William Sealy Gosset, writing under the pseudonym “Student,” developed the t-test for working with small samples. He published the underlying method in 1908, and it became known as Student’s t-test and Student’s t-distribution.
Gosset was not a university statistician working only with abstract examples. He was a chemist and brewer at Guinness, where he needed reliable ways to compare barley varieties and assess brewing quality when experiments produced relatively few observations. Standard large-sample methods were poorly suited to those situations, so Gosset developed a distribution that accounted for uncertainty in estimating a sample’s standard deviation.
Guinness required researchers to publish under pen names to protect company information, which explains “Student.” Karl Pearson helped Gosset with some of the mathematics, while Ronald Fisher later clarified, extended, and popularized the method. That history creates a common confusion: Fisher made major contributions to the modern use of the t-test, but Gosset was the statistician who first developed and published the work under the name Student.
The t-test remains useful for comparing means when sample sizes are small and the population standard deviation is unknown, provided its assumptions are reasonably satisfied.