This is a guest post by Eugenie Samuel Reich, a contributing correspondent for Nature.

Last time I posted on The Intersection, a couple of commenters were curious about a disclosure I made about having brought a Freedom of Information Act (FOIA) lawsuit against the US Department of Energy (DOE), to obtain a report into alleged scientific fraud at Oak Ridge National Lab (ORNL).

So I thought that Intersection readers might be interested in my article, out today in Nature, reporting on the lawsuit’s progress.

The fraud allegations in the case date back to 2006. That year, a group of researchers led by Stephen Pennycook were accused of fabricating data; which they have strongly denied. Pennycook’s research involves developing cutting-edge techniques for imaging materials using an electron microscope, which can solve problems in nanotechnology, energy research, and condensed matter physics. His group receives about $2 million per year from DOE.

The data that prompted concerns included one example, discussed in this 2006 Boston Globe article, where the group had mirrored and spliced datasets together to represent a scan of a sample that had allegedly never been studied for real. In another example, the group had replaced some, but not all, of the electron energy loss spectra taken from a sample with allegedly completely different ones when a reviewer suggested the scientific conclusion that atomic-scale resolution had been achieved were not fully supported by the originals.

Following federal policy on research misconduct, ORNL held an investigation. The investigation panel of three scientists exonerated the group of misconduct, finding they were guilty of no more than errors of judgment and careless errors. In order to understand how this conclusion was reached, I sought the investigation report under FOIA, and went through several rounds at the DOE appeals process (with at one point, the Office of Hearings and Appeals ruling in my favor, only for the Office of Science, which had the report, to dispute its conclusions. In 2008, a summary statement by the investigators was released, but I continued to pursue the original report. By 2009, I had received a final answer that DOE wouldn’t release it, and so sued the agency in district court in Boston, where I am based.

In fighting the case, the government has gone to great lengths to argue that the investigation report it received is not a government record subject to FOIA. Its rationale for this is that ORNL is run by a private contractor, UT Battelle, which owns the report, and that DOE officials did not read or rely upon it when they approved the investigation. The court filings include sworn declarations from officials involved in approving the investigation saying they didn’t read the final investigation report, and later gave it back to UT Battelle. They say they approved the exoneration through phonecalls and meetings with an ORNL manager. They imply this is typical for science oversight at the DOE national labs and consistent with federal policy that requires the investigation report and evidentiary record to be documented by the funding agency for oversight purposes.

An editorial in Nature accompanying my article characterizes the problem this way:

Important decisions were taken informally by a small group of officials and an adviser who apparently shared a common interest: to see the matter quietly resolved. The procedure ought to be more formal, better documented and even adversarial, with the institutional managers required to satisfy officials whom they do not know and who have no stake in the case outcome. This is closer to the more careful oversight of alleged misconduct by both the Office of Research Integrity at the Department of Health and Human Services (DHHS) and the inspector-general of the National Science Foundation.

Not reading the report apparently had an advantage for DOE: the judge ruled in March that the record was not subject to FOIA in part because the government had never read it. I’ve filed a motion to reconsider, and the case is still pending.