cc: Timo Hmeranta <timo.hameranta@pp.inet.fi>, "Ross McKitrick" <rmckitri@uoguelph.ca>
date: Thu Jul 22 09:51:30 2004
from: Phil Jones <p.jones@uea.ac.uk>
subject: Re: ROG Data
to: "Steve McIntyre" <stephen.mcintyre@utoronto.ca>

    Steve,
       The two files show the cross-dating of the polar urals cores (ring width and density).
   Should
    be relatively self-explanatory. Output comes from a program developed here, that is
   loosely
    based on cofecha. Program takes the set of cores, averages them (with various options on
    the smoothing to get residuals and also periods, this one using 50-year segments), then
    correlates each with the average. It also does some averaging across the rows and columns.
       Finally it has the option of using another chronology, so here we've used the density
   one
    for ring width and vice versa.
      We use the program to check dating and help with dating when starting with a new set of
    trees. There are loads of plot options as well, as although the numbers help plots are
    necessary as well.
       Correlations are lower in the early years, but the plots confirm the dating is correct.
   Take
    it from me these are good. We are using the program on much of the material we have from
    northern Eurasia and also from the world data bank. We are finding some simple mistakes.
    What I mean here is that the chronology produced from the cores in WDCP don't give the
    chronology also given there. If you alter a few cores - give them the dates they should
   have -
    then they do. So, it seems that archive problems like the one you've seen with these polar
    ural series have occurred. WDCP is not that well resourced - the last I heard was that
   NOAA
    may cut much of its paleo funding, so I'm not surprised.
       On a different matter, I can understand why people like Tom Crowley don't respond to
    you and others in the skeptic camp (can't think of a better simple term at the moment). We
    are all busy people and we've also spent a lot of time (years) making contacts to get a
   lot
    of the paleo data we use. The same applies to instrumental data also. Also constraints are
    put on us by a lot of the groups/individuals who send data not to pass it on. I'm going to
    try and put more data onto our web site, but this is not high on my agenda. I would like
   it
    to be higher, but there are other things I have to do (some of which I'm much less keen
   on)
    that take precedent.
       Your aim is to check the work we do, with the explicit intention of trying to find
    mistakes or conclusions that aren't valid. If you found we were right would you still
   write a
    paper?  I will commend you for writing a paper first and then putting something on your
    web site, rather than doing it the other way round like David Legates. I've always dealt
    with the scientific literature and believe this is the best way. I would recommend that
   you
    publish your results in journals that climatologists read. Web sites are just places for
    more details - not substitutes for formal publications.
      I think I've done and said enough over the past week or two, so can we have a break
    for a while.
    Cheers
    Phil
   At 09:14 21/07/2004 -0400, Steve McIntyre wrote:

     Dear Phil,

     Thank you once again for your continual courtesy and energy. I had surmised that the 400
     year gaps in russ021w and russ021x were probably due to identification number
     duplication.  However, in the files as archived with WDCP, these have been actively
     filled with 0 values and are not merely latent curiosities of identification number
     duplication.  Any mechanized handling of the data under present circumstances will lead
     to incorrect results and accordingly the archived datasets should be promptly amended.

     I would appreciate a copy of the cross-dating statistics report and the rwl version that
     you actually used, since it differs from the version archived at WDCP.

     As a matter of interest, I am still puzzled by your criteria for "too few trees". I
     would like to test the 10th and 11th century portions of the Polar Urals chronology
     against explicit CRU policy. I am not presently suggesting that these portions do not
     meet such criteria, but they are obviously in a danger zone and, given the wide use of
     this indicator, it seems worthwhile checking. I think that, for millennial dating, one
     might reasonably wish for a site with more than 3 trees in the late 10th and early 11th
     century (of which 2 are short-lived with low crossdating values). Hence my interest in a
     check against explicit criteria.

     I am also interested in the crossdating checks for these early trees. A couple of the
     early 11th centuries are very short. My initial crossdating calculations for these few
     trees show values significantly lower than later values, which would seem to make this
     period a little precarious.

     BTW Jacoby has recently archived the Sol Dav data (but not before AD900) and states that
     the early portion of the published series does not meet replication standards.

     Again, thanks for your continuing courtesy in this. Replication and checking are fact of
     life in business (where most of my experience lies), especially when you are
     communicating with the public. There are very formal processes for this; auditors and
     securities lawyers, who are among the most highly paid professionals in our society, do
     little more than replicate and check. If you wish to offer securities to the public, you
     get used to dealing with questions from them. The corresponding processes in
     paleoclimate studies (and probably most academic pursuits) seem very casual to me. When
     studies get used for policy purposes, it seems to me that there is a material change in
     the level of due diligence is required.  This leads to a conundrum: many scholars seem
     quite happy to have their studies quoted in big reports (like IPCC), but then fail to
     make arrangements for public archiving of their results and methods and become defensive
     if they are asked for their data. In the offering of securities, there is an interesting
     stage that deals with this - if a report by an indepenedent professional (e.g. a
     geologist) is used in a prospectus, the independent professional has to provide a
     consent letter authorizing the promoters of the prospectus to refer to his report and to
     supply the consent letter to the securities commission. The terms of the consent letter
     impose disclosure requriements on him.  This would deal with the situation of someone
     like Crowley, whose study is quoted by IPCC, but who repeatedly and persistently refuses
     to disclose his proxy data versions. In a prospectus situation, if the IPCC wished to
     use Crowley's report, Crowley would have to agree to make his data pulic if asked (which
     he should probably do on alternative grounds).

     There are some interesting articles about replication studies in economics and political
     science by Dewald, Gary King and McCullough/Vinod, which  might interest you. Some
     journals have now adopted policies requiring the archiving of data and source code as a
     condition of publication. I would encourage you to contribute the source code for Jones
     and Mann (2004) and Mann and Jones (2003) to your planned data contribution. Procedures
     like this with avoid frustrating efforts after the fact to obtain data and methods.

     Regards, Steve McIntyre

     Prof. Phil Jones
     Climatic Research Unit        Telephone +44 (0) 1603 592090
     School of Environmental Sciences    Fax +44 (0) 1603 507784
     University of East Anglia
     Norwich                          Email    p.jones@uea.ac.uk
     NR4 7TJ
     UK
     ----------------------------------------------------------------------------
