see attached supplement file

Review of: Internal climate variability and spatial temperature correlations during the past 2000 years

I think this paper is in very good form. I only have a few minor comments.

Lines 10-12: I found this sentence confusing: 'However, combining the iLOVECLIM results with CMIP5 model results and various PAGES-2K temperature field reconstructions, we find that neither model results or reconstructions are robust.' The models and reconstructions are not robust in what sense? I think this sentence could be more clearly re-worded to be more specific or perhaps even cut entirely given the content that follows it.

All of the main text figures have quite small font sizes on nearly all the labels and axes. Their readability would be much improved by increasing the font sizes of everything in the figures.

Could you comment in the paper on how the spatial resolution of the different datasets may influence the comparisons you've done? LOVECLIM is 3 degrees, the PAGES2k reconstructions are 5 degrees, the CMIP models are ~2 degrees, and the proxy data is based on individual points, so there's a lot of spatial averaging differences that could affect the comparisons.