This fascinating article in The New Yorker describes the phenomena of the unrepeatability of scientific studies. The scientific method depends upon the proposition that an experiment, properly designed and executed, yields results that can be replicated if the experiment were followed again at a later time. We regard results from a properly conducted experiment to be scientific truth, so the extent that such a thing can be ascertained.
We're accustomed to seeing results of experiments challenged due to faults in experimental protocol. The author of this article, however, describes something more unnerving - the apparently common problem of being unable to replicate results of an experiment without faults in its protocol, thereby raising doubts about the ability of the scientific method to reveal any truth.