Author Contribution paragraphs are often a good idea. In this case, where the authors have changed in the revision, I would like you to provide one. I think it may be positioned before the Acknowledgments.

Why is some of the data password protected? If you can provide a good reason for this, please also send me a password (or put it in the manuscript?!), so I can view the data.

For the model output, I like the way you have done it, specifying a minimum in the paper but also including a file with the full set of desired variables. For people who can provide a bit more than the minimum, but not everything, is there any value in providing high frequency output for just part of the run? Would there be specific parts that would be of most interest?

Note that the citation to Kageyama et al overview paper in the CMIP6 special issue is only to the GMDD paper in review. This is unlikely to improve since that paper is on hold pending the acceptance of this manuscript (and the other 2)!

I have managed to access the forcing data through the password protected site! Thus, the only addition I have to the issues common to all the PMIP experiment description papers is that you should email me to let me know the password has been lifted so that I can check this before I finally accept the paper. Aside from that, there are just the queries raised by me, James Annan and Didier Roche, editors of the 3 PMIP papers, which I already emailed to you, Masa and Bette. For convenience and transparency, these queries are repeated here below.

Dear Masa, Bette and Johann,

Sorry for the long time that the I-G and LM PMIP papers have been with the editors. We want to make sure we are being consistent across the papers. We have all looked through the manuscripts and found a number of things that need clarification, or that are inconsistent across the papers. We’re emailing you all today so that you can start to coordinate your response, and then you should expect that most of these comments will also be included in the individual editor’s revision for each manuscript.

Some papers mention the entry cards for CMIP, but others do not. Some bring in the “Tier” nomenclature without defining it, while the LGM discusses “sensitivity” experiments, leaving it uncertain as to how these fit into the CMIP system.

The documentation requirement seems to be different in all the papers (and the LGM is actually inconsistent with the overview paper). The documentation requirement is really presented as part of the protocol, so should be included in each of the individual papers.

* All papers should explain the entry card system and contain full and consistent information on the documentation requirement for the runs. Many groups will do all the Tier 1 runs, so it really will be very confusing if the documentation requirements are different! All the papers should make sure to define CMIP concepts such as "Tier 1".

We are all confused about what the comment that groups are responsible for finding their own ESGF space for Tier2-3 experiments means in practice, and note that there is no indication as to whether the LGM sensitivity experiments should be uploaded or not.

* Please clarify what, in practice, this group responsibility means. Will modellers be able to upload Tier 2 and 3 experiments to ESGF or not, and how are the LGM sensitivity runs to be made available?

Are all boundary conditions to be uploaded to ESGF? This is promised in the LGM paper, but not in the others.

* Please make it clear whether all boundary conditions will be uploaded to ESGF. It would be best if they were, as the current situation where everything is available on the PMIP website is sub-optimal (it seems to me OK for modellers doing the runs now, but it is not future-proof: web addresses change!).

The LGM experiment requires 100 years of equilibrated run to be stored on ESGF, and defines which variables are to be used by participating groups to ensure sufficient equilibrium is attained before the other experiments are started. There is less detail on this in the other papers.

* Please can all experiments require the 100 years of equilibrated run on ESGF. If possible, please define a common metric for assessing equilibrium for all the runs. If there are good reasons why different metrics should be used for the different experiments, please provide an explanation and guidance for the modellers.

jules, Didier and James

Thanks for your patience, and for all the clarifications. I am considerably less confused after the latest revision, and hopeful that readers may find the protocol straightforward to implement!

jules