A curious challenge between colleagues ended up shaping the rules that today underpin nearly any scientific experiment.
AI Brief
Your highlights
Image source, Getty ImagesPhoto caption, In the UK, tea is usually served with milk, always cold, but there is great disagreement over whether it is added before or after. Byline, Author's title, BBC News MundoPublished 58 minutes agoReading time: 8 min
In 1906, a botanist arrived at Rothamsted Research in Britain, one of the oldest and most prestigious agricultural research centers in the world, who would stand out not only for her research and leadership but who, without intending to, changed the place's routine.
Winifred Brenchley was the first female scientist to join the staff and, as the then director of Rothamsted, Sir John Russell, would later recall, "in those days, nobody quite knew what to do with a woman working in a laboratory." They thought there must be something she would surely like: drinking tea.
"From the very day of her arrival, every afternoon, at exactly four o'clock, a tea tray and a tin of biscuits would appear. The five members of the scientific staff—all there were at the time—were invited to share the break," Russell recounts.
Years later, that same tradition would be the setting for another courtesy toward a female scientist that would eventually give rise to one of the most famous experiments in modern science.
One afternoon in the early 1920s, the mathematician and biologist Ronald A. Fisher offered a cup of tea to the algologist Muriel Bristol. To his surprise, she refused it, explaining that she preferred the taste when the milk was added before pouring the tea, not after.
"Nonsense! Surely there's no difference," Fisher replied, according to his daughter Joan Fisher Box in the biography she wrote of her father.
Bristol insisted that she could taste the difference. "Let's test it," suggested the chemist William Roach, Bristol's future husband.
That playful challenge was the spark that would eventually shape the mathematics that today underpin much of science and technology, from vaccine approval to artificial intelligence algorithms.
## From the clockwork universe to chaos
The debate over whether to add the milk before or after had been heated in the UK since tea arrived in England in the mid-17th century, and what happened that afternoon of the meeting between the mathematician, the algologist, and the biochemist was not going to silence it.
But by that time, something was indeed beginning to change: the worldview inherited from Isaac Newton, whose scientific revolution had begun shortly after tea conquered the British.
Image source, Getty ImagesPhoto caption, The idea of the clockwork universe was born during the Scientific Revolution, driven especially by Newtonian physics and mechanism.
According to that vision, the universe was like a clockwork mechanism, a perfect and predictable system in which, if one knew the physical laws and initial conditions of any phenomenon, its outcome could be calculated with complete certainty. Any discrepancy was due to our temporary ignorance; errors would diminish as more exact information was obtained.
Obtaining it, however, had not been easy. Although science had relied on experiments for centuries, there was no standardized method for conducting or interpreting them. If they yielded obscure or contradictory results, it was usually attributed to some error in execution or to "noise" (data that make it difficult to clearly see the real pattern).
Only that noise had become increasingly deafening, and it was particularly evident in fields such as agriculture, biology, or medicine, where too many variables were involved—the climate, the soil, genetics, biological chance—for rigid formulas to fit.
At Rothamsted, precisely, they had been studying crop yields for decades, and often found little more than confusion: the same treatments, applied to nearly identical plots, produced different results year after year.
Nature resisted fitting into that scientific vision that promised certainty. And that afternoon, over several cups of tea, a path would begin to be traced between the observable chaos of reality and the rigor that science demanded.
## The tea taster
By the time Roach issued his "let's test it," Fisher had already been at Rothamsted for some time, thinking about how to design agricultural experiments that would give reliable answers amid all that natural noise.
Testing whether Bristol could really distinguish the taste of tea with milk depending on when the ingredients had been poured was a problem similar to the one he faced daily.
What they had to do was make sure the result of the test was real, not mere chance.
They started from skepticism. Neither believed that, however refined Bristol's palate might be, she was truly capable of such a feat, among other things because they doubted there was any difference to detect: thermodynamically, it made no difference how the tea was prepared.
So what would convince them otherwise?
Image source, Getty ImagesPhoto caption, Ronald Aylmer Fisher is often considered the founder of modern statistics.
If they gave Bristol a single cup of tea, and she guessed right, that would prove nothing: anyone would have a 50% chance of guessing. Two cups was better, but not enough to dispel doubts, since the probability of choosing the right ones at random was 25%.
The only way to convince themselves—and any skeptic—was to find a number of cups that would make chance nearly impossible.
That number turned out to be 8: four with the milk poured first and four with the tea poured first, presented in random order. Bristol had to identify which four of the eight cups had the milk first.
Why 8? It could have been many more, of course, but they calculated that this amount was sufficient to reach a highly probable conclusion. Remember that the taster knew in advance that there were four cups of each type, and it was not a matter of guessing each cup individually, but of choosing a group of four out of the eight.
It turns out there are exactly 70 different ways to choose four cups from a total of eight. If she had no real gift and chose her group of four at random, she had only one chance in those 70 of getting exactly the right ones: a mere 1.4%.
After tasting each cup, to Fisher and Roach's surprise, Bristol identified all of them correctly, say some, or at least almost all, say others; in any case, enough that from that day on, neither of them, in a gesture of gallantry, ever offered her a badly made cup of tea again.
And just enough for that challenge to become more than an anecdote.
## From tea to data
That afternoon, Fisher had stumbled upon a logic that could be applied to any scientific question where it was necessary to separate a real effect from mere chance.
Over the following years, he polished and systematized it until turning it into a complete method, which he would set down in two works that would change the way science is done: Statistical Methods for Research Workers (1925), translated into Spanish in 1949 as "Métodos estadísticos para investigadores," and The Design of Experiments (1935).
In the latter, he introduced the tea cup experiment at the beginning to illustrate several of the fundamental ideas that scientists around the world use today.
One of them is the null hypothesis, something that, he wrote, "is never proved or established, but is possibly disproved in the course of experimentation. Every experiment may be said to exist only in order to give the facts a chance of disproving the null hypothesis."
In the case of the tea taster, it meant assuming by default that Bristol had no ability to detect how each cup had been prepared and that any correct guesses would be pure luck.
Image source, Getty ImagesPhoto caption, Analysis of Variance (ANOVA) is one of the most powerful and widely used tools Fisher bequeathed to science.
The null hypothesis became a fundamental tool for scientific discovery in all branches of science, from psychology to virology and cosmology. Always accompanied by what is known as the p-value, the probability threshold that something occurs by chance which determines whether a result is significant. If the p-value is less than 5%, the null hypothesis is rejected and a real effect is considered to exist.
In the experiment with Bristol, the probability was comfortably below Fisher's threshold, at just 1.4%.
In addition, he introduced experimental methods that seem obvious to us today, such as random sampling and the use of control groups, to confirm whether a change is a consequence of the intervention, and even double-blind experiments, where neither the researchers nor the volunteers know who is receiving the placebo.
He also developed analysis of variance, or ANOVA, a method for establishing whether differences in the results of an experiment reflect a real effect.
Thus, he turned statistics into a methodology for drawing conclusions from experiments: he took what had until then been a rather scattered collection of calculation techniques and turned it into a rigorous and universal scientific framework.
His contributions would become the invisible skeleton of almost all modern scientific research.
## The taste of tea
Image source, Getty ImagesPhoto caption, Decades later, the British Royal Society of Chemistry would assert that in the case of tea with milk, the order of the factors does indeed alter the result.
By the mid-20th century, the world had stopped pursuing absolute certainties and had learned to live with probabilities.
Fisher was not the only one responsible for that shift, but he was one of its most influential architects, and the change he helped drive was as profound as the one that, centuries earlier, Newton's clockwork universe had represented, though in the opposite direction.
The purpose was no longer so much to seek absolute laws and fixed truths behind every phenomenon; science could be done without that certainty, accepting uncertainty as part of the method and learning to measure it rather than eliminate it.
That same principle is, in essence, what sustains data science today. Machine learning algorithms or models that evaluate whether a vaccine is effective do not pursue absolute laws: they identify patterns, correlations, and deviations within the noise, just as Fisher did that afternoon with the eight cups of tea.
Ah! And Muriel Bristol, it seems, was right.
In 2003, a statement from the Royal Society of Chemistry reported that "if milk is poured into hot tea, individual drops separate from the bulk of the milk, and come into contact with the high temperatures of the tea for enough time to allow significant denaturation. This is much less likely to happen if hot water is added to the milk."
In other words, adding milk, which is always served cold, second tends to change its taste, making it more like that of boiled milk.
That by no means means that the debate over the correct way to prepare tea has concluded.
Download the new BBC World Service app. Choose BBC News Mundo and receive news alerts, documentaries, and analysis, all in one place. Important: the new app is not available in the UK or the United States. You can also follow us on YouTube, Instagram, TikTok, X, Facebook, and on our WhatsApp channel. And don't forget to subscribe here to our newsletter to receive a special selection of our best content every Friday.