---
title: "The hand that beat the machine — until the machine measured rounder: an Avogadro-sphere craft legend that thinned as its sources were chased, over a metrology that stayed Tier 1"
type: "moc"
writer_model: "warden/claude-opus-4.8"
tags: ["metrology","craftsmanship","tactile-cognition","measurement-history","avogadro-constant","verification-method","sourcing-tiers","silicon","si-units"]
date_created: "2026-08-15T00:00:00.000Z"
updated: "2026-08-15T00:00:00.000Z"
provenance: "Warden pass 2026-08-15 (warden/claude-opus-4.8), run per 00-meta/specs/seek-warden-spec.md on a different engine than the notes' writers. Discharges the 2026-08-13 [entity] flag 'the Leistner/Avogadro-sphere metrology cluster crossed the MOC threshold' (six claim-notes, one theme). Grounded in a direct read of all six member notes, not in cosine."
audits: ["2026-08-16 claude-opus-5"]
seek_code_commit: "17d9798"
---


The recurring argument in this cluster is not "Achim Leistner was a great optician" (the
most-mentioned figure) and not "the Avogadro spheres are round" (the most-mentioned object).
It is that **one object carried two claims of very different sourcing weight, and only one of
them survived being chased to primary sources.** The measurement — enriched silicon-28 spheres
built to count atoms and pin the Avogadro constant for the 2019 SI kilogram redefinition — is
Tier 1 from NIST and from the International Avogadro Coordination's own papers, and it holds. The
*legend* stacked on top of it — that a craftsman's unaided fingertips, the "atomic feeling,"
achieved a roundness "found to be superior to any purely machine-based method" — is Tier-3
magazine-and-museum prose, and every step toward a primary either fails to find one or turns up a
document pointing the other way.

Titled for the argument's arc — a superiority claim and its Tier-1 reversal — not for Leistner
or the spheres, per the 2026-07-25 lesson. What makes this a map rather than a list is a
*load-bearing bridge*: the single Tier-1 document that speaks to hand-versus-machine
([[claim-ptb-repolish-out-rounded-leistners-original-avogadro-spheres]]) revises the exact
roundness figures the legend decorates ([[claim-avogadro-silicon-spheres-among-roundest-objects-ever-made]]).
The metrology sub-theme and the craft-legend sub-theme are not two themes circling — they are one
story about the same two spheres, which is what distinguishes this from a held-open two-sub-theme
case like `moc-musical-instruments`.

## The bedrock — a measurement that redefined the kilogram

Tier 1, uncontested, load-bearing for everything else. This is the part that does not thin.

- [[claim-avogadro-silicon-28-spheres-realized-2019-kilogram-redefinition]] — the metrology.
  The 2019 SI kilogram redefinition moved mass off the platinum-iridium artifact and onto a fixed
  Planck constant, realized by two independent routes; the silicon route counted atoms in spheres
  of "99.9995 percent silicon-28," each "about a 94-mm diameter sphere" with "a mass of 1 kg," to
  determine the Avogadro constant. Tier 1 (NIST), re-fetched and cross-corroborated at arXiv.
- [[claim-avogadro-silicon-spheres-among-roundest-objects-ever-made]] — the roundness, and why it
  was load-bearing rather than decorative: to back out the Avogadro constant the sphere's volume
  had to be known to matching precision, so its diameter had to be uniform to nanometer tolerances.
  NIST's own scale analogy: blown up to Earth's size, the surface would deviate only "3–5 meters."
  Tier 1, `verified_verbatim`. **This is the note the reversal below revises.**

## The legend — a craft-superiority story that will not source

The vivid part, and the exposed part. Each note here is honest about its own tier and is held
`seedling` and flagged pending corroboration it has not received.

- [[claim-leistner-atomic-feeling-hand-polished-avogadro-spheres]] — the anecdote. Per the
  Powerhouse Museum's object record, Leistner (Zeiss-Jena-trained, emigrated to Australia 1957)
  did the final figuring by hand, using "his hands to feel for any imperfections," a technique the
  team nicknamed "atomic feeling," which the museum states "proved superior to purely machine-based
  methods." Tier 3, `flagged [unverified-historical]`: a surprising, load-bearing craft-superiority
  claim resting on a single museum caption.
  [Corrected 2026-08-16, cross-model audit — the record was re-fetched this session and three things
  in the sentence above are not in it, which also put this bullet at odds with the map's own
  open-threads section. (a) The Zeiss-Jena training and the 1957 emigration do not appear in the
  object record; their source is elsewhere and unlocated. (b) The nickname "atomic feeling" does not
  appear in the object record either — the map says so four sections down, and should not have
  attributed it here. (c) The museum's actual sentence is "Leistner's fabrication method has been
  found to be superior to any purely machine-based method," not "proved superior to purely
  machine-based methods"; the map's own opening paragraph already quotes it correctly. What the
  record does carry, and what this bullet is entitled to: the hand-figuring, and "his hands to feel
  for any imperfections." The member note carried the same three defects, flagged by the 2026-08-14
  audit as a carried finding and fixed there this session.]
- [[claim-leistner-atomic-feeling-corroborated-beyond-powerhouse-museum]] — the careful half-win,
  and the half-loss. Two independent named journalists (Keats, *Wired* 2011; Powell, *New
  Scientist* 2008) corroborate the hand-polishing *practice* without paraphrasing the museum — so
  the anecdote is not one caption dressed as fact. But a 2026-08-14 cross-model audit found the
  nickname "atomic feeling" appears in *neither* new source and *not even in the Powerhouse record*
  itself; its only traceable home is a Tier-4 Wikipedia article citing a BBC Four documentary no
  vault note has read. The practice corroborates; the nickname does not.
- [[claim-leistner-hand-beats-machines-claim-lacks-tier1-primary]] — the superiority claim, twice
  sourced and still short. Keats (*Wired*) writes that "the most advanced computer imaging has
  confirmed … no machine can match his touch"; the museum makes the same claim. Two Tier-3 venues
  independently asserting superiority establish the claim is *real* (someone at CSIRO said it, more
  than once) — but a specific technical-mechanism claim needs Tier 1–2, and neither venue cites the
  "advanced computer imaging" study. `flagged [unverified-mechanism — needs primary]`.

## The reversal — the one Tier-1 test cuts the other way

The bridge that makes this an argument. The strongest primary found while investigating the "beats
machines" claim measures against it.

- [[claim-ptb-repolish-out-rounded-leistners-original-avogadro-spheres]] — Tier 1, the load-bearing
  reversal. Azuma et al., *Metrologia* 52 (2015), reports that spheres AVO28-S5 and AVO28-S8 — the
  actual spheres Leistner hand-finished — were later etched and *repolished at PTB* in 2013–2014,
  and the peak-to-valley roundness deviation dropped (S5: 98 nm → 69 nm; S8: 90 nm → 38 nm) below
  "the initial spherical state produced by the Australian polishing." A later process made these
  specific spheres *rounder* than Leistner's hand-finish. The note is scrupulous about what it does
  *not* claim: the method paper for the PTB rework was unread, so whether that later process was
  machine-automated or still hand-guided is not established. It claims only the measured outcome —
  which is enough to unmake "hand beat machine, permanently."

## Why the pieces belong together

The legend is a claim *about the metrology* — it asserts that hands achieved the roundness the
Tier-1 notes measure. So the sourcing gap is not incidental: the map's whole point is that the
measurement and the story about how it was achieved sit at different tiers, and the one document
that can adjudicate the story is itself a metrology paper reporting on the same two spheres. Read
the cluster for that gap, not for the romance.

## The method's home, cross-linked not folded

Touch standing in for — or exceeding — instrumented precision is a thread with its own lineage;
kept as a cross-link so the join is visible rather than silently merged into the sphere story. The
member note itself draws the distinction: those cases substitute touch for *absent* sight, while
Leistner's hands reportedly beat *available* machines.

- [[claim-saunderson-palpable-arithmetic-tactile-calculating-device]] and
  [[claim-blind-mathematicians-independently-reinvented-private-notation]] /
  [[moc-blind-mathematicians-working-methods]] — the tactile-cognition thread this episode bridges
  into, at a different constraint.
- [[moc-the-number-that-outran-its-evidence]] and [[moc-too-clean-to-be-real]] — methodological
  cousins: a vivid claim deflated by reading the record rather than re-running an experiment. This
  cluster is a metrology instance of the same shape, cross-linked, not claimed as those maps' spine.

## Open threads (honest caveats, not hidden)

- **The superiority claim has no primary and its one Tier-1 test fails it.** "Found to be superior"
  is double-Tier-3; the only Tier-1 document on hand-vs-machine roundness for these spheres reports
  a later repolish beating the hand-finish. The claim is real but unproven, and the record leans
  against it. Tracked at [[question-verify-leistner-atomic-feeling-superior-to-machines]].
- **The nickname is unsourced at both ends.** "Atomic feeling" is in neither corroborating source
  nor the museum record; its home is a Tier-4 Wikipedia citation of an unread BBC documentary.
- **The reversal's method is unestablished.** Whether PTB's rounder result came from a machine or
  from other opticians' hands is not known — the note refuses to say, and so does this map.
- **Australia's NMIA/CSIRO — where the spheres were polished — is the one collaborator NOT on the
  Tier-1 NIST roster** (it rests on Tier-3 reporting), which is precisely the corner of the story
  the craft legend occupies.
- **Every member note is still `seedling`.** A live cluster; this map records its current sourcing
  footing, not a frozen verdict.

> [!note] Warden's commentary:
> The tell that this is one argument and not "everything about the Avogadro spheres" is that the
> two halves are load-bearing on each other: the legend is a claim about the roundness, and the
> only Tier-1 document that can test the legend is the one that revises the roundness. I was
> careful not to overreach in either direction. The metrology is genuinely Tier 1 and I said so
> plainly — this is not a "gotcha" map that pretends the whole story is soft. And the reversal is
> genuinely narrow — one metric, two spheres, method-of-the-rounder-result unknown — so I let it
> unmake "permanently superior" without letting it swing to "the machine always wins," which the
> document does not support either. What the cluster actually teaches is the gap between a
> well-sourced measurement and the under-sourced romance a good story pins to it: the number that
> redefined the kilogram is bedrock; the fingertips that supposedly beat the machine are a museum
> caption the primary record talks back to. — warden/claude-opus-4.8, 2026-08-15