A tea bag is an extraordinary thing. Each small sachet contains a mix of leaves from different producers and different places. Hundreds of factors can affect the flavour of each leaf, from the amount of sunlight and rainfall to the type of soil it was grown in, how it was plucked and how it was dried. Yet when you drink a cup of your favourite brew, you expect it to taste exactly like the last one. How is that possible?

Tetley, a British teamaker, boasts that its basic blend has had the same distinctive taste since the company was set up in 1822. Every day some 45m cups of Tetley tea are drunk across the world. Reproducing that dependable, unvarying flavour is the job of Sebastian Michaelis, a master taster for Tetley. He tastes up to 250 different brews each day, looking for subtle differences.

The subtlety of palate required to do this is hard for a lay person to grasp. The softly lit room in west London where Michaelis works has row upon row of porcelain cups laid out alongside carefully measured mounds of dried camellia sinensis, or tea, as most of us call it. The mission of his team, says Michaelis, is “to make order out of the chaos”. Michaelis’s taste buds are so valuable to Tetley that it has insured them for £1m.

Sitting around drinking tea all day sounds like a cushy job, but the tasting process is not about enjoyment. Tasters double the ratio of tea to water that you’d get in a normal cup, and steep each sample for around six minutes, then take a spoonful of the strong, bitter brew, slurp it noisily (to heighten the aroma) and spit it into a large spittoon.

Michaelis, now 38, was not always a super-taster. He didn’t even consider himself a tea connoisseur. After studying philosophy at university he happened to see an ad for trainee tasters at Tetley and decided to apply. At the start of his apprenticeship – which essentially involved tasting different batches of tea for several years – he struggled to tell the difference between different leaves. He was guided by other tasters, some of whom had 40 years of experience, who would tell him to look out for particular elements of a brew.

Sebastian Michaelis’s taste buds are insured for £1m

Together they built up a common vocabulary, and he gradually learned to pick out particular notes. “There is a flavour of raspberry jam in some teas,” he says. “I’m not sure if you would be able to perceive it. But sometimes I take a sip of certain batches and there it is, a very noticeable flavour of RJ.”

Today Michaelis understands the aromatic properties of a tea leaf better than almost anyone else in the world. A well-trained taster learns to discern flavours without consciously thinking about it, he says. There is no room for subjectivity. “The question is not whether I like the tea but where it sits within our internal flavour language,” says Michaelis.

Some 15 years ago, Tetley created its own flavour taxonomy to describe the tea arriving at its tasting facility each day. This system is called Uhuru, the Swahili word for freedom. Black tea is rated on a scale of 1-40 across categories such as zing, body, sparkle and colour. The leaves are also graded numerically in each of these categories, with, say, a seven for zing and a 25 for body. This enables Tetley to predict the result of blending thousands of batches of tea.

The tasting process sounds like the kind of thing you’d associate with high-end sophisticated products, but most mass-manufactured food items go through something similar. When firms want to work out if a foodstuff is pleasant to consume they tend to use focus groups drawn from the general public. To know what a product consists of, by contrast, food and beverage companies call on the experts.

“We are here to make order out of the chaos”

Most firms rely on external panels of trained tasters. Tetley is unusual in having seven full-time in-house professionals like Michaelis, as well as part-timers and trainees. There are other valuable mouths in the business too. Dreyer’s, an American ice-cream company, insured the taste buds of John Harrison, who invented the cookies-and-cream flavour, for $1m. Hayleigh Curtis, a chocolate scientist at Cadbury, has a mouth that is so valuable the company’s insurance terms ban her from eating certain chillies in case they damage her taste receptors.

In almost every other industry, machines are disrupting the human quality-controllers. Musicians are turning to artificial-intelligence tools to balance the tone and volume on tracks, a job that used to be the preserve of highly skilled sound engineers. Some Hollywood studios use algorithms to help producers work out which elements of a film – and which stars – might yield the biggest box-office returns. But a computer still can’t process the taste of a cup of tea. Yet.

Flavour is an elusive concept, even for scientists. It has multiple elements: touch, taste, sight, smell, plus the wild card of subjective psychological association. Smell alone is an incredibly complicated sense. Neuroscientists reckon the human nose can pick up more than a trillion different scents. All of this means the vocabulary that most of us use to describe how a cup of tea tastes – sweet, say, or bitter – is woefully imprecise.

With enough training, we can learn to ask more focused questions. How woody is the tea? How burnt? But we can’t assume that our own sense of woodiness and burntness is the same as others’. This is one of the hardest things to teach, says Sue Langstaff, a professional taster of olive oil and tasting trainer. It’s hard enough learning to spot all the flavours in a drop of olive oil, but it takes years for trainee tasters to align their experiences of peppery, pungent and spicy with the consensus.

A computer still can’t process the taste of a cup of tea. Yet

You need a highly specialised vocabulary to discern and describe flavours at the level of complexity needed by the food industry. Some food producers, like Tetley, have their own bespoke systems of categorisation. The International Olive Council, a standards body, certifies more than 100 flavour panels around the world to ensure that tasters adhere to an agreed definition when identifying a bottle as “extra virgin”.

Professional tasters organise their lives around their palates and the need to keep them as pure as possible. Most do their work in the morning, when the mouth is most sensitive. They cannot taste if they have a cold, and are forbidden from wearing perfume or cologne on the job. Though tasters strive for scientific accuracy, there is some art to the work too. Michaelis says he uses his subconscious associations with certain flavours to help fix them in his mind.

Michaelis finds it hard to switch off his super-tasting sense when he leaves the office. Sometimes that’s useful: a few days before we spoke, he had cooked a joint of meat but realised that it was off the moment that it touched his tongue (his wife noticed nothing). He also finds certain flavours too overwhelming to enjoy in civilian life, such as smoked meat or strong cheese.

Now Michaelis is training a new generation of tasters for Tetley. Collectively, they form the company’s institutional flavour memory. “When we fly to conferences,” he says, “the tasters have to fly in separate planes, just in case something happens.” These precious taste buds are currently facing a more serious threat than plane travel. Some people who get covid-19 lose their sense of smell and taste – and not all of them get it back. Despite careful precautions, two members of Tetley’s tasting team contracted the virus, though neither suffered lasting consequences. For the industry the pandemic has been another reminder that tasters are expensive instruments; difficult to calibrate, and not easily replaced.

When you think about how much we value our sense of taste – the effort a company puts into developing a new flavour of ice cream or the amount we’re prepared to pay for an unseen chef to cook us a gourmet meal – it seems impossible that a machine could replicate the experience. But in Silicon Valley, a biotech startup called Aromyx is trying to do just that. The company is trying to clone and isolate the genes behind each of the 400 olfactory receptors in the human nose in order to visualise on a computer the reaction of each receptor to different aromas. At Tetley, Michaelis translates taste into words; Aromyx hopes to turn flavour into numbers.

“Tasters have to fly in separate planes, just in case something happens”

Aromyx wants to produce a sort of taste fingerprint based on which elements of a flavour trigger different receptors, and how great each receptor’s reaction to that element is. It is already working with a drinks company on a lemonade tasting project. Like tea, lemonade involves natural ingredients that can vary from batch to batch. The company will see whether Aromyx’s technology can replace the tasting panel which normally tests each haul of lemon juice for quality, consistency and sweetness. Josh Silverman, the boss of Aromyx, says his firm has a clear aim: “turning the subjective into objective”.

Other tech firms are trying to do something similar. For Silicon Valley’s disrupters, flavour is one of the final unbreached frontiers. IBM, an IT firm, is developing an electronic, AI-assisted tongue called Hypertaste, which is already able to determine the chemical similarity between some liquids, an important step towards more complex forms of analysis.

The machines have a lot of catching up to do, however. Flavour is a multidimensional experience dependent on many variables. Even sound can affect how you taste something: a recent study at Oxford University found that people are more likely to pick up bacon flavours in a dish when a recording of a sizzling frying pan is played in the background.

Robot tongues and food fingerprints are still worth pursuing. Elusive though they are, their potential applications are far-reaching. If the multidimensional world of flavour can be captured by machines, then it could ultimately be recreated too. That would revolutionise food production: lettuce leaves could be made to taste like beef burgers, saving the Amazon (and our hearts).

Aromyx is cloning each of the 400 olfactory receptors in the human nose

That day remains far off – we may not get there. Some people, like Silverman, operate on the assumption that you can translate a particular flavour into computer code. Others reckon that you can never divorce the subjective human experience of tasting, complete with its unpredictable variables, from the objective properties of a flavour.

This is the same philosophical question that Michaelis pondered as a student. How can anyone be sure that their perception of something – the colour blue or the smell of bacon – matches another person’s? Michaelis doesn’t think it’s possible. For him, there is only trained consensus.

Back in London, Michaelis has recently taken on a graduate student who is learning Tetley’s training vocabulary. Together, they attempt to identify and describe the subtlest of flavours, one word at a time, eliminating all subjective associations. “A good taster is able to switch off the internal monologue when slurping a tea,” says Michaelis. “I suppose, ironically, we actually train ourselves to be more robotic in our tasting.”■

Barclay Bram is an anthropologist at the School of Global and Area Studies at the University of Oxford

ILLUSTRATIONS: MATTHEW RICHARDSON