The manuscript presents the linear depolarization ratio of small ice crystals (< 70 µm) at 178° scattering angle under controlled laboratory conditions. The size and morphology of the ice crystals are well characterized in the laboratory. Therefore, the chamber studies present a valuable piece of information for atmospheric observations of cirrus clouds with lidar. Furthermore, it serves as constraint for optical models in the description of the complex shape of ice crystals. It is worth to analyze the AIDA measurements for this purpose and the topic is well suited for ACP. My concerns mostly regard the modelling approaches used to fit the observations. Therefore, it should be published after major revisions.

Major comments

1. A major point of critics is that you compare your measurement results to CGOM calculations for columns or even hollow columns, but only 19% (-40°C) and 40% (-50°C) of your observed ice crystals actually have a columnar shape. And of these, only 46% (-40°C) and 40% (-50°) are hollow columns (Tab. 2). It means that you found hollow columns only in 0.19*0.46 = 9% (-40°C) and 0.4*0.4= 16% (-50°C) of your ice crystal samples. Hollow columns represent only a minority of your lab measurements and it is not comprehensible why the CGOM modelling results of hollow columns should explain your data. The morphologically more complex particles which represent the majority of your ice crystals are more difficult to reproduce in optical models but are dominating your signals.
2. It seems that the IITM simulations were added later to the study. Sect. 2.4 firstly focuses only CGOM and T-Matrix and much later, you mention that IITM was used additionally. Also, Fig 8 and 9 are split between the models. Actually, the T-Matrix and the IITM both are used to describe the smaller particles up to 9 µm and for AR=1, IITM was extended up to around 17 µm which gives an overlap to CGOM. However, this overlap is not used in your discussions. Also, T-Matrix and IITM are not compared. I am wondering, if the increasing depolarization for small particles as seen for IITM is present in the T-Matrix results as well. How does it look like if you start your calculations at GMD = 0 as done for IITM? A more comprehensive use of the 3 models and corresponding discussion is needed.
3. The T-Matrix method is used for spheroids. However, the shape of ice crystals is far away from a spheroidal shape. You need a strong motivation to use the spheroidal shape for the representation of the small ice crystals. The agreement of the results for the spheroidal shape might be pure coincidence. Because if it agrees already with spheroidal simulations (L447/448), then you don’t need to ask for more complex optical models.
4. Do you have to consider possible orientation/alignment effects of the ice crystals? Maybe not, because you’re observing them from the side in a turbulent environment. However, in lidar, specular reflection caused by horizontally oriented ice crystals lead to much lower depolarization ratios in vertically pointing lidars. To avoid the effect of specular reflections in cirrus clouds, most lidars are tilted to an off-zenith angle of 3-5°.

Minor comments:

• An outline at the end of the introduction is common to guide the reader through the manuscript. Furthermore, one or two introductory sentences at the beginning of each section would be helpful. It would help to keep the track of your study.
• Eq 1+4: Why do you name it delta_parallel and not just delta?
• A sketch of the setup and the involved instruments would be nice. I agree that it is already presented in previous publications, but a visualization (even very schematic) would help to better follow the methods section.
• Fig 1 Does the scattering pattern depend on particle orientation? And how do you deal with randomly oriented regular particles?
• L119-149: These paragraphs should be carefully checked again, because there are some small careless mistakes, e.g., twice a reference to Fig. 1c, a different uncertainty of sigma in Fig 1e and the text (L135), in bracket references not correctly formatted, and some rather slang formulations.
• In Sect. 2.3, I don’t like the formulations like “minute 25” in a written text. A suggestion would be to write “at t=25 min”. Furthermore, Fig 2 should be introduced once before presenting the procedure. Is it a typical experiment to give an example? Or were all experiments done like this?
• L186: What is mu? Overall, this note is a bit tricky to understand.
• L215: What exactly do you mean with spherical equivalent diameter? Which quantity is equivalent to a sphere (volume, surface, …)?
• And why you are using the maximum dimension of a spheroid (L215) or the column length (L228) for comparison to the spherical equivalent diameter?
• L234: Please provide the refractive index.
• It would be recommended to better link Fig 4 with Tab 2 as some images belong to cases presented in the table.
• The caption of Fig 4 is quite long. You may consider to add some information, e.g., the supersaturation, already to the images.
• L275 and Fig 5: You split your results by campaign. It is important for your analysis. However, for the reader it is a bit confusing as I don’t see the differences between RICE 01 – 03. An overall number for each regime might be sufficient. It is up to you how to best harmonize the description.
• L353/354: Please add a reference to the appendix as well.
• L392: Actually, it is not in line with your chamber studies, because in the colder temperature regime (CIRRUS-ML, -58 – -63°C), you observe a constant depolarization ratio and then a increase towards smaller particle sizes.
• The difference in the scattering angle and the possible implications should also be mentioned in the conclusion.
• L460: A better reference for EarthCARE would be Donovan et al., AMT 2024 as they describe the lidar measurements and the retrieval of ice crystal properties.
• References to Sato and Okamoto are missing. In several publications, they discussed the modelling of the depolarization ratio of cirrus clouds. Furthermore, you can search for more lidar observations of the depolarization ratio of cirrus clouds.
• The plots of Fig B2 and B4 are swapped. It really created a lot of confusion for me until I’ve got that the figures are not the ones explained in the caption.

Technical Corrections

• L 436: Skip the word “about”. Or is the number of experiments not sure?
• Fig B3: According to the figure, C is shown for hollow columns not solid ones. The caption is probably wrong.

This is a very nicely written discussion of what I consider to be a very important subject matter, i.e. interpretation of polarization ratios derived from both passive and active remote sensors. The authors have conducted very thorough research and have approached the problem from multiple directions in order to achieve some semblance of closure.

What I think is missing, or perhaps has been overlooked, is a thorough uncertainty analysis, not only of the measurements but of the models, as well. There are also some open questions that I have about the SIMONE measurement system that are related to my request for the analysis of uncertainties.

Before evaluating the differences in the models and measurements, I think that it is imperative to discuss uncertainties. I am skeptical of the depolarization uncertainty that is stated as 1.4%, based solely on what is stated as calibration cycles. I didn't read the Schnaiter (2012) paper but my opinion is that referring the reader to this paper is insufficient due to the importance of this derived ratio to the main focus of the paper. Although it is not clearly explained in this paper, I believe that the SIMONE does not measure single particle scattering but detects scattering from an ensemble of particles. That being said, what impact do the following have on the derived polarization ratios: 1) number concentration, 2) size distribution, 3) crystal orientation, 4) spatial distributions within the sample volume (if not random, then how will this impact the measurement?).

My second concern is that there does not appear to be any mention of the difference between 178 degrees and 180 degrees backscattering on the polarization ratios. From my own modeling of polarized backscattering from single particles, there can be quite a difference in polarization ratios when looking at 178 degrees versus 180 degrees. Was a sensitivity analysis done in order to evaluate the differences between lidar at 180 and SIMONE and models at 178 for ensembles?

Once a detailed uncertainty analysis is include in this submission, I will be ready to accept it for publication.