Nature has an article comparing Wikipedia to Encyclopaedia Britannica. It focuses on science articles, but the results are great news for fans the Wiki concept in general and the whole “knowledge wants to be free” crowd.
Wikipedia is only marginally more inaccurate Britannica, at least for science-related articles. Not surprisingly, the most error-filled articles were the history of science ones; after all, history of science is much more cognitively sophisticated and complex than science itself.
(I say that only half-jokingly. Recovering the complexities of past knowledge is quite a bit harder than finding the current scientific version. However, the history of science articles were biographical, so factual errors and mistatements were mainly the issue.)
Then again, when you have Michael Gordin reviewing Mendeleev articles, he’s bound to be able to find errors in just about anything not written by him. I have his Mendeleev book (and it’s excellent); I wish I had time to go through it again to fix the article, as there aren’t many historians of science editing Wikipedia.