How often does it happen that you hear a story about a great result, that turns out to be irreproducible only a year later?  Well, pretty often.  But how often do you hear that the reason for the irreproducibility was in fact, discovered, and the 'culprit' cleared of any wrongdoing?  About once a decade is my guess.  I heard inspirational stories like this during my undergrad actually.  Here is another one for you, in a bit more technical detail than the blurb in CEN News.

http://pubs.acs.org/cgi-bin/sample.cgi/orgnd7/2007/26/i04/pdf/om061009y.pdf

First off, here's the link to the CEN article.

http://pubs.acs.org/isubscribe/journals/cen/85/i04/html/8504reproducibility.html

Now, I read the original article before.  It involved a Zirconium mediated CC bond breakage and reforming reaction, which is significant since the CC bond being broken was not that special at all.  It was the CC bond in tetraphenyl ethane and since the CC bond was  a simple, non-special sigma bond, it had implications for alkane metathesis.  I was actually researching this topic at the time.  And if you are interested in alkane metathesis yourself, then you'll be in for a treat since  Alan Goldman will be here very soon, and from what I gleaned by reading the propaganda brochure for the talk, he'll be talking about alkane metathesis chemistry that led to that Science article last year.

Anyways, back to the issue at hand.  The Eisch group were able to observe the intermediate where Zr was in between the CC bond and did some nice deuterium labelling experiments.  The CC bond could be reformed, suggesting that maybe if one diphenylCH could be replaced by another alkane while it is attached to the zirconium, alkane metathesis could occur.  Now, the reaction was never catalytic, but it is a very difficult field and any new advance is much appreciated.  Another interesting result that the researchers can report is the stitching together of two THF molecules and capping them with hydrolysis to form OH-(CH2)8-OH. 

However, when the research done by Dutta was continued by the Gitua, he found out that, in fact, no CC bond breaking occurs at all.  You should really read the article since it is written in a very good, conversational style, with long and descriptive footnotes.  You get a sense for the frustration that was felt by the researchers when they found out that "no such C-C bond cleavage occurred whatsoever".  In Italics!  After about three months of guessing and second guessing, during which a less experienced group head would have pulled the plug on the whole thing, they decided to see if maybe the stir bar PTFE coating had come off.  So they threw in some catalytic iron.  And the reaction worked.  Not too well, but it worked.  So maybe the stir bar that Dutta used was a little worn out and some iron was showing.  However, it must have had special 'nanoparticle' Fe(acac)3 deposits, since catalytic iron still didn't really increase yield to what they were earlier.  But since Eisch has been a professor forever and has seen pretty much everything, they also decided to put the flask next to a light bulb to see if that did anything.  And indeed it did.  In fact, the "light-bulb near the flask and catalytic iron" method yielded CC cleavage products in 100% yield!! (well, 99.9% if I was the one reviewing the article).  In fact, they state in a footnote that Dutta had a light-bulb only 1.5 meters above his desk where he did a lot of his reactions, while Gitua, who 'likes to operate in the dark' (my quote), had a light-bulb that was almost three meters away.  And on a happy, final note, the Eisch group got another Organometallics paper out of the whole saga with better and reproducible results.  (An unhappy note unfortunately, is that it's a photochemical reaction with iron catalysis.  Good luck trying to figure that mechanism out).

To me, this story shows that there are many sources of irreproducibility when trying to repeat a reaction.  Be it poor lab technique, or simply something in your set-up that was vital to the whole process, but that the authors did not mention due to oversight, their assumption that you are familiar with "this simple technique", or their ignorance of what the hell is going on in reality.  Just four years ago, I used to curse to myself in frustration when I found that another Tetrahedron Letters procedure failed for me.  Today, I am older and wiser and perhaps I will not curse the procedure.  I will simply put it aside and move on the alternative Chem Comm. method.  After all, it's always easier to accuse than to delve into the details.  The Eisch group story shows that there are very tangible benefits for those who choose the latter route however.