The Repository Deep Time Archives

The Doushantuo Embryos Never Grew Up Into Animals

September 28, 2026 · Runtime 11:31 · Watch on YouTube

In the Weng'an phosphorite of South China, balls of cells less than half a millimetre across, split into two, four and eight, were published in 1998 as the oldest animal embryos. In 28 years they have also been read as giant sulphur bacteria, as encysting protists and as stem animals. In March 2026 a count of their cells came back in powers of two, with no gastrulation.

Chapters

  1. 0:00 Waves over a shallow sea
  2. 0:29 The Weng'an Biota
  3. 0:59 The Doushantuo Formation
  4. 1:31 A beach of reworked grains
  5. 1:59 Out of the rock with acid
  6. 2:28 1998: animal embryos
  7. 2:57 What a real embryo does
  8. 3:28 Bigger and bigger claims
  9. 4:01 The worm that was a slice
  10. 4:30 Giant sulphur bacteria?
  11. 4:58 The decay test
  12. 5:30 Inside, without cutting
  13. 5:59 Nuclei, and germination
  14. 6:30 Spheres inside cysts
  15. 7:02 The black-layer reading
  16. 7:33 A borrowed date
  17. 8:05 13 of 21 zircons
  18. 8:36 609 million years, or older
  19. 9:06 2,048 and 4,096 cells
  20. 9:33 No gastrulation
  21. 10:04 Measured, borrowed, open
  22. 10:31 Detail is not identity
  23. 11:02 Next: Dickinsonia's molecule

Transcript

Waves over a shallow sea 0:00

Six hundred million years ago, waves rolled across a shallow sea over what is now southern China, pushing grains of sand back and forth. Some of those grains are balls of cells, less than half a millimetre across, split into two, four, eight. For a quarter of a century they were the oldest animal embryos on Earth. This year, someone counted their cells. And the count did not come out the way an animal embryo does.

The Weng'an Biota 0:29

This is the Weng'an Biota, in Guizhou Province, and these fossils are preserved in calcium phosphate, down to single cells. In twenty-eight years they have been read as animal embryos, as giant bacteria, as single-celled organisms in resting cysts, and as early relatives of animals. By the end, you will know how many cells they reach, where their date really comes from, and why the most detailed fossil of its age still cannot tell us what it is.

The Doushantuo Formation 0:59

The rock is the Doushantuo Formation. It begins directly on glacial debris from the last Snowball Earth, about six hundred and thirty-five million years ago, and runs up to about five hundred and fifty-one. The fossils sit in one unit of it: the Upper Phosphorite. At Weng'an that unit has a black layer about five metres thick, and above it a grey one about ten metres thick. In places the grey layer is so full of microfossils that it looks like a rock made of beads.

A beach of reworked grains 1:31

And those beads were not buried where they lived. The phosphorite beds carry wave ripples and cross-bedding, the marks of water shallow enough for ordinary waves to stir the bottom. The fossils are abraded and rounded. They were turned to phosphate first, then broken loose, rolled, carried across the basin and sorted by the waves. What we collect is not a graveyard. It is a beach of reworked grains.

Out of the rock with acid 1:59

Getting them out is chemistry. The grey rock is soaked in weak acetic acid, which dissolves the carbonate and leaves the phosphate. The residue is sorted by hand under a microscope, grain by grain. The black layer does not give them up this way: there the fossils sit inside reworked phosphate clasts, and the only way to see them is to cut the rock into thin slices. Remember that difference. It will matter.

1998: animal embryos 2:28

In 1998, Shuhai Xiao, Yun Zhang and Andrew Knoll published them in Nature. Spheres inside an ornate envelope, each one divided into two cells, then four, then eight, without growing larger. Their size, around half a millimetre, was what you would expect for the egg of an early animal. Their conclusion was that the lineages leading to complex animals may have split well before those animals left any large fossil.

What a real embryo does 2:57

Here is what that comparison means. This is a living sea urchin embryo. It divides the same way, two, four, eight, sixteen, each cell smaller than the last, inside the same envelope. But an animal embryo does not stop at a ball. Later, one side folds inward and the cells start to move into layers. That fold is called gastrulation, and it is the beginning of a gut. Keep that word too.

Bigger and bigger claims 3:28

What followed was a run of bigger claims from the same deposit. Larvae of animals in 2000. In 2004, a tiny adult with a mouth, a gut and an anus, described as a bilaterian forty to fifty-five million years before the Cambrian. In 2006, embryos with polar lobes, a trait of some bilaterians. Each claim made the first animals more complex than the last. And several of them came from flat slices through round objects.

The worm that was a slice 4:01

The tiny adult was named Vernanimalcula. It was seen only in thin sections. In 2012, a team scanned Weng'an fossils in three dimensions and showed that ordinary spheres, cut at a particular angle, can look exactly like it. The mouth, the gut and the body cavities could all be mineral cement that grew in the fossil after burial. The paper's title called it a merciful death for the earliest bilaterian.

Giant sulphur bacteria? 4:30

The embryos themselves faced a different challenge. In 2007, Jake Bailey and colleagues pointed to a living organism with the same size and the same trick: Thiomargarita, a giant sulphur bacterium. Its cells can divide in three planes into symmetrical clusters, and today it lives where phosphate minerals form on the sea floor. Maybe, they argued, the oldest animal embryos were bacteria.

The decay test 4:58

That idea could be tested, and it was. In 2012, John Cunningham and colleagues let Thiomargarita decay in the lab and compared it with decaying embryos. About ninety-eight percent of a Thiomargarita cell is one central vacuole, and the cells simply collapse. Embryos are different: bacteria eat them and replace them with a three-dimensional film that holds their shape. The fossils have structures spread through their cells, like embryos, not one great empty vacuole.

Inside, without cutting 5:30

To see inside without cutting, the fossils went to a particle accelerator. The Swiss Light Source, at the Paul Scherrer Institute, makes X-rays bright enough to scan a half-millimetre sphere in three dimensions, slice by virtual slice. In 2006 one such scan showed embryos of about a thousand cells, still with no sign of cells arranged into layers. Five years later, the same method went further.

Nuclei, and germination 5:59

In 2011, Therese Huldtgren and colleagues reported structures inside the cells that they read as nuclei. They also found later stages in which cells break out of the envelope, the way spores germinate. Their conclusion: these fossils are neither animals nor embryos. They probably belong to the wider group that contains animals and their single-celled relatives. Within months, Xiao and colleagues replied that the structures were too large to be nuclei.

Spheres inside cysts 6:30

There is one more clue, and it points at eggs. In 2007, Leiming Yin and colleagues found the same embryo-like bodies inside large spiny cysts, in the Yangtze Gorges. That makes them organisms with complex cells, not bacteria, and it suggests resting eggs that wait out bad conditions. Cysts of that type first appear just above a volcanic ash dated to six hundred and thirty-two and a half million years, barely three million years after the ice.

The black-layer reading 7:02

In 2014, Lei Chen and colleagues went back to the black layer, the one you can only see in slices. They described large spheres with clusters of cells nested inside, and read them as cells set aside for reproduction, like the division between body and germ cells in animals. Their conclusion was a stem animal or an alga. A 2017 review answered that the steps between single cells and those clusters are missing, so the clusters could be something else.

A borrowed date 7:33

So when did all this live? The phosphate at Weng'an was dated directly, by lead isotopes, to 572 and 599 million years, ages now considered unreliable. The number now quoted, six hundred and nine million years, plus or minus five, was not measured at Weng'an at all. It comes from a volcanic ash at Zhangcunping, in Hubei, hundreds of kilometres away, lying just above a phosphorite unit that matches the one at Weng'an.

13 of 21 zircons 8:05

Even that ash had to be sorted. Twenty-one zircon crystals were measured. Eight were older grains swept into the ash, some about eight hundred million years old, two nearly two billion. The other thirteen agreed on six hundred and nine. Then that age travels to Weng'an by matching the layers, the microfossils and the carbon isotopes between the two sites. The date is measured. Its link to these fossils is argued.

609 million years, or older 8:36

If that correlation holds, the Weng'an fossils are about six hundred and nine million years old, or older. That is some twenty-six million years after the Snowball Earth ended, nearly thirty million years before the Gaskiers ice, and some thirty-four million years before the oldest complex fossils at Mistaken Point, from the last episode. That is why these spheres matter so much. If they are animals, they are the oldest animal fossils anyone has.

2,048 and 4,096 cells 9:06

Which brings us to this year. In March 2026, Kirsten Flett and colleagues scanned hundreds of the large many-celled spheres and counted cells. The counts ran from fourteen to ten thousand two hundred and one, and they bunched up at two thousand and forty-eight and at four thousand and ninety-six. Those are powers of two. Each cell kept splitting in half, eleven times, then twelve.

No gastrulation 9:33

But not in step: inside one sphere, cells differ in size by two to three times. The overall volume wanders as the cells multiply. And there is no fold, no movement into layers, no gastrulation. The small spheres and the large ones turn out to be early and late stages of one organism. The authors' verdict: incompatible with an animal of any living group, and not a fossil to set the clock of animal origins by.

Measured, borrowed, open 10:04

So what does the rock let us say? The preservation is measured: cells, and perhaps what was inside them. The age is borrowed, from an ash hundreds of kilometres away. The identity is open. One rarer Weng'an fossil, Caveasphaera, does move cells inward in a way its describers compare to gastrulation, and even they stop at the wider group that includes animals, not at animals themselves.

Detail is not identity 10:31

The story most people know is simple: the oldest animal embryos, frozen mid-division, proof that animals were already here. The rock gives something stranger. A cell that divides into two, four, eight is doing what an egg does, and also what a spore body does. Twelve rounds of it, with no fold, are not the embryo of any living kind of animal. The most detailed fossils of their age are also the ones that will not say what they are.

Next: Dickinsonia's molecule 11:02

Next time, we go to the White Sea coast of Russia, about five hundred and fifty-eight million years ago, to a fossil called Dickinsonia. It was declared an animal not by its shape, but by a molecule. On the largest specimen, ninety-three percent of the steroid molecules were the animal kind. In the sediment around it, about eleven percent. The next episode asks what one molecule can settle.

Description and sources

Weng'an Biota, Upper Phosphorite of the Doushantuo Formation, Guizhou: fossils phosphatized to cellular detail, reworked by waves in shallow water. Xiao, Zhang & Knoll (1998): embryos in early cleavage. Vernanimalcula (2004) failed in 3D (Bengtson et al. 2012). Thiomargarita (Bailey et al. 2007) failed a decay test: its cell is about 98% vacuole (Cunningham et al. 2012). Huldtgren et al. (2011): nuclei and germination, neither animals nor embryos. Chen et al. (2014): cell differentiation, contested. The age, 609 ± 5 Ma, is a zircon date from an ash at Zhangcunping, Hubei, carried to Weng'an by correlation (Zhou et al. 2017): 13 of 21 zircons. Flett et al. (2026): 14 to 10,201 cells, clusters at 2,048 and 4,096, asynchronous division, no gastrulation — incompatible with a crown-group animal.

PRINT-READY, FROM THIS CHANNEL

The Deep Time Record, Vol. 01 — the cores, the curves and the clocks

https://therepository.gumroad.com/l/deep-time-record

PRIMARY SOURCES

ARCHIVE IMAGES — author, date, archive and licence.

  • John A. Cunningham et al. — 2017 | Wikimedia Commons | CC BY 3.0
  • www.sqfp.info photograph | Wikimedia Commons | CC BY 2.0
  • Zongjun Yin et al. — 2022 | Wikimedia Commons | CC BY 4.0
  • Laurent Formery et al. — 2022 | Wikimedia Commons | CC BY 4.0
  • Ryan Bodenstein — 2018 | Wikimedia Commons | CC BY 3.0
  • Zongjun Yin et al. — 2022 | Wikimedia Commons | CC BY 4.0

... and 3 more, each credited on screen in the video and listed in this project's manifest.

The 2 generated plates were made for this channel with FLUX.1-schnell. No person is depicted in any of them.

Deep Time Archives rebuilds the pre-human Earth from the physical record — outcrops, drill cores, golden spikes, catalogued specimens. Every episode starts from an object you can go and check, and the sources are listed above.

About the images. The fossil images are scientific figures, credited on screen: SEM images of Weng'an fossils from Cunningham et al. 2017 and Yin et al. 2022 (two panels cropped from Yin et al. fig. S7; the cover is a crop of panel A with the authors' annotation box removed), and Caveasphaera. Also archive photographs: a living sea urchin embryo in cleavage (Formery et al. 2022), the Swiss Light Source, the top of the Doushantuo Formation near Sancha, Hubei (cropped), and a Dickinsonia from South Australia (same genus as the White Sea specimens in the next episode). The shallow sea and rock ledge are AI-generated reconstructions; no organism or cell is generated. This video is labelled as containing altered or synthetic content. The diagrams are drawn in code from the published numbers.

#DeepTime #EarthHistory #Ediacaran #Paleontology #Evolution