Sir Ronald Fisher wrote a famous short paper in 1937 called "The Mathematics of a Lady Tasting Tea". He relates the story of a fictitious lady who claims she can tell if cream was poured into a cup before or after tea. His paper proposes a test of her ability. His procedure is to assemble four cups with tea added to the cup first and four cups to which cream is added first, then all 8 cups are presented to the lady randomly. Her task is to taste the cups and sort them without error. According to his standard, if she makes any mistakes, she fails the test. He explains where he gets his numbers. His reasoning is based on the idea that we could make a mistake whether we say the lady has the ability or that she does not have it.
You can get all philosophical in assessing the lady, but a common idea in British and American approaches to statistical analysis of experimental data is to aim at having no more than a 5% chance of thinking the lady has the ability she alleges she does when in fact she does not.
This paper is quite famous. You can read it here: https://en.wikipedia.org/wiki/Lady_tasting_tea
