Comment by jonathan-re

Well generally yes, but that isn’t how it works there. Since the things they want to measure in their experiments are so atomically small, sensor noise becomes a huge problem. So it’s not enough to find the sensor readings for NewMysteriousParticleX to be sure that it actually exists, it could just have been noise. So you have to run the experiment again and again until your datapoint is statistically significant enough that you are sure, it wasn’t just noise. A couple of years ago there was this case where they almost found a new particle, the significance was pretty close to the threshold - the problem was that this particle was not expected and would have shaken the foundations of particle physics. Some weeks later the particle has vanished back into the abyss of noise.