In June 1999 a group at the Lawrence Berkeley National Laboratory announced in Physical Review Letters that they had made elements 118 and 116 by firing krypton-86 at a lead target. The reaction had been predicted to have an unexpectedly high yield, the result was consistent with that prediction, and it was accepted.
Nobody could reproduce it. Berkeley itself could not reproduce it on repeat runs; neither could GSI in Darmstadt nor RIKEN in Japan. In July 2001 the authors published a retraction. A subsequent internal investigation found that the decay chains had not been present in the raw data at all and had been inserted during analysis. The principal author, Victor Ninov, was dismissed in 2002. He denied wrongdoing.
The case is worth recording precisely because superheavy element work is the kind of science in which fabrication is hardest to catch: the events are single atoms, the analysis is a search for rare patterns in a very large dataset, and the number of people able to check is small. What caught it was replication, which is the only thing that ever does.
It caused real damage beyond the retraction. Berkeley had proposed the name ghiorsium for element 118, after Albert Ghiorso; that was withdrawn along with the claim. And the episode remains a standing reason for the caution the international commission now applies to superheavy claims, including its rejection of both competing bids for element 113 in 2011.