l28,l93. SAT is never defined - in fact using "temperature" may often be more appropriate.

l108. I feel you need a reference for the quotation.

l151. please remove the space between 1 and n. It confused me

l270. What sort of equilibrium do you mean here? I feel it must be a Charney/climate equilibrium, rather the full carbon-cycle equilibrium of the cGENIE model. I think you could rephrase to avoid this ambiguity.

l361. Why do you say emulated-preindustrial here? At this point, is it not simulated?

l364-369. I found this paragraph awkward to read. Did Manoj Joshi publish on this issue?

l422. "estimated" not "calculated"

l449. What about the fact that the climate sensitivity may be dependent on the orbital configuration.

l478. I cannot see the regression line getting shallower during the years 21-500 in your figure.

l511. Is the slab model of Williams et al. (2001) not equilibrated?

l520. Please either use ln or log_n rather than log. I assume base 10 when written like that.

l527. Was there a reason that you did not choose to use the overlapping ensemble members from the high/lowCO2 ensembles to validate each other?

l559. How have you calculated the standard deviation for the emulator validation shown in Fig. 8 and what does it mean? I had initially assumed that it was that spatial field computed from the final 50 years of each GCM run. This would effectively sample the local internal variability (ok, at annual resolution not multi-decadal, but still pertinent). But then you state that high internal variability in the Arctic could explain your problems, which made me doubt my assumption. And you also give a single number of 0.3oC - the global average of my assumed variable would be nearer 0.5oC.

l586. You note that you are not area-weighting. I can see little justification for this choice however.

l595. Why do you state that the ice-sheets should not change outside the range considered? Surely you have provided 2 different scenarios with invariant instead of a range of ice-sheet changes.

l606. Why even bother state than modice does something similar to lowice. Isn't the only difference between a constant offset?

l614. timeslices -> each 1kyr

l615/616. Repetition of CO2 settings.

l620-632. I found this paragraph heavy-going. Could you please revise for readability.

l635. This sentence just repeats the data from the table.

l644. you may want to add that the methodology you present here could be extended to other variables to investigate this issue.

l733. Surely this sentence is better placed in the acknowledgements.

l747-760. I cannot help but feel that a single table could readily replace both Fig. 13 and the majority of this paragraph.

l762. Up until -> Until

l774. Given your discussion of global mean SAT, you might want to consider adding it to Fig. 14. I believe that it would be of more interest for most readers than the 4 potential nuclear sites.

l779. Would this not be obvious from the emulator function/sensitivity itself and not require the construction of timeseries.

l827. Perhaps you would want to mention the applicability of a mean annual precipitation vs seasonal - and then stress that the methodology presented could create a function of that just as readily.

l831. This final sentence reads poorly. Please tighten up.

l890. I feel you should demote GCM bias and emulator error in your list of potential problems. I suspect that the other issues are more important.

Table 2. Please add the parameter settings for the preindustrial control. Please mark the ensemble members that you were not able to run due the ozone issue. The only time you use the ensemble numbers are in Fig. 8 and they do not correspond - perhaps remove them altogether.

Table 4. Perhaps you could add the value of the ice correction at each location from Fig 4.

Fig. 2 As you finish up treating them as a single ensemble, it may be better to plot as just one.

Fig. 4. Please improve upon the caption. The final sentence is not necessary: when plotting (a-c)-(b-c) is irrelevant what c is.

Fig. 6. Given this plot and the fact that $\Delta F=0$ for orbital changes, would it be fair conclude that orbital changes do not matter for global mean climate?

Fig. 7. If you keep this plot, please relabel the x-axis to something more meaningful: either actual CO2ppm or 1x,2x,4x, etc.

Fig. 8b. Why do you differentiate between the high and low CO2 ensembles? You don't in panel A or the emulator.

Fig. 9. I found it hard to understand this figure from the caption alone. Please try to refine the caption - you can refer to a table for the core sites.

Fig. 10. The y-axes are temperature change.