Galaxies and the Milky Way

A wide band of sky filled with countless small stars, denser and yellower on the left, with streaks of dark dust and bluish clusters. — Part of the disc of the Andromeda galaxy (M31), in the largest and sharpest image of it ever taken by the Hubble Space Telescope: more than 100 million stars can be seen. NASA, ESA, J. Dalcanton, B.F. Williams, and L.C. Johnson (University of Washington), the PHAT team, and R. Gendler, CC BY 4.0, source

A little over a hundred years ago, nobody knew whether other galaxies existed.¹

This was not simply a debate: nobody knew. It was the great open question of astronomy, and it was settled in five years — from 1920 to 1925.

The debate of 26 April 1920

On that day, in Washington, two astronomers take the stage at a museum to debate before their colleagues from the Academy of Sciences.²

The question: what are the fuzzy spiral patches that can be seen in the sky?³

Harlow Shapley, an astronomer at Mount Wilson, argues that the spirals are much smaller than the Milky Way, and therefore that they are part of it. For him, our galaxy is the Universe.⁴

Heber Curtis argues the opposite: they are other galaxies, immensely far away. People spoke at the time of “island universes”.⁵

Yet Shapley was far from narrow-minded. That day he defended a Milky Way three hundred thousand light-years across — an overestimate, but much larger than what was accepted at the time. On another point, moreover, he was right and Curtis, who placed the Sun very close to the centre, was wrong: the Sun is not at the centre.⁶

The plate of autumn 1923

In 1923, the astronomer Edwin Hubble spends several months scanning the Andromeda nebula with the most powerful telescope of the era — the Hooker, one hundred inches in diameter, on Mount Wilson, in California. The target is so large that he needs dozens of photographic plates to cover all of it.⁷

On the night of 5–6 October 1923, in poor conditions, he makes a forty-five-minute exposure that yields three suspect stars. He writes “N”, for nova, next to each one.⁸

Comparing this plate with the earlier ones afterwards, he finds that one of the three brightens and dims far too quickly for a nova. He derives a period of 31.4 days from it: it is a Cepheid.⁹

On the plate, he crosses out the “N” and writes: “VAR!”¹⁰

Then he writes to Shapley to tell him, enclosing the star’s light curve.¹¹

Why this settles the question

Our page on light and distances tells the story of Henrietta Leavitt, who discovered that the length of a Cepheid’s cycle of variation reveals its true brightness.

This was precisely the tool Hubble needed. The period gives the true brightness; compared with the apparent brightness, it gives the distance.¹²

His first estimate places the star at one million light-years — more than three times the diameter Shapley gave the Milky Way. By the end of 1924, with thirty-six variable stars found in Andromeda, twelve of them Cepheids, he obtains nine hundred thousand light-years.¹³

Current measurements place Andromeda at about two million light-years. Hubble was therefore below the true value, which did not alter his conclusion.¹⁴

Andromeda was not a cloud within our galaxy. It was another galaxy.

One instrument for all

It is worth noting who used Leavitt’s law in this story.

Shapley used it to measure the Milky Way. Hubble used it to measure Andromeda. Both sides of the debate, and the man who settled it, were using the same instrument.

That instrument had been forged by a woman whose result appeared under the signature of the director of her observatory: the 1912 paper that states the law bears the name of Edward Pickering, and its first sentence specifies that the measurement was prepared by Miss Leavitt.¹⁵

That is the cost of an erasure. It is not only unjust to the person: it makes the story we pass on false. We say that Hubble discovered the galaxies, without saying what he was measuring with.

Shapley’s words

When Hubble’s letter reaches him, Shapley comes round at once: the evidence seems solid to him.¹⁶

A young doctoral student, Cecilia Payne, is in his office that day. Much later, she would recount what he said to her as he folded the letter:

“Here is the letter that destroyed my universe.”¹⁷

He does not dispute it and looks for no way out. He observes that his view of the world has just collapsed, and says so out loud.

There is, however, a second reaction, which nicely complements the first. In the letter he sends to Hubble, Shapley writes that his letter is the most entertaining piece of literature he has seen for a long time.¹⁸

Crushed in private, gracious in writing. The two attitudes do not contradict each other: it is possible to acknowledge one’s defeat while refusing to make a drama of it in front of one’s opponent.

It is Shapley himself and the astronomer Henry Norris Russell who then urge Hubble to present his results to their colleagues, at the end of December 1924. The paper is read in his absence and shares the prize for the best paper. The news causes no stir in the room: Hubble had informed the leading astronomers several months earlier.¹⁹

One last, surprising point

An image search for the Milky Way brings up dozens of pictures of it: a beautiful spiral, seen from above, with its arms and its bright centre.

None of these images is a photograph.

We are inside our galaxy. You cannot photograph a house from its living room. No probe will ever travel far enough to turn round and take its picture: New Horizons, for example, is more than seven billion kilometres from Earth after eighteen years of flight — a tiny distance on the scale of a galaxy.²⁰

All these images are therefore reconstructions: drawings or computed models, based on measurements made from the inside — where the stars are, how far away, in which direction.

They are not false. But they do not mean “here is our galaxy”. They mean: here is what it would look like seen from elsewhere, if our measurements are right.

How many are there?

One would expect a number. Here are three, with the reasons why all three are honest.

1990s: one hundred to two hundred billion galaxies, based on the deep images from the Hubble telescope.
2016: at least two trillion — ten times more. The figure then goes round the world.
2021: the estimate comes back down to a few hundred billion.²¹

In twenty-five years, the number rose tenfold, then fell back by roughly as much.

The universe did not change during that time. It is the instruments and the methods that changed.

Why these two figures are so different

They do not answer the same question, and they are not obtained in the same way.

The 2016 figure is not based only on galaxies that have been seen. The visible ones are counted, then extrapolated: from the mass of the galaxies observed out to the visible limits, one estimates how many small galaxies too faint to be seen elude the telescopes. The numbers add up on one condition only: that 90% of galaxies are still invisible to our instruments.²²

The 2021 figure rests on an entirely different, and particularly elegant, idea. The New Horizons probe, at more than forty times the Earth–Sun distance, sees a sky about ten times darker than the darkest accessible to the Hubble telescope. From there, the background light of the sky is measured — the sum of the light from everything that exists beyond our galaxy. Yet this light is not enough to account for two trillion galaxies. One of the lead authors summed it up simply: take all the light from the galaxies the Hubble telescope can see, double it, and that is what we see — but nothing more.²³

This measurement has since been refined. In 2021, the team found a background slightly brighter than expected, without knowing where the extra came from; in 2024, with better data, it showed that this was stray light from our own galaxy.²⁴

On one side, an estimate by extrapolation; on the other, a measurement of total light. These are not two counts that contradict each other: they are two ways of approaching the question.

What holds despite the disagreement

This point matters more than the number itself.

On every estimate, a large share of the galaxies in the universe remains too faint for us to see. The disagreement is about how many — not about the fact that we glimpse only some of them.

The number of galaxies is not an established fact written in the sky: it is an estimate, which changes every time a new instrument sees a little further into the dark.

That is why this page does not give a single number. When a figure moves this much, the honest thing to pass on is not the figure: it is the series, with its dates, and the reason for its variations.

Note 1

Though the universe is filled with billions upon billions of stars, the discovery of a single variable star in 1923 altered the course of modern astronomy. […] But in the early 1900s, most astronomers considered the Milky Way a single "island universe" of stars, with nothing observable beyond its boundaries. […] Were these spiral nebulae part of the Milky Way or were they independent island universes lying outside our galaxy? Astronomers didn't know for sure

NASA, “Hubble Views the Star that Changed the Universe”, communiqué du 23 May 2011