I've heard about the dog eating homework excuse, but this article in the Wall Street Journal on a 1992 study about heart attack outcomes and diet is something else. From the article:
A year later, Richard Smith, the journal's editor, received two letters questioning the findings. What followed was an extraordinary inquiry stretching over a dozen years and 5,000 miles. Along the way, Dr. Singh contended that termites had eaten crucial data and Dr. Smith spent four years begging a busy statistician to deliver a report.
Finally this July, the journal, called BMJ, lowered the boom. Editors said they had "reasonable grounds to doubt the validity of the 1992 paper" and published an article calling the data in another paper submitted in 1994 "either fabricated or falsified."
Dr. Singh, while acknowledging some missteps, says he is the victim of scientists who don't understand what it's like to work without steady electricity or research grants.
The article goes on to talk about the difficulties involved in journal editors' finding and exposing questionable or downright fake results. Definitely worth a read for anyone involved in enduring materials.