Readers Journal     weblog/wEssays     home

The Hockey Stick Breaks   (Protagoras, June 25, 2007)

A truly dramatic event in contemporary intellectual history occurred unnoticed in April. To understand it, you have to understand the history of the famous Hockey Stick climate chart, which has decorated IPCC documents since the late nineties, appeared in lots of Green publications all over the world, and has been used as vivid proof that human caused global warming (AGW) is real and an imminent threat to human life and civilization.

The event was that Dr Phil Jones was finally persuaded by requests, including some under the Freedom of Information Act, to reveal the sites used in his 1990 paper [1] There are no clues on the page to the meaning of this event, and the history of the attempts to get access to this data is itself an interesting chapter which will repay study. For now, let us focus on the Hockey Stick and this paper's role in it.

The Hockey Stick chart appeared to show that we live in extraordinary times. The researchers, Michael Mann and his coauthors, had assembled a reconstruction of temperatures going back about 1,000 years. There were two elements to this. First, there were temperature records from weather stations. These however only go back about 100 years, if that. So you need some other way of estimating temperatures from earlier periods.

The method chosen was tree ring data. Trees respond to rising temperatures by growing faster. This faster growth is laid down for posterity in the form of wider annual growth rings. By drilling cores into old enough trees, and by calibrating the width of the ring with observed temperatures and width for that species, it is possible to reconstruct what must have been the temperature in the year that ring was laid down. Use was made of a great many such series in a seminal paper, which reconstructed global temperatures going back about 1,000 years. It is usually referred to as MBH98 after Dr Mann and his coauthors.

This reconstruction had some striking features. It appeared that the Medieval Warm Period, during which the Vikings had colonised Greenland, had not existed. There had been no unusual warmth then. In addition, the Little Ice Age, during which the Thames had frozen repeatedly, had also not existed. There had been no unusual cooling then. What had happened was very different: temperatures globally had been more or less constant for about a thousand years, until the mid twentieth century.

The paper then went on to overlap the reconstructed climate record onto the twentieth century weather station measured record, and the combination showed what seemed to be, after this period of extraordinary stability, a sudden hockey stick shaped rise in global temperature. This rise coincided with the spread of the Industrial Revolution.

CO2 emissions were later measured and found to have risen throughout the twentieth century as a result of the Industrial Revolution and the burning of fossil fuels. CO2 is a greenhouse gas. That is, it traps heat rather as a greenhouse does. The conclusion was clear. We were in the middle of a catastrophic runaway process of global warming, which threatened to destroy civilization as we know it. And the cause was the industrial revolution and its emissions of CO2.

Almost at once there were some doubts, and efforts to verify the findings. Jones' paper was about the reality or otherwise of the UHI (urban heat island) effect. It was argued that previous attempts to use the instrumental record to show that the planet was warming had been corrupted by the fact that too many of the stations were subject to creeping urbanization. The warming trend they showed could therefore be dismissed.

When a previously rural station becomes surrounded by a city, it will show warming, because cities are warmer than rural areas, because of the UHI effect. Concrete absorbs heat, cities consume huge amounts of energy, so cities will always be warmer. This was alleged to be the sole or main reason why the instrumental record showed runaway warming. Jones' paper was believed to decisively destroy this argument. He had used series from Russia, China and Australia, only used carefully vetted stations whose history was known, and it was clear that UHI was a negligible quantity in the instrumental record of warming. The phenomenon was real, and it was serious.

Unpleasantly suspicious people demanded to verify this by examining a list of the stations whose data was used. Jones declined to provide them. An interesting extract from correspondence between Jones and Sherrington on the Australian data will be found at [8]. The Australian data is, well, surprisingly difficult to come by.

The reasons given for the refusal varied. On Feb. 21, 2005, Jones is reported as having replied to Warwick Hughes:

"I should warn you that some data we have we are not supposed to pass on to others. We can pass on the gridded data - which we do. Even if WMO agrees, I will still not pass on the data. We have 25 or so years invested in the work. Why should I make the data available to you, when your aim is to try and find something wrong with it."[9].
Not discouraged by this memorable and touchingly honest refusal, enquiring minds eventually resorted to using the Freedom of Information Act to apply pressure to Jones, and whether for this or some other reason, the data appeared - in April of this year. It had only taken 17 years. The detailed inspection of the validity of the claims could now begin.

Ealier, in another part of the wood, efforts had been underway to determine exactly what samples MBH98 had used, and what methods they had applied to them, in order to produce the famous graph. Requests for the data were refused. After twists and turns worthy of a Wilkie Collins novel, including an anonymous tip-off that the missing data could be found on an ftp site in a folder called 'censored', MacIntyre and McKittrick finally reconstructed the statistical method used and pronounced it faulty.

A couple of years after this reconstruction, the topic was examined by the Wegmnan Committee, at the request of the US Congress. Also at the request of Congress, the information behind the paper was finally revealed. You can get the Wegman report at [4]. Wegman's separate reply to later questions is also very worth reading. If you do read this, read his remarks on social networks. It is one of the great masterpieces of understatement in scientific literature.

It may not be clear why any statistical methods were needed, or what they could possibly be. The problem is that the tree ring data, if simply plotted in a spreadsheet for instance, in raw form, does not show any clear signal or trend. This is because it is very noisy in a mathematical sense. There may or may not be a signal in there, that is, consistent movement of the variable, but there is also a great deal of random movement. So the series is laid out and then adjusted according to various algorithms in order to eliminate the random and reveal the signal.

The same kind of thing is done, for instance, with software which attempts to determine whether a sonar probe is picking up, in the middle of random ocean noises, the faint signals which will tell the presence of a submarine propeller. Chartists have the same problem with stock market price movements. This is a process in which it is easy to produce spurious signals. So there are recognised valid algorithms and procedures which statisticians use to do it. The question Wegman and his colleagues asked was: had MBH98 used them? And if not, was the method they had used statistically valid?

While all this was going on, on another front, the CO2 question was being examined. If you drill ice cores, you can extract temperature records and CO2 content of the air from the time of the samples. It was clear that the history of the planet has been marked by pronounced fluctuations in temperature in the past. The ice ages of the past are proof of this, and we know there were ice ages by the evidence of the landscapes in which we live today. Geology has mapped the extent of the ice fairly precisely. For the Greenhouse Gas theory to be correct, it seemed that it would be necessary for CO2 rises to precede the warmings. That is how causation usually works.

However, what the record, particularly the Vostok record, shows is that the warmings precede the rises in CO2 levels by several hundred years. It gets worse. The prescription is that we should stop this warming by stabilizing CO2 levels, and if necessary produce cooling by lowering CO2 levels. This proposal runs into a number of logical difficulties: at no point in the past has rising CO2 preceded warming. At no point has stabilized CO2 accompanied stable temperatures. At no point has falling CO2 preceded falling temperatures. Enquiring minds began to wonder. It might work. But how do we know it will?

Enquiring minds also tried to get to the ice core records raw data. Alas, this turned out to be impossible in the case of much of the data, particularly the data on which Dr Thompson's studies depended.... The sorry saga of attempts to find the detail on these records, or one consistent account of what they show, or even to establish whether they are proxies for temperature or precipitation, can be found at [7]. Readers will begin to lose their sense of humor a little when they realize that the Hockey Stick curve in An Inconvenient Truth appears to have been based on these studies.

After the dust settled on the Wegman inquiry a report was produced. A number of things then became clear about tree rings and statistics and Mann's use of them. Wegman is an authority on statistics, and had essentially evaluated the criticisms made by MacIntyre and McKittrick.

It turns out that the hockey stick effect depends on two procedures. First, on using a certain method of Principal Components Analysis. Second on including some tree ring series which are generally agreed not to be temperature proxies. Essentially what M&M had argued, and what Wegman confirmed, was that a home made, statistically invalid, method of analysis had been used which would generate hockey stick shaped curves from random data. It did this by overweighting large values at the end of the series. If you added some series which are not about temperature at all, interpolated a few more values where some were missing, used series from one tree where best practice is to insist on several, you got the Hockey Stick.[10]

With the publication of the Wegman endorsement of the M&M work, the tree ring issue was closed. The reconstruction of the temperature record from tree rings was, scientifically speaking, a fantasy. It gave no information one way or the other about what the temperatures had been in the past.

A further slight difficulty with tree rings now appeared. The work of Keith Briffa also reconstructed temperatures from tree rings in 2000. It was included in the IPCC latest material. A slight variant on it is referenced; it is JGR 2001. The interesting thing about this is not apparent from the IPCC graphs. It is the problem of divergence. The instrumental record appears to show alarming warming. The proxy data appears to support this. One validates the other.

Unfortunately what Briffa's work shows is that the two started to diverge in some important cases in about 1960. Since then, in Briffas reconstruction, the proxy record shows temperatures falling. There's a problem, isn't there? Either the proxies are not telling us the truth about temperature, or the temperature record is wrong. Or could it be that the proxies had been right up to recent times, and then inexplicably became wrong?

The alert reader will be wondering why this would not be apparent in the IPCC charts. Alas for human nature, while Briffa's work extends further than 1960, and the IPCC reference is to the article showing the full series, and while some IPCC reviewers wanted the series shown in full, it appears in the IPCC documents, yes, sadly... truncated! Does this really make the problem go away, you ask. No, not really. Read all about this distressing episode at [5].

At this point there was a choice. You now had to admit that the Hockey Stick was nonsense, as regards the earlier period. You could adopt for a fundamental rethink. Or you could take the usual path which committed AGW advocates have taken. This is to argue that all that tree ring stuff was never of any great importance. After all, we have the instrumental record, and what that shows is dramatic and threatening levels of warming. Who cares what happened or didn't happen in 1300? What counts is what we know is happening today.

It is at this point that it becomes more and more interesting to validate Jones' station data. It becomes more and more important to prove that Jones' station data is not contaminated by UHI effects. Jones' study was the one that showed beyond doubt that ground station data was not contaminated, and proved the existence of a runaway warming trend. So it was very frustrating that it was so hard to get at the data. Even if, as Jones feared, those who wanted the data wanted it so as to try to refute it.

Various people tried to reconstruct what must have been the stations, with varying success, and expressed varying levels of scepticism about the validity of the series. However until April of this year, it was impossible to be certain. In April the data was revealed for the first time, and the results regarding China have turned out to be a shocker.

It always did seem to border on the miraculous that detailed and accurate station records should have been kept through the Cultural Revolution, a time of famine, purges, mass murders and vast population movement. It has been pointed out that during this egregious episode of human depravity tens of millions died, but it is not known how many tens. One might expect that the maintenance of weather station data could suffer. However, Jones said: "The stations were selected on the basis of station history: we chose those with few, if any, changes in instrumentation, location or observation times."[6]

So that was all right then.

It turns out however that, according to Keenan in the above paper. 84 stations were used. Of these, the histories of 49 are unknown. Of the remainder, half are said to have undergone major moves and changes. Their records are thus not homogenous and cannot be strung together to make a valid series indicative of temperature trends.

At this point, the original Hockey Stick work is broken beyond repair. We have an invalid temperature proxy record spliced to an invalid measured instrument record. It's surely over?

Not quite. We can now move to a country in which there have been no revolutions, wars or major upheavals resulting in changes in civil service approaches to data in the time period which interests us. A country in which there is a Freedom of Information Act, so that we can obtain data collected by the government. This country is the USA.

We can then start to examine the physical stations, find out if they are urban or rural, find out if their surroundings or land use patterns have changed, and take pictures of how well they are maintained. This is now being done on, and you can find comments on the work on the blog at The results are being analysed statistically and graphically on Climate Audit. However, similar doubts and concerns have been expressed in other quarters, notably by Roger Pielke on his blog and in peer reviewed papers. You may wonder, if the world is being invited to spend billions upon billions on theories based on this data, why it has not been critically examined and validated before. Its a good question.

It is too early to be definite about what this is showing, but it seems clear already what it is not showing. It is not supporting the hypothesis that there is a warming trend in the instrument record when we eliminate heat island effects. It turns out that on examination of the details, for the few stations that have been examined, if the stations are 'good', that is, properly maintained, away from extraneous heat sources, in properly constant rural areas, they seem to show little or no temperature rises.

Many of the stations however are in areas where there has been major land use change. Many have been classified as rural when they are plainly not. Many have extraneous items such as lightbulbs, barbecues, jet exhausts, parking lots, in places where they would influence readings. So for a great many stations it seems likely that their record is contaminated by UHI.

So let us now go back to where we came in, and you see that we are in the middle of a managed retreat which threatens to turn into a rout. You will recall that we came in on an argument that had three pillars. The first was the proposition that CO2 was causing greenhouse global warming. The second was that the proxy record showed that nothing like the present warming had happened for a thousand years. The third was that the instrument record showed that there was recent warming which was proceeding with catastrophic speed. What does all this show about those arguments? It shows they are all unsupportable in their original form. The position we have to retreat to seems less and less plausible or defensible.

First we have to argue that this episode of warming is unique. It is the only one ever in the history of the earth that has been caused by CO2. Now, this could be. Models purport to prove that CO2 is a greenhouse gas, and that current levels can produce warming. Its just that we have lost the support of the historical record to back this proposition up. We can no longer argue that there is any precedent for it. This means it has gone from being a plausible and well evidenced hypothesis, to being, well, an hypothesis.

Second, you have to argue that the proxy record tells us nothing relevant. It does not matter whether the medieval warm period happened. Or the little ice age. All that, you have to concede, was a red herring. We do not know whether this period is unprecedentedly warm. It may have been warmer in 1300. So we have to find some other reason for finding the alleged warming alarming, and it seemed like the instrumental record did that. But, but....

Third, in the light of the Jones revelations, we finally have to admit that the instrumental temperature record cannot be relied on as an indicator of global temperature changes this century. We have no reliable evidence from surface measurement stations which can show that the global temperature is doing anything remarkable.

But, you still believe that there is catastrophic warming. How to proceed?

One gambit is to argue that what counts is not the measurement but the models. The logical difficulty is that if we cannot verify the models against measurements it is hard to see how we know they are valid, and that their predictions can be trusted. If you have this much faith in models, you need to know more about how they are made. Read [11].

Another is to argue, what we have to rely on to prove the existence of warming is other things than proxies or the measured temperature record. Perhaps ecological evidence from time at which plants flower, and geographical extent of species. These show warming. We have to work at proving that they are dramatically different from in the past. Hurricanes? Well, we cannot rely on them, they have been proved no worse and no better than in the past. Arctic ice melts?

Promising, very promising. The difficulty is, maybe they are caused by particulate deposits changing the albedo, rather than general warming. Glacier retreat? Again, very promising. The trouble is, we have surveyed such a small fraction of the thousands of glaciers, and there are real debates about whether the retreats that have been observed are due to increased rainfall or increased warmth. Rise in temperature of lower troposphere in the tropics, faster than measured surface temperatures? Yes, this one would be a killer proof. If this is happening, it could only be caused by greenhouse warming.

Is it happening? Well, no. At least, not for the last few years. Oh dear.

The jury is still out to some extent. It is still possible that we will find evidence that there is catastrophic man-made global warming. It is true that CO2 levels have risen, most likely due to human activity, and that this is indeed a first in planetary history. Whether and when it will lead to warming, and if so of what magnitude, is an interesting and important question.

What we can say for sure however about the Hockey Stick is that after 10 or 15 years struggle to get to the underlying data from the first seminal paper on the subject, now that we have it, we can be sure that the Hockey Stick is totally discredited. The point at which this happened is, we predict, likely to be recorded later as a turning point in 21st Century intellectual history, and attitudes to science and public policy.

It will be the point which people later remember as the time when we came to realize that the story of the great Global Warming panic so far has belonged to the history of hysteria, not of climate. Whether this moment will be succeeded by the application of science to the issue, well, right now that hangs in the balance. We hope so. But we are not holding our breath.


[1] Jones, P.D., et al., 1990: Assessment of urbanization effects in time series of surface air temperature over land. Nature, 347, 169-172 The posting of the data is at:

[2] Mann, M.E., R.S. Bradley, and M.K. Hughes, Global-scale temperature patterns and climate forcing over the past six centuries, Nature, 392, 779-787, 1998.




[6] Quoted in the public draft version of "Wei-Chyung Wang fabricated some scientific claims", Douglas J. Keenan, 19 June 2007.



[9] Do a search in the browser for Hughes Jones. The posting with the quote is by MacIntyre, about halfway down the page.

[10] For a clear explanation, read McKittrick's account: Ross McKitrick, 'What is the Hockey Stick Debate About?', APEC Study Group, Australia, April 4, 2005. It predates the release of all the data, but its clear about the statistics.

[11] 'Seductive Simulations? Uncer tainty Distribution Around Climate Models', Myanna Lahsen, Social Studies of Science 35/6(December 2005) 895–922

Keenan's draft paper

For more on a wide array of other topics, please visit the weblog.


HTML, format and art copyright © 2007 Charles Hugh Smith, copyright to text and all other content in the above work is held by the author of the essay as of the publication date listed above. All rights reserved in all media.

The views of the contributor authors are their own, and do not reflect the views of Charles Hugh Smith. All errors and errors of omission in the above essay are the sole responsibility of the essay's author.

The writer(s) would be honored if you linked this Readers Journal essay to your site, or printed a copy for your own use.


  Readers Journal     weblog/wEssays     home