Here’s a strange thing: you spend your whole life moving through a world that feels solid and continuous—a desk you can touch, air you can feel, water that flows smoothly through your fingers. Yet for more than two thousand years, some philosophers and scientists have insisted that behind all of this solid, smooth stuff, there is actually nothing but tiny, invisible particles moving around in empty space. They call these particles atoms, and they claim that everything—solid rock, liquid water, your own body—is just atoms arranged in different patterns.

The weirdest part? They turned out to be right. Sort of.

How did anyone ever get from the world we can touch and see to the conclusion that there are invisible particles making it all happen? And what does it even mean to have evidence for something you can’t see? This is a story about what counts as a good reason to believe in something invisible—and how that changed over time.

The First Argument: It Just Makes Sense

In the 1600s, a group of thinkers called the mechanical philosophers dusted off an old Greek idea and gave it new life. The most careful among them was Robert Boyle, who was both a philosopher and a pioneer of experimental science. Boyle argued for atoms this way: when someone says an object is “elastic” or “sweet” or “hot,” they’re describing something real, right? But what are those things, exactly? Aristotle had said objects have “forms” and “qualities” built into them, but Boyle found this completely mysterious. How can something that isn’t matter be stuck inside matter? That seemed like magic dressed up in fancy words.

Boyle’s solution: strip it down. The only things matter must really have, he said, are shape, size, and motion (plus the ability to be solid and take up space). Everything else—color, taste, elasticity, heat—is just what happens when lots of tiny particles with different shapes and sizes bump into each other. A key opens a lock because of its shape, not because it has “key-ness” built into it. A clock works because of how its gears fit together. So why not explain everything that way?

This was an argument from intelligibility: it just seemed clearer and more sensible than the alternatives. Boyle thought atoms had to exist because the rival explanations were literally unintelligible.

But here’s the problem. Boyle’s examples weren’t actually pure shape-and-size explanations. A key works because it’s rigid. A clock works because its springs are elastic and its pendulum bob is heavy. Those are properties Boyle wanted to explain away, not build upon. And when he looked around for purely mechanical examples—things that could be explained by shape, size, and motion alone—he couldn’t find very many. The world is messier than the philosophy.

There was another problem too. How do you get from things you can see to things you can’t? Philosopher Maurice Mandelbaum called this the problem of transdiction: how do you jump from the observable to the unobservable? Boyle thought you could solve it by noticing that some laws apply at every scale—heavy objects fall the same way whether they’re the size of a boulder or a pebble—and then assuming those laws keep working all the way down to the atomic level. But this doesn’t work either, because the laws we can observe involve properties (elasticity, weight, rigidity) that atoms supposedly don’t have. You can’t use a law about springs to prove that atoms aren’t springy.

A Different Kind of Atom

While Boyle was pushing his mechanical atoms, a separate tradition was coming from chemistry and from Aristotle. The question was: when you mix copper and tin to make bronze, what happens to the copper and tin? They don’t seem to be in the bronze the way raisins are in a muffin—you can’t just pick them out. But they must still be there somehow, because you can recover them. So the copper and tin must exist as tiny “least parts” inside the bronze, combined in some way that preserves them while also making something new.

These were natural minima: the smallest bits of a substance that still count as that substance. A minimum of copper isn’t a generic little ball—it’s copper-flavored. It has copper’s properties. And when copper minima and tin minima combine, the result is a bronze minimum with bronze properties. This is very different from Boyle’s atoms, which were all the same kind of stuff, just arranged differently.

The natural minima approach was less ambitious than Boyle’s. It wasn’t trying to explain everything in the universe. It was just trying to make sense of chemical change, and it did that by giving atoms the properties of the substances they came from. This turned out to be more useful for actual science.

Newton Adds Forces

Isaac Newton improved on mechanical atomism in one obvious way: he could actually write down precise laws for how atoms move. Before Newton, philosophers just said atoms “move around” and “bump into each other.” Newton gave them his three laws of motion, inertia, mass, and precise collision mechanics. This was a real upgrade.

But then Newton did something that would have horrified Boyle. He added forces. Gravity pulls atoms together; other forces push them apart. For Boyle, that would have been cheating—forces are invisible, mysterious, and not part of the simple, clear ontology he wanted. But for Newton, forces were what made the theory work. Without gravitational force, you can’t explain why things fall. Without chemical affinities (forces between chemical atoms), you can’t explain why some substances combine and others don’t.

The problem was that Newton had no way to measure or verify these atomic forces. He could measure gravity at the scale of planets, but the forces between individual atoms were pure speculation. Chemists in the 1700s paid lip service to Newtonian atomism while actually doing their work in a completely different way. They figured out which chemicals combined with which by mixing them in the lab, not by speculating about invisible forces.

The Big Breakthrough: Dalton Weighs Atoms

Everything changed in 1808 when John Dalton proposed that atoms of different elements have different weights. For the first time, there was a property of atoms—relative weight—that scientists could actually measure. Here’s how it worked:

If hydrogen and oxygen combine to form water, and you know that water contains 1 part hydrogen to 8 parts oxygen by weight, then if you assume water is HO (one hydrogen atom + one oxygen atom), the oxygen atom is 8 times heavier than the hydrogen atom. If you assume water is H₂O (two hydrogens + one oxygen), then oxygen is 16 times heavier.

The problem: you don’t know which formula is right. Dalton’s solution was simplicity: assume the simplest possible formula unless you have evidence for something more complicated. This was a reasonable guess, not a proof.

But Dalton’s theory made predictions that could be tested. He predicted that if two elements form more than one compound (like carbon and oxygen forming both CO and CO₂), the weights in which they combine should be in simple ratios to each other. This was confirmed. He predicted that if element A and element B both combine with element C, then the weights of A and B that combine with the same amount of C should be in a simple ratio to how A and B combine with each other. Also confirmed.

These successes were real, but they weren’t slam dunks. As Dalton’s rival Berzelius pointed out, all the testable predictions of Dalton’s theory could also be stated without mentioning atoms at all. You could just say “water contains 1 part hydrogen and 8 parts oxygen by weight” without saying anything about invisible particles. And in chemistry up through the mid-1800s, that’s basically what happened. Chemists developed an incredibly powerful system of chemical formulas that let them predict and manipulate reactions—without any agreement on whether atoms actually existed.

The Trouble with the Kinetic Theory

In the 1860s, physicists working on the kinetic theory of gases took a different approach. They assumed gases were made of tiny molecules moving around randomly, bouncing off each other and off container walls. From this, they could explain the gas laws (how pressure, volume, and temperature are related), predict how fast different gases would diffuse, and even predict a wild thing: that the viscosity (stickiness) of a gas should be independent of its density. That prediction was confirmed, which was impressive.

But the kinetic theory had serious problems too. It predicted that the ratio of two specific heats of a gas would have a certain value, and experiments disagreed. It also faced the reversibility problem: the laws of physics work the same forward and backward, but gases always spread out and never spontaneously gather—so why does time have a direction? The theory could only say this is “very unlikely” rather than impossible. To many scientists, that sounded like an excuse.

The Turning Point: Brownian Motion

Here’s where the story gets really interesting. In 1827, botanist Robert Brown looked at pollen grains floating in water through a microscope and noticed they jittered around randomly, never stopping. Nobody knew why. By the late 1800s, physicists suspected the jittering was caused by water molecules bumping into the pollen, but they couldn’t prove it.

Then in 1905, Albert Einstein showed that if you assume the pollen is being knocked around by molecules, you can make precise predictions about how far the pollen should move over time. In 1908, French scientist Jean Perrin tested these predictions. His experiments showed three remarkable things and got three consistent numbers out of them.

First, Perrin measured how the density of Brownian particles varied with height. He found it decreased exponentially—the higher you went in the solution, the fewer particles there were. From that distribution, using Einstein’s equations, he calculated Avogadro’s number (the number of molecules in a given amount of gas): about 68 × 10²².

Second, he measured the average distance the particles moved in a given time, and calculated Avogadro’s number again. Same answer.

Third, he measured how fast the particles rotated, and calculated it again. Same answer.

Three different measurements, all based on the assumption that molecules exist, all pointing to the same number.

Perrin then pointed out something clever. Before his experiments, what would you have predicted about the density distribution of Brownian particles if you didn’t believe in molecules? You might guess they all sink to the bottom (since they’re heavier than water)—that would give you a meaningless infinite number. Or you might guess they spread evenly—that would give you zero. Or you might guess something in between. The range of possible outcomes was enormous. But the actual outcome matched the molecular theory’s prediction. As Perrin put it, a coincidence that perfect could only be explained by the theory being correct.

Did that Settle It?

Not entirely. The kinetic theory was clean and successful, but it also had problems that pointed toward deeper weirdness. The specific heats problem, for instance, was a real anomaly that couldn’t be ignored. And the statistical explanation of the second law of thermodynamics—that entropy increase is just very probable, not necessary—made some scientists very uncomfortable.

What Perrin’s experiments did was shift the burden of proof. Before Brownian motion, a smart person could reasonably doubt the existence of atoms. After Brownian motion, the evidence was strong enough that opposition became mostly philosophical rather than scientific. The anti-atomist chemist Wilhelm Ostwald conceded the point, saying Perrin’s experiments had convinced him.

But here’s the twist. The atoms that Perrin’s experiments confirmed were not Boyle’s atoms. They weren’t featureless little balls with only shape and size. They had weight. They had structure. They had properties that depended on what kind of atoms they were. Modern atoms turned out to be more like the natural minima tradition than the mechanical atomism tradition. They have specific chemical properties built into them.

And modern physics has pushed even further: atoms themselves are made of smaller particles (electrons, protons, neutrons), and those particles obey laws that aren’t mechanical at all. At the subatomic level, things don’t behave like tiny billiard balls. They behave according to quantum mechanics, which is deeply weird and not mechanical in the least.

So Where Are We Now?

The core idea of atomism—that macroscopic matter is built from tiny particles—is now as established as anything in science. There is no serious scientific doubt about this. But the version of atomism that turned out to be true is not the version any philosopher before the 19th century imagined. The properties of atoms were not discovered by reasoning about what must be “intelligible.” They were discovered piece by piece, through experiments designed to probe what atoms actually do.

This is a strange outcome. The ancient Greek atomists guessed the basic picture without any evidence. The mechanical philosophers argued for it on grounds of clarity and simplicity. Neither of those arguments was strong enough to actually establish atomism. What finally worked was designing experiments that forced the invisible world to leave traces in the visible one—and then following those traces wherever they led.

The atom we ended up with is stranger and more complicated than anyone expected. It has properties—like quantum spin and wave-particle duality—that would have seemed completely unintelligible to Boyle. But it does the job he wanted: it explains why the world behaves the way it does, all the way down.

---

Appendices

Key Terms

Term	What it does in this debate
Atom	A tiny, indivisible particle that combines with others to make up all material things
Mechanical atomism	The view that atoms have only shape, size, and motion, and everything else is just how those atoms are arranged
Natural minima	The smallest bits of a specific substance that still have that substance’s properties
Problem of transdiction	The puzzle of how to get knowledge about unobservable things from knowledge about observable ones
Kinetic theory of gases	The explanation of gas behavior by assuming gases are made of many tiny molecules moving randomly
Avogadro’s number	The number of molecules in a fixed amount of gas (about 6 × 10²³), used as a test of atomic theories
Brownian motion	The random jittering of tiny particles suspended in a liquid, caused by molecules bumping into them

Key People

• Robert Boyle (1627–1691): A philosopher and early experimental scientist who argued that atoms must have only shape, size, and motion because other properties were “unintelligible.”
• John Dalton (1766–1844): The chemist who first connected atoms to measurable weights and worked out how atoms combine in simple ratios.
• Jean Perrin (1870–1942): The scientist whose experiments on Brownian motion provided the first really strong empirical evidence for the existence of atoms.

Things to Think About

1. Is there a difference between “this theory is clear and makes sense” and “this theory is true”? Can an explanation that is very clear and neat be completely wrong? What about clear but incomplete?

2. Perrin’s argument from coincidence seems very strong—three different measurements giving the same number. But how many coincidences does it take before you must accept the explanation? Is there a number, or is it more of a feeling?

3. Some scientists (like Ostwald) rejected atoms even when the evidence was quite good, because they thought science should stick to things you can directly measure. Were they being too strict, or were they protecting science from unreliable speculation? Where is the line between necessary caution and stubbornness?

4. Modern physics says atoms are mostly empty space, and the particles inside them behave according to rules that seem to violate common sense. Does this mean the mechanical philosophers’ demand for “intelligibility” was a bad guide all along, or does it mean we just have to get used to a new kind of intelligibility?

Where This Shows Up

• In science class: Understanding that chemical formulas represent actual structures, not just convenient fictions. The periodic table is a map of atoms with different weights and properties.
• In everyday life: When you see dust motes dancing in a beam of sunlight, you’re watching something like Brownian motion. The jittering is caused by air molecules you can’t see.
• In other fields: The question “how do we know about things we can’t see” comes up everywhere—in medicine (how do we know viruses exist?), economics (how do we know about supply and demand?), and history (how do we know about events before anyone alive was born?). The same basic problem of transdiction shows up in all of them.
• In current debates: Questions about “unobservable entities” still matter. When scientists talk about dark matter or string theory, they’re dealing with things they can’t directly detect. The arguments for and against these things echo the old atomism debates.