United Nations officers inspect what is alleged to be a North Korean torpedo. Credit: SONG KYUNG-SEOK/GETTY IMAGES

It seemed like an open-and-shut case. In May, two months after the South Korean warship Cheonan suddenly sank in the Yellow Sea, the country released the findings of its investigation, blaming a torpedo attack by its northern neighbour.

The evidence included a smoking gun: parts of a torpedo found near the ship had the same dimensions as those in North Korean munitions pamphlets, and ink marks identified the torpedo as North Korean.

But North Korea has consistently denied the attack, demanding that the United Nations (UN) coordinate a joint North–South investigation. The UN released a long-awaited statement on 9 July, condemning the incident, which led to 46 deaths, but it conspicuously failed to blame North Korea.

On the same day, at a press conference in Tokyo, scientists raised issues about the South Korean investigation that they believe cast doubt on North Korea's involvement. Seung-Hun Lee, a South Korean-born physicist who works at the University of Virginia in Charlottesville, says that the most problematic point is the claim that material from the Cheonan was adsorbed onto the torpedo's propeller. Although electron-dispersive spectroscopy found a match between the adsorbed material and the ship, X-ray diffraction analysis found no sign of the expected aluminium or aluminium oxide from the ship.

To explain this discrepancy, the investigation team suggested that, after the explosion, the material cooled rapidly into an amorphous form that cannot be detected by X-ray diffraction. But this supercooling is a delicate process, says Lee. "It's impossible that 100% of it would be amorphous." Lee's experiments show that aluminium left over from a detonation would mainly be crystalline.

Lee's analysis, posted online on 3 June (S.-H. Lee and P. Yang, preprint at http://arxiv.org/abs/1006.0680; 2010), was soon bolstered by independent work from Panseok Yang, a laboratory manager specializing in mass spectrometry at the University of Manitoba in Winnipeg, Canada. Yang's data, added on 28 June to Lee's report, show that the ratio of oxygen to metal in rapidly cooled aluminium oxides would be much lower than that found by South Korea's investigation, and that the material identified could be the torpedo's own corroded aluminium.

Lee also says that it is unclear how the ink, which apparently identified the torpedo as North Korean, could have survived the heat of the detonation. He suggests that the Cheonan might have been hit by a mine or rammed by another ship — a suggestion also raised by Shin Sang-chul, a former officer in the South Korean navy who was part of the investigation team.

This week, a US-led UN team plans to meet North Korean military representatives to discuss the sinking. The South Korean government has adamantly denied fabrication or major problems with its interpretation of the data. Others also doubt that there is an alternative interpretation. "Aside from the science, it is consistent with North Korea's behaviour in the past," says James Schoff, an Asian regional security expert at the Institute for Foreign Policy Analysis in Cambridge, Massachusetts. Referring to South Korea's investigation, he says: "I have no doubts personally that the conclusion is correct."