The Foxtails and the Hounds

There is something weird going with Esper’s delivery of data, where he delivered quite a lot of material – but left out some key data. Here ‘s the status on data – there are some puzzling methodological issues which I’l’l return to.

Chronologies: Esper sent 13 of 14 chronologies in February. This was very interesting and very helpful. However, he failed to deliver the 14th chronology from Mongolia. Despite a couple of iterations, no progress has been made on finishing the job.

Measurement Data: In March, Esper sent 10 of 14 measurement files in March, omitting Mongolia, Polar Urals, Boreal and Upperwright. Why would these have been omitted? The Polar Urals measurement file as used was sent in April. Again, why wouldn’t the others have been sent at the same time? In April, Science cited the existence of the Mongolia file at WDCP. I’m quite sure that this relates to the Esper version, but the dates do not match so the Esper version is different somehow.

Hanson said that he was pursuing the "original authors" for the foxtail data. The context is unclear whether he is trying Graumlich, having been unsuccessful with Esper. It’s hard to tell. But what does Graumlich have to do with it? We’ve seen that Esper changes the versions, so the Esper version is the one that needs to be examined.

If he’s trying to get the measurement data from Graumlich, then he may be in for a rough go. Andrew Bunn, who published on these sites (Bunn et al, 2005 discussed last year on this site) said recently that he did not see the foxtail measurement data during the course of writing Bunn et al. even though chronologies for Boreal and Upperwright are published in Bunn et al 2005 (indeed this is the only publication of them, as the earlier Graumlich study cited in Esper et al 2002 does not report chronologies for these two sites. Bunn said that Graumlich lost the measurement data when she moved from Arizona to Montana a number of years ago – something Crowley would understand, and a good reason for archiving the data in the first place. Someone paid for it.

When Bunn told me that the data had been lost, I hypothesized that Esper might have had a grey version and the data might thus have been preserved despite Graumlich’s loss. But with every week that goes by, I’m beginning to wonder. It could be that Esper’s just being difficult, but now it’s not just me, Science is involved. Every time that Hanson has to pick up this file again, he’s going to get mad at both Esper and me. I’m sure that initially it was me, but as time goes on, and the production of a simple data set by Esper seems to be more and more problematic, Hanson’s surely going to get increasingly irritated with Esper and the Hockey Team.

The foxtail series are not just used in Esper et al 2002; they are also very prominent in Osborn and Briffa 2006. Wouldn’t it be amusing if hypothetically an IPCC lead author was using lost data?


  1. jae
    Posted May 10, 2006 at 3:56 PM | Permalink

    Wouldn’t it be amusing if hypothetically an IPCC lead author was using lost data?

    Amusing, but not surprising….

  2. Pat Frank
    Posted May 10, 2006 at 4:55 PM | Permalink

    I don’t understand how someone can be using lost data. So, if Graumlich lost the data, and Esper has a copy, then where is the problem with him saying, “I have a copy of the data set you lost, isn’t that wonderful?” and, “I hope you don’t mind if I use it, and Oh! By the way, here’s a copy of my copy for you, too.” Graumlich then laughs with happy gratitude and we all get on with life. I can’t see that Esper is holding on to one or another data set merely to jerk you around, Steve.

    If Esper is holding back data, then I see three substantive possibilities: 1. He’s since lost that data, too, and is either ashamed to admit it or can’t be bothered to track it down. Or, 2., that single data set plays such a big part in the total reconstruction that releasing it would endanger the credibility of the final proxy. Or, the data set has been modified in some non-standard way that leads back to the outcome of “2.”

    It could be as simple as laziness or lack of consideration. Or it could be serious. The only way to know is to get the series. One question, Steve: If you used Esper’s methods to reconstruct his proxy using all of the 13 series you have got, and then scanned Esper’s final proxy reconstruction, would their difference produce the shape of the missing series?

  3. per
    Posted May 10, 2006 at 6:45 PM | Permalink

    Actually, this is amusing.

    In forensic work, there are concepts like “chain of custody” for evidence. You have to be able to show where the evidence was at all times to ensure that there was no mix-up, or adulteration of evidence.

    So scientist A loses original data. Scientist B pipes up and says, “okay, I have a gray copy of the original measurements”. Already, I am having difficulty establishing that scientist B didn’t just make this stuff up, never mind the minor possibilities of data corruption, accidental manipulation, or even misidentification of the data set. In terms of quality control, this is an apocalypse.

    Now if this data set should transpire to be the basis for a whole series of science and nature papers; well, it is very difficult to see how Science could take a relaxed view on this, given its recent Korean adventures.


  4. Steve McIntyre
    Posted May 10, 2006 at 9:20 PM | Permalink

    The “unlocatable” foxtail data plays a very important part in both Science studies. As bristlecone cousins, the series have big HS shapes. Esper has only 7 or so series going back to the MWP of which 2 are foxtails. The foxtails are hardly northern sites to start with; one would bad enough, but two series – really!

    Now Esper’s reconstruction has a somewhat elevated MWP so if you only had one foxtail series, you probably change the modern-MWP location. Whenever you have such knife-edge accounting, it’s hard not to be suspicious.

    I’m going to post up some methodological responses from Esper. Again, they are very odd. All of these obfuscations merely make one more determined to run it to ground.

    If people wonder why I don’t get more done with the other studies, I’ve been trying to get this data for almost two years now and didn’t progress one inch until the Hwang fiasco got Science to get off the dime.

  5. per
    Posted May 11, 2006 at 3:17 AM | Permalink

    i was going to suggest that this sounded like another Korea, but refrained. It is good that Science are taking their responsibilities more seriously.

    If the original data has been lost, that has got to be a corrigendum at least… Will that get into the AR4 ?

  6. Gary
    Posted May 11, 2006 at 7:18 AM | Permalink

    Is it possible the Dharma Initiative is involved with this?

    Seriously, it’s easy to lose data unless one is meticulous about archiving and few are that compulsive. Even if a P.I. personally has good habits, delegated work rarely gets returned in the same fastidious form. Source data and the meta-data describing it are less familiar to the P.I. and so may get misplaced in the rush to work up the information into a publication. Ultimately, sloppiness is easier to overlook than cleanup; you just hope nobody will notice and if they do you stonewall until they give up. As kids we learned we didn’t have to clean our rooms unless mom made us do it. Even if this is learned behavior and generally successful – although unexcusable in serious matters – only steady serious pressure will change it. It may be frustrating, but Steve is doing it the right way.

  7. John A
    Posted May 11, 2006 at 7:23 AM | Permalink

    Re #6

    That’s why the WDCP is there. Wouldn’t it be easier to log the data first at the WDCP and then do your calculations?

%d bloggers like this: