A nervous system, in eight movements
Scroll, and a water bear's brain is found, measured, taken apart, and switched on. Every number here was measured from a published X-ray scan or a public transcriptome. Four of them had never been measured by anyone.
Loading about eleven megabytes of water bear…
One hundred and fifty-two micrometres
This is a real individual, scanned by X-ray nano-tomography and published in 2019 under a licence that let anyone rebuild it. Nobody had.
It is about a sixth of a millimetre long. Everything that follows is inside this.
The brain sits over the gut like a saddle, wrapped around the tube the animal eats through. In a body this size it is one per cent of the volume.
For comparison, yours is about two per cent.
Four ganglia nobody had measured
Behind the brain, four swellings sit under the gut, one per pair of legs. The paper that published this scan states plainly that it did not measure them.
So we did, by cutting the cord where its cross-section passes 10 µm²: 385, 430, 379 and 194 µm³, front to back. The fourth is half the size of the others, and stays half the size however the cut is made.
The brain is 1,389. The four ganglia together are 1,386.
We have no explanation for that and we are not going to invent one. It survives every threshold we tried, and it is the kind of thing that is either meaningless or the whole point.
Two hundred and fourteen cubic micrometres of cable
Paired connectives run the length of the animal, joining each ganglion to the next. They are thin — 2.5 to 6 µm² in cross-section against the 47 to 55 of a ganglion — which is what made cutting the cord into pieces possible at all.
That is the entire central nervous system: 2,728 µm³. A cube 14 µm on a side.
Shape is the easy half
All of that is one adult, fixed and dried. It says nothing about how any of it got built.
So: sixty-two embryos, each killed and sequenced at a known moment across the four days from laying to hatching. Nine thousand genes, measured by somebody else in 2016, and left alone since.
The pattern fires in twelve hours
The genes that lay out the nervous system peak between 41 and 53 hours — a twelve-hour window in a four-day development — and then go quiet.
We reported at first that they did nothing at all. They failed every test we ran. They were failing because we were testing whether they trended, and a pulse does not trend. The null was a property of our ruler.
And one gene that turns on
Inside that window a gene comes on that is absent from almost every earlier embryo: 288-fold, from nothing to fifty counts per million.
It is the tardigrade's version of unc-86 — the gene that, in a nematode, decides which cells become which neurons. No POU gene has ever been reported in any tardigrade. It sits with unc-86 at 93% bootstrap.
Then the synaptic machinery climbs, slowly, from 45 hours until the animal hatches. The pattern is laid down fast and the wiring is built slowly, and you can watch the difference rather than take it on trust.
What this is, and what it is not
Every number on this page is a measurement, and most of them came out of data that was already public. The scan was published in 2019, the transcriptome in 2016. The four ganglion volumes, the coincidence with the brain, the twelve-hour pulse and the unc-86 orthologue are ours, and each is recorded in the graph with the evidence attached and the failures kept.
What it is not: this is one adult animal, fixed and dried, and bulk expression from whole embryos. It cannot tell you which cells anything is in. The four ganglion volumes depend on a threshold, and we report the range it survives. And an orthologue is not a function — what unc-86 does in a nematode is a reason to look, not a result.