For every active lab scientist, the magic number is 3. N=3 is the nirvana for every experimenter and constitutes the publishable proof!

But when is a n=3 a n=3. Of course, one has to repeat the same experiment three times – independently (triplicates don’t count!!!).

The question is however, what do you do with the experiments that didn’t work. Let’s propose you design a critical experiment that you hope will support your hypothesis. You do the experiment three times, the first time you get really nice results and your findings are exactly what you expected. When you repeat the experiment the next day, you are surprised, because nothing happened. Oh well, you probably were tired, or the lab was too warm, or it was too noisy that day, or you must have made some tiny error somewhere along the way. No problem, just do it again. And so you repeat it a third time and to add to your frustration, the results are the opposite to what you found the first time – totally contradictory to your hypothesis.

You go, present the data to your supervisor and you are told to do it again. Of course! So you repeat the experiment a forth, fifth, sixth time. The results are still variable, but you also manage to repeat the results from the first experiment a couple of times.
Finally! The solution is clear. You have a n=3 that shows nicely what you set out to prove, but what to do with all the other experiments and all that contradictory data in your lab book. Well, something must have gone wrong, you were probably tired, had too much to drink the day before, the chemicals were old, the lab was too warm, or you must have forgotten some reaction steps all together.

N=3 is n=3, so lets draw up a nice, shiny figure and submit the sucker! It took long enough!

The big question that remains: How do you know something went wrong?

I believe that this is common practice in all labs and that there is a humongous collection of unsuitable results that will be forgotten. Only the published truth remains!