This paper takes a very detailed look at bubble free layers in the Law Dome ice cores and their potential drivers. A particular care is given to the identification of the timing of tthe occurence of these layers, with respect to the season, which is essential for the correlation with climate variables like temperature, accumulation (or hiatus), and atmospheric circulation.

I am impressed with the level of details and care given in the paper. Not much more could have been done.

In the reading, I still struggle a bit with how things are presented, and in the section describing the exploration of mechanisms, some extra care could be put on describing hypotheses more clearly, and then, what data can be brought to support them or not.

A diagram/drawing showing the potential formation mechanisms would greatly help the reading of the paper.

These are mostly cosmetic comments, on what is an excellent, thorough paper.

I sugest minor revisions.

line 35: bubble number density and paleoclimate: cite Fegyveresi 2016:

https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2015PA002851

also potentially:

Spencer, M. K., R. B. Alley, and J. J. Fitzpatrick (2006), Developing a bubble number-density paleoclimatic indicator for glacier ice, J. Glaciol., 52(178), 358– 364, doi:10.3189/172756506781828638. 

Fegyveresi, J. M., R. B. Alley, M. K. Spencer, J. J. Fitzpatrick, E. J. Steig, J. W. C. White, J. R. McConnell, and K. C. Taylor (2011), Late-Holocene climate evolution at the WAIS Divide site, West Antarctica: Bubble number-density estimates, J. Glaciol., 57(204), 629– 638, doi:10.3189/002214311797409677. 

Melt layers: consider citing Keegan 2014: www.the-cryosphere.net/8/1801/2014/

doi:10.5194/tc-8-1801-2014

line 97: missing space before citation.

line 225: When you introduce the synoptic types, in section 2.5, you can include here the description of the two important types that you will use later, with a figure (potentially in the supplement if you feel you have too many figures) of the synoptic types, and their relationship with accumulation. 

In general, I think that a description of the relationship between synoptic type and accumulation (or hiatus) is missing in a quantitative sense. 

As I understand we could make a schematic:

synoptic type --causes --> accumulation or hiatus --causes--> bubble free layer.

In this causal chain, you would find a causal relationship between synoptic type and BFL, but you would not be learning anything more about their formation than when you were looking at accumulation/hiatus. 

Figure 5: I can't see the grey for layer 2

Figure 7: the aws data is too tiny, Can you make it a bit thicker?

parenthesis issue in caption

section 3.4

line 330 : you mean table 4

worth adding that som1 is asociated with high precip, or describing te significant modes a bit better

line 340+:trend: once you have analyzed the drivers, can you say something about the trends in the drivers themselves?  Would we be expecting any trend?

line 390: you have here a great comment about formation vs preservation of BFL.

I suggest that you restructure section 4 with the investigations of mechanisms, spelling them one by one, and detailing the supporting data/correlation, and the complicating factors: For instance, make a section on vapor flux upward condensing on the surface, as a formation mechanism, then show evidence for formation during snow-warmer-than-air times, during times of long surface exposure (your hiatus), seasonality, radiation, etc. And discuss evidence where it doesn't quite work. 

Repeat this effort for other mechanisms of formation and preservation, aided by a schematics, so that we can more clearly see where your thinking is.

line 419 to 435: not sure it belongs in the paper.

start of 4.4 to line 450: should go into the result section.

Section 4.4: Give more info on the relationship between SOMs and accumulation/hiatus, temperature inversions, wind scours, etc..

Appendix A1: ERA is overstimating snowfall by a factor of 2. Did you consider scaling ERA, and plotting on figure A1 the scaled ERA to highlight the temporal accuracy of hiatuses, or perhaps even plot the derivative.