The investigation of the causal relationships among variables or the testing of a hypothesis is called Experimentation. Experimentation under controlled conditions is fundamental to the ideal of science. Both the scientific method and repeatability works best with controlled experiments but not all area of scientific study lend them selves to experiments. There are a lot of real world observations that are beyond the possibility of a controlled experiment do to things like distance, size and time. In some cases such as in historical sciences experiments can only be used to test the possibility of a hypothesis by trying to replicate a past event to show it could have happened.
In an Ideal Experiment the number of variable should be
reduced to one and be easily repeated by other scientists. This increases the
likely hood that someone will try to repeat it and makes it more likely that
the attempt to repeat it will succeed. It should also be limited in complexity
and cost which also increases the likely hood that someone will try to repeat
it making it more likely that the attempt to repeat it will succeed.