See attached pdf.

This manuscript reports on a valuable topic in the glaciological community, and I appreciate the care taken with the experimental design and execution. These are clearly difficult experiments, as indicated by the challenges in achieving similar results as past work, so these additional test with different methodologies are important to help the community move towards a common solution for quantifying grain growth.

The conclusions seem to be mostly well supported by the data, so I suggest only minimal adjustment to them, as indicated below. More significant to me is the treatment of n within the manuscript, and the relationship between the conclusions and the manuscript body. My comments here are about providing the strongest statements the data can support.

1. Many places in the manuscript and in the conclusions (e.g., lines 198, 209, 230, 231, 286, 294, 472, 473) n is described as varying. Yet Line 232 (similar to line 364) claims that "n can be regarded as relatively independent of both temperature and concentration". I find those statements incompatible, but perhaps I am missing something. In my limited time spent with these data, I do not see a systematic relationship, but I have difficulty seeing n as independent, particularly given how far outside of uncertainty the values lie according to Table 1. Possible resolutions are that the uncertainties are higher than reported or that the data aren't able to capture some additional complexity.

2. Throughout the manuscript, including in the conclusions, n is characterized as having an average value. I do not have a sense of how that average is calculated, and how much the experimental condition choices affect the determination of that average. Meaning, if 5 experiments had been run above the eutectic but still in the same temperature range, I expect the average n would be different. So it is hard for me to see how reliable the average is.

3. The manuscript spends significant time discussing the grain growth constant, k, yet no conclusion point relates to this value.

4. In conclusion point #4, I suggest separating the observations from the inferred cause (i.e., that soluble impurities enhance grain growth above the eutectic and the presence of a melt phase; similar comment for below the eutectic).

5. In Figures 6, 7, and 9, please describe how the best fits are calculated.