The manuscript focuses on the Holocene section of the South Pole ice core, in particular on dust concentrations and size distributions. The time series of associated variables are presented and interpreted in combination with other proxy records, as well as based on indicators of potential transport of dust, through analyses on back-trajectories in modern times, and analysis of some variables from CMIP6 and PMIP4 simulations with global climate models.

In my opinion, the data presented here is very interesting; however, there are two main issues that need to be addressed. The first issue that warrants attention is the lack of sufficient clarity/details/justifications in some passages. The second issue is the limited comparability of some of the metrics used here with those typically presented in the literature; I think that the authors should make an effort to include more mass/volume-based metrics, at least in the supplement, to enhance comparability with previous work.

Specific comments

68. Which subset? What were the criteria for selection? Say something briefly also in the main text.

Text S1, p2. How are all the bins defined? You report “additional” bins by numbers … are they bin centers or lower/upper edges? How are those bins spaced?

Text S1, p2. “The CFA system has an effective resolution of 3 mm with a dispersion signal of 1

cm”. What are the implications?

72. “we do not use metrics …” I suppose you refer only/specifically to the calculations of CPP, since your section 3.1 is called “Holocene dust flux”. Please clarify. 

74. You should probably specify “we use particle number concentration” 

75. Remove “<” before “5.1”

Text S1, p2. “We use a spherical particle shape assumption to facilitate comparison to other

Antarctica ice core particle records that include a volume assumption (i.e., flux)”. You could compare your “best” fluxes, i.e. assuming spheroids, with Coulter Counter based fluxes (and for calculating CPP). Those measurements are deemed accurate with respect to the actual volume of the particles, without the need of any assumption on their shape. Using spherical equivalent diameters to identify their sizes does not conflict with this notion. In fact, I think you should add in the main text a description of how you calculate dust mass flux, and move Figure S1 to the main text. The shape is not the only possible source of uncertainty, as there might be an unconstrained bias, in terms of a systematic underestimation of dust fluxes, due to the limited size range of the laser counter; the comparisons with coulter counter measurements in Chesler et al. (2023) over the same size range do not allow a full evaluation this aspect. Also, you could compare different assumptions on particles shape (spheres and spheroids) to calculate mass-based FPP/CPP for the South Pole record and allow an easier comparison to previous published work using the same metric. At least this should be added as part of the supplement.

78-80. Please provide a little more information on how you set up HYSPLIT runs, e.g. the starting vertical level of back trajectories, etc.

83. “pre-industrial (1850 – 2014 AD)”: this not pre-industrial. The indicated period suggests that you used the output from historical CMIP6 experiments, rather than the pre-industrial equilibrium control experiments with 1850 CE boundary conditions. Please clarify, also by adding the experiments IDs, and add references for both periods referring to the CIMP6 and PMIP4 experimental setup (or synthesis of the main results) publications, e.g. for the mid-Holocene: Otto-Bliesner et al., 2017, https://doi.org/10.5194/gmd-10-3979-2017 (or Brierley et al., 2020, https://doi.org/10.5194/cp-16-1847-2020 ). In fact, you should also justify why you chose the historical rather than the (more logic option) piControl experiments.

84. Do you mean “We replace all positive values, …” with ones? 

Table 1. I believe it’s NCAR - CESM2

95-98. I would still prefer to read “number concentrations” for the sake of immediate clarity to the readers

97. “the timing of relative change within each metric is relatively similar”: I do not understand this sentence, please clarify.

97-98. What I see is a generalized inverse relationship between number concentrations and CPP along the entire record, which breaks between ~2,000 and ~4,500 Years BP, when the two records run almost parallel to each other. The phrasing of the description here should be more accurate. In addition, this feature should be discussed when positing a significant relationship between d18O and dust in the following section, i.e. at lines 112-114. Perhaps, you could use a running window approach for calculating correlations.

Figure 3. “No increase” should read “All decrease”, and a symmetrically diverging red-white-blue palette might be more effective at depicting the actual situation.

156-160. Those two sentences are unclear. Please rephrase.

178-179. Not surprisingly you see a spatially consistent warming in all model simulation: you are comparing, apparently, mid-Holocene simulations with the average of historical simulations, which include the response to increased greenhouse gases emissions up to the present-day.

Data. You should acknowledge the HYSPLIT website, World Climate Research Programme’s Working Group on Coupled Modelling, which is responsible for CMIP, the Earth System Grid Federation, as well as indicate the specific ESGF node and access date of your downloads for CMIP6/PMIP4 model output data. 

In this manuscript Chesler and colleagues present a new dust particle record from the South Pole ice core. The data is very interesting and worthy of publication in CP. However, their interpretation of the data is in my opinion incomplete and should be revised in light of the comments below. At this stage, I recommend major revisions.

Major Comments:

I think the fact that CPP rises because the fine fraction drops faster than the coarse fraction is an important result. This could easily be shown in an additional figure, or even within Figure 1 using a stacked plot instead of the current good looking but not so informative shaded areas. Within that context, I don’t think the authors have made a strong case for a strengthening of the SHWW. If that were the case, one would expect increasing coarse particle concentrations. But this is not the case, CPP increases only because the coarse fraction decreases less than the fine fraction. Basing the conclusions on the CPP alone may therefore be deceptive. Similarly, a southward shift of the SHWW should increase the local (Antarctic) contribution. In line 180, the authors write “A southward shift in the core of the SHWW can simultaneously explain the opposing trends in CPP and total particle concentration …”, but do not proceed to explain this in actual words. I think the authors should take a good look at Lamy et al., 2010 Nature Geoscience, and rethink their interpretation of the data, looking more at the absolute concentration trends and less at the relative CPP. This paper potentially offers a framework that may better explain the trends apparent in the absolute data, and which is still consistent with the studies cited in lines 183-196.

The authors attribute the 3-5 um coarse fraction of their measured dust particles to local, and the 1-3 fine fraction to remote sources. However, recent research has shown that coarser particles can travel a very long way and should not be excluded from remote sources (see recent papers by Jasper Kok and Yemi Adebiyi). That interpretation should therefore be revisited taking into account the recent development in coarse particle transport research.

Minor Comments:

Abstract: Not sure if the guidelines for CP abstracts have changed recently, but if allowed, it would be good to add 1-2 sentences after the first sentence about the state-of-the-art of SHWW changes through the Holocene. Also, I recommend against elevating the 20 particles /mL /kyr average number to the abstract. This will be picked up by many people and used as is. The real behavior shown in Figure 1 and discussed by the authors is significantly different than this average value.

Lines 68-70: Please indicate the time resolution of the resulting dataset.

Lines 72-76: This first sentence doesn’t make much sense (particle shape does not cover the Holocene?). And the second sentence doesn’t follow from the first one (therefore,…). Please clarify these two sentences. Also, the choice of 2-3 um for the FPP or CPP cutoff in the Delmonte papers is based on the size-distributions shown in Fig. 3 of Delmonte et al., 2002 (Climate Dynamics). To justify the same cutoff (and generally, for good measure), it would be good to show the size distribution from this core as well.

Line 83: 1850 – 2014 AD is the historical simulation, not PI. Did you compare to PI or to historical?

Line 84-85: That method is very confusing and badly described. What were the positive values replaced with? Positive differences mean that pi values are higher than MH, not a change in time. If all negative values are replaced by zero then how can you assess if there is an increase/decrease in a variable? Please clarify.

Table 1: Add length of simulation used for pi and mh.

Line 95: Here would be a good place for a size-distribution graph. In particular to indicate how much of the dust is finer than 1.1 um (even if it’s not actively measured in the Klotz device) and how much is coarser than 5.1 um. Incidentally, naming the 1-5 um fraction “fine” dust is a bit confusing when the authors name the 3-5 um fraction “coarse”. Below, the authors use mostly the word “dust” (like dust flux), so it would probably be best to just define here that dust means the 1.1-5.1 fraction for the rest of the document.

Lines 153-154: I don’t understand these grids very well. Could you show them in Figure 2?

Line 159: Fluctuations are not advected. Please rephrase

Figure 3: An aesthetic suggestion: If you copy-paste your data from 0 deg longitude to an additional column at 360 deg longitude, you avoid the white slice between 358 and 0 deg in the figure.