Why the UN should apologize for its discredited “hockey-stick” graph

The UN’s second assessment report, in 1996, had displayed a 1,000-year graph showing that temperature in the Middle Ages was warmer than today’s. But the 2001 report contained a new graph showing no mediaeval warm period. It wrongly concluded that the 20^th century was the warmest for 1,000 years. The graph looked like a hockey-stick. The wrongly-flat 1000-1900AD temperature line was the shaft: the uptick from 1900 to 2000 was the blade. Here is how they falsified the graph:

• They gave one technique for reconstructing pre-thermometer temperature 390 times more weight than any other (but did not say so).
• The technique they overweighted was one which the UN’s 1996 report had said was unsafe: measurement of tree-rings from bristlecone pines. Tree-rings are wider in warmer years, but pine-rings are also wider when there’s more carbon dioxide in the air: it is plant food. This carbon dioxide fertilization distorts the calculations.
• They said they had included 24 datasets going back to 1400. Without saying so, they left out the set showing the mediaeval warm period, tucking it into a folder marked “CENSORED_DATA”.
• They used a computer model to draw the graph from the data, but scientists later found that the model almost always drew hockey-sticks even if they fed in random, electronic “red noise”.

The large, full-colour “hockey-stick” was the key graph in the UN’s 2001 report, and the only one to appear six times. The Canadian Government copied it to every household. Four years passed before a leading scientific journal would publish the truth about the naff graph. Did Kofi or Ottawa apologize? Did they heck. The UN still uses the graph in its publications.

After the UN’s graph was exposed, several scientific papers apparently confirming its abolition of the mediaeval warm period appeared. The US Congress asked independent statisticians to investigate. They found that the graph was unmeritorious, and that known associates of the scientists who’d compiled it had written many of the later papers supporting its conclusion.

The UN, echoed by the Stern review, says the graph is not important. It is. Scores of scientific papers show that the mediaeval warm period was real, global and up to 3C warmer than now. The following few have been chosen because each illustrates the existence of the mediaeval warm period with a graph:

Bjorck et al. (2006) inferred general climatic conditions from "sedimentology, geochemistry, diatom analyses, magnetic properties, and multivariate statistics, together with ¹⁴C and ²¹⁰Pb dating techniques" applied to a core obtained from the center of a small crater lake on the Azores island of Pico (38°26'N, 28°12'W). The mediaeval warm period was broadly characterized by the adjoining "cooler/drier periods" of 400-800 and 1300-1800 cal yr BP, but the authors found it most strongly expressed between AD 1000 and 1100.

Chu et al. (2002) derived a climatic history from the geochemistry of dated sediments taken from Lake Huguangyan (21°9'N, 110°17'E) on the Leizhou Peninsula of South China, as well as from information contained in historical documents.  The Mediaeval Warm Period occurred between AD 900 and 1300.

Gupta et al. (2005) derived a high-resolution record of variations in the Indian summer monsoon from relative abundances of the planktic foraminifer Globigerina bulloides which they had obtained from a sediment core off the coast of Oman in the northwestern Arabian Sea (18°03.08'N, 57°36.56'E),  indicating that southwest monsoon winds were stronger during the Mediaeval Warm Period (AD 800-1300), coincident with a period of high solar activity.

Hallett et al. (2003) compared the dendrochronological fire history of the forested area surrounding Dog Lake (50.77°N, 116.1°W) with a high-resolution charcoal record derived from a sediment core extracted from the lake to reconstruct the region's fire history over the past 1000 years. In addition, the authors constructed a proxy record of lake-level change based on accumulation rates of Chara globularis-type oospores in the lake sediment core. These analyses revealed the presence of frequent forest fires and lowered lake levels during the Mediaeval Warm Period (AD 980-1270), which in the words of the authors support evidence of "warmer and drier climate than today."

Holzhauser et al. (2005) present a high-resolution record of glacial variation for Gorner glacier, in the Swiss Alps (~46.05°N, 7.62°E), as part of an effort to develop a 3500-year climate history of west-central Europe. In their estimation, "at no other glacier in the Swiss Alps ... [is] the Mediaeval Climatic Optimum so well documented as at the Gorner glacier," especially when the glacier retreated to levels beyond that of the present-day between AD 800 and 1100. Because glaciers in mountain areas are "highly sensitive to climate changes and thus provide one of nature's clearest signals of warming or cooling and/or dry and wet climate periods," as they describe it, "one can say that the quasi periodical fluctuations of Alpine glaciers were driven by glacier-hostile (warm/dry) and glacier-friendly (cool/wet) periods." On this basis, therefore, one can cautiously conclude that temperatures at Gorner Glacier were likely warmer during the Mediaeval Warm Period than they have been recently.

Holzhauser et al. (2005) also present a high-resolution record of glacial variation for the Great Aletsch glacier in Swiss Alps (~46.38°N, 7.75°E), as part of an effort to develop a 3500-year climate history of west-central Europe. As they describe it, the Mediaeval Warm Period occurred between AD 800 and 1300; and based on data presented in their Figure 2 (reproduced below), glacial extension between AD 800 and 1000 was at a level equal to that of today. Furthermore, because glaciers in mountain areas are "highly sensitive to climate changes and thus provide one of nature's clearest signals of warming or cooling and/or dry and wet climate periods," in their estimation, they state that "one can say that the quasi periodical fluctuations of Alpine glaciers were driven by glacier-hostile (warm/dry) and glacier-friendly (cool/wet) periods." On this basis one can logically, albeit cautiously, conclude that temperatures during the Mediaeval Warm Period were likely as warm as they are today at the Great Aletsch Glacier.

Holzhauser et al. (2005) also present a record of glacial variation for the Lower Grindelwald glacier, in the Bernese Alps, Switzerland (~46.58°N, 8.00°E), as part of an effort to develop a 3500-year climate history of west-central Europe. The Mediaeval Warm Period was identified in their Figure 2 as a period of significant glacial recession between AD 800 and 1300.

Khim et al. (2002) inferred general climatic features from a study of the grain size, total organic carbon content, biogenic silica content and, most importantly, magnetic susceptibility of ²¹⁰Pb- and ¹⁴C-dated sediments retrieved from the eastern Bransfield Basin (61°58.9'S, 55°57.4'W) just off the northern tip of the Antarctic Peninsula.  Most of the Mediaeval Warm Period (AD 1050-1550) was warmer than the Current Warm Period.

Kitagawa and Matsumoto (1995) analyzed δ¹³C variations of Japanese cedars growing on Yakushima Island, southern Japan (30°20'N, 130°30'E), in an effort to reconstruct a high-resolution proxy temperature record over the past two thousand years. The Mediaeval Warm Period occurred between AD 800-1250 and from the authors' Figure 3, peak warmth during this time was about 1°C above that of the Current Warm Period.

Mangini et al. (2005) developed a highly-resolved 2000-year record of temperature with better than decadal resolution from a stalagmite recovered from Spannagel Cave in the Central Alps of Austria (47.09°N, 11.67°E). The highest temperatures of the past two millennia occurred during the Mediaeval Warm Period (AD 800-1300) and were “slightly higher than those of the top section of the stalagmite (1950) and higher than the present-day temperature.” In fact, at three different points during the mediaeval warm period, their data indicate temperature spikes in excess of 1°C above present (1995-1998) temperatures of 1.8°C.

Noon et al. (2003) inferred primarily summer climatic conditions from a δ¹⁸O record preserved in authigenic carbonate retrieved from sediments of Sombre Lake (60°43'S, 45°38'W) on Signy Island, maritime Antarctica.  The Mediaeval Warm Period (AD 1130-1215) was warmer than the Current Warm Period.

Pla and Catalan (2005) analyzed chrysophyte cyst data collected from a sediment core obtained from Lake Redon (42.64°N, 0.77°E) in the Central and Eastern Pyrenees of northeast Spain, producing a history of winter/spring temperatures for this region throughout the Holocene. The mediaeval warm period (~ AD 875 to 1000) was categorized as the "warmest period" of the record, with temperature about 0.25°C warmer than it is currently.

Qiang et al. (2005) conducted stable carbon isotope analyses on sediment cores taken from Lake Sugan (38°51.19'N, 93°54.09'E), in the northeastern region of the Tibetan Plateau, to produce a proxy of winter temperatures over the past 2,000 years. The results indicated a warm and dry period between 580 and 1200 AD, which they state "corresponds to the Mediaeval Warm Period." A view of the data in the author's Figure 3 reveals the mediaeval warm period was probably at least as warm between ~AD 1100 and 1200 as it is presently .

Rein et al. (2005) derived sea surface temperatures from alkenones extracted from a high-resolution marine sediment core retrieved off the coast of Peru (12.05°S, 77.66°W), spanning the past 20,000 years and ending in the 1960s. From their Figure 11, adapted below, it can be seen that the warmest temperatures of this 20,000 year period (~23.2°C) occurred during the late Mediaeval time (AD 800-1250). Taking this value, 23.2°C, and comparing it with the modern monthly long-term means in sea surface temperature, which the authors characterize as between 15°C and 22°C, we estimate the peak warmth of the Mediaeval Warm Period for this region was about 1.2°C above that of the Current Warm Period.

Seppa and Birks (2002) reconstructed July mean temperatures from a pollen profile of the sediments of Toskaljavri (69°12'N, 21°28'E), a tree-line lake in the continental sector of northern Fennoscandia.  The mediaeval warm period occurred between AD 600 and 1000 and was 0.8°C warmer than today.

Tyson et al. (2000) reported that maximum annual air temperatures in the vicinity of Cold Air Cave (24°1'S, 29°11'E) in the Makapansgat Valley of South Africa were inferred from a relationship between color variations in banded growth-layer laminations of a well-dated stalagmite and the air temperature of a surrounding 49-station climatological network developed over the period 1981-1995, as well as from a quasi-decadal-resolution record of oxygen and carbon stable isotopes.  The mediaeval warm period (AD 1000-1325) was as much as 3-4°C warmer than the Current Warm Period (AD 1961-1990 mean).

Williams et al. (2005) inferred temperatures from δ¹⁸O data obtained from four stalagmites found in caves at Waitomo (38.3°S, 175.1°E) on New Zealand's North Island for which 19 TIMS uranium series ages were measured.  The Mediaeval Warm Period occurred between AD 1100 and 1400 and was warmer than the Current Warm Period.

Wilson et al. (1979) derived temperatures from an ¹⁸O/¹⁶O profile through a stalagmite found in a New Zealand cave (40.67°S, 172.43°E), revealing the mediaeval warm period to have occurred between AD 1050 and 1400 and to have been 0.75°C warmer than the current warm period.

Christopher Walter, Third Viscount Monckton of Brenchley, is a former policy advisor to Margaret Thatcher during her years as Prime Minister of the United Kingdom.  He may reached through SPPI, or directly at (+44 1882 632341)( This e-mail address is being protected from spam bots, you need JavaScript enabled to view it )

Robert Ferguson, President

This e-mail address is being protected from spam bots, you need JavaScript enabled to view it

209 Pennsylvania Ave., SE

Suite 299

 Washington, D.C 20003

www.scienceandpublicpolicy.org  (202) 288-5699

*Views here expressed are those of the author, and not necessarily those of SPPI.