Out at CERN, the highest energy machine in human history (yes, it unfortunately outdoes what we had at Fermilab by several times now), a hope for a new, unpredicted particle seen in some of the data last year, has statistically gone away as new data runs have been added. This is a good example of how science works, where one does not just take some new little spike in a data set as a new discovery and, in this case, a sign for new physics beyond the Standard Model (which is the theory that explains and describes what we know about particles, matter, and forces).
Each branch of science has standards that must be met before one can claim a discovery, and in particle physics it is a '5-sigma rule' where there must be, effectively, small enough uncertainties that will allow a 1 in 3.5 million chance that the discovery is something other than what we think it is, i.e. a miniscule chance that the data are showing a fluke. I know about this firsthand when, back in grad school and on an experiment searching for the top quark, we had a mass peak but, due to the statistics of the data set, we only were in the 2 sigma or so range that this was the top quark. We felt pretty confident that it really was, since there were multiple checks and re-checks of the detector, software, and data quality, but we could not claim discovery - we instead published a paper on the 'evidence for the top quark.' It took another year of millions more collisions before we reached the 5-sigma standard, and then claimed discovery. I am glad to see these scientists are doing things the right way, and upon further review and new data, found out the peak seen earlier is apparently a statistical fluctuation.
See a Scientific American article here.