The Cloud Phase

A big star does not simply stop. It explodes, and what is left is called a remnant — the star's own material, thrown outward at thousands of kilometres a second.

The version most of us were handed ends about one step past that. The shell expands, thins, cools, and disperses into the interstellar medium. Fade out. The star is finished, and the sentence is finished with it.

But the physics runs further than the sentence does, and there is another word waiting at the end of it.

In 2024 Leonard Romano, Manuel Behrendt and Andreas Burkert followed remnants through their long cold decline in three-dimensional simulation. They found the late evolution divides into four stages — and the one at the end they call the cloud phase.

That is the whole reason this zine exists. We did not go looking for a metaphor and bend the astronomy to fit it. The astronomers got there first and wrote the word down. What is left of a dead star is understood as a cloud, studied as a cloud, and given a stage of its own in the literature.

The intuitive picture of a remnant is a bubble that just keeps growing until it is too thin to see. Outward, all the way down.

That is not what the simulations do at the end. Romano and colleagues find four stages inside the radiative phase:

The third stage is the turn. Once the shell is no longer being pushed from behind, and the pressure inside has fallen to match the pressure of the medium outside, nothing is holding the cavity open. So the remnant falls back into itself.

And what settles in the middle is not thin leftover. In these simulations it is highly chemically enriched, can reach a thousand to ten thousand solar masses, and can become self-gravitating within a few million years — which the authors call an attractive, novel pathway for supernova-induced star and planet formation.

The timescales are long. The implosion takes a few hundred thousand to a few million years to launch. The cloud takes a few million to about ten million more to form. Nothing here is quick, and nothing here is idle.

On 23 February 1987 a star exploded in the Large Magellanic Cloud, close enough to see with the naked eye. It was the nearest supernova since 1604, and we have been watching it continuously ever since.

In 2014 ALMA resolved the millimetre emission from the middle of that remnant and measured what was sitting there: at least 0.2 solar masses of dust — the largest mass then measured in any supernova remnant — concentrated at the centre, in the cold inner ejecta, not yet reached by the reverse shock. Later work with Herschel and further ALMA data puts the cold dust nearer half a solar mass.

Here is the part worth sitting with. That dust is new. Silicate and carbon grains condensed out of the cooling wreckage, assembled from atoms forged in the star's last seconds and in the explosion itself. Before the star died, that material did not exist in that form anywhere.

Dust is not incidental, either. It is what lets a cloud cool, what shields molecules from starlight, what eventually accretes into rock. The grains in your bones were made in places like this. Some fraction of them, in something's ejecta, in some galaxy's slow afterwards.

This is the middle of the argument, and it is not a simulation. It is a measurement.

In 2011 Norman Murray took the thirteen most luminous free-free radio sources in the Galaxy — responsible for a third of the Milky Way's free-free emission — matched them to their host giant molecular clouds, and asked how much of each cloud had actually been turned into stars.

The star formation efficiency ran from 0.002 to 0.2, with an ionizing-luminosity-weighted average of 0.08.

Eight percent. In the most vigorously star-forming clouds in the Galaxy — the bright ones, the ones you would pick deliberately if you wanted star formation to look efficient — ninety-two percent of the cloud does not become a star.

And that is the flattering figure. Murray gives the Galactic average efficiency as 0.005.

Read that as a shortfall and you have read it wrong. There is no rate the cloud is failing to hit. Nobody set a target. There is no efficient cloud somewhere, meeting expectations, against which these are underperforming.

Ninety-two percent is not what the cloud failed to do. It is what a cloud is.

The same paper measures the other half of it: how long a giant molecular cloud lasts.

27 ± 12 million years, at the mean mass of that sample — a bit less than three free-fall times.

Tens of millions of years of being a cloud. And the bright episode inside that span is short, late, and terminal: Murray finds the luminous clusters in his sample are actively tearing their host clouds apart. The star formation does not crown the cloud's life. It ends it.

So the arithmetic runs the opposite way from the story we get told. Put the two measurements side by side and the shape is unmistakable:

Which means the in-between is not the gap between the real states. It is the majority state — by mass, and by duration — and the luminous thing is the exception that terminates it.

We are not offering that as consolation, and it would be a weaker piece if we were. It is simply what the measurements say. The refusal to rush this to a bright resolution is not editorial restraint. It is the data.

Helen Edgar names this stage from the inside, and she named it before we went looking for the physics.

In My Monotropic Galaxy (More Realms, June 2026) she maps twenty constellations of her own Autistic self. The second one is Supernova Remnant, and it is burnout.

She is precise about the direction of travel: “Each significant burnout has remade me and formed new constellations.” Remade. Not restored, not repaired, not returned.

The third part of her burnout series makes the word carry weight — Autistic Burnout Recovery as Ecological Re-assembly — where recovery is the rebuilding of relational and attentional systems rather than a journey back to a previous configuration.

Which is exactly what a remnant does. The material persists; the arrangement does not. Every atom is still there, and not one of them is where it was, and the thing they now compose is not the star.

So when we are asked to come back — the same output, the same tolerances, the same mask, the same person we were before we broke — we are being asked for a return that the physics of the thing does not contain. There is no configuration to go back to. There is only what assembles next, out of the same material, differently.

Reading Deleuze and Guattari's body without organs through Autistic experience, Helen Edgar glosses deterritorialisation in six words.

In the same piece: “attentional resources redistribute in order to survive.” The familiar routes into engagement stop working, and other ways of regulating and relating become available in their place. Reorganisation, not damage.

A cloud does this too — and, crucially, it does it relationally.

What a cloud does next is not set by anything internal to it. It is set by the gravity of what is nearby, the radiation falling on it, the pressure of the medium it sits in, the winds and explosions of its neighbours. Romano's remnant implodes at a precise and entirely external moment: when the pressure inside finally matches the pressure outside. The trigger is the surroundings.

Ombre Tarragnat's ethodiversity extends the same reading past our own species — the diversity of ways living beings perceive and engage their environments — and Helen reads Autistic experience through what Tarragnat calls weather-bodies: shaped by a total climate that is sensory, social, technological and ecological at once.

So the honest question when one of us reorganises is not what is wrong with them. It is what is the pressure, and where is it coming from. To read the reorganisation as an individual deficit is to read the cloud with the field deleted — and the field is the entire explanation.

The twentieth and last constellation in Helen Edgar's galaxy is Emergence Point.

Now the care, because this is where a piece like this usually goes wrong.

It would be easy to run that together with the astronomy and land somewhere warm: the cloud becomes a star, so hold on, your star is coming.

We are not saying that, and spread five forbids it. Most of the mass never lights. Ninety-two percent — or ninety-nine and a half — of a cloud is not quietly gestating anything. It is a cloud, at rest, cold and structured and enormous, for tens of millions of years, and that is a complete and finished description of it.

The claim we are actually making is smaller, and it holds:

A waiting room is defined by what happens next; nothing in it is the point. A molecular cloud is defined by what it is — cold, dense, dusty, chemically rich, gravitationally structured, and among the most massive objects in the galaxy's disk. It is not a lesser version of a star. It is a cloud, and a cloud is a thing to be.

And if something does surface, it surfaces on its own timescale — millions of years for a cloud, and whatever it is for you. Neither timescale takes instruction, and neither one owes anybody an arrival.

Our own solar system began in a cloud, and there is a fingerprint of what was in it.

In 1976 Typhoon Lee, Dimitri Papanastassiou and Gerald Wasserburg found an excess of magnesium-26 in calcium–aluminium-rich inclusions in the Allende meteorite — the decay product of aluminium-26, which has a half-life of about 717,000 years.

That half-life is the entire argument. ²⁶Al is made in massive stars. For its decay product to be locked into the oldest solids in the solar system in that quantity, live ²⁶Al had to arrive in our birth cloud within a few half-lives of those solids forming. Something nearby, and freshly dead, was in the material we are made of.

Fifty years on, which something is openly contested:

Tang and Dauphas measured the early solar system's ⁶⁰Fe in 2012 and found it steady and low — around 1 × 10⁻⁸ relative to ⁵⁶Fe. A nearby supernova should have brought both isotopes. The missing iron is why a wind is now a serious contender.

That dead stars' material is in us is not in doubt. Which star, and by what route, is not settled — and we are not going to pick one to make the ending tidier. The contested part gets its name, out loud, on the last page.