According to the Beasley report, Schön used many different techniques to fabricate and falsify data for his papers. These ranged from the use of mathematical formulae instead of data to plot points on a graph, to reusing data from one experiment in plots associated with another one. In addition, says the report, he often changed his account of how data were collected, when questioned by investigators.

One experiment used large electric fields to induce superconductivity in carbon buckyballs (J. H. Schön, Ch. Kloc & B. Batlogg Nature 408, 549; 2000). The report catalogues how Schön used a mathematical function to create a graph of the buckyball's electrical resistance as a function of temperature.

“Figure 32 [Fig. 1 of the paper (right)] shows an amazing systematic variation of the resistance versus temperature,” the report says. “Even the deviations are reproduced with high precision between the various curves... It is clear that these are not real data: they have been generated using a mathematical function.”

Initially, Schön described how the apparent smoothness of the curve could result from smoothing of real, noisy data points. But, when faced with the reality that such smoothing could not explain the extraordinary fit, he acknowledged that the data were in fact spliced-in analytic functions. “I thought that a smoother curve would look much better,” he told the committee.

After admitting this falsification, Schön supplied a plot showing more realistic curves, but not the original data. Schön told the panel that he had deleted all his primary data because his computer at Bell Labs was short of memory.

→ http://www.lucent.com/news_events/researchreview.html