cc: Eduardo Zorita , hegerl@duke.edu, Jan Esper , Keith Briffa , Myles Allen , Nanne Weber , t.osborn@uea.ac.uk date: Fri, 31 Mar 2006 16:08:09 +0100 from: Martin Juckes subject: A new draft to: Anders Moberg Hello, a new draft is attached. I've shortened a lot of the discussion and added a few more results. There is now a table showing the coherence of each of 34 proxy records which extend back to 1000 -- this answers a suggestion by McIntyre and McKitrick that MBH's N. American PC1 is a statistical outlier -- it isn't. I think one of the figures in the previous version I sent round had a slight error in that I was using a composite of proxies: for the Mann et al. proxy collections this included principal component series with arbitrary sign. I have now made sure that the sign is determined by the correlation with temperature. The draft is still rough in places, but I think it is more readable and compact now. I'll be away for two weeks: if people can get comments back to me within the next 3 weeks, I'd like to submit it soon after that. An obvious next step would be to do a reconstruction using all the 32 long proxy records, but I haven't done that yet because there are sampling issues to deal with. cheers, Martin Attachment Converted: "c:\eudora\attach\cpd-2006-xxxx.pdf" Attachment Converted: "c:\eudora\attach\cpd-2006-xxxx.tex" Content-Type: text/plain; charset="iso-8859-1"; name="cpd-2006-xxxx.bbl" Content-Disposition: attachment; filename="cpd-2006-xxxx.bbl" X-MIME-Autoconverted: from 8bit to quoted-printable by balin.rl.ac.uk id k2VF4xxb029885 Attachment Converted: "c:\eudora\attach\cpd-2006-xxxx.bbl"