Science that can prove itself.
An unbroken chain of custody from the instrument to the publication. Every version, every transformation, captured at the source.
- SampleTB-041 · batch 2026-07-12
- Structureα-Fe₂O₃ · a = 5.04 Å
- ScanXRD · CuKα · 20–80° 2θ
- Signalcounts vs 2θ · 4,801 pts
- Raw filescan03.xy · sha256:66ca679f…
- Figurefigure_3.svgtraced ✓
The chain, forming
Sample · TB-041 · batch 2026-07-12
Every result begins somewhere physical.
A reaction. A pair of hands or a robot arm. An unremarkable Tuesday.
Something forms. A sample. An artefact that did not exist earlier that morning.
Structure · α-Fe₂O₃ · a = 5.04 Å
Within it, a structure. Governed by the same fundamental rules as everything else — atoms, arranged the only way they can be.
Structure means distances — coordination, spacing, repeating motifs. Fixed, specific, indifferent to what anyone hoped for.
Measurable. Therefore knowable.
Scan · XRD · CuKα · 20–80° 2θ
So we measure it. X-rays, electrons, light, voltage, current, mass.
Physics, chemistry, biology — the same move every time.
Every measurement collapses the same way: signal, then number, then a point on a plot.
Signal · counts vs 2θ · 4,801 pts
The plot becomes a pattern, an inference, a finding.
Every claim that follows now rests on this one artefact being exactly what it says it is.
Raw file · scan03.xy · sha256:66ca679f…
Months later — sometimes years — it becomes a figure in a paper.
A claim other people will review, cite, and build on.
The chain, broken
By then the path back is gone.
Which artefact. Which sample. Which conditions. Which of the eleven versions on the shared drive — renamed, moved, re-exported, cleaned by a script nobody kept.
Sometimes not even the person who made it can say.
The chain, recovered
Now multiply that by every result ever published. We are generating results faster than we ever have — and losing their origins just as fast.
This is the substrate of modern science. A few have an intact chain of custody. The rest are unverifiable by anyone, and we are training machines and models on it.
Some hold.
What if every result did? The lineage captured at the moment of origin, before a human hand or a script could intervene.
A verified chain is not a record of what happened. It is proof that nothing else did.
Not certainty that the result is right — science does not offer that. Certainty that the result is untouched.
In an age where data can be generated as easily as it can be measured, that distinction is the whole ballgame.
Three things an unbroken chain makes possible.
Signed at the source
Data and metadata are fingerprinted at the instrument, before a human hand or a script can touch them.
Every version survives
Not the latest file. Every state it passed through, and the path between them.
Fabrication has a signature
Synthetic peaks, enhanced micrographs, generated datasets. They leave traces. We look for them.
Stemmatics — the study of how manuscripts descend from one another. Which copy came from which. Where the errors entered.
We do it for data. Measured or generated, human or model, first result or thousandth.
Everything science builds next will stand on data we can no longer trace. We think that is worth fixing.
early access · waitlist open
Early access opens for a small group of researchers.
If your figures deserve a history, leave your email.