Why is it important to repeat the experiment

Quality assurance in researchRepetition in science as a safety check

University of Frankfurt, Department of Molecular Microbiology and Bioenergetics. The hallways are long, the rooms crammed with strange appliances, and people wear smocks.

"We can get the culture out of that for now."

Doctoral student Lisa Hitschler opens a heating cabinet and takes out a small bottle in which a slightly cloudy liquid sloshes. A bacterial culture.

"Well, they're here in the little serum bottles, they have a gas atmosphere in them. In that case, that's CO, carbon monoxide."

Lisa Hitschler has been working for weeks on getting the Thermoanaerobacter kivuii bacterium to grow to 100 percent carbon monoxide, i.e. pure poison. A simple measurement can show whether the experiment was successful.

"What we're doing now is measuring OD, so optical density is measured to track growth."

The less translucent the liquid, the more bacteria it contains. The biologist draws up a sample with a syringe and transfers the cloudy liquid into a small vessel. She measures the light transmission in a photometer.

"It is currently at 0.220 after about 27 hours. At zero hours, when I vaccinated her, she had an OD of 0.02. That means that it has now risen by a factor of ten."

Repeat as a security check

She looks satisfied and not a bit surprised. On the contrary: you already know how much the bacteria will have grown by tomorrow. Her former colleague Marie Weghoff published two years ago how Thermoanaerobacter kivuii can be made to live on carbon monoxide. Lisa Hitschler is just re-cooking this experiment. Waste of time? No way, says Mirko Basen, who oversees Lisa Hitschler's work.

"I took up the project and we are now trying to understand how the metabolic pathway works. And we are doing studies in the laboratory so that we can now repeat the experiments in the case of Ms. Weghoff and then we want to find out whether, for example have changed things in the genome. So whether there have been mutations that might have led the bacterium to grow on carbon monoxide. "

Reproducing an old experiment is the first step in answering a new question. This principle applies in biology as well as in psychology or physics. But repetition also plays an important role beyond that.

"We also very often reproduce our own data in the laboratory."

Lisa Hitschler sets out to prepare a new culture bottle. The bacteria should prove their ability to grow one more time. A kind of security check.

"If we come to significant deviations, then we would repeat that and repeat it again to see if we simply made an individual mistake - was the medium perhaps not well prepared or the tribe not well prepared and if we always did would find a deviation again, then we would begin to investigate why this deviation is there. "

Coincidence or Law?

Multiple repetitions of an experiment should ensure that the results are not unique products of chance. After all, it is about creating generally valid, resilient knowledge. To distinguish between coincidence and regularity is often more difficult than you might think. In a survey by the journal Nature in 2016, 50 percent of the scientists questioned said that they were unable to reproduce one of their own results. Repeating the experiments of colleagues even caused problems for 70 percent of those questioned at least once.

"In fact, I also tried in another laboratory to exactly repeat a protocol for the genetic modification of a microorganism, and I just never succeeded as well as the working group that originally developed it. They just have their own Handles and tricks that you can't see, where the group didn't even think that this might be important. "

Lisa Hitschler takes care of her bacteria and then adds the measurements of the day in her laboratory book.

So you can see here - the values ‚Äč‚Äčthat I have taken correspond to those of Ms. Weghoff at the respective probationary times. For example, at 30 hours it has a value of 0.08 and I have a value of 0.09. So it fits in with the values. We were able to reproduce the experiments.

For the doctoral candidate, the real work is only just beginning. Reproduction may be an important part of scientific work. In the long term, what counts is what you have found out. Results that can be published in a specialist magazine. Confirmation of an underlying experiment is not one of them.