Science · Scientific discovery & evidence

How science works

Science tests ideas against the world and against other scientists. See how a guess becomes a test, what 'theory' really means, how sciences without labs test ideas, and how errors get caught.

4 short lessons · about 16 minutes · 4 sources · updated November 2026

What you will learn

  • Tell a hypothesis apart from the prediction it produces
  • Explain why a good test is one whose result could show the idea is wrong
  • Define fact, law and theory as scientists use the words
  • Explain why a scientific theory isn't a guess waiting to become a law

Lesson 1 of 4 · 4 min

How a guess becomes a test

Before you readIn 1846 an astronomer aimed a telescope at a spot where math said an unknown planet should be. How long did he search before finding it?

Correct.Not quite. The answer is Less than an hour.

He found it in under an hour, almost exactly where the math pointed. You're about to see how that kind of prediction puts an idea to the test.

By the 1840s, the planet Uranus was misbehaving. Astronomers used Newton's laws to predict where it should be, and it kept drifting off that path. Something was wrong, and there were two candidates.

Either Newton's law of gravity failed that far from the Sun, or an unseen planet was tugging on Uranus. The French mathematician Urbain Le Verrier backed the second idea. That idea is a hypothesis, a proposed explanation for something observed.

A hypothesis alone can't be checked against the sky. So Le Verrier worked out what it implied, the spot where the hidden planet should be right then. That's a prediction, what you'd expect to observe if the hypothesis is true.

A handy way to keep them apart is the sentence "I expect to see A because of B." B is the hypothesis. A is the prediction it produces.

Check yourselfLe Verrier said an unseen planet was tugging on Uranus, so a new planet should appear at a certain spot in the sky. Which part of that is the hypothesis?

Correct.Not quite. The answer is An unseen planet is tugging on Uranus.

The hypothesis is the proposed explanation. The spot in the sky is the prediction it produces, and Uranus drifting is the observation it explains.

Le Verrier sent his numbers to Johann Galle at the Berlin Observatory. On the night of 23 September 1846, Galle found Neptune in less than an hour, within 1 degree of the predicted spot.

What made that a strong test is that it could have failed. An empty patch of sky would have counted against the idea. A test only tells you something if some result would show the idea is wrong.

Le Verrier's method did fail once. In 1859 he blamed odd wobbles in Mercury's orbit on another unseen planet, later called Vulcan. Astronomers searched for decades and never confirmed it. Einstein's 1915 theory of gravity later explained Mercury's orbit without any extra planet.

Say a friend claims a lucky charm works, but only when it wants to. No result could ever count against that claim, so no test can check it.

Try to recallWhy was Le Verrier's prediction of where Neptune would be a strong test of his idea?

It named a specific spot in the sky, so the idea could have failed. If Galle had found nothing there, the unseen-planet idea would have been in trouble. The later Vulcan search shows such predictions really can fail.

A hypothesis is a proposed explanation, and a prediction is what you'd expect to see if it's true. You can now judge a test by asking whether some possible result would have shown the idea was wrong.

Lessons in this topic

  1. How a guess becomes a test

    Follow the hunt for Neptune to see how a hypothesis turns into a prediction, and why the best tests are ones an idea could fail.

    Essentials · 4 min

  2. What scientists mean by fact, law and theory

    Learn what the three words mean in science, and why calling something "just a theory" gets the word backwards.

    Essentials · 4 min

  3. Testing ideas you can't put in a lab

    See how geologists and astronomers test ideas about the past and the far away, and why no single step-by-step method fits every science.

    Deeper · 4 min

  4. How scientists catch each other's mistakes

    Follow the 2011 faster-than-light scare to see how outside checking catches errors, then see what a scientific consensus is and how it changes.

    Deeper · 4 min

Three quick stories from this topic

4,000 feet of coral

A 1951 drill went through more than 4,000 feet of coral before reaching volcanic rock

Coral only grows in shallow, sunlit water. So a coral pile that thick on Eniwetok atoll meant the volcano below had slowly sunk while the reef kept growing upward.

Two ideas, one drill

A reef on an underwater mountain would sit as a thin layer. A reef around a sinking volcano would stack up hundreds of feet thick.

Testing the past

Geologists can't rerun an island's history, but each idea predicted different rock underground, and the drill decided between them.

The loose fiber cable

A 2011 result showing particles beating light came down to a loose fiber cable

The OPERA experiment clocked neutrinos arriving about 60 billionths of a second early over more than 730 km. A loose link between a GPS receiver and the master clock made the trip look 73 billionths of a second shorter.

Asking for help

The team found no fault itself, so CERN invited physicists everywhere to scrutinize the work and repeat the measurement.

Checked and settled

In 2012 four experiments at the same Italian lab measured neutrinos moving at the speed of light, as physics predicts.

Neptune found by math

Neptune was found in under an hour, within 1 degree of where math said it would be

In September 1846 Johann Galle in Berlin aimed a telescope where Urbain Le Verrier had calculated an unseen planet must be. Le Verrier had worked backward from the way Uranus strayed from its predicted path.

Two possible answers

Uranus's drift meant either Newton's law of gravity failed far from the Sun or a hidden planet was pulling on it. Finding Neptune showed Newton's law still held.

Some luck too

Later critics argued both Le Verrier and John Couch Adams overestimated the planet's distance and got its position nearly right partly through a fluke of timing.

Sources

  • What makes an experiment convincing?

    Set up a fair comparison, reduce bias, and judge how far an experiment's result can support a claim.

    4 lessons · 14 min

  • What a statistical test actually decides

    Set up a null and an alternative hypothesis and see which errors a test can make. Then read a result for the narrow claim it actually supports.

    4 lessons · 21 min

  • Albert Einstein

    A patent examiner published four papers in 1905 that changed physics, then spent years on a new theory of gravity. See what he did, how he argued with quantum theory, and his life as a refugee and campaigner.

    4 lessons · 16 min

  • The Moon's changing phases

    See each phase as a changing view of a sunlit sphere, then use that geometry to explain the Moon's schedule and its eclipses.

    4 lessons · 22 min

  • Natural selection without a plan

    See how inherited differences and unequal reproduction change a population, with no intention, foresight, or universal ladder of progress involved.

    4 lessons · 14 min

  • Acids, bases, and what pH measures

    See what makes something an acid, how water manages to be an acid and a base at once, and how to read a pH number, including why neutral is 7 only at 25 degrees Celsius.

    4 lessons · 20 min

All science topics · Take the quiz