Case Study 2 — The Margarine Reversal

This case study is about the largest dietary recommendation reversal of the twentieth century. It is also a cautionary tale about how chemistry, public health, food industry, and well-intentioned medical advice can compound into a generation-long mistake — and how that mistake was found and corrected. The chemistry is in this chapter (Chapter 11). The history is its own lesson. Pat Hammond's classroom demonstration started teaching the chemistry in the 1990s. By the 2010s she was teaching the reversal as well. By the 2020s her students were watching their grandmothers throw out tubs of margarine that had once been the sensible substitute for "deadly" butter.

How margarine got invented

In 1869, the Emperor Napoleon III of France offered a prize for anyone who could invent an inexpensive butter substitute for the working class and the French navy. Butter was expensive, perishable, and uneven in quality; the navy in particular needed a fat that could last through long voyages. A French chemist named Hippolyte Mège-Mouriès won the prize with a process that used beef tallow, skim milk, and a small amount of cow udder — yes, that's literally what was in the original recipe — to produce a spreadable fat with a pearly white color. He named it after the Greek word for "pearl," margarites. Margarine was born as a public-good project.

Margarine took off in the United States in the early 1900s, originally still based on animal fats. The breakthrough that made margarine into the cultural object Pat would later teach about was the invention of partial hydrogenation by a German chemist named Wilhelm Normann in 1901 (patented in 1902, scaled industrially through the 1910s by Procter & Gamble in the United States and Lever Brothers in Britain). Hydrogenation took inexpensive liquid plant oils — first whale and seal oils, then cottonseed, then soybean and corn — and converted them, by adding hydrogen across some of their double bonds in the presence of a metal catalyst (usually nickel), into solid or semi-solid fats that behaved like animal fat. The process was selective and partial: it would saturate some double bonds completely, but it would also isomerize others from the natural cis configuration to the trans configuration. The trans isomers were straight chains, like fully saturated fats, and they packed tightly to give the partially hydrogenated oil its solid texture. The chemists who designed the process knew about the isomerization. They did not know what trans fats would do to the human cardiovascular system. Nobody did. The science of cholesterol metabolism was decades away.

By 1950, margarine had largely replaced butter in working-class American kitchens because it was cheaper. It was sold as "vegetable oil margarine," with the implication that vegetable oil was inherently more wholesome than animal fat — though the connection between animal fat and disease was, at this point, an unproven hypothesis.

The fat-cholesterol hypothesis takes over

In the 1950s, an American physiologist named Ancel Keys proposed what would become known as the diet-heart hypothesis: that dietary saturated fat raised blood cholesterol, and that elevated blood cholesterol caused atherosclerosis and heart disease. Keys ran a landmark study, the Seven Countries Study, comparing the diets and heart-disease rates of populations in seven countries (Italy, Greece, the former Yugoslavia, the Netherlands, Finland, the United States, and Japan). The study found a correlation: countries with more saturated fat in the diet had more heart disease.

The Seven Countries Study had problems. It was a correlational, observational study (not a randomized trial). The seven countries were selected from a larger pool of twenty-two for which Keys had data, and the subset showed a stronger correlation than the full sample would have. The dietary recall data was unreliable by modern standards. The hypothesis was that saturated fat was the cause; the evidence supported only that saturated fat was correlated with disease, in a sample of seven countries selected to make the correlation visible.

But Keys was a forceful advocate, the data was the best then available, and the hypothesis caught on. Through the 1960s and 1970s, the U.S. medical establishment converged on the recommendation that Americans should reduce saturated fat (animal fat, butter, lard, eggs, full-fat dairy) and increase polyunsaturated fat (margarine, vegetable oils, polyunsaturated oils). The American Heart Association formally endorsed this advice in 1961. The U.S. government's Dietary Goals for the United States in 1977 codified it. By 1980, the public message was unambiguous: butter is bad, margarine is good.

This message coincided exactly with the rise of partially hydrogenated margarine as the dominant kitchen fat in middle-class American households. Pat Hammond, born in 1972, grew up with margarine on the table because that was what nutritionists said you should have on the table.

The trans-fat alarm

In 1990, Dutch researchers Mensink and Katan published a remarkable controlled feeding study. They enrolled 59 adults, fed them three different diets for three weeks each (in a crossover design — every participant ate every diet), and measured the effects on blood lipids. The diets differed only in fat composition: one was high in oleic acid (the dominant fatty acid in olive oil); one was high in trans fatty acids (from partially hydrogenated soybean oil); one was high in saturated fat. The result was alarming. Trans fats raised LDL ("bad") cholesterol and lowered HDL ("good") cholesterol. Saturated fats raised LDL but also raised HDL. Trans fats were producing the worst possible lipid profile — the opposite of what they were intended to do.

The study was small but well-controlled. Replication came quickly. By the late 1990s, large prospective cohort studies (the Nurses' Health Study, with over 80,000 women followed for years) confirmed the association: trans fat consumption was independently and strongly associated with cardiovascular disease, more strongly than saturated fat consumption. The mathematics began to be done. By 2002, an Institute of Medicine report estimated that trans fats were causing tens of thousands of preventable cardiac deaths annually in the United States alone.

The chemistry was clear in retrospect. Trans fats interfered with cellular cholesterol metabolism in ways that natural fats did not. The unnatural geometry of the trans double bond meant that when the human body's enzymes tried to process trans-fatty-acid-containing phospholipids, the processing produced abnormal signaling and abnormal lipid trafficking. The body had evolved to handle cis double bonds and saturated bonds. Trans bonds were a chemical novelty — a fact pattern that human metabolism had never seen before partial hydrogenation invented them.

The slow reversal

The science of trans-fat harm was settled by the early 2000s. Action by regulators took longer.

• 2003: Denmark became the first country to substantially restrict trans fats in the food supply, capping them at 2% of fat content in any food.
• 2006: the U.S. FDA required trans-fat content to be listed on nutrition labels. Industry began voluntarily reformulating away from partial hydrogenation, replacing it with palm oil (saturated, but at least not trans), interesterified fats, and fully hydrogenated oils blended with liquid oils.
• 2013: the FDA tentatively determined that partially hydrogenated oils were not Generally Recognized as Safe.
• 2015: the FDA finalized the determination. Industry was given until 2018 to remove partially hydrogenated oils from food.
• 2018: the U.S. ban on added partially hydrogenated oils took effect.

By 2020, trans-fat consumption in the United States had dropped by approximately 80% from its peak. Population-level cardiovascular indicators began to improve in ways that matched the predictions made by epidemiologists in the early 2000s.

The first wave of reversal — that trans fats were harmful — was clean and decisive. The second wave — that we may have over-blamed saturated fat — has been slower and more contested.

The saturated-fat reconsideration

Through the 2000s and 2010s, a series of meta-analyses began to question whether the broad recommendation to reduce saturated fat had ever been supported by strong evidence. A 2010 meta-analysis by Siri-Tarino and colleagues, published in the American Journal of Clinical Nutrition, pooled data from 21 prospective cohort studies (over 347,000 subjects) and concluded that saturated fat intake was not associated with cardiovascular disease in their data set, after adjusting for trans fats. A 2014 meta-analysis by Chowdhury et al., published in Annals of Internal Medicine, reached similar conclusions. The 2017 PURE study (the Prospective Urban Rural Epidemiological Study) followed 135,000 people in 18 countries on five continents and found that higher saturated fat intake was associated with lower total mortality and lower stroke risk; it also found that high carbohydrate intake was associated with higher total mortality.

This evidence was contested. Defenders of the older orthodoxy pointed out methodological limitations of observational studies, residual confounding, and the difficulty of disentangling specific fat sources from dietary patterns. They argued that randomized trials replacing saturated fat with polyunsaturated fat had shown cardiovascular benefit. Proponents of the revisionist view argued that the older observational studies were the basis for the original consensus, and if those same studies do not show what they were once taken to show, the recommendation needs revision.

The current state of the evidence (as of 2024–2025) is something like this:

• Trans fats from partial hydrogenation are clearly harmful. Settled.
• Replacing saturated fat with refined carbohydrates does not improve cardiovascular outcomes. Probably worsens them. Reasonably settled.
• Replacing saturated fat with polyunsaturated fat may modestly improve cardiovascular outcomes, particularly when the polyunsaturated fats include omega-3 sources. Reasonable confidence.
• Replacing saturated fat with monounsaturated fat (Mediterranean pattern) shows reasonably consistent benefit. Strong confidence based on multiple lines of evidence including the PREDIMED trial.
• Total saturated fat per se, with no specification of replacement, is probably not the dominant dietary lever for cardiovascular health. Contested but increasingly accepted.
• Specific foods matter more than abstract macronutrient percentages. Butter from grass-fed cows, fatty fish, full-fat fermented dairy, and unrefined olive oil all contain saturated fat (in varying proportions) and all show favorable or neutral cardiovascular signal in observational studies, while industrial sources of saturated fat (highly processed meats, sweetened full-fat dairy, fast-food fryer oils that have been hydrogenated to extend life) show different signals.

This is what an evolving science looks like. It is not satisfying. It is honest.

Pat's classroom now

Pat Hammond's chemistry classes start the unit on lipids by passing around two pats of butter and two slabs of margarine — the latter saved from the discontinued formulations of the early 2000s, kept in her freezer as artifacts. The kids pass them around. They look the same. They are not the same.

Pat tells the story this case study tells. She reminds the kids that the chemistry was elegant — partial hydrogenation was a brilliant solution to the shelf-life problem of margarine, and it ran for nearly a century before the medical evidence caught up. She reminds them that consensus is not always right. She reminds them that the right move, when consensus and evidence diverge, is to look at the evidence, not at the consensus.

Then she does the demo. She melts butter on one hot plate, margarine on the other. The butter foams and browns. The margarine melts and sits there. The kids smell the butter from across the room. Cow, Pat says. Plus water. Plus milk solids. The cow is the part you can smell. The Maillard reaction is making the milk solids brown. The reaction is the same one that happens when you sear a steak.

The kids remember. They go home and tell their parents that margarine was a public-health failure. Their parents say but we were told, and the kids say the science was wrong. This is what high-school chemistry is for.

Analyze this

The trans-fat reversal raises a question that every dietary recommendation will eventually face: how do you act on imperfect evidence when waiting for perfect evidence will leave the question unanswered for decades?

1. The Seven Countries Study was correlational, not causal. The recommendation that came out of it became a generational orthodoxy. With the benefit of hindsight, what kinds of additional evidence would have been needed before the recommendation was issued? Why were they not collected before the recommendation was made?

2. From 1990 (Mensink and Katan) to 2018 (the U.S. ban on partially hydrogenated oils), the trans-fat reversal took 28 years. Walk through that timeline. Where did the bottlenecks live? What does this tell you about how dietary advice changes?

3. The current view is that what replaces the saturated fat matters more than the saturated fat itself. Apply this principle to a current dietary debate of your choosing — sodium, fiber, cholesterol, alcohol, eggs, red meat. What is the evidence on the food itself? What is the evidence on the replacement?

4. Pat Hammond's demonstration is unchanged from 2003 to 2025; only her framing changed. What does this teach about teaching science to high schoolers? When the evidence updates, do the demonstrations need to update too?

5. Imagine a chemist invents a new fat substitute today, with a clear safety profile in 6-month rat studies and an attractive cost structure. Should it be adopted into the food supply? On what evidence? Who decides? What conditions would have to be met for the same regulatory pathway not to repeat the trans-fat story?