An ongoing split within the HARP experiment at CERN in Geneva has come out into the open – with fierce differences of opinion between rival groups within the collaboration over how to analyse their data. One of the groups accuses the other of research that "violates standards of quality of work and scientific ethics on several counts", and its leader, CERN’s Friedrich Dydak, believes this to be a reflection of a more general decline in scientific standards at the lab.

HARP was designed to investigate a major technological challenge in the construction of a possible multi-billion dollar facility known as a neutrino factory. By slamming protons into targets made of heavy nuclei such as tantalum, HARP created sub-atomic particles known as pions, which in a neutrino factory would decay into muons and then into neutrinos. The idea was to measure the rate of pion production and therefore work out how powerful a neutrino factory’s proton source would have to be in order to generate the desired neutrino flux.

HARP ran in 2001 and 2002 but severe disagreements over the quality of data analysis within the group led to a split in 2004. Some 100 members of the collaboration remained as the "official" HARP group, while the remaining 20 formed a new group under Dydak, who had been spokesman of the collaboration as a whole but was not re-elected in 2003.

A rushed job?

The focus of the dispute was the experiment's "time projection chamber", which measures the positions and momenta of the collision products. After being approved by the CERN management in 2000, HARP was up and running in just 17 months. Dydak says that such an experiment would typically need three or four years before it could start taking data, but that HARP's development time was slashed because the management wanted to get the experiment out of the way in time for the lab's headline project, the Large Hadron Collider. The result was that the projection chamber didn't work as intended and the group had to try to compensate for its shortcomings by making painstaking corrections to the data.

Dydak's group argued that the corrections made by the official group were inaccurate, and proposed its own data analysis. The experiment's two primary funding agencies – the Italian National Institute of Nuclear Physics (INFN) and CERN – set up a review board to resolve the dispute. Chaired by Lorenzo Foa of the University of Pisa, the board came down in favour of Dydak's analysis in a report in March 2007. This position was further vindicated a few months later by CERN's SPSC advisory committee, which said that its own review of the data "calls into question the validity of the results in recent publications" of the official HARP collaboration, and also noted that the collaboration had not fully cooperated with its investigations.

According to Dydak, the official group has overestimated the pion production rate inside the detector by about a factor of 1.5, which, he points out, would lead to the beams in an eventual neutrino factory being only two-thirds as intense as they should be. He is angry that a paper published earlier this year by the official group (arXiv: 0903.4762) was allowed onto CERN’s preprint server even though the group’s analysis had been previously discredited, and that the rebuttal by his group was not allowed onto the server (arXiv: 0909.2745). "The authors of the official group mislead the reader by avoiding any reference to relevant published work that is critical of their analysis," Dydak told "This is unethical."

The official HARP group said it was too busy to comment.

'Disturbed by the statements'

However, Sergio Bertolucci, who was vice-president of the INFN at the time of the SPSC review, believes that Dydak has overstepped the mark. "I am disturbed by the statements of Dydak's group. It is unacceptable to accuse about 100 other colleagues to be scientifically unethical. These are very respectable people." He points out that Dydak's paper was not allowed onto the preprint server because it did not meet one of the basic criteria for acceptance – that papers should not contain offensive or inflammatory statements. "If the statements are removed, we will be happy to publish it," he adds.

In fact, Bertolucci, who is now director of research at CERN, believes that the Foa committee was wrong to come down so clearly in favour of Dydak’s group. He thinks that "both groups are suffering from the poor quality of the raw data," and that "both analyses are trying to regain as much quality as possible" from the flawed experiment. He strongly believes that the differences should have been sorted out within the group, a position that is reflected in a change of policy about to be introduced by CERN. From now on, says Bertolucci, papers from experimental groups will be posted on the preprint server with CERN’s knowledge but without requiring the lab’s explicit approval. Until now CERN has generally filtered papers appearing on the server to ensure they fulfilled basic criteria of scientific quality, but Bertolucci says that will not be possible owing to the size and complexity of the collaborations involved with the Large Hadron Collider. “Why should someone arrogate to themselves the right to judge these results?” he asks.

Rules of democracy

Dydak believes this new approach is seriously flawed. "I feel extremely badly about the way that things are going here," he says. "Science is no longer of prime importance. The rules of high-quality scientific work are being replaced by the rules of democracy. That is appropriate in politics but not in science."