The Indonesian volcano Mount Merapi continues to rumble away. But perhaps the most famous volcanic eruption ever is that of Merapi’s neighbour, Krakatoa, in 1883. The sound was heard over 3,000 miles away, and giant waves at least 40 metres high killed over 35,000 people. Yet the explosion of Tambora in 1815, also in Indonesia, was much more powerful. It was the most violent volcanic eruption of historical times. Vulcanologists have developed a scale to compare the eruptions. The volcanic explosivity index (VEI) is similar to the Richter scale for earthquakes. Each point on the VEI represents an explosion some ten times bigger than the previous point. Mount St Helens in Washington state in 1980 measured five on the VEI, Krakatoa six and Tambora seven.
A statistical law has been discovered which links the size and frequency of volcanic eruptions. Tom Simkin and Lee Siebert of the Smithsonian Institution in Washington DC have compiled a database of eruptions over the past 10,000 years, focusing on ones with VEI of two or more. For every point increase on the VEI, explosions are around one sixth less frequent. So for every 15,000 eruptions of VEI two, there are 2,500 of VEI three, 400 of four, 60 of five, ten of six and only one—Tambora—of seven.
The size and frequency of earthquakes are linked by a similar statistical relationship: the number of earthquakes of any given size on the Richter scale reduces quite rapidly with the size. The discovery of the relationships does not mean that we can predict when the next super-eruption will take place. But it does tell us how likely we are to see an event of any given size within a given time interval: the next century for example.
Wars have a pattern
The size and frequency of wars appear to be connected to each other in the same statistical pattern followed by volcanic eruptions and earthquakes. In the 1920s, the British physicist Lewis Fry Richardson drew up estimates of the number of people killed in 82 wars since 1820. He converted the data into logarithms, so that a conflict with 1,000 deaths would be graded as a three (ten times ten times ten=1,000; in other words, ten to the power of three), and a massive one with 1m deaths would be a grade six. He plotted the grade of the war on one axis and the number of wars of a given grade on the other. The plotted data followed a very clear relationship. The number of wars of any given grade falls away rapidly with the grade.
Later updates of the list, including the second world war confirmed the connection between size and frequency. The first and second world wars remain the only wars of grade seven, and between them account for 60 per cent of all deaths in the list of 315 wars.
Even military historians might struggle to name all wars that reached grade six: the Taiping rebellion (1851–64), the American civil war (1861–65), the great war of the triple alliance in Paraguay(1864-70), the civil war in Russia (1918–20), the first Chinese-Communist war (1927–36), the Spanish civil war (1936–39) and the communal riots in the Indian peninsula (1946–48).
Again, we cannot use these patterns to predict how many people will be killed in Iraq. But we can estimate how likely is an event of any given size in any particular year or decade.