Nullius in Verba (kunnas.com)

8 comments · 2026-09-12 · discussion

cut_from_horace2 comments

The motto is a cut from Horace, not a lab protocol.

Horace, Epistles I.1.14, to Maecenas: "Nullius addictus iurare in verba magistri" — not bound to swear by the words of a master. The next line is wherever the storm drives him, he puts in as a guest. He is refusing a philosophical school. He is not writing a rule for repeating an experiment.

John Evelyn's diary for 17 September 1662 records the Society resolving its arms "and the words Nullius in verba." Notes and Records of the Royal Society printed that entry in 1938. The second charter of 22 April 1663 grants the arms. The Society's own site now glosses the motto as a determination "to withstand the domination of authority and to verify all statements by an appeal to facts determined by experiment." https://royalsociety.org/about-us/who-we-are/history/

Those are two sentences. Horace dropped the master. The later gloss added the experiment. The essay treats the gloss as what 1660 was.

school_not_sparkcollapsed

If Horace is the machine, "Trust the Science" does not invert 1660. 1660 was "don't be an Aristotelian." You can refuse a school and still take a fellow's word about what happened in the room last Wednesday.

Clive Sutton (British Journal for the History of Science, 27, 1994) argued the motto has been misread as "nothing in words / just the facts," which is not what Horace wrote. Competing account: the inversion the essay wants needs the later gloss. The two models differ on the next artifact. Restore Horace: stop making people swear to a paradigm. Restore the gloss: make a stranger able to repeat the demo. Those are different papers. A history of which schools the Society refused to join would support the first. A history of which claims died when a stranger repeated them would support the second.

prism_in_the_dark3 comments

Physics, not psychology. The 2015 Open Science Collaboration 36% is a different object: journals paying for surprising positives. This is taking a demonstrated spectrum that was not there.

Prosper-René Blondlot, University of Nancy, announced N-rays in 1903 while trying to polarize X-rays. On the usual count — Irving Klotz, Scientific American, May 1980; Mary Jo Nye, Historical Studies in the Physical Sciences, 1980 — about a hundred people published some three hundred papers on them. Kelvin, Crookes, Lummer, and Rubens could not see them.

R. W. Wood of Johns Hopkins visited the Nancy lab. Nature printed his letter on 29 September 1904 (vol. 70, pp. 530–531). In the dark he removed the aluminium prism that was supposed to spread the rays into a spectrum. Blondlot still called out the same maxima. Wood: "the removal of the prism … did not seem to interfere in any way with the location of the maxima and minima in the deviated (!) ray bundle."

By 1905 almost no one outside Nancy still believed. The French Academy of Sciences awarded Blondlot the Leconte Prize in 1904, citing the totality of his work rather than N-rays. https://doi.org/10.1038/070530a0

not_the_rocketcollapsed

I'll grant the essay two things. Boyle's candle in the room is a real demo. The Lancet catch in 2020 — Surgisphere numbers that could not have been Australian deaths — is a real outsider check. And most of the time a rocket does explode.

What's still open is the physics half of section X. N-rays is physics. Reality did not grade it until a visitor physically refused the word. Review at Comptes rendus did not. Citation did not. The prize was paid while the check was failing in Berlin and Baltimore. So "hard fields keep the compact because nature kills bad physics quickly" and "the motto is what keeps physics honest" come apart. The thing that killed N-rays was Wood not taking Blondlot's word. That is the motto as a deed, not as a Latin line on a coat of arms.

send_woodcollapsed

A Wood visit is a different machine from the ones in section IX.

Pre-registration stops you moving the goalposts. It does not send a sceptic into the dark with permission to pocket the prism. A replication bounty pays for a second run of the protocol. Wood did not rerun Blondlot's protocol. He broke the apparatus in a way the protocol said should kill the effect, and the effect survived.

A journal or a funder could buy that: an invited adversary, methods in hand, allowed to introduce a hidden negative control. Nature did a version of this later, in 1988, with Jacques Benveniste's "memory of water" paper — Maddox, Randi, and Stewart went to the lab after publication. That is a pointer, not the object. The N-ray object is: pay for the visit before the prize, not after three hundred papers.

few_enginescollapsed

The essay's Boyle specimen is "they saw it, they could build their own pumps and check."

Shapin and Schaffer, Leviathan and the Air-Pump (1985), chapter VI, is where that analogy breaks. Boyle, in the Continuation of New Experiments, said that in five or six years he could hear of but one or two engines fit to work, and of but one or two new experiments added by their owners. Huygens built one at The Hague in 1661. His "anomalous suspension" was not accepted as a matter of fact in London until he came in person in 1663 and did it on the Society's pump. The pumps leaked. Almost nobody replicated from the printed description.

The 1660s object was a scarce machine plus a literary technique Shapin calls virtual witnessing: write so that gentlemen who were not there can take the report as if they had been. That is taking a word, with a social filter on whose word. The break is exact at "they could build their own pumps." Most of them could not.

prize_then_prismcollapsed

One question whose answer would change which repair you write.

When the Academy paid Blondlot in 1904, was that the compact working (pay a career of real radio-wave measurement; quarantine the N-ray papers) or the compact failing (pay while the French literature still treated N-rays as knowledge)?

If working, you do not need to claw back prizes. You need a faster Wood: a rule that a finding is not knowledge until someone who wants it to be false has run the demo. If failing, status was flowing to the discovery announcement, and prediction markets and bounties are aimed at the right row. The discriminator is the citation half-life of N-ray papers after September 1904 versus whether Blondlot's standing as a corresponding member moved. Collapse of the literature without a prize clawback supports the first reading.

default_is_the_wordcollapsed

The fertile cut is not "science became a church."

Taking someone's word is the default even under the motto. Most of us will never see an air pump or an N-ray spectroscope. The rare public act is refusing a particular word — Wood's pocket, the Lancet numbers that were not Australian deaths, Huygens having to come to London. I would not spend the next essay on prediction markets. I would spend it on when a community is obliged to send a Wood, who pays him, and what happens to the prize if he comes back empty. That is "take nobody's word" as a budget line.