Stimpunks Foundation × More Realms · Working Paper

Edition of 4 September 2026

The second edition of Glimmer Wire, the scan that sits behind Glimmers, and the thin week. It was compiled in a session where every source was refused at the network layer before the request reached the site, published saying so — and then amended hours later when the block lifted: one item verified, and corrections in all six, of which the two sharpest turned out to be ours rather than the coverage’s. Items are sorted by lag, and the section that matters most is Held, which lists what we declined and why.


6 filed · 4 held · 3 leads · 1 verified · median lag 40.5 d

What was actually read: nothing, then most of it

Not one source in this edition was opened. Every publisher, preprint server, aggregator, university newsroom, press wire and DOI resolver this scan reached for was refused at the network layer — a 403 on the proxy tunnel, before any request arrived at the site. Refused when tried, by name rather than by assumption: science.org, sciencedaily.com, phys.org, eurekalert.org, sciencenews.org, arxiv.org, api.crossref.org and doi.org. The publishers behind four of the six items below — Nature Materials, PNAS, Cell, Annual Reviews — were not individually attempted after the DOI resolver itself came back refused, which is a limit of the machine this scan ran on rather than a proven fact about those hosts, and is said here rather than rounded up into a longer list. The sourcing therefore stopped one step further out than the 2 September edition's, which at least reached institutional press releases. Everything below rests on search-result summaries, which this project's own rules class as leads and never as citations.

That made this the weakest edition on the page, and the tally said so: no item was graded verified. It is published rather than skipped because the archive is meant to record the cadence including the thin weeks, and because a scan that states its evidentiary floor is more use than a gap where a week should be. The consequence is strict, though. Nothing here may be promoted to a zine, a Glimmers entry or a card without the primary source first, a changelog entry and its row in the fact-check ledger. Every figure below was marked as summary-sourced; several were withheld outright; and the two dates the lag axis leans on were flagged where they were unconfirmed. A number that could only be had from a summary is named as such rather than laundered into a citation line.

Amended the same day, and most of the edition moved. The network restriction described above was widened hours after this edition was compiled, and every source was tried again. Crossref, PubMed, Europe PMC, arXiv, PubMed Central and the EPA all answered. Four publishers still refused, and the distinction matters: getting past the network was not getting past a paywall. Nature Materials served its abstract and priced the full text at $39.95; PNAS, Cell and Annual Reviews returned 403 at the article page itself. So one item is now verified — the bats paper turned out to be open access in PubMed Central and was read in full — and the other five keep the grades they had, now standing on the papers' own abstracts instead of on search summaries.

What the primaries mostly bought was not upgrades. It was corrections, in every one of the six entries, and the two sharpest are ours rather than the coverage's: we misread which noun the Science paper's title makes its subject, and we described the Hall result using the wrong Hall effect. Three more were errors of attribution — a single-author review called a group, three dates that turn out to describe cooling in sediment cores rather than encounters with clouds, and a species count that is nowhere in the paper it was credited to. Each entry below now names what it got wrong. The rest of the edition's standing does not change: five of six are still not read at source, and nothing here except the bats item may be promoted to a zine, a Glimmers entry or a card without the primary first.

Filed

Sorted by lag. Not one of these six carries a date inside the last seven days — which is a correction to what this section said when it was filed. The newest item looked one day old against its printed issue date; its online-first date, obtained afterwards, makes it nine. The median gap between publication and the wire that carried it here is 40.5 days either way, and the oldest item is 146 days old — an April paper arriving on a September feed. Last week's distribution said the same thing, and this is the second data point rather than a fresh observation. Two of the six were public far earlier than any date below, as preprints: the bats result from August 2024 and the Hall result from May 2025. The lag axis measures journal publication, so it cannot see either, and both entries say so.

Verified Primary source read and confirmed.
Plausible Finding sound and metadata confirmed; paper text not read.
Contested Specialists disagree, or the claim outruns the method.
Held Could not be traced to a source that supports it.

Plausible Published online 2026-08-24 Journal PNAS Read the abstract, at Europe PMC

Lag 9 d

The oldest way we can show an animal made a living, and it was a partnership

Conotubus was a small tube-dwelling animal in the Ediacaran seas of what is now China, about 550 million years ago. Extremely low molybdenum isotope values — δ98Mo, in pyritised fossils — are the evidence, and they are otherwise known only from modern cold-seep tubeworms that host sulfur-oxidising bacteria. Sulfur isotopes do a different job in the paper: they establish that rapid pyritisation preserved the original signal, and that microbial hydrogen-sulfide oxidation was going on in the animal's habitat. The authors offer it as the earliest geochemically supported case of chemosymbiosis in the fossil record, and say the strategy likely enabled Conotubus to thrive in a stratified ocean where sulfide and oxidants sat side by side.

The star stuff frame A hydrothermal-vent tubeworm doing this reads as a specialisation — an animal that adapted its way into an impossible place. Push the same arrangement back to the first chapter of animal life and it stops looking like an adaptation and starts looking like one of the ways animals began. The partnership is not a late accommodation bolted onto an independent creature; on this reading it is load-bearing from the start. Nothing about needing another organism to survive puts an animal in deficit. It is how the oldest one we can read managed at all.
What is inference and what is ours Isotope ratios in a tube this old are a proxy, not an observation of a partnership — and the paper's own hedging, which we could not read when this was filed, is now legible and is stronger than ours was: Conotubus may have engaged in chemosymbiosis. Three corrections the abstract forces. The proxy is molybdenum; we had sulfur and molybdenum doing the same work. The bacteria drawing down a lethal sulfide load is not in the paper — that was a summary's addition, and we repeated it. And the modern comparison is cold-seep tubeworms, not the hydrothermal-vent animal our own frame reached for. The 550-million-year figure is still not the paper's; it says terminal Ediacaran, and the number stays approximate. The frame below is ours — the authors are reporting a trophic strategy, not making a claim about deficit.
Where it goes Checked against the shelf rather than assumed: chemosymbiosis is not unspent here. A Field Guide to Life in Impossible Places already carries Riftia, the vent tubeworm, living on “oxygen and sulfide at once.” What is unspent is the age, and that the evidence is isotopic rather than anatomical. So: a companion entry to that card, or the evidence spread of a piece — not a new argument. No. 27, Every Nervous System, is the only other page here that says Ediacaran.

Wang, Z., Peng, Y., Tang, Q., Schiffbauer, J. D., Xiao, S., Fike, D. A., Jia, Z., Feng, D., Crockford, P. W., Cai, Y., Yuan, X. & Pratt, L. M. “Chemosymbiotic trophic strategy in an Ediacaran tubular animal.” PNAS 123(35), e2526201123, published online 24 August 2026, issue of 1 September 2026. DOI 10.1073/pnas.2526201123 · PMID 42636363. Amended 2026-09-04: the full author list is now confirmed at Crossref and Europe PMC, and the online-first date — 24 August 2026 — resolves the lag from “1 d, a floor” to a measured 9 d against the 2 September wire that carried it. Three claims corrected in the caveat above. Read: the paper's abstract, at Europe PMC. Not read: the paper. PNAS refused the article page with a 403 and Europe PMC lists it as subscription-only, with no PMC copy. Still not the paper's own: the 550-Ma figure.

Contested Published 2026-08-21 Journal Annu. Rev. Astron. Astrophys. Read the abstract, at Crossref

Lag 11 d

Where the solar system happened to be standing may be part of Earth's climate record

A review of the heliosphere — the bubble of solar wind the whole solar system sits inside, reaching about 120 astronomical units in the direction of travel — over the Sun's last ten million years. The Sun moves at 19 parsecs per million years and has crossed many different structures in the interstellar medium, at times, the review says, collapsing the bubble to sub-astronomical-unit scales: smaller than Earth's orbit, leaving the planet briefly outside its own star's shield. Encounters like that could be as often as every couple of million years, which would make them a major external disturbance to the development of life here. The review then sets that beside something measured independently: stepwise cooling intervals in deep-sea sediment cores at 13–14, 6–7 and 2–3 million years ago, whose causes are openly under debate, and suggests such cooling might be triggered by heliosphere collapse.

The star stuff frame Umwelt at galactic scale. The atoms in a body were forged in dying stars; on this account the conditions were partly set by whichever piece of the galaxy the solar system was passing through at the time. An organism's environment does not stop at its skin — the whole site keeps arriving at that — and it turns out a planet's may not stop at its orbit either. Nothing here was ever self-contained, at any scale we have looked.
Why contested, and three things we had wrong This is still a review article built on a modelled trajectory, not a measurement, and the aggregator headline — Earth lost the Sun's protective shield — is far more definite than the review, which says such cooling might be triggered by collapse and that the driving mechanisms are “subject to ongoing debate.” The grade holds. But reading the abstract corrected us three times. The three dates are not cloud encounters. We printed them as when the Sun crossed cold clouds; they are cooling intervals read out of deep-sea sediment cores, which is the independent half of the argument — and getting that backwards turned a proposed correlation into a reported observation. It is not a group. The review has a single author. And the superflare and nitrous-oxide strand is not in it — nothing in the abstract mentions Kepler, superflares or the early atmosphere, and the review's stated span is the last ten million years, so we cannot attach that strand to this paper at all. It was very likely a second piece of work the summaries folded into this one. On the figure itself: nitrous oxide is 273 times as warming as carbon dioxide over a hundred years, per IPCC AR6 via the US EPA — not 300, and the missing part was never the digit but the timescale, without which a warming potency is not a number.
Where it goes Unspent: no page here says heliosphere, checked by grep. (Interstellar does appear on five pages, The Cloud Phase and Bodies of Water among them; we have not re-read them, so treat the overlap as unchecked rather than absent.) It is a lead, not a piece. A zine built on this would have to carry the modelling caveat in the argument itself rather than in the colophon, and that is a harder piece to write than it looks.

Opher, M. “Understanding the Heliospheric Shield: Laying the Groundwork to Predict Habitable Astrospheres.” Annual Review of Astronomy and Astrophysics 64, 139–177, 2026. DOI 10.1146/annurev-astro-120425-053711. Amended 2026-09-04: title, author, volume, pages and DOI all now obtained from Crossref, where the publisher has deposited the structured abstract. The 21 August 2026 date is the whole of volume 64 — every article in it carries that date — so it is a volume date rather than this paper's own online-first date, and the 11 d lag inherits that looseness. Three corrections in the caveat above, one of them an inversion of what the three date windows describe. Merav Opher directs NASA's SHIELD DRIVE Science Center; the earlier work this entry could not identify is almost certainly Opher et al., arXiv:2502.16360, on the solar system's passage through the Radcliffe wave in the middle Miocene. Read: the review's abstract, at Crossref. Not read: the review. Annual Reviews refused the article page with a 403; there is no preprint of the review itself. Unattributable: the superflare/nitrous-oxide strand, which is no longer credited to this paper.

Verified Published 2026-07-29 Journal Science Advances Read the paper, in full

Lag 34 d

Twenty-six bat species build antibodies from two gene sets, where every other mammal has one

An antibody's heavy chain is assembled from one immunoglobulin heavy-chain locus. Every mammal anyone had looked at has exactly one, humans included, plus up to two light-chain loci. Working through the largest bat family, Vespertilionidae, this group found two heavy-chain loci — on separate chromosomes — in 26 bat species. In one of them, the big brown bat Eptesicus fuscus, single-cell transcriptomes confirm that both loci are functional: both rearrange, both are expressed, and they generate antibody diversity by different mechanisms. The authors call it extreme variation in immunogenetic architecture across the order Chiroptera, and offer it as a foundation for studying humoral immunity in bats.

The star stuff frame This is No. 33 run backwards. The Ick Is a Border is about anglerfish that lost functional antibody-maturation genes so that two bodies could fuse — adaptive immunity given away to permit contact. Here twenty-six mammals have doubled the same stretch of genome, onto a second chromosome, and in the one species checked closely both copies work — by different means. One lineage surrendered the machinery; these kept two of it. Same locus, opposite directions, and neither is the standard, which is the whole of it: the phrase the authors use is extreme variation, and they use it about architecture rather than about tuning. That is the difference-first claim arriving in a place nobody was looking for it, in a peer-reviewed immunology paper that is not arguing about people at all.
The verb is doing work — and two things we had wrong Tolerate is not resist, and the gap matters: a bat is not better at winning a fight with a virus, it is differently arranged. This is precisely the shape of finding that gets rewritten into a superpower on the way out the door, at which point it stops being a claim about architecture and becomes a claim about merit. Reading the paper made that caution sharper, because the tolerance story is not the paper's. Bats as virus reservoirs is the paper's motivation, in its first sentence; its conclusion is that these results “provide a foundation for analysis of humoral immunity and pathogen response in bats.” Nowhere does it offer the duplication as the reason bats carry viruses without getting seriously ill. We filed that as the group's own claim. It is the coverage's. And the species count was wrong. We printed vesper bats as “the largest bat family at more than 500 species,” implying the duplication is a family-wide trait. The paper calls Vespertilionidae the largest family but gives it no number; the only count it prints is “over 1,500 known species of bats,” which is the whole order. The measured figures are 26 species with dual loci, one confirmed functional. Neither 500 nor a family-wide claim survives.
Where it goes A card or a field-guide entry in the same family as No. 33 rather than a zine of its own, because the pairing is the finding and one half of it is already published here. Nothing on this site says immunoglobulin or vesper; No. 33 owns the loss direction and would need a link back. This is now the one item in this edition that clears the bar to be built — the paper is read, open access, and citable. Any card must carry the twenty-six, not five hundred.

Pursell, T., Reers, A., Mikelov, A., Kotagiri, P., Lam, B., Ellison, J. A., Boyd, S. D. & Frank, H. K. “Immunoglobulin heavy chain locus duplication in bats.” Science Advances 12(31), eaeb6714, published online 29 July 2026, issue of 31 July 2026. DOI 10.1126/sciadv.aeb6714 · PMID 42525754 · PMCID PMC13418733. Stanford, Tulane, Monash and the CDC. Upgraded to verified 2026-09-04, when the paper was found to be open access in PubMed Central and read in full. The author list we filed from summaries is exactly right, all eight in order — and the date is right too. Two corrections, both in the caveat above: the “more than 500 species” figure is in no part of this paper, and the virus-tolerance explanation is the coverage's claim rather than the authors'. One institution was missing: Prasanti Kotagiri is at Monash, which we had not listed. And the lag axis cannot see this item honestly: the finding was public as a bioRxiv preprint on 9 August 2024 — the paper says so itself — which is 753 days before the wire carried it, against the 34 we print from journal publication. Read: the paper, in full, at PubMed Central. Not read: the Tulane release.

Plausible Published 2026-07-16 Journal Science Read the abstract, at PubMed

Lag 47 d

Organ ageing traced to one cell type losing the knack of clearing up after another

Tissue-resident macrophages are the immune cells that live permanently inside an organ and clean up after it. This group identifies them as central coordinators of age-related organ decline, through failing to clear away senescent neutrophils — a failure regulated by EP2, a receptor for prostaglandin E2. Reducing that one signal in aged mice preserved youthful mitochondrial fitness and prevented cognitive decline, frailty, sarcopenia, adiposity, cardiac impairment and systemic inflammation. Plasma proteomics points at the liver as a major source of the age-associated immune change. And elevated EP2 and senescent neutrophils were also seen in aged and diseased human tissues — observed, not intervened on. The abstract's closing words: impaired macrophage clearance is “a reversible driver of organ decline in aging.”

The star stuff frame The failure is not in either cell. It is in the relationship between them — one cell type's job is to take care of what another leaves behind, and what breaks is the taking care. That is a shape this site keeps arriving at from the other direction: a deficit that reads as located in a body is often located in a relation, and the repair is to the relation rather than a correction to the part. Stated in plain words on purpose, and not as a model — the argument for keeping it that way is already a page here.
Mice — and we misread the title These are mice, and the distance from a mouse organ to a human life is still the whole argument. But the verb complaint was ours and it was wrong. We wrote that “the paper's own title claims the impairment limits organ ageing,” and used that to say the coverage's reverses ageing had overreached. Read the title again: Restored clearance of senescent neutrophils by tissue-resident macrophages limits organ aging.” The subject is the restored clearance, not the impairment — the title says restoring the clearance limits ageing, which is much closer to the coverage than we allowed, and the abstract closes on reversible driver. We had the grammar backwards and built a correction on it. The honest caution is the ordinary one: mice for the intervention, humans only for the observation. The withheld author list is resolved — the ordering that looked garbled was correct, and all thirteen are printed below. Ages and durations are still not confirmed: neither the 22-month-old mice nor the two-month course appears in the abstract, and the full text is paywalled, so both figures are struck from the finding above rather than carried on a summary. This remains the item most exposed to the cure frame, which the Not block refuses outright.
Where it goes Not a Glimmers entry — a mouse result is not an occasion. At most the evidence spread of a piece about deficits that turn out to live in relations, and that piece would need the human end of the claim to be honest about how far away it is. Nothing here says macrophage, neutrophil or efferocytosis.

Tan, Y. J., Conley, T. E., Yao, F., García-Marqués, F. J., Akinyemi, D. E., Dinh, V. V., Wang, Q., Bermudez, A., Kim, J., Belk, J. A., Soehnlein, O., Pitteri, S. J. & Andreasson, K. I. “Restored clearance of senescent neutrophils by tissue-resident macrophages limits organ aging.” Science 393(6808), eaea3075, 16 July 2026. DOI 10.1126/science.aea3075 · PMID 42462036. Stanford, with the University of Münster and the Chan Zuckerberg Biohub — not Stanford alone, as we had it. Amended 2026-09-04: the full author list is no longer withheld; the ordering we distrusted was right, and it is confirmed at both Crossref and PubMed. Our reading of the title was wrong and is corrected in the caveat above — this is the sharpest thing the primaries found, because the error was not in the coverage we were checking but in the check. A companion Perspective ran in the same issue: Science 393(6808), 244–245, DOI 10.1126/science.aei9816, also unread. Read: the paper's abstract, at PubMed. Not read: the paper, which Science paywalls, or the Stanford release. Still summary-only, so struck from the entry: the mouse ages and the treatment duration.

Contested Published 2026-05-28 Journal Nature Materials Read the abstract; the authors’ preprint in full

Lag 96 d

A Hall response that follows the magnetisation sideways, in a material built so it can

The anomalous Hall effect is the magnetic cousin of the textbook one: the sideways voltage comes from the material's own magnetisation rather than from an applied field, and symmetry normally requires it to track only the magnetisation's out-of-plane component. This group stacked a low-symmetry topological semimetal (TaIrTe4) on a ferromagnetic insulator (Cr2Ge2Te6) to make a heterostructure with just one mirror plane left. Give the magnetisation a component inside that plane and the last symmetry breaks — and the Hall response then follows both the in-plane and the out-of-plane components. It is gate-tunable across several devices, and the authors put it forward as “a pathway for engineering tunable, symmetry-driven Hall effects,” which is a considerably quieter sentence than the one the coverage ran.

The star stuff frame Take the framing carefully and the interesting shape is not novelty at all: a response may have been there the whole time, hidden by the one geometry everybody measured in. That is the recurring How We Got Here move — a rule that turns out to describe the instrument rather than the world, and a variation that was invisible because of where the apparatus was pointed. It is a good shape and this is not yet a good enough source for it.
Why it stays contested — the guess was right, and one of the errors was ours The entry said the thing to check first was whether prior parallel-field work made “century-old assumption overturned” a press-release artefact. It does. The in-plane anomalous Hall effect has a literature of its own reaching back to at least August 2022, with observations published in 2024 — one attributed to orbital magnetisation, one to an octupole in magnetisation space — and a run of further papers through 2025 and 2026, including a unified symmetry framework for the effect posted six weeks ago. The assumption was under active revision years before this result. So the honest story is the narrow, real one: a well-made measurement inside a live field, and a release sentence that does not appear anywhere in the paper. The larger error is ours. Our own opening described the ordinary Hall effect — current, perpendicular applied field, voltage across the sides, Edwin Hall in 1879, hence “century-old” — and then reported this paper as contradicting it. It does not. This is the anomalous Hall effect, which depends on magnetisation and on symmetry, not on an applied field, and nothing here touches the 1879 result or the sensors built on it. We repeated the release's conflation in the act of complaining about it. Two smaller things: the response follows both magnetisation components, not the in-plane one instead of the out-of-plane one; and the author list runs to 21 names across multiple institutions, so “a Carnegie Mellon group” undersells it — we have not verified the individual affiliations at source, so none is named here. What is now settled: the date. Crossref and the publisher's own page both give 28 May 2026, so the 96 days is measured, and the “provisional” comes off.
Where it goes Nowhere yet, but the reason has changed. It is no longer “the literature history is unread” — it is read well enough to say that the shape we liked is not in this paper. A rule that turns out to describe the instrument rather than the world is a real How We Got Here move, and a symmetry argument about what an engineered interface permits is a different and less quotable one. If a chain link ever comes out of this, the joint would be marked contested and it would be about how a press office turns a symmetry result into an overturning — not about the physics, which is fine.

Kao, I.-H., Bandapelli, R. K., Cui, Z., et al. (21 authors) “In-plane anomalous Hall effect in a low-dimensional system.” Nature Materials, published 28 May 2026. DOI 10.1038/s41563-026-02611-9. Amended 2026-09-04. Date confirmed at Crossref and on the publisher's page, so the lag is measured, not provisional. The contested question is answered and the caveat above carries it, together with a conflation of our own. Read: the published abstract, at the publisher; and the authors' own preprint in fullarXiv:2505.06829, “Magnetization Dependent In-plane Anomalous Hall Effect in a Low-dimensional System,” posted 11 May 2025, 20 of the 21 authors, same experiment. Not read: the version of record, which Nature Materials prices at $39.95; the institutional release; the coverage. Not graded verified for that reason — a preprint read in full is the authors' text but not the published paper, and this page does not blur the two. The prior literature, named rather than gestured at: arXiv 2208.14251 (2022, PT-symmetric antiferromagnets), 2402.15741 and 2405.16722 (2024, both reporting observations), 2601.05462, 2606.10063 and 2606.14272 (2026), and 2607.29452 (July 2026, a unified symmetry framework). And the lag axis undercounts here too: public as a preprint 478 days before the wire, against the 96 printed.

Plausible Published 2026-04-09 Journal Current Biology Read the abstract, at PubMed

Lag 146 d

The merger at the root of complex cells, watched instead of reconstructed

Asgard archaea are the closest known relatives of the archaeal lineage that gave rise to complex cells. One model of how those cells began has sulfate-reducing bacteria and hydrogen-producing archaea entering a partnership in ancient microbial mats. Working in the living mats of Shark Bay, Western Australia, this group obtained a highly enriched — 89% — culture of a new Asgard archaeon, Nerearchaeum marumarumayae, alongside a new bacterium, Stromatodesulfovibrio nilemahensis. Electron cryotomography shows the archaeon budding chains of envelope vesicles held on extracellular fibres, and tube- and cage-like structures inside the cell. And it shows the two species directly interacting, through tubular fibres assembled by the bacterium. The genomes say what could pass along them: the archaeon can make hydrogen, acetate, formate and sulfite; the bacterium makes amino acids and vitamins. The authors say these features “may reflect an early step” in the symbiotic evolution of complex cells.

The star stuff frame No. 7 and No. 84 already argue that you are a merger, and that the merger was a long negotiation rather than one swallow. What is new is the verb. The origin story's usual verb is engulfed; here are two intact cells with tubes between them, both still themselves. Partnership without absorption is a different biological picture and a different political one, and it is the version that does not require one party to disappear. Note what is seen and what is not, though, because the frame leans on the difference: the connection is imaged. The trading is read off the two genomes, and the honest verb for it is could.
One bay, one imaging result — and two over-reads of ours A model for a hypothesis is not evidence of an event that happened once, something under two billion years back — No. 84 carries the dating and its hedges, and this entry borrows neither. The species name is now confirmed and printed: Nerearchaeum marumarumayae, with the bacterium Stromatodesulfovibrio nilemahensis. Two things we filed are corrected. We said the group reported isolating the archaeon; the paper reports a highly enriched culture at 89%, which is not an isolate, and the gap between those two words is a real one in microbiology. And we said the cells were imaged “apparently exchanging material.” They were not. What is imaged is the connection — and the fibres making it are assembled by the bacterium, not jointly. The exchange is inferred from complementary gene content: the paper says these products “could be exchanged in a syntrophic partnership.” We had a genomic inference standing where an observation should be, which is exactly the substitution this page exists to catch. The lag is still the largest in this edition: 146 days, an April finding reaching a September feed.
Where it goes The origin argument is spent — No. 7 holds it and No. 84 holds the “long negotiation” reading, citing the Asgard literature by name. Unspent: partnership without engulfment as the picture, and the fact that it was imaged rather than inferred. A coincidence worth noting and not a pairing: Shark Bay is already on this site, in No. 36, for dolphin sponging in the water above the same mats.

Nobs, S.-J., Johnson, M. D., Williams, T. J., Meltzer, J., Vázquez-Campos, X., MacLeod, F. I., Rowell, K., Pitt, M., Paul, B., Shepherd, D. C., Michie, K. A., Duggin, I. G., Ghosal, D. & Burns, B. P. “An Asgard archaeon from a modern analog of ancient microbial mats.” Current Biology 36(8), 2090–2103.e7, published online 9 April 2026, issue of 20 April 2026. DOI 10.1016/j.cub.2026.03.041 · PMID 41962538. UNSW Sydney with UTS and the University of Melbourne — all three confirmed. Amended 2026-09-04: DOI, full author list, volume, pages and title all now obtained, and the species names confirmed and printed. The 9 April 2026 online date we filed from a summary is right. Two corrections in the caveat above, both ours. Read: the paper's abstract, at PubMed. Not read: the paper — Cell's article page returned a 403 and there is no PMC copy — nor the UNSW release. Surfaced on a science feed on 2 September 2026, 146 days after publication.

Held

Four this week, and the first of them is the item most on-frame for this whole site — which is the reason the section exists. In an edition where nothing could be read at source, the declines are the part of the record we can actually stand behind: they rest on what the sourcing does and does not say, and that much was legible.

Held — superpower framing, and the source is a congress presentation

Attention patterns in ADHD as a creative advantage

The most on-frame item of the week and the one we are least willing to file. It goes straight at the paradigm: the attention we get measured as deficient for having, doing something the deficit account cannot explain. Three problems, any one of which would hold it. The sourcing is not a study. What we could reach traces to a congress presentation and a cross-field synthesis, and a conference abstract is not a paper. The summaries conflate at least two pieces of work under one headline — a congress item from 2025 and a separate 2026 design-task study — and a digest that cannot tell you which finding it is describing has not traced it. And the framing on offer is the trap. Creative activity as a therapy that “channels defocused attention” while strengthening focus and self-control sits one inch from regulate yourself out of your material conditions, which this collection refuses. Note too that the design-task result, as summarised, cuts both ways: more novel ideas, fewer high-quality ones, with participants good at picking their own best. That is a more interesting finding than the headline and it needs the primary. Here, the superpower frame is not a risk downstream — it is the headline.

Could not verify — the sources disagree on the journal

Each extra hour of prolonged sitting linked to 9% higher cancer death risk

A UK Biobank analysis, reported as 91,292 participants wearing accelerometers for seven days and followed for a median of 12.38 years, with each hour of uninterrupted sitting associated with 9% higher cancer mortality, and an hour of light activity substituted for it associated with 12% lower risk. Held for two reasons. First, the sourcing does not agree with itself: the summaries we could reach name PLOS One and PLOS Medicine for the same paper. A digest that cannot say which journal published a result has not traced that result, and the rest of its numbers inherit the doubt. Second, and this is the editorial reason: an observational hazard ratio about how a person's day is broken up quietly converts a question about who is permitted to stand up at work into personal advice. The finding may be perfectly sound. The framing does the opposite of what this collection is for.

Held — cure framing, and a mouse lifespan claim

A single dose of engineered immune cells protecting aged mice for a year

Surfaced alongside the macrophage item above: a report in Advanced Science that CAR-T cells directed against uPAR, a cell-surface marker, can selectively kill senescent cells, improving metabolic function and physical performance in aged mice, with one dose said to protect for up to a year — all of that from a search summary. Declined for the reason the Not block gives — this page is not a cure list, and a durable-single-dose claim in mice is the exact register where enthusiasm outruns the evidence fastest. It also could not be opened. Filed here rather than dropped because the filed macrophage item is the same field seen from a better angle, and a reader is owed the neighbouring claim we chose not to repeat.

Held again — undated round-up sourcing

Octopus, whale and corvid cognition

The same fault as last edition's cow with a brush, and it is held deliberately a second time so the pattern is on the record twice. Looking for this week's More Than Human material returned undated listicles, review articles, and an analysis of humpback song from 2025 — nothing from the last seven days with a paper behind it. There is real work in this area and some of it is genuinely on-frame for ethodiversity. But “surprisingly intelligent animals” round-ups are seedbed material, not wire material, and this is the category where our sourcing standard comes under the most pressure — because the items are the most fun.

Seedbed

Thin on purpose. In a week where nothing was read at source, almost nothing has earned an hour's build — so there is one seed here, and it is the only claim in this edition drawn from something we did read: our own pages.

Field guide entry · companion to No. 33

There is no standard immune system

Fifteen guides in the collection arrive independently at there is no standard star, nest, egg, shark, rainbow, bird, way to rest — re-derived today, because the figure in our own working notes says twelve and the guide to untranslatable words still says thirteen in its own refusals. The grep now returns sixteen guides; three of them are newer than either count, and were read here to confirm they arrive at the formula rather than merely mention it. The sixteenth, the unfinished-animals guide, matches while explicitly declining the formula, which is why fifteen and not sixteen. That is the standing lesson about counts, arriving on schedule. This would be the same formula pointed at immunity, and the two halves are already in hand: No. 33 holds anglerfish that gave up functional antibody-maturation genes so two bodies could fuse, and the bats item above holds twenty-six species carrying the same locus twice, on separate chromosomes. Surrender at one end, duplication at the other, and the textbook arrangement is neither. Amended 2026-09-04: the reason given here for holding it at seed was that one half was only a search summary. That reason is gone — the Science Advances paper has been read in full and the bats entry is graded verified. What still holds it back is the second reason, and it was always the better one: it needs the rest of the vertebrate range, because a guide built on two poles is an argument wearing a catalogue's clothes. Note also that the guide would have to say twenty-six species, not the five hundred we first printed.

Derived from the-ick-is-a-border-zine.html, read directly in the repository, plus the unread bats item above. Not obtained: anything at source for the bats half.

Leads · unchecked

Three more off this week's wires

A three-dimensional map of a mouse ovary across a reproductive lifespan, reported to show that variation in egg number is already present before puberty and so originates in embryonic development — the no standard clock shape, and the strongest of the three. A Basel group reported to challenge the decades-old view that coronin proteins do the cytoskeletal job they have been credited with, having failed to find the actin binding across several model systems. And a proposed law for early embryonic developmental timing across species, framing embryogenesis as paced by resource consumption rather than by particular biochemical kinetics.

All three off press-release and preprint feeds, 31 August – 3 September 2026, and none of them opened. Journal, authors, dates and DOIs not obtained for any of the three, and not chased in the 4 September amendment either — that pass was spent on the six filed items, and a lead nobody has opened is still a lead. Listed so that next week starts from somewhere.

Not

Not a citation. This is still the refusal that governs the edition, with one item now excepted. Corrected 2026-09-04: this paragraph said nothing here was read at source, which was true when it was written and is not now — the bats paper has been read in full and four more entries stand on their papers' own abstracts. The refusal survives the correction, because five of six are still not read at source and an abstract is not a paper. Anything lifted out of those five and printed elsewhere without the primary would be laundering a summary into a fact — the precise failure this page exists to make visible.

Not a cure list. The macrophage item is filed for what it says about a relationship between two cell types, not because it promises anybody a longer life; the engineered-immune-cell item next to it is held for going the other way.

Not the superpower frame. Twenty-six bat species with two antibody loci — corrected from “a family,” which the paper does not claim — is a fact about architecture, not about merit. And the paper does not offer it as the reason bats survive their viruses; the coverage does. The ADHD item is held because in that story the superpower frame is not a risk downstream — it is the headline.

Not “science says you are valid.” A finding that makes disbelieved people credible also makes them revocable. The belonging claim was never waiting on a result, and it does not move when one fails to replicate.

Not anti-science. The Hall-effect complaint is about a press release, not about a measurement. If a draft off this digest starts reading as a case against physics, or against evidence, it has gone wrong.

Not because of. An animal 550 million years dead living on the work of bacteria is a rhyme, not a proof. The universe declining to pathologize its own variation is why the phrase carries — down to your star stuff measures how far the loving reaches, never why it is owed.