Dear Authors,

Thank you very much for taking a great effort to respond to the referees' comments in quite some detail! Overall, I strongly encourage you to submit a revised manuscript, which I will be glad to reconsider after a final round of peer review.

I actually welcome the planned addition of Section 2.1 in order to discuss the requirements and possible "weak points" of the plateau-tuning method. However, the revised manuscript should not be longer, but shorter than the original manuscript, which may require more economical wording and possibly deletion of sentences or even short paragraphs. Furthermore, as a side note, I am not satisfied by the way you dealt with the results by Franke et al. (2008) and Butzen et al. (2017) in your response to the second referee. These are publications in their own right, representing the state of the art at the time of publication. Instead of simply dismissing them as "outdated" or "discouraging", I ask you to actually name the data-model differences and argue for possible reasons, on both, the data and the model side of the problem.

Yours sincerely,

André Paul

Dear Authors,

Thank you for responding to the referee's comments and submitting a revised manuscript. It has not been easy to review the revised manuscript, partly because of its length that does not always seem to be justified. I received one new referee report, and the referee has a main issue with the length of the revised manuscript, too ("some paragraphs are still long-winded and contain a lot of detail that are likely not necessary for the main message of that paragraph"). Therefore I ask you to take the opportunity and seriously shorten the manuscript, closely following the suggestions of the referee.

The second larger issue is with the significance of the adjustment of the Suigetsu age scale: The referee finds it "hard to distill the main message(s) from it, and in particular to separate previous findings from new findings that result from the update in the Suigetsu age scale, and accordingly the 14C plateau-tuned surface ocean reservoir age estimates". In addition, not all data are described and not all arguments are explained, there is merely a reference to a supplementary figure (S2) or unpublished data, which I do not find satisfactory. So please highlight the novel knowledge gained form adjusting the chronology of the records, as proposed by the referee, also in relation to what is already known from other proxies such as benthic d13C, eNd or Pa/Th, and delete that would require an even more detailed presentation.

Regarding the model-data comparison, I agree with the referee that the comparison to studies other than the one by Muglia et al. (2018) is limited. References to personal communications are not very helpful in my mind and should be limited to a minimum. My recommendation is to delete the model-data comparison and leave it to a manuscript in its own right that could do better justice to other modeling studies. This would also be a chance to shorten the manuscript a bit further.

The same applies to the unpublished data and the data from the submitted manuscript by Küssner et al.: If a satisfactory solution cannot be found until the submission of the final version of this manuscript, then I recommend to leave these data out as well.

In summary, please provide a concise and pointed response to the comments and suggestions and follow them as closely as possible, as I expect this to lead to a further improvement of your manuscript, paving the way for the attention and reception that it clearly deserves.

Sincerely,

André Paul

Dear Authors,

Thank you very much for submitting your response to the referee report and your revised manuscript. There are three remaining major points, the first one is actually critical for finally accepting your manuscript for publication:

(1) Unpublished data and manuscripts "in prep."
There are two unpublished data sets related to two manuscripts by Ausin et al. and Küssner et al. that are both still "in prep." (by the way, the manuscript by Ausin et al. is not even cited). The data sets are stored at PANGAEA, but "under embargo as the papers are still in preparation", hence they are not yet available in any way, and therefore it is impossible to evaluate them. Both the referee and myself expressed that we are not satisfied with this situation. We understand that it is highly frustrating for you as authors of this manuscript, too. However, in its current form it cannot be published using those data.

There are the following options to deal with this situation: (a) The manuscript is put on hold, you wait until the embargo on the unpublished data is lifted and remove the references to the unpublished manuscripts. (b) You try to rewrite the paper, removing the references to the unpublished data and data anywhere in the paper. (c) You invite more co-authors and have the embargo lifted for the relevant data. (d) You decide to withdraw the manuscript.

(2) Personal communication ("pers. com")
On several occasions, you refer to unpublished, personal communications. As there is no way to check those for correctness, I ask you to remove all personal communications from the text and rephrase the text accordingly. If you want, you can still mention the persons you talked to in the Acknowledgment section.

(3) Length of manuscript
Another issue is the sheer length of the manuscript, which is 850 lines plus 100 lines and a couple of figures in the supplement. Attempts to shorten the manuscripts as recommended by the referees and myself have only been partly successful. Therefore I would welcome further attempts to shorten, for example, the abstract and the introduction.

Two minor points are:
- Lines 572 to 577 regarding model-data comparison may possibly be shortened and rephrased as follows:
"Further improvements are expected from a model configuration that properly resolves the topographic details of the continental margins and adjacent seas, which frequently form the origin of our sediment-based data sets (Butzen et al., 2020). For the time being, we compared our empirical estimates with estimates from a coarse-resolution GCM, using the results by Muglia et al. (2018) as an example."
A formulation along these lines would also prevent the impression of picking only those model results that fit your data and leave ample room for an independent manuscript devoted to a broader and more detailed model-data comparison.
- Lines 489-490: "The locations of the 18 (20) cores are shown in Fig. 7."
It is not clear why you add the number 20 in brackets. Please clarify.

Please let me know how you plan to deal with the unpublished data and manuscripts "in prep.". I am looking forward to hearing from you soon.

Yours sincerely,

André Paul

Dear Authors,

Thank you for submitting a revised version of your manuscript. Unfortunately, I think that it is not quite the final version, because it differs in a few points from your response to the points (1) and (2) that I had raised before:

(1) Unpublished data and manuscripts "in prep."
I accept you decision to put your manuscript on hold until the two manuscripts by Ausin et al. and of Küssner et al. become publicly available (which probably means until they are accepted for publication) and the embargo on the related data sets on PANGAEA will be lifted.

In the revised version of the manuscript, the following citations would need to be corrected:
Line 264: Ausin et al., 2020 subm.
Line 637: Ausin et al. (in prep.)
Line 171, 263, 476, 481, 503, 591-592, 636, 669, 847, 1338:
Küssner et al., 2020 subm.

In addition, the manuscripts by Ausin et al. would need to be added to the list of references, while the reference to Küssner et al. in lines 973-975 would need to be completed.

(2) Personal communication ("pers. com")
In contrast to your response, I cannot find the reference to the paper by Bronk Ramsey "now published under 2020"". Instead, in lines 228-229 of the revised manuscript, there is still a reference to personal communication.

Similarly, I cannot find the reference to the abstract by Abé-Ouchi "for the OC3/IPODS Meeting held in Cambridge in September 2018"". Instead, in lines 656-657 of the revised manuscript, there is still a reference to personal communication. I suppose that in both cases the modifications to the manuscript were simply lost or
forgotten.

(3) Length of manuscript
I am fine with your efforts to slightly shorten the abstract and the introduction.

Finally, when reading your manuscript again it occurred to me that the model used by Gebbie (2014) is not the MIROC model, but a simple "water mass decomposition model" or "global tracer transport model", as the author himself calls it. The paragraph leading up to line 653 of the revised version of the manuscript would need to be rephrased accordingly.

Yours sincerely,

André Paul

Dear Authors,

Thank you for submitting the revised manuscript, which I am happy to accept for publication subject to two minor technical corrections:

(1) Please update the references, especially those by Küssner et al. and Ausin et al., to reflect the correct title and journal. In the main text as well as in the bibliography, refer to these two manuscripts by "Küssner et al. (under review)" and "Ausin et al. (under review)".

(2) Please cite the two datasets associated with the manuscripts by Küssner et al. and Ausin et al. in the "Data availability" section of your revised manuscript. Take special care of the DOIs (Digital Object Identifiers) and check whether they correctly link to the datasets published in the PANGAEA database.

Yours sincerely,

André Paul