisons, as requested? I would expect that this would be most appropriate as an extension to the DGR approach. As well as ensuring comparability with other works, this will also allow the dataset to be more accessible for any future meta-analyses. Even if these comparisons turn up null results, this would still be an important observation to report in its own right. If the authors still feel that generating these statistics would be inappropriate, they are still entitled to make that argument, but such an argument should probably be reflected in the text of the paper, so as not to leave it open to accusations of deliberate omission. I should state however that my personal inclination is to err on the side of generating these statistics unless there is a technical reason that prevents this.

Finally, I think the point referee #2 had in their point 3 was not how the PCT events had been defined, but that there was no comparison with similar results in the literature. I think whether these the observations defined this way definitely represent NPF is likely to be a moot point, however a paper of this nature should be placed in the context of other works where possible. A quick comparison with other works would be welcome.

As a separate issue concerning referee #1's point regarding trajectory accuracy, I do not regard the issue of back trajectory validation to be one that is easily solved. Because the arctic region is so data-poor, it is highly likely that the wind field used has assimilated (and placed great weighting on) local weather observations. As a result, good agreement here is not surprising but does not necessarily validate the fidelity of the model field away from the observations. I think it should be sufficient to state that there are inherent uncertainties with this technique.

I thank the reviewers for taking the comments on board and calculating the formation rates in a very timely fashioin. However, I would remind them that the referee twice very specifically asked that the formation rates be correlated with the CS/GR ratio, rather than simply comparing Jx percentiles with GR. I still stand by my decision that this should be attempted and given that now the Jx data has been generated, I would expect that this should be fairly trivial to do. This is important because this correlation pertains to possible underlying mechanisms behind NPF and growth and as I have explained, even a lack of correlation would still be reportable within the context of the paper. I would forsee that two scatter plots would be generated and discussed, one for events where J11 is available and one for J22.

Note that in the event that the plots are inconclusive and you consider their inclusion superfluous to the core content of the paper (and associated publication charges), these can very easily be included as supplementary material.

The manuscript now looks close to being done. I have given it one last read and identified the following issues that should be resolved prior to publication, mainly concerning the referencing of technical detail and general presentation. I have classed these corrections as 'technical', so it should not have to be reviewed again.

Line 56 & 154 (and possibly elsewhere): Writing as '1970s', '1980s' and '1990s' would be regarded as more conventional English.
Page 8: Please add technical references for the DMA geometries and EUSAAR inlet specification (as distinct from to the overall EUSAAR project), if available.
Line 182: Please refer to "10 nm" rather than "ten nanometres".
Line 215: Please provide a direct link for the sampling protocol, if available.
Line 375: Please provide details of the specific NSIDC data product used.
Figure 4: What are the units of the bottom axis? Reposition the colour scale so that it does not obscure the bottom axis label (possibly related to the previous point). Also, specify what base of logarithm is used and what 'relative number concentration' means in the caption.
Figure 8: For consistency, I would suggest plotting all the sub-figures on the same colour scale range. However, if the authors feel that some detail would be lost, feel free to leave this as is.
Supplement: Please add a title, author list and affiliations as a header for the supplement.