The Mystery in Kenya

Today’s Mannian mystery takes us to Kenya Tanzania, not to Kilimanjaro itself, but to the great plains, still home to prides of lions, herds of wildebeest and giraffes.

On the right is Mann et al Fig 1, showing the location of its proxy series. On the left, I’ve provided a blow up Africa, where three proxies are shown. The figure on the right color codes proxies by start date and type. The light blue square is an ice core series starting in the mid-first millennium: it may be identified as Thompson’s Mt Kilimanjaro series, about which we’ve posted in the past. The light green triangle is a sediment series starting around 1000 and can be immediately identified as Verschuren’s Lake Naivasha series.

The interesting puzzle for you is the series denoted by the orange star. The star symbol denotes a documentary series, the orange color a start date somewhere around AD1400. The BBC program mentioned that Mann had been scouring the world for proxies, but his discovery of documentary climate records for the plains of Kenya Tanzania (Update – OK, OK) going back to the time of Columbus are important not just for climate studies, but for world history. Just so that readers do not under-estimate the importance of Mann’s discovery: I have verified that this is not a simple mislabeling of a local African proxy series as a documentary record. This is a genuine documentary series.


PS. I have provided several clues to assist readers in the identification of this series.

Update: Here is a plot of the proxy against Mann’s gridcell temperatures for the corresponding gridcells in Spain and east Africa.


  1. Steve McIntyre
    Posted Sep 22, 2008 at 8:08 PM | Permalink

    bender figured most of this out from a remark on another thread before I’d even posed the question in this post (I had a teaser in the other thread). I’ve put bender’s comment in suspense and will post his correct guess in time sequence.

    bender- there’s something even sweeter about this that you haven’t picked up yet. See if you can figure it out, but send anything offline since you’re way ahead of others on this.

  2. Mark Duffett
    Posted Sep 22, 2008 at 8:53 PM | Permalink

    Is this a case of the rain in Spain being on the plains of Kenya?

    It looks suspiciously close to being exactly on latitude zero…maybe in the wrong hemisphere as well? Or have latitude and longitude been swapped?

  3. Paul29
    Posted Sep 22, 2008 at 8:54 PM | Permalink

    He flipped latitude and longitude, and its really in Algiers?

  4. Keith
    Posted Sep 22, 2008 at 9:07 PM | Permalink

    Hmmm, maybe a transcribed copy of an oral history from some Bushman tribe? Assign a generational life span to the people in the history, combined with interpretations of growth and rainy seasons based upon the descriptions in the history. But wouldn’t that be anecdotal exidence, which the Team discounts with regard to the MWP? That would truly be a case of “Do as I say, not as I do,” by them.

  5. Steve McIntyre
    Posted Sep 22, 2008 at 9:07 PM | Permalink

    Yes, he flipped latitude and longitude and it’s really from Spain. The flipping is very odd because the coordinates are correct in the PNAS Supplementary Information (the SI where the Briffa MXD series are wrong), but they are reversed in rtable1209 which seems to be used for the plot.

    bender spotted one clue when he looked at the tagline for the graphic – the figure is denoted rainin16.jpg.

    But the sweetness doesn’t just end there. There’s more. The actual proxy being shown really is just that – the rain in Spain. The full citation is : Rodrigo, F. S.; Esteban-Parra, M. J.; Pozo-Vázquez, D.; Castro-Díez, Y. 1999. A 500-year precipitation record in Southern Spain. International Journal of Climatology, vol. 19, Issue 11, pp.1233-1253

    After the amusement that we had with “the rain in Maine falls mainly in the Seine”, it’s pretty funny that the “rain in Spain falls mainly in the plain” – but in the plains of Kenya.

    It does seem to detract somewhat from the physical interpretation of the proxy “teleconnections” when rain in Spain debouches through the Zambezi.

    • Mark Duffett
      Posted Sep 22, 2008 at 11:52 PM | Permalink

      Re: Steve McIntyre (#5),

      Really nitpicking now, but I’m pretty sure the plains of Kenya aren’t in the Zambezi watershed (having been in the headwaters of the latter region not all that long ago). They might actually drain into the Nile? In that case, Steve, you really are a de-Nile-ist.

  6. Steve McIntyre
    Posted Sep 22, 2008 at 9:09 PM | Permalink

    I tried to move bender’s comments from another thread, but the move function doesn’t seem to be working since our recent update and merely eats comments up.

  7. Larry T
    Posted Sep 22, 2008 at 9:11 PM | Permalink

    Mann found The Ark of the Covenant.

  8. Steve McIntyre
    Posted Sep 22, 2008 at 9:14 PM | Permalink

    #7. I think that was Jones, I. et al 1981.

  9. J.Hansford.
    Posted Sep 22, 2008 at 10:32 PM | Permalink

    Good grief, hopefully not some long lost climate records from the Chinese Admiral Zheng He, in about 1405 to 1433 and our intrepid amateur historian Gavin Menzies…?

    Surely not!

    • kim
      Posted Sep 22, 2008 at 10:53 PM | Permalink

      Re: J.Hansford. (#8),

      Surely not a Suzuki documentary.

  10. Posted Sep 22, 2008 at 10:45 PM | Permalink

    I think that was Jones, I. et al 1981.

    Ok. That is bleeding hilarious!

  11. Posted Sep 22, 2008 at 10:57 PM | Permalink

    Could the rain from Spain be falling mainly on the plain? And Lucia, has she helped precipitate this event?

  12. Dishman
    Posted Sep 22, 2008 at 10:57 PM | Permalink

    The real value of peer review is an additional set of eyes to catch stupid mistakes…

    so you don’t embarass yourself.

    Trying to prevent embarassment after publication leads to … well…

  13. Geoff Sherrington
    Posted Sep 22, 2008 at 10:59 PM | Permalink

    Temperature reconstructions from the skeletal length of giraffe necks?The hotter it was, the fewer the trees and the higher the giraffe necks had to be to reach leaves. Robust, otherwise why do giraffes have backs and necks that resemble hockey sticks?

  14. MarkR
    Posted Sep 22, 2008 at 11:05 PM | Permalink

    It’s a cave drawing of a primitive Hockey Stick, c1400, used as a totem by the N’ASA tribe, and found with a scrap of bristlecone pine parchment on which the words “Tu bene dicis. Ego contra erravi. Mihi, quaeso, ignosce”, in early FORTRAN by Homo Defectus, or Stupid Mann.

  15. J.Hansford.
    Posted Sep 22, 2008 at 11:24 PM | Permalink

    LOL… “You are correct. I was wrong. Please forgive me : “Tu bene dicis. Ego contra erravi. Mihi, quaeso, ignosce”. :-) MarkR U dag.

  16. Posted Sep 22, 2008 at 11:34 PM | Permalink

    Funny stuff!

  17. Steve McIntyre
    Posted Sep 22, 2008 at 11:46 PM | Permalink

    Interestingly, this is one of the 484 proxies with a statistically “significant” relationship to gridcell temperature. I think that this “relationship” becomes less convincing when latitudes and longitudes are accidentally reversed, but perhaps this depends on the teleconnection channel – FM899 spiking to you from the plines.

    • Paul
      Posted Sep 23, 2008 at 5:00 AM | Permalink

      Re: Steve McIntyre (#18),

      Can you confirm it is in fact the case that Mann computed the correlation for this series against a local Kenyan temperature record. i.e. this isn’t simply a case of putting the star in the wrong place, but a true data error.

      • Jean S
        Posted Sep 23, 2008 at 5:15 AM | Permalink

        Re: Paul (#27),
        I can confirm taht the location is wrong in every data file I’ve checked, and it is correct only in the SI file. So I don’t see a way the correlation was calculated with respect to the correct grid cells unless it was manually calculated separately from the rest of the proxies. I’m sure Steve will confirm this as soon as he wakes up :)

    • Michael Jankowski
      Posted Sep 23, 2008 at 9:25 AM | Permalink

      Re: Steve McIntyre (#18),
      I think it would be interesting for someone to randomize the locations of the proxies and see how many have a “significant” relationship to the corresponding grid cell temps. Maybe it would be as simple as flipping the lat-longs of all the proxies (not sure how one would fairly treat something that would land in the ocean upon flipping).

  18. Tim Davis
    Posted Sep 23, 2008 at 1:50 AM | Permalink

    Now we can add ‘plain’ old fashioned incompetence to deliberate rejection of unfavourable data, sloppy data storage, wacky statistical methods, unethcial review and publication processes etc etc

  19. Jean S
    Posted Sep 23, 2008 at 2:05 AM | Permalink

    I have some great news for all you hockey fans in Kenya: Dr. Mann has kindly donated a beautiful hockey stick to you! It’s inverse, but don’t worry: it will be inverted in CPS and RegEM does not care!

  20. Posted Sep 23, 2008 at 2:53 AM | Permalink

    If I’m going to guess, (and nobody carp because McIntyre “He who must not be named” tells me nothing), then I’d guess than Mann has put a series for Southern Spain into East Africa by transposing the latitude and longitude.

    So the rains in Spain fall mainly on the Plains (of the Serengetti)

  21. Louis Hissink
    Posted Sep 23, 2008 at 3:05 AM | Permalink



    You are kidding, Mann made that an elementary error? Good grief.

    • Jean S
      Posted Sep 23, 2008 at 3:55 AM | Permalink

      Re: Louis Hissink (#23),
      I agree that this was probably originally an unintentional elementary error (see the location of all Rodrigo’s series (originals; PNAS version)). However, there is an interesting twist to the story: the location is correct in the SI! Not only that, this series is the only series whose latitude (or longitude) is written with comma (37,5) instead of a dot (37.5). So it seems that the location error was spotted at some point and possibly manually corrected to the SI without bothering to rerun the reconstruction algorithms (I doubt it really “matters”; but tells a lot about the attitudes).

  22. Demesure
    Posted Sep 23, 2008 at 3:36 AM | Permalink

    Next time, Mann should be peer-reviewed by tectonic plates expert.

  23. Geoff Sherrington
    Posted Sep 23, 2008 at 4:43 AM | Permalink


  24. Alan Bates
    Posted Sep 23, 2008 at 5:01 AM | Permalink

    Had a quick look at the Spanish paper (Caveat emptor – I am not an expert! Also this is simplified.)

    Starting around 1400 is optimistic – the first records start 1500. There are 65 records, none of which extend over the entire period. Rainfall is split into 5 categoies from -2 with drought and loss of springs/river flow to +2 for rain produced floods. Where there was no comment the rainfall was assmued to be normal and given category 0. Most (317) but not years (350) had information. 25% of the total years had values different from 0.

    Likewise, 500 years is also optimistic – the period is 1500 – 1850 which then overlaps with instrumental records starting 1791 (from Gibraltar – i.e. not in the region of the rest of the data.

    There is an overlap from 1791 – 1850 which is further split into 2 x 30 year periods, one for calibration (i.e. comparison with the instrumental record) the other for “Verification of results”.

    Overall, the two series (65 data sources with none extending over the period) and the instrumental Gibraltar data are fused together to produce the 1500 – end 20th century, 500 year anomalies plot (the 2 parts are clearly marked and in different line style.)

    Sounds like as good as they could do but I wonder what Mann made of it?

    Also, is there evidence that rainfall on a seasonal basis acts as a temperature proxy?

  25. David P
    Posted Sep 23, 2008 at 5:40 AM | Permalink

    #26: Bravo, sir.

  26. Yorick
    Posted Sep 23, 2008 at 6:03 AM | Permalink

    “Chinese Admiral Zheng He”

    Hey, I didn’t know it was a fraud when I bought it! Not until he started describing a navigation method that concievably could have worked with an atomic clock and precise astronomical instruments and a supercomputer executed with a mud puddle and a sheep stomach. Which fortumately was the first chapter.

    I’ve asked that this book not be discussed here. It wasn’t used by IPCC (nor should it have been.)

  27. Posted Sep 23, 2008 at 6:20 AM | Permalink

    The orange star is actually plotted a little south of Mt. Kilimanjaro, placing it in Tanzania, not Kenya. Even Mt. K is in Tanzania, although right on the border and “plainly” visible from Kenya.

    Also, the orange is a little redder than 1400 — I think this was correctly coded as circa 1500.

    RE Jean S, #25, Excel interprets “37,5” as being greater than 80.98, the highest other latitude — when I sorted the spreadsheet by latitude, this ended up at the end. So even the SI would not be read correctly, except perhaps by Euro-Excel.

  28. Steve McIntyre
    Posted Sep 23, 2008 at 6:54 AM | Permalink

    #25, 32. Now that I think of it, this value had to be manually corrected when I collated the SI into R. In an R read, the column was interpreted as characters because of the one comma. So I had to go back, manually compare the read to the original, correct the comma in my version of the spreadsheet and re-compile into an R table. I was wondering how they got past that stage. Now we know that they had duplicate accounts, something that we’ve seen before in MBH. The existence of multiple data versions in a project developing over time is something that you need to guard against in a final archive. One more reason to archive working code – as required by econometrics journals.

  29. stan
    Posted Sep 23, 2008 at 6:55 AM | Permalink

    It’s the sloppy science. Same song, how many verses now? Temperature sites measuring A/C exhaust temps, surface temperature “adjustments”, SST wild-ass guesses, the various Mannian versions of alternate reality, lost data, “private” IPCC records, peer-review rubber stamps, photo-shopped government reports ….. the hits just keep on coming!

    It’s all about the sloppy, sloppy science.

  30. Mike S
    Posted Sep 23, 2008 at 7:06 AM | Permalink

    Perhaps the Dendros are adopting scatological techniques and are examining fossilized elephant droppings. Anyone who has driven around a corner and had to brake hard to avoid a small mountain of elephant excretia can testify that they do contain partially digested tree limbs.

  31. Posted Sep 23, 2008 at 7:07 AM | Permalink

    Re Alan Bates, #28,

    There are 65 records, none of which extend over the entire period [ending in 1850]. Rainfall is split into 5 categoies from -2 with drought and loss of springs/river flow to +2 for rain produced floods. Where there was no comment the rainfall was assmued to be normal and given category 0. Most (317) but not [all] years (350) had information. 25% of the total years had values different from 0.

    It sounds like this series of necessarily stationary perceived differences from “normal” is a guaranteed HS shaft out to 1850: If the backyard that was always dry when I was a child is now often full of water, I’ll regard the present as “flood” conditions. But my grandchildren will just think of it as “the pond”, and won’t complain of “flooding” until the front yard fills up as well. In fact, they will complain of “drought” if the backyard goes dry again. Any true long-term swings in climate will get filtered out as they are incorporated into the perception of the norm.

    If then, as Alan relates, the series was spliced by the authors onto the Gibraltar instrumental (precipitation?) series 1791 – 1850, and then continued to the present, you could get a pretty good HS.

    The correlation with temperature over 1850-95 is -.1494, which passes screening. But is this with Spanish temperature or East African?

  32. Jeff Alberts
    Posted Sep 23, 2008 at 7:11 AM | Permalink

    “Welcome to de middle of de film”

    Hey, it’s an Africa reference…

  33. Neil McEvoy
    Posted Sep 23, 2008 at 7:15 AM | Permalink

    Today’s Mannian mystery takes us to Kenya, not to Kilimanjaro itself

    Er … Kilimanjaro is in Tanzania.

    • Luis Dias
      Posted Sep 23, 2008 at 8:34 AM | Permalink

      Re: Neil McEvoy (#38),

      Come on now. Kilimanjaro is in the border between Kenya and Tanzania. Audit is fine, but please!

  34. Posted Sep 23, 2008 at 7:48 AM | Permalink

    RE Steve, #33: Truly a “commady of errors”!

  35. Steve McIntyre
    Posted Sep 23, 2008 at 8:41 AM | Permalink

    You know the style of joke that David Letterman sometimes uses – where you give the punchline and then you have to make up a question. Today’s punchline:

    The rain in Spain falls mainly in the plain.

    Here’s my candidate question:

    Why are the snows of Kilimanjaro receding?

    Readers are invited to submit their own candidate questions to which the answer is : The rain in Spain falls mainly in the plain. Or others, I guess. For example, if you have a good question to which the answer is: The rain in Maine falls mainly in the Seine, I’d accept that. But this is a highly structured discussion. Answers cannot go to far from the format.

    • Luis Dias
      Posted Sep 23, 2008 at 8:59 AM | Permalink

      Re: Steve McIntyre (#41),

      I’m sorry, mr. Steve, I understand the need to be light sometimes and that to make fun at dire things is sometimes the best choice. Rather laugh than cry. But doesn’t this start to seem a little of too much poking fun at little details that aren’t useful at all in the big picture? Isn’t this type of post the kind of which gives credit to realclimate apologists’ claims that you’re just here trying to discredit them at any costs, ad hominems and stupid jokes included?

      I say this with the awareness that the quality control of this study may be lacking. But without context, this Kilimanjaro joke is but a pixel in the big picture. Unless these kind of errors are proved to be more than trivial, but really distorting the end results, isn’t this just a futile exercise at schadenfreude?

    • Dave Dardinger
      Posted Sep 23, 2008 at 11:33 AM | Permalink

      Re: Steve McIntyre (#41),

      How come my entry into the “contest” got erased? If you’ve given up on it then modify the post to say so.

      Steve: dunno. TRy again.

    • Dave Dardinger
      Posted Sep 23, 2008 at 1:06 PM | Permalink

      Re: Steve McIntyre (#41),

      Answer: “The rain in Spain falls mainly in the Plain”

      Question: What typo by Joe Biden made his claim that “the Republican VP candidate is causing global warming” fall flat.

      BTW, it this were a clue in a British style crossword it’d read -43 VP candidate falls flat.

  36. Jeff Alberts
    Posted Sep 23, 2008 at 8:59 AM | Permalink

    My candidate question is:

    “Orange water given bucket of plaster”


    Makes as much sense as any other.

  37. Steve McIntyre
    Posted Sep 23, 2008 at 9:28 AM | Permalink

    We started out trying to replicate Mann’s reported table of correlations and got stuck at several points, which I’ve documented.

    This post is a bit of an in-joke because of the incorrect locations of precipitation in MBH98, where we observed incorrect locations of precipitation series memorialized as the “rain in Maine falls mainly in the Seine”. You’ll have to admit that it’s at least a bit funny that they have a geographical mislocation for the rain in Spain.

    Second, my practice is to work through examples to see if they make sense. There’s far too much glossing over the details in these studies. For example, isn’t it a bit interesting to think about what happens in Mannian methods if you incorrectly locate one series thousands of miles away? Does it matter? And if not, why not?

    Also, I don’t see any ad hominems in the post. There is sarcasm, however there is nothing in the post that is not within the four corners of the original article.

    • Luis Dias
      Posted Sep 23, 2008 at 9:44 AM | Permalink

      Re: Steve McIntyre (#45),

      As I said, I did understand the in joke thingy all along. And I do see your hard work, you don’t need to repeat your badges to me. It doesn’t feel right, though. Mann’s study is considered mainstream the very moment it is launched, with all the implications. Given the pressure that many skeptics are having at the moment (you included, while I know you don’t want to be labeled as such, they did so nevertheless), isn’t this a little too light-hearted? It’s as if (and mind the limits of the analogy!) you were a chess player poking fun at the fact that your opponent just wiped out (by cheating of course) the inconvenient pieces laid out in the board. It’s not funny, for if no one does anything about it, you are going to lose. The loser in here may be science itself.

      The ad hominem thingy is across the post, it is implicit (the automatical dismissal of it all, the joking of wrong details, the contest of making questions to the punchline above).

      Sorry, it doesn’t feel right. But hey, you’re the boss here!

      • bender
        Posted Sep 23, 2008 at 9:50 AM | Permalink

        Re: Luis Dias (#46),
        Luis, you are free to start your own blog and repackage everything that Steve M says so that it is PC.

        Steve: Luis’ comments were presented in good faith and he has a point. It’s hard to strike a balance.

  38. Luis Dias
    Posted Sep 23, 2008 at 9:58 AM | Permalink

    Oh well, forget it. Poke fun at the bully while he’s destroying your house. Like I care.

  39. Steve McIntyre
    Posted Sep 23, 2008 at 10:01 AM | Permalink

    A separate but relevant issue here is special handling: why are the rtable1209 lat-longs reversed for this one series but not others? Maybe the special handling doesn’t matter, but sometimes it does e.g. the Gaspe extension. Special handling needs to be explained.

  40. bender
    Posted Sep 23, 2008 at 10:02 AM | Permalink

    The more the bully bullies the more obvious it is what he is doing. You lose a house, you gain a neighborhood.

  41. Andy
    Posted Sep 23, 2008 at 10:07 AM | Permalink

    One effective way of dealing with oppressive bullying is to laugh in its face. SteveM’s posts get the balance just right between self evident scientific and technical proficiency and sarcasm.
    Considering the brown stuff coming the other way, he deserves a medal for continuing in the same vein, without resorting to the standards of some folk.

  42. Luis Dias
    Posted Sep 23, 2008 at 10:09 AM | Permalink

    Yes, mr Bender, except that this bully has been around for ten years straight and is still mainstream. Even mr. Joliffe acknowledges his work is very confusing, but he doesn’t have the “balls” to call it for what it is.

    And to reinforce my point (and this is my last comment on this for it’s OT, I promise) everyone saw how of a bully Genghis Khan was, but it didn’t matter much, now did it?

    • bender
      Posted Sep 23, 2008 at 10:17 AM | Permalink

      Re: Luis Dias (#52),
      The pressure for RC to throw Mann overboard (err, “take on other duties”) is mounting. There will be a turning of the tides at some point. Perhaps within a year.

  43. bender
    Posted Sep 23, 2008 at 10:14 AM | Permalink

    Mann has upped the ante at every opportunity and Steve M has called his bluff every time. Now the cards are down and we see Mann was indeed bluffing the whole time. Sad.

    Fastforward a half-decade. Now, it is simply a matter of weeks before this later paper is exposed as a similar pot of rubbish. The fatal error there is the inadequate robustness test. This is a serious problem, not at all trivial like some of the other minor, humorous issues. More sadness.

  44. W F Lenihan
    Posted Sep 23, 2008 at 10:26 AM | Permalink

    I suggest a new hypothesis to explain the nearly systemic, pattern of Mann’s erroneous selection, corruption and use of proxy data sets. An AGW skeptic has infiltrated Mann’s lab and is sabotaging code writing and data entry.

    I recognize that this is just another form of “my dog ate my homework” excuse for failure. snip

  45. Richard deSousa
    Posted Sep 23, 2008 at 10:26 AM | Permalink

    Actually, the rain in Spain falls mainly in Galicia, the Iberian peninsula, which is northwestern Spain.

  46. Mark T.
    Posted Sep 23, 2008 at 10:27 AM | Permalink

    Yeah, but in many ways, Gavin is just as bad, and he’s been one of Mann’s biggest defenders. I think you are overly optimistic on this point.


    • bender
      Posted Sep 23, 2008 at 10:35 AM | Permalink

      Re: Mark T. (#57),
      I can’t agree with that. Gavin’s tone is often ingratiating, and he is frequently a bit dodgy. But he doesn’t make all that many mistakes. And that’s what matters. gavin answers way more questions – and far more difficult questions – than mike.

      [William Connolley left RC "to take on other duties" at a time when he was becoming a liability. mike is getting there.]

  47. Luis Dias
    Posted Sep 23, 2008 at 10:32 AM | Permalink

    Perhaps that’s true. I’ll wait for that in my seat though. We all know their typical response. It’s not about Mann, it’s about the Team, and they have now a bunch of HSs (Moberg et al, etc. It’s like that game of cards 52 catch, where you throw away all the cards and have to catch them all). Those results reinforce the HS, and while the HS itself will be discredited in due time, the other HSs will shout out “but hey, Mann’s work is not alone, you see? So that’s not important“, and the charade goes on, even at the scientific level, as the BBC documentary so well demonstrated.

    Next year, Kyoto 2009. Mann may fail to win the postcard graphic, but it’s a close call and if no paper comes out in 2008 to discredit it, then we’ll have a lot of this HS stuck into our TVs for the entire political show in 2009. Even if Mann gets discredited, perhaps Moberg’s graph will enter the scene.

    • bender
      Posted Sep 23, 2008 at 10:42 AM | Permalink

      Re: Luis Dias (#58),
      I have asked Dr. Loehle if he has any plans to write a paper “on the independence of paleoclimate recons”. I think the community could use a paper like that.

      • Craig Loehle
        Posted Sep 23, 2008 at 1:40 PM | Permalink

        Re: bender (#61), Bender–no plans to do that particular topic. I noted on another thread that it is interesting how carefully Gavin critiqued my proxies and how he smiles proudly at Mann’s 1000+ proxies, as if no such problems lurk beneath the depths.

  48. Kusigrosz
    Posted Sep 23, 2008 at 10:36 AM | Permalink

    Could a documentary series be carried by an African swallow?

  49. Alan Bates
    Posted Sep 23, 2008 at 11:25 AM | Permalink

    Re: #28

    I hesitate to ask the same question again:

    But the Kenyan / Tanzanian / Spanish series is not temperature. It is a subjective assessment of rainfall from many differemnt sources.

    Is there evidence that the subjective seasonal rain in Spain calibrated against rainfall in Gibraltar actually is a proxy for temperature? If it is not then why is it being used by Mann 2008 anyway?

  50. MarkW
    Posted Sep 23, 2008 at 11:34 AM | Permalink

    snip – please do not post this sort of stuff

  51. Steve McIntyre
    Posted Sep 23, 2008 at 11:45 AM | Permalink

    Here is a plot of the proxy against Mann’s gridcell temperatures for the corresponding gridcells in Spain (37.5N, 5W) and east Africa (5S, 37.5E).

    The Spanish rainfall series has a 0.013 correlation over 1850-1995 to the Spain gridcell temperature (Mann temperatures) and -0.049 to the east African gridcell temperature, while the two gridcell temperatures (Mann version) have a 0.607 correlation to one another. The correlation to the “next nearest” east African gridcell (42.5E, 7.5S) was -.193, higher than reported. A question here: for a proxy located on a gridcell border e.g. 5S or 5W here, there are two equidistant “nearest” neighbors. This situation is not described in the text, but might be clarified in the code.

    But there’s something even more interesting arising out of this. Readers of last years’ Where’s Waldo series will doubtless be just as surprised to see an instrumental temperature series in East Africa going back to AD1850, as we were to see an “east African” documentary series going back to AD1500. If you look at the linked post, you’ll see that there are no non-urban temperature station in Africa as one goes back.

    The answer here is that the correlations are not being calculated against original temperature data, but against RegEMed temperature data. So the African temperature data is being “infilled” from somewhere else – but from where? I wonder if it’s Spain. :)

    • Luis Dias
      Posted Sep 23, 2008 at 12:43 PM | Permalink

      Re: Steve McIntyre (#65),

      Now that’s funny! I also chuckled at your new title in the top.

    • Posted Sep 23, 2008 at 1:58 PM | Permalink

      Re: Steve McIntyre (#65), you might have hit the nail on the head. The chuckle was nice while it lasted but something stinks here.

      I am afraid I don’t have a good Jeopardy style question for you, but here is something from Baltasar Gracian which, I feel, sounds pertinent:

      BECAUSE AN ASS once, not twice. All too commonly, in order to repair one foolish step, four more are taken; or excuse is made for one dumb trick with a second, and a greater; folly is either of the house of lies, or lies are the house of folly, for in order to stand, each needs the support of many; worse than the defense of evil, is its protection, and even worse than the evil, the inability to hide it; the bequest of one vice is the bestowal of many others at interest: the wisest of men may slip once, but not twice, and that only by chance, and not by design.

      — Sinan

    • Jean S
      Posted Sep 23, 2008 at 3:04 PM | Permalink

      Re: Steve McIntyre (#65),

      A question here: for a proxy located on a gridcell border e.g. 5S or 5W here, there are two equidistant “nearest” neighbors. This situation is not described in the text, but might be clarified in the code.

      It is done in ‘propick.m’, see the part described as ‘great circle distance’. As far as I can see it works as follows:
      1) calculate the distance (see the code for the formula) between the proxy location and all instrumental grid cell centres (loop ‘i’). The values are stored in vector ‘d1′. (BTW, notice here that comments for lat1/lon1 and lat2/lon2 are wrong (reversed):
      ‘X’ contains instrumental data and ‘Y’ proxies)
      2) use build-in sort-function to sort ‘d1′. Sort-function arranges values from smallest to the largest and in the case of equal values, the value appearing first comes first (FIFO).
      Hence, the order of equidistant neighbours depends on the order the intrumental grids appearing in ‘X’. ‘X’ comes from file ‘clidata’. ‘climdata’ is prepared in ‘doannsst.m’, where there is a description how the grid cells are arranged. Data is coming from ‘instfilled’, and the order should be the same (I’m too tired to check the code).

      However, notice that there are additionally flags ‘enough’ and ‘toofew’ (found in intrumental data directory), which control if the grid cells are used: the locations in ‘enough’ are used and the ones in ‘toofew’ are not. According to ‘Readme.txt’ in the ‘instrument’ data folder, those grid cells which have less than 10% of annual data are disregarded.

  52. stan
    Posted Sep 23, 2008 at 11:48 AM | Permalink

    Luis Dias,

    Humor is the best possible way to make the point. The work is sloppy. Laugh at it. Other people (scientists and non-scientists alike) are far more likely to notice the laughter and check in to see what’s up (or watts up). If you want the world to know the truth, invite them to come join in the laughter. If you wait for them to prop open eyelids with toothpicks to slog through the statistics, you will have a very long wait.

  53. The Pedant's Apprentice
    Posted Sep 23, 2008 at 12:00 PM | Permalink

    At least no-one referred to “parallels of longitude”.

  54. Alan Bates
    Posted Sep 23, 2008 at 1:19 PM | Permalink

    Re: Steve McIntyre (#65) & my #62

    Thank you.

  55. aurbo
    Posted Sep 23, 2008 at 1:24 PM | Permalink

    When criticism turns from serious and sometimes acerbic contradictions to derision, the fat lady is clearing her throat.

  56. Sam Urbinto
    Posted Sep 23, 2008 at 1:32 PM | Permalink

    The opera is over when the East African Eland hunting lady sings.

  57. Steve McIntyre
    Posted Sep 23, 2008 at 3:44 PM | Permalink

    I’ve located enough and toofew.

    The cells are arranged S to N, minute hand dateline to dateline.

  58. nevket240
    Posted Sep 23, 2008 at 4:40 PM | Permalink

    snip – please do not discuss motives

  59. nevket240
    Posted Sep 23, 2008 at 11:22 PM | Permalink


    I wonder if he got the idea from the recent “discovery” of weather records from the British Navy logs?
    Which reminds me. Quite a few Old dears here have commented lately on how the local Aussie Rules games have been played in conditions like the 60’s & 70’s. Particularly the Southern ranges, Otways.
    But, they are not Climate Scientists so it doesn’t mean anything.

  60. PaddikJ
    Posted Sep 24, 2008 at 12:13 AM | Permalink

    Could a documentary series be carried by an African swallow?

    Only if it’s packed in coconut shells.

    But wait – African Swallows en’t migratory.

    Oh. Roight – maybe it’s the Giant Iberian Swallow, eh?

    (Come, Patsy; let us leave this silly place.)

  61. TAC
    Posted Sep 24, 2008 at 5:32 AM | Permalink

    SteveM: What a thread! Incredible comedy! How fabulous; how rich!

    I hereby nominate you and Michael Mann for an Ig® Nobel Prize — this year’s ceremony is October 2nd — which surely belongs alongside that other prize you shared a piece of in 2007.

    In any case, this thread made my day!

    Time to visit the Tip Jar.

  62. steven mosher
    Posted Sep 27, 2008 at 8:42 PM | Permalink

    in case you missed the update on the 25th to fix a lat/lon problem

  63. steven mosher
    Posted Sep 27, 2008 at 8:59 PM | Permalink

    I posted a thanks to gavin, see if it gets through.

    I was glad to see this File.

    As the file indicates a lat/lon error in one series was fixed. So kudo’s to ever fixed the error ( There’s no indication it impacts the results that I know of) It was just
    good to see the attention to detail and the fixing of little mistakes even if they don’t matter.

    I believe the error was first pointed out here on Sept 22.

    So again, kudo’s to whoever updated the files in such a timely fashion.

  64. Steve McIntyre
    Posted Sep 27, 2008 at 9:11 PM | Permalink

    Their revision removed the disclosure of prior corrections to all the Schweingruber series previously noted here

    This sort of subprime accounting should be criticized,

  65. steven mosher
    Posted Sep 27, 2008 at 11:33 PM | Permalink

    RE 83. Argg I missed that.

  66. steven mosher
    Posted Sep 27, 2008 at 11:42 PM | Permalink

    83 any wager on whether my complimentary post giving you credit will make it through?

  67. Steve McIntyre
    Posted Sep 28, 2008 at 6:24 PM | Permalink

    Steve Mosher, I looked at this again and they’ve done this totally the wrong way. Their SI should be exactly what was used to generate their results. They can’t go around diddling their SI after the fact. Now their SI isn’t consistent with the data used to generate their graphics and results. Instead of changing their SI, they should have a placed a List of Errata on their website and issued a Corrigendum.

  68. Steve McIntyre
    Posted Sep 28, 2008 at 6:27 PM | Permalink

    #84. Now they say:

    1209proxynames.xls: Microsoft Excel file containing a list of all the
    1209 proxies used and information pertaining to each proxy. The last
    column in the file corresponds to the proxy data file name in the
    “data/proxy” directory. **UPDATED: 25 September 2008 to correct
    errors with the lat/lon values**.

    If you’re amending financial statements, you need to show both the before and after. Why shouldn’t that apply here?

  69. Steve McIntyre
    Posted Sep 29, 2008 at 8:58 AM | Permalink

    I noticed something further that’s anomalous about this series. In Mann’s proxyinfill program, he uses (or says that he uses) a dataset 1227proxynamenolu.txt. “nolu” is a shorthand for “no Luterbacher”. The prefix “1227” doesn’t tie into present nomenclature which has things like 1209proxy … etc and presumably comes from an unreported earlier iteration with 1227 proxies. However this data set has 1137 series. All 1137 are in the 1209. HOwever 1209 minus 71 Luterbacher = 1138. What’s the missing series? The rain in Spain/Kenya (rodrigo).

One Trackback

  1. By The Trouet Ocean Proxies « Climate Audit on Sep 11, 2010 at 10:34 AM

    [...] This is Vershuren’s sediment series, mentioned in passing in connection with Mann et al 2008 here , also here. The data set was also used in the discussion of a warm Warm Pool MWP by Newton et al [...]


Get every new post delivered to your Inbox.

Join 3,191 other followers

%d bloggers like this: