Reviewer comment by Gabor Vali on "Aggregation of ice-nucleating macromolecules from Betula pendula pollen determines ice nucleation efficiency." by Florian Wieland, Nadine Bothen, Ralph Schwidetzky, Teresa M. Seifried, Paul Bieber, Ulrich Pöschl, Konrad Meister, Mischa Bonn, Janine Fröhlich-Nowoisky and Hinrich Grothe.

The paper presents a rather extensive set of measurements and interpretations aimed at obtaining better understanding of the ice nucleating ability of birch pollen. The topic is of interest both from the point of view of basic nucleation concepts and as a potential contribution of the pollen to atmospheric cloud processes.

The ice nucleating ability of macromolecules from birch pollen has been the subject of various prior publications over the past two decades with a focus on freezing temperatures near -15℃. Some activity was found at higher temperatures as well and that diversity prompted studies to identify its causes. That's the main focus of the current paper. Activity at temperatures around -9℃ is shown more definitely than previous work. The authors conclude that aggregates of macromolecules may be responsible for this.

The main feature of the empirical work is the use of two different instruments with differing sample volumes and thus obtain freezing nucleus spectra over a broad temperature range. While data over this large temperature range could also have been obtained by successive dilutions of the sample and fusion only the instrument with larger sample volumes, the use of two devices was no doubt a practical help, and the agreement in the results from the two techniques adds further confidence to the results. The authors overstate to some extent that the use of two devices was crucial.

As mentioned above, the results of the freezing assays are reassuringly solid. The spectra shown have relatively little scatter. While the authors indicate (lines183-188 and in the Supplement) that variations results with time, sample origin and other factors, they do not indicate what conditions were in place for the results shown in the different graphs.

The basis for discussing multimodal freezing activity is the presence of three areas of steep slopes in the observed cumulative spectra, resolved into three components of the differential spectrum via the HUB method (line 196). The advantages and disadvantages of this approach, compared to direct calculation of the differential spectrum from observed freezing temperatures of individual drops are debatable, but are not essential for the present topic. The intermediate leveling of the spectrum is clear evidence for separating activity into different temperature regions. The authors call this A, B and C class nucleating agents. By subjective examination of Fig 1, classes A and C seem to have stronger evidence than B. Uncertainties attached to the resolved spectra are not specified in the paper.

The principal focus of the paper, as indicated by the title and tb he explanations offered in the Conclusions is the notion that aggregates of smaller units are responsible for activity found at temperatures higher than about -15℃. Why that process would lead to activity at distinct temperatures (spectral peaks) is not discussed. To this reviewer, that question is so fundamental. Perhaps the preservation of the spectral shape (Fig. 5) is the main justification for discrete sizes and discrete regions of activity. If so, that point should be made emphatically, beyond the statement on line 356. Better yet, the point should be made explicit by plotting differential spectra for the samples with different filter cutoffs. However, the sample to sample variations and the small number of freezing events for class A probably make such analysis impractical. Therefore, the issue seem to have no clear answer in the data so far obtained. Perhaps the authors do have other arguments for why there are preferred sizes or shapes of aggregates which correlate with the preferred nucleation temperatures; if so the readers need to find it in the paper. Neither of the four possible explanations offers clarity on this. Neither does the widely used assumption that larger sizes of a nucleating substrate offer greater chances of finding sites of given activity (a theory not entertained by the authors) explain the discreteness observed. That notion, coupled with assumed frequency distributions of sizes would explain the results for any one peak but is hard to separate from the effects of random size fluctuations of embryos.

Results presented on the effects of freeze-drying , heat treatment, and of freeze-thaw cycles are good material for assembling evidence on the chemical composition of the pollen INMs, but do not constitute significant additions to the main point of paper.

A comment on writing style: Section 3 is titles "Results and discussion" and the content follows that double intention. The mixed presentation of measurement results, and discussions of their possible implications can be helpful in digesting the significance of data, but it is carried a little too far in this paper. Conclusions are pre-announced, references to other work are not restricted to facts but also to speculations, and in general much is included in the Results section that should be left to the Conclusions. The authors might want to re-examine this.

Minor comments:

Key figure: this illustration is overly simplistic in that aggregation in the atmosphere is not likely to result in units of the same species. Many different components of the atmosphere are likely to participate in forming any aggregates. Coagulation, in general, is driven by forces that do not distinguish one substance over another., or does s to a very minor degree.

Line 39: location and time of year drive many of the factors mentioned.

Line 43: nucleation is indicated to be at -15℃. Usually there is a spread in temperatures rather than a fixe value.

Line 58: " .. in the supplement ..." is meant to refer to the publication by Dreischmeier et al.

Line 66: Unclear whether the authors here mean INP or INM.

Line 68: "Therefore " hardly follows from the preceding sentence.

Lines 79-84: For readers not familiar with the process, it might be helpful to clarify how the sample preparation leads to macromolecular ice nucleators, not pollen grains or particles. "Extraction time" is not explained.

Line 92: RT presumably means room temperature. Not a generally known acronym.

line 117: A correction factor for cooling rate effect is reasonably well known: about 2℃ for a factor ten in cooling rate. This correction could be applied rather than dismissing the influence altogether.

Sections 2.2.2 to 2.2.5: most readers would benefit from some explanation of what is expected to be learned from these measurements.

Line 181: The sentence "The differential spectrum ...." seems to be in error, as C class is more frequent by far.

Line 326: This sentence is an early and incomplete statement of the results of the filtering tests.

Line 326-327: "We also note ..." - this is unclear.

Lines 334-335: Concentration of the suspensions combines with drop sizes to determine freezing temperature ranges.

Lines 335-336: Why does size uncertainty bring into question the 'nucleation mechanism"?

Lines 342-342: An example of jumping into partial conclusions.

Line 351: " .. radii range from 240℃ ... ????

Figure 5: Is Nm expressed with respect to the mass of pollen in the unfiltered sample or with respect to the mass of material left after filtration. Clearly, the latter would be more informative.

Line 394: Citation missing. More importantly , information on what materials are present in the samples in addition to the INMs would be welcome.