Almost Shadow

Johannes Kepler was trying to draw an eclipse properly, and the language stopped him.

The problem is geometry. If the thing casting a shadow is lit by a point of light, the shadow has one crisp edge and you need one word. But the Sun is not a point. It is a disc — an extended source — and every part of it throws its own shadow in a slightly different direction.

So the shadow comes out layered. There is a core where the whole disc is hidden. There is an outer region where some of the disc is hidden and some of it still shines through. Between them there is no edge at all, just a handover.

Kepler had umbra for the core. For the handover he had nothing. So, in the middle of a diagram in Ad Vitellionem Paralipomena, he made one:

Let what lies between KE and NM be called penumbra — and LN, indeed, umbra.

The verb is dicatur: let it be called. That is not a man using a word. That is the word being made, in one clause, because a drawing demanded it.

He built it out of paene, almost, and umbra, shadow. Almost-shadow. And notice what gets named: not the dark, not the light. The thing between them.

A few pages on he is already using it as an ordinary working term — decrescente, crescit penumbra, as the one shrinks the other grows — and noting that well before an eclipse looks like it has begun, nearly the whole Moon is already inside the almost-shadow.

Which is the first thing this zine wants: the partial state arrives long before anyone would say anything is happening.

Kepler's two words became three, and the third one is the one people have never heard of.

Because the Sun is a disc and the Moon is a ball, the Moon's shadow is a cone — it narrows to a point and then stops. Which means the shadow behaves differently depending on how far down it you stand:

the Moon's shadow, per NASA's eclipse glossary

zone	what you see from inside it
umbra	the darkest part of the shadow. The Sun is completely blocked — a total eclipse.
penumbra	the weak or pale part. The Sun is only partially blocked — a partial eclipse.
antumbra	past the cone's apex. The Sun appears larger than the Moon, which is seen in complete silhouette — an annular eclipse, the ring of fire.

The antumbra is worth sitting with, because it is the case a two-word vocabulary genuinely cannot describe. Stand there and the Moon is entirely in front of the Sun — nothing is sticking out, nothing is off to one side — and it is still not a total eclipse, because the Moon is simply too small in your sky to cover what is behind it.

Two words would have forced that case into one of the other two, and both would have been wrong. The vocabulary has three zones because the world has three.

Here is the fact that rearranges the picture, and it is purely a matter of footprint.

When the Moon's shadow lands on Earth, the two zones are not remotely comparable in size. The umbra touches down as a narrow track — on the order of a hundred to a few hundred kilometres across, a stripe you can drive out of in an afternoon. The penumbra lands as a vast disc, thousands of kilometres wide, covering a good fraction of a hemisphere.

Run the consequence. Everyone inside that huge disc is having an eclipse. Almost none of them are in the track.

And this is exactly why the word had to be invented rather than borrowed. The condition Kepler named is not an edge case or a rounding error at the border of the real thing. It is the ordinary condition, and the total case is the exception.

Which sets up the trap this zine spends the next spread walking into on purpose. If partial is the normal state, and partial is a gradient, it is very tempting to conclude that the difference between 99 percent and 100 percent is just... one percent.

It is not. It is not close to one percent. And the reason is not poetic.

Stand at 99 percent obscuration and you are not almost at totality. You are in a slightly odd afternoon.

The reason is that the Sun's surface is outrageously bright and the corona is not. The corona — the thing you have come to see, the pearl-white structure that only exists for the eye during totality — is roughly as bright as a full Moon.

Now leave one percent of the photosphere uncovered. That surviving sliver is still hundreds to thousands of times brighter than a full Moon. It does not politely dim the scene. It obliterates the corona and keeps the sky blue.

Photometry bears this out as a cliff rather than a slope: measured illuminance falls roughly a hundred-fold in the last two minutes before totality, by another order of magnitude in the final half minute, and by more than four orders of magnitude in total between full Sun and totality.

And then the Moon covers the last sliver, and the world does something else entirely:

A fermata is the mark that tells a musician to hold a note past its written value — to suspend the count. It is the right word, and it is not available at 99 percent. At 99 percent the birds are unbothered and the count runs on.

So both of these are true at once, and the zine will not drop either:

The shadow is a gradient. There is no edge; there never was; Kepler needed a third word precisely because the handover is smooth.

And there is a threshold inside the gradient. Somewhere in that smooth handover sits a point where the birds stop.

Most arguments about difference pick one of these and lose the other. The shadow does not let you.

Pick gradient only and you get the flattening move: everyone's a bit like that, it's all a spectrum, we're all somewhere on it. True as far as it goes, and it dissolves every threshold that matters. If it is all one smooth ramp, nothing in particular is owed to anyone in particular. The word for this is not solidarity. It is erasure with a friendly face.

Pick threshold only and you get the gatekeeping move: either you qualify or you don't; either you're disabled or you're fine. Also recognisable. It is the shape of every eligibility form any of us has ever been failed by, and it makes the enormous partial zone invisible by construction — which, as spread four showed, is where nearly everyone actually is.

Kepler's word is precise about this in a way our usual vocabulary is not. Penumbra does not mean a bit shadowed. It names a specific region with real boundaries — you are in it or you are not — whose defining property is that the blocking there is incomplete.

It is a category, not a hedge. A named place where a partial state is the actual condition, and the partial state is not a failed attempt at a total one.

Two other fields worked this out independently, went looking for a word for it, and arrived at Kepler's. They arrived there because in both cases treating the partial zone as though it were the total one costs people their lives.

When an artery in the brain closes, the damage is not a disc. It is a shadow, with zones.

In 1981, Jens Astrup, Bo Siesjö and Lindsay Symon published a paper in Stroke with a title that is half physiology and half borrowed astronomy: "Thresholds in cerebral ischemia — the ischemic penumbra."

Working in a baboon model of middle cerebral artery occlusion, and measuring blood flow alongside electrical activity, pH and extracellular potassium, they found not one threshold but two:

two thresholds, and the zone between them

as blood flow falls	what happens
first threshold	electrical failure. The tissue goes silent. It stops doing its job.
second threshold	ion-pump failure. Potassium floods out. The tissue stops being tissue.

Between those two lines sits a region that is functionally silent but structurally intact. It is not working. It is also not gone. Restore the blood supply and it can come back.

They called it the ischaemic penumbra — and the word fits with an exactness that is almost uncomfortable. A region of partial deprivation, wrapped around a core of total deprivation, distinguished from it not by how bad it looks but by whether anything can still be done.

The core is the umbra. Nothing is coming back from there. The penumbra is the part still holding on, and it is holding on on a clock.

This is the part that matters, and it is not linguistic. Naming that zone changed what happens to people.

Consider the alternative vocabulary — the two-word one. In it, brain tissue after a stroke is either alive or dead. Under that description, the urgency is unclear and the target is unclear. There is damage; you assess the damage; the damage is what it is.

Now describe the same brain with three words. Suddenly there is a region that is currently non-functional and currently salvageable, and it is shrinking, and it is shrinking as a function of time.

That description does something the other cannot: it makes a clock visible. And a visible clock is the thing an entire field of emergency medicine is organised around — the reason acute stroke has a treatment window at all, the reason imaging tries to distinguish core from penumbra before deciding, the reason the phrase time is brain exists.

Sit with the shape of that, because it recurs everywhere and is almost always got wrong.

The zone that looks least dramatic — not the catastrophic core, not the healthy surround, the ambiguous ring in between — is the only zone where anything you do makes a difference. The core is past help. The healthy tissue does not need it. All of the leverage is in the part that a two-state description cannot even name.

And a two-state description does not merely fail to help there. It actively hides where the help would go.

The second borrowing is stranger, more contested, and we are going to say so on the page rather than in a footnote.

In Griswold v. Connecticut, 381 U.S. 479 (1965), the Supreme Court struck down a law banning contraceptives. The difficulty was that the Constitution does not mention privacy. Justice Douglas, writing for the Court, reached for Kepler's word:

The claim is that a written guarantee casts something beyond its own literal edge. Protect the home against quartered soldiers, protect papers against search, protect association — and around each of those, in the region where the light is partly blocked, sits something the text implies without stating.

Now the honest part. This is among the most criticised passages in American constitutional law, and not only from one direction. The objection is real and worth stating fairly: if a text casts penumbras, who decides how far they reach? A shadow with no measurable edge can be extended to cover whatever the reader wanted covered.

We are not resolving that here; it is a live argument among people who have spent careers on it. We include it because of what the borrowing shows: twice now, in fields with nothing else in common, someone confronted a situation where a binary would produce an obviously wrong answer — and both times the word they reached for was the astronomer's.

Both times, for the same reason. A hard edge would have been easier to administer and false.

Here is where we live, and both halves of the physics are load-bearing.

Take the gradient half first. Most of us are neither fully blocked nor fully in the light — and the next spread is about who gets told they are the former. We are Disabled and Autistic and chronically ill in ways that fluctuate — good days, bad weeks, a capacity that is real on Tuesday and gone on Thursday. There is no crisp edge to point at, and every system we deal with demands one.

So we get told the partial state is not a state. You managed last month. You're doing fine now. The penumbra gets read as a failed umbra — as a claim that hasn't finished proving itself — rather than as the place where most people actually are.

Now the threshold half, which cuts the other way and is the one more often forgotten by people who mean well.

Support is not proportional. Ninety percent of a ramp is a wall. A venue that is accessible except for the toilets is not 95 percent accessible; it is a venue you cannot stay in. Captions that drop out for the last two minutes did not deliver 98 percent of the talk. The last increment is where the phase change is — and up to that point the sky stays stubbornly, uselessly bright.

And then hold them together, because separately each one is a well-worn mistake. All gradient gives you the flattening, and nobody is owed anything specific. All threshold gives you the eligibility form, and the enormous partial zone vanishes.

The shadow insists on both. Smooth, and with a place in it where the birds stop.

The last thing to carry over is the 1981 one, which is the most hopeful sentence in this zine and the least sentimental. In the penumbra, what looks like loss is often deprivation — and deprivation is a thing you can reverse, if you can see that the zone exists and get there in time.

Everything so far has been about shadows falling on things. Now the question that decides whether any of it is safe to use: where does an umbra actually live?

Go back to the geometry on spread 3. An umbra is a relation between a source, an obstruction, and a place. It is not a substance and not a property. The ground in an umbra has not changed; it is simply where the blocked light would have gone. Move the source and the umbra moves. Take the obstruction away and it is not there at all.

So the threshold on the previous spread is real, and it is an umbra of support. Support can be partial, and it can be total — ninety percent of a ramp is a wall, and behind that wall nothing gets through at all. That umbra is a fact about an arrangement, and arrangements can be changed by whoever built them.

Functioning labels do something else entirely. They take the shadow the arrangement casts and paint it onto the person standing in it. High-functioning. Low-functioning. Mild. Severe. Levels. Each one relocates a fact about circumstances into a fact about a body.

And “profound autism” is the umbra version of that move. Proposed in the Lancet Commission on the future of care and clinical research in autism (2021) for Autistic people who are minimally or non-speaking, require round-the-clock care, and are “not able to advocate for themselves.”

Hold that last criterion against what the research actually finds. Steven Kapp's 2023 review concludes that minimally speaking Autistic people can communicate — particularly with responsive partners — that they need testing which lets them show what they have rather than testing built to miss it, and that they can self-advocate.

Now put spread 7's sentence where it belongs. In the ischaemic penumbra, what looks like loss is often deprivation — and reading deprivation as loss is not a small error there. It is the exact mistake that costs the salvageable zone.

Profound reads deprivation as loss, and does it on a person.

So let this be flat and unhedged. This piece recognises an umbra in support — support that is wholly blocked, and can be unblocked. It recognises no umbra in an Autistic bodymind at all. Not a small one. Not a rare one. Not one reserved for somebody further along than you. There is no zone of person that light does not reach and cannot be made to.

And that refusal is not a denial of anyone's support needs, which would be its own way of writing people off. Some people need constant, skilled, expensive, round-the-clock support, and saying so plainly is part of the argument rather than an exception to it. Helen Edgar puts the distinction exactly:

Which is why support-needs language restores the penumbra. It is specific, it names domains, it varies by day and by room, it can go up and down, and — the part that matters here — it never declares any part of a person past reaching.

The word came from an eclipse, and an eclipse is the right lender. Medicine's umbra is a genuine property of matter: that tissue is dead. That is precisely why it must not travel any further. Borrowing the astronomer's word does not license borrowing the surgeon's verdict.

Two fields reaching for the same word is a striking fact. It is not evidence for anything about us, and we want to be exact about that before closing.

What the borrowings actually establish is narrow: that in at least two serious domains, practitioners found a two-state description inadequate to something real, and the vocabulary they wanted already existed in optics. That is a fact about the usefulness of a distinction. It is not a finding about disability, and no amount of elegance makes it one.

The physics on spreads three through five is exact and checkable. The turn on spread ten is an analogy, and analogies license nothing on their own. A rhyme, not a proof.

What we would defend without the metaphor: that fluctuating capacity is real, that partial support frequently delivers nothing, and that describing people in two states hides the place where help works. Those stand on their own evidence, and the eclipse is how we show them, not why we believe them.