A cup of tea and statistics

Now interestingly (I am not sure if this story is true or not) but I was rather taken by a story I read in Dave Trott's book, Crossover Creativity (2023). As Trott does not name his sources, it may lack a bit of credibility. But I think it is a good story!

Here goes. A "mathematician working at Rothamsted Experimental Station in Hertfordshire, in 1920", Ronald Fisher, offered "his colleague, biologist Muriel Bristol, a cup of tea" putting the milk in first (Trott, 2023, 77%) and topping it up with tea from a tea urn (Wikipedia, 2025). All good so far. However, Muriel was not impressed with Fisher's tea boy skills, telling him that she didn't like tea made with the milk in first.

Fisher felt that there was no way that Muriel could tell the difference. He felt "this was a matter of simple thermodynamics: liquid A added to liquid B is exactly the same as adding liquid B to liquid A, the order is irrelevant" (maybe Fisher had probably only ever had porcelain cups - which would not crack when boiling water was poured into them, unlike cheaply made cups, which do). Muriel said that if milk was added first, the tea tasted differently to when milk was added last (Trott, 2023, 77%). And personally, I agree with her. I think tea with milk in first tastes 'softer' than the other way around.

A colleague, William Roach, said they should test Muriel. So "Scientists gathered round as Fisher made eight cups of tea, identical in every way except one. In four of the cups the milk was added first, in the other four it was added second. As a blind test, Muriel Bristol had no way of knowing which was which. But everyone watched as, one after the other, she identified immediately from taste alone which cup of tea was which" (Trott, 2023, 77%). Good on Muriel. What's more, she was bang on, 100% accurate. 

However, while Muriel may have made her point, Fisher went looking for a formula to see if the result could be attributed to chance, sample size, or "random variations", shifting focus "from simply analysing a tea test into devising the correct way to run tests to arrive at a more accurate statistical analysis" (Trott, 2023, 78%). Fisher ending up by formulating what is now known as Fisher's exact test (Wikipedia, 2025), followed by the publication, in 1925, of "Statistical Methods for Research Workers"; apparently still considered a foundational statistics text (Trott, 2023, 78%). 

I will be interested to see if anything comes out of the woodwork to prove or disprove this story!

Sam

References:

Trott, D. (2023). Crossover Creativity: Real-life stories about where creativity comes from [ePub]. Harriman House Ltd.

Wikipedia. (2025). Muriel Bristol. https://en.wikipedia.org/wiki/Muriel_Bristol