Jump to content

Control Studies Project: Difference between revisions

From POSIWiki
Content deleted Content added
Test the shape half of the morphological claim against the deposited originals; record the membrane measurements and the live refutation condition
Flag 3: name the comparison sources (Zhu 2020, CDC Pink Book) in place of uncited ranges; note the measles LLM-screenshot reference; add preparation and range caveats
 
Line 62: Line 62:
The preliminary electron microscopy study claims particles "of exactly the same size, shape and protein inclusions" as reference images of three agents. None of the three reference images carries a scale bar; the author's own crops do. Two of the three references are photographs of a screen — one of the Centers for Disease Control image library page, one of a '''news article''', the latter described in the post as "the CDC image".
The preliminary electron microscopy study claims particles "of exactly the same size, shape and protein inclusions" as reference images of three agents. None of the three reference images carries a scale bar; the author's own crops do. Two of the three references are photographs of a screen — one of the Centers for Disease Control image library page, one of a '''news article''', the latter described in the post as "the CDC image".


In the Vero study the measles comparison is sourced lower still. Its "MAINSTREAM CONSENSUS" block opens with a screenshot of a conversational language model answering the prompt "What is the Size and Morphology and overall appearance of the measles virus. Short Answer", and the size and morphology it returns are the standard against which the control-culture images are then declared identical. A model's summary is not a measurement, carries no method and no provenance, and cannot be checked by a reader.
⚫
'''This assessment measured the particles.''' Against the author's own 500 nm bars, the particle labelled SARS-CoV-2 is approximately '''119 nm''' and the particle labelled measles approximately '''185 × 195 nm'''; in the Vero study, structures labelled measles measure approximately '''210 nm''' and '''295 × 350 nm'''. Published ranges are 60–140 nm and roughly 100–300 nm respectively. '''The size claims are therefore supportable.''' What is absent is the measurement, and a reference with a scale against which a reader could perform it.

⚫
'''This assessment measured the particles.''' Against the author's own 500 nm bars, the particle labelled SARS-CoV-2 is approximately '''119 nm''' and the particle labelled measles approximately '''185 × 195 nm'''; in the Vero study, structures labelled measles measure approximately '''210 nm''' and '''295 × 350 nm'''.

The figures these are compared against are named here, because the corpus does not name them. For SARS-CoV-2, Zhu ''et al.'' report that "Diameter varied from about 60 to 140 nm", measured on '''negatively stained''' particles ([https://doi.org/10.1056/NEJMoa2001017 doi:10.1056/NEJMoa2001017]; the sentence was read in the [https://www.ebi.ac.uk/europepmc/webservices/rest/PMC7092803/fullTextXML Europe PMC full text], which unlike the publisher page serves to an automated request). For measles there is no single published figure. The CDC's ''Epidemiology and Prevention of Vaccine-Preventable Diseases'', chapter 13, states "It is 120 to 250 nm in diameter" ([https://www.cdc.gov/pinkbook/hcp/table-of-contents/chapter-13-measles.html cdc.gov], updated April 2024; archived [http://web.archive.org/web/20261020061451/https://www.cdc.gov/pinkbook/hcp/table-of-contents/chapter-13-measles.html 20261020061451]); image-library captions for thin sections give 100 to 200 nm; and general reference works give 100 to 300 nm. The Zhu article and the CDC chapter both refuse automated retrieval at the publisher and open normally in a browser; the Europe PMC and archive routes given above serve to a script.

Measured against those figures the particles fall inside, so '''the size claims are supportable'''. Two things limit what that is worth, and neither is in the corpus. The comparison is not like for like: the published figures come from negatively stained or vitrified preparations, while the images here are thin sections, and preparation changes apparent size. And for measles the range depends on which source is consulted, spanning 100 to 300 nm across the sources named.

What is absent from the corpus is the measurement itself, and a reference with a scale against which a reader could perform it.


The claim is also weaker than it appears for a reason the corpus does not address: a size range that admits many cellular structures is precisely what a non-specificity argument requires. Falling inside the published range for an agent is simultaneously a real finding and a weak identification.
The claim is also weaker than it appears for a reason the corpus does not address: a size range that admits many cellular structures is precisely what a non-specificity argument requires. Falling inside the published range for an agent is simultaneously a real finding and a weak identification.
Line 130: Line 138:
Two of the deposited originals are '''observed''': <code>TEM_Path2_035_Mag_ 18500.tif</code> and <code>TEM_path1_043_Mag_ 11000.tif</code>, retrieved from the deposit, verified against the checksums it publishes, and read at full resolution. The remaining files in the two transmission-electron-microscopy archives are '''not inspected'''.
Two of the deposited originals are '''observed''': <code>TEM_Path2_035_Mag_ 18500.tif</code> and <code>TEM_path1_043_Mag_ 11000.tif</code>, retrieved from the deposit, verified against the checksums it publishes, and read at full resolution. The remaining files in the two transmission-electron-microscopy archives are '''not inspected'''.


The size measurements at flag 3 are this assessment's own, taken from the published scale bars with edges judged visually, and are good to roughly ±10–15%. The membrane measurements are taken from the originals at 0.380 nm per pixel by intensity profile across the boundary, and the structures measured were located by mapping the post's annotations onto the original by the ratio of their dimensions, a mapping confirmed by resizing the original to the published width and comparing.
The size measurements at flag 3 are this assessment's own, taken from the published scale bars with edges judged visually, and are good to roughly ±10–15%. The comparison figures it is measured against are '''observed''' for SARS-CoV-2 and for the CDC statement on measles, both read in the cited text. The membrane measurements are taken from the originals at 0.380 nm per pixel by intensity profile across the boundary, and the structures measured were located by mapping the post's annotations onto the original by the ratio of their dimensions, a mapping confirmed by resizing the original to the published width and comparing.


== See Also ==
== See Also ==

Latest revision as of 21:22, 28 October 2026

The Control Studies Project is a body of work published by Jamie Andrews, with protocols written and financed by Albert C. Mathews, which runs the uninoculated control arm of the cell-culture virus isolation procedure and documents what that arm produces. It is assessed here against its own claims.

It is a corpus rather than a single document — five posts and one deposited image dataset — and it is assessed as one, because its most consequential features exist between the documents rather than inside any of them. Several of its methods are better than those of the literature it criticises. Several of its claims exceed what its own published evidence supports, and in two cases the evidence that contradicts a claim appears in the author's own material. The assessment is on the first of the two grounds at Which Documents Get An Entry, and it belongs to the isolation track described at Two Tracks.

Citation and Access

A Substack post is deletable and unversioned, so each is cited both as published and as captured on the date of reading. The archived copy is the citable form; the timestamp is the capture date.

Document Published at Archived copy
Cell Culture Isolation Control Experiment (HEK293T study) controlstudies.substack.com/p/cell-culture-isolation-control-experiment 20261027163115
Transmission Electron Microscopy (preliminary TEM) controlstudies.substack.com/p/transmission-electron-microscopy 20261027163139
Controls 2 (Vero cell-culture study) controlstudies.substack.com/p/controls2 20261027163202
Vero Control Culture Transmission (Vero TEM) controlstudies.substack.com/p/vero-control-culture-transmission 20261027163230
We Have Proved Viruses Don't Exist controlstudies.substack.com/p/we-have-proved-viruses-dont-exist 20261027163005

The only citable primary is Mathews A, Vero Cell Culture Image Dataset and Procedure Description, Zenodo, 30 July 2025, doi:10.5281/zenodo.16619228, CC-BY-4.0 — a five-page procedure description plus approximately 6 GB of light-microscope and transmission electron microscope images.

The Standard Applied

The corpus is measured against its own stated claims and, where it states them, its authors' own stated requirements. It is not measured against the standards of peer-reviewed publication, which it does not claim; nor is its publication venue a basis for any flag here.

Where this assessment contributes a measurement of its own, that is marked.

What the Corpus Establishes

Stated against this assessment's own interest. Several of these are methodological choices the literature under criticism frequently does not make.

  • An independent laboratory, blinded to the purpose, wrote the culture descriptions. The Controls 2 study states that the contract lab "were not aware of the aims of this study, they were briefed that the Cell Culture protocol was to starve the Cell Line, so that it may produce Extra Cellular Vesicles". Its day-by-day descriptions are then published in full, for both arms.
  • Image selection in the preliminary electron microscopy was blinded by the same device, and the reasoning stated: the microscopist was briefed to find extracellular vesicles, "at no point were they instructed to look for viral-like particles… as we didn't want to insert any bias into our findings". All nine supplied images are published.
  • The HEK293T study measures its outcome rather than describing it. Viability was read on a Countess 3 FC automated counter across 105 test cultures and seven conditions, crossing three serum concentrations with three antibiotic concentrations, with a half-density seeding control, a nutrient check, a single-step passage test and a human-sample arm.
  • The reagent premise is sourced. The reduction of fetal bovine serum on inoculation is attributed to European Centre for Disease Control protocols and to the Clinical & Laboratory Standards Institute manual M41, named and quoted.
  • Raw data is deposited openly, with a persistent identifier, an open licence, a procedure description specifying equipment to catalogue number and software to revision, and an explicit invitation to reinterpret: "Please download and interpret and disseminate your interpretations."
  • Acquisition metadata is burned into the micrographs — filename, culture path, calibration in nanometres per pixel, accelerating voltage, magnification, exposure, and the processing state (Gamma: 1.05, No Sharpening, Normal Contrast). Most published micrographs carry nothing comparable, and it is what makes the assessment below possible.
  • Results are repeatedly reported against the authors' own thesis. Antibiotic concentration had "negligible effect"; the 1% against 2% serum difference was "negligible at day 5"; a reading is flagged as "a potential misread by the Countess"; an added human sample changed nothing "in any meaningful way"; syncytia and plaque formation were not observed in the Vero study, with a reason given.

Flags

1. The central count is not supported by the laboratory text the corpus itself publishes

Controls 2 states that the blinded laboratory "verified in the description of the Cultures 6 out of 9 of the CPE morphologies said to indicate viral presence".

Tested against that laboratory's own reproduced descriptions of the reduced-serum arm: rounding, vacuolation and dying cells or debris are supported, the latter two hedged in the original. Detachment is not: the laboratory's standing sentence, repeated nine times across five days, is that the cells "look healthy, with intact membranes, good attachment to the substrate, and no obvious signs of detachment or cytoplasmic granulation", and the evidence offered for detachment is the phrase "poor adhesion", which appears in that text as one of three alternative explanations for rounding. Plaque formation is not: the quotations offered describe dying cells, and the study's own conclusion lists plaque formation among the morphologies the technician "didn't explicitly note". Cytoplasmic granulation appears on both sides — denied nine times in the standing sentence, recorded as "dense/granular" in the day-three to day-five specifics.

A defensible count from the published laboratory text is three supported and one contested.

2. A stated hypothesis contains a comparison the design does not make

The HEK293T study's hypothesis is that uninoculated cultures produce cytopathic morphologies "to the same extent and in the same time as infected cultures". No infected cultures were run. The conclusion nonetheless asserts that the breakdown "occurred at the same speed, to the same severity", on the basis of what the post calls a positive control: published images, compared "by imaging comparison".

Severity in that study is a measured percentage; the compared publications do not report comparable viability figures. The first half of the hypothesis is well supported by the design. The second is not tested by it.

3. The morphological claim is asserted where it could have been measured, and only its size half holds

The preliminary electron microscopy study claims particles "of exactly the same size, shape and protein inclusions" as reference images of three agents. None of the three reference images carries a scale bar; the author's own crops do. Two of the three references are photographs of a screen — one of the Centers for Disease Control image library page, one of a news article, the latter described in the post as "the CDC image".

In the Vero study the measles comparison is sourced lower still. Its "MAINSTREAM CONSENSUS" block opens with a screenshot of a conversational language model answering the prompt "What is the Size and Morphology and overall appearance of the measles virus. Short Answer", and the size and morphology it returns are the standard against which the control-culture images are then declared identical. A model's summary is not a measurement, carries no method and no provenance, and cannot be checked by a reader.

This assessment measured the particles. Against the author's own 500 nm bars, the particle labelled SARS-CoV-2 is approximately 119 nm and the particle labelled measles approximately 185 × 195 nm; in the Vero study, structures labelled measles measure approximately 210 nm and 295 × 350 nm.

The figures these are compared against are named here, because the corpus does not name them. For SARS-CoV-2, Zhu et al. report that "Diameter varied from about 60 to 140 nm", measured on negatively stained particles (doi:10.1056/NEJMoa2001017; the sentence was read in the Europe PMC full text, which unlike the publisher page serves to an automated request). For measles there is no single published figure. The CDC's Epidemiology and Prevention of Vaccine-Preventable Diseases, chapter 13, states "It is 120 to 250 nm in diameter" (cdc.gov, updated April 2024; archived 20261020061451); image-library captions for thin sections give 100 to 200 nm; and general reference works give 100 to 300 nm. The Zhu article and the CDC chapter both refuse automated retrieval at the publisher and open normally in a browser; the Europe PMC and archive routes given above serve to a script.

Measured against those figures the particles fall inside, so the size claims are supportable. Two things limit what that is worth, and neither is in the corpus. The comparison is not like for like: the published figures come from negatively stained or vitrified preparations, while the images here are thin sections, and preparation changes apparent size. And for measles the range depends on which source is consulted, spanning 100 to 300 nm across the sources named.

What is absent from the corpus is the measurement itself, and a reference with a scale against which a reader could perform it.

The claim is also weaker than it appears for a reason the corpus does not address: a size range that admits many cellular structures is precisely what a non-specificity argument requires. Falling inside the published range for an agent is simultaneously a real finding and a weak identification.

The shape half was tested against the deposited originals, and it does not hold. The question is whether the labelled structures are bounded by a limiting membrane, which is what separates a particle from a region of dense cytoplasm. In a thin section a membrane reads as a narrow dark line with cytoplasm returning bright on both sides of it; a density reads as a gradual step from bright into sustained darkness.

In TEM_Path2_035_Mag_ 18500.tif, the original of the measles control-culture figure at 0.380 nm per pixel, a membrane elsewhere in the same field measures 3.0, 8.0 and 9.1 nm across three parallel cuts at 40 to 58 grey levels of contrast. A boundary of comparable contrast anywhere in that field is therefore resolvable. Of the three structures arrowed as measles, two show no limiting membrane: the profile runs from bright cytoplasm into sustained darkness with no bright return, over 21 and 45 nm at one and 16 and 29 nm at the other. The third carries a 9.1 nm line at 44.5 grey levels with bright cytoplasm on both sides, indistinguishable from the control membrane; whether that line is the structure's own boundary or a separate membrane passing nearby is unresolved.

The reference images the post compares these to show discrete, membrane-bounded particles standing against an electron-lucent background with the envelope visible. For two of the three structures the labelled object is a region of cytoplasm rather than a particle with a boundary, and that difference is about boundary rather than size.

4. Structures in an uninoculated culture are annotated with viral protein names

A crop from a control culture is labelled "Viral Membrane", "Matrix" and "Phosphoprotein" — identifications of named viral proteins applied to a structure in a culture that received no inoculum. This is the class of label the corpus argues elsewhere is applied without justification.

5. The claim that the cultures could not have contained an agent is an argument from compliance

The Vero study states that its cells "could not possibly have contained these viruses according to the handling and best practices of the cell banks". No mycoplasma test, no culture for pleuropneumonia-like organisms and no adventitious-agent screen is reported. The contract laboratory's visual note of "no clear evidence of contamination" is not a screen.

A continuous Vero line is precisely the material on which the historical literature records this problem — see Virology Methodology on latent agents in primate kidney culture.

6. "Falsification" is applied to agents where the design reaches only a method

The Vero study concludes with a list of thirty-two named agents presented as falsified, including several — human papillomavirus, hepatitis B, hepatitis C, adenovirus — that have no section in the post. The design supports a claim about what transmission electron microscope morphology cannot establish. Genomic, serological and reverse-genetics evidence is untouched by a morphology comparison and is not addressed.

7. The strongest-titled post rests on an invalid inference

We Have Proved Viruses Don't Exist argues that if particles indistinguishable from virions occur where no virus could be, then physical attributes do not define the agent, "Therefore it DOES NOT EXIST". That morphology is insufficient to identify something does not entail that the thing does not exist; almost nothing in biology is identified by electron-microscope appearance alone.

The argument also contradicts its own post. Extracellular vesicles were the microscopist's brief and are identified in two of the nine images; they are not defined by appearance either, yet the post treats them as real and uses them as the baseline the other particles differ from.

Findings That Exist Only Between Documents

This section has no analogue in a single-document assessment. It is where the corpus's most consequential features are.

  • The same cultures are characterised oppositely in two posts. The Vero study states that the reduced-serum arm "showed great amounts of Cell Death and clinical morphology of CPE". The laboratory text published in Controls 2 describes that arm, at each of five days, as cells that "look healthy, with intact membranes, good attachment to the substrate", with "some sign of stress" by day five. The clearest difference the laboratory records between arms is confluency — reduced proliferation under reduced serum, which is the expected result and a different observation from cytopathic effect.
  • A claim in one post is refuted by evidence in another. The "6 out of 9" count appears in Controls 2; the laboratory text that does not support it appears in the same post, reproduced by the author.
  • A premise flagged as unsupported in one post is properly sourced in another. The Vero study assumes reduced serum replicates infection medium without citation; Controls 2 cites CLSI M41 and the European Centre for Disease Control for exactly that.
  • The means to settle a claim exist in one post and go unused in another. The micrographs carry nanometre-per-pixel calibration. The size claims are asserted without measurement.
  • Figure attribution does not track the stated design. Of thirty-five Vero micrographs with a readable path label, eleven are from the 10% serum negative control, and three of those sit under an explicit claim about the reduced-serum arm. Two of the three understate the authors' own case, since a morphology found in a healthy control needs neither an agent nor starvation to appear.

Claims Not Supported by the Corpus

Two readings circulate and neither survives it.

  • That it demonstrates viruses do not exist. Its own strongest-titled post rests on the invalid inference at flag 7, and that post's final paragraph concedes that "thousands of claimed particles… are needed to be falsified" before such a statement could be made.
  • That it lacks controls. It carries a purpose-blinded independent laboratory, a blinded image selection, a crossed dose–response design, a seeding-density control, a nutrient control, a passage test and an instrumented viability measure. Several of these are controls the literature it criticises frequently omits. Dismissing it on this ground misdescribes it, and a reader who checks will find so.

What the Authors State as Limitations

The corpus discloses a good deal against itself, and this is recorded because it bears on how its other claims should be read.

Lead citrate is a positive stain and "would render the claimed Spike Proteins invisible" — so the preparation cannot show the feature that defines coronavirus morphology. Scraping rather than trypsinisation may distort what is seen outside the cell, and the effect "could be speculated to have a fair impact". The two-hour microscopy session was not systematic: "we were not being specific in our approach, rather looking for overall information". Increased filopodia appeared in the negative control as well as the test arm. Antibiotic concentration had negligible effect. The experiments that would settle several questions are named — negative against positive staining of the same samples, and scraped against trypsinised cells.

What Would Weaken This Assessment

  • The shape finding in flag 3 rests on intensity profiles taken at three locations in a single micrograph. A profile across either of the two structures reported as unbounded, showing a dark line with bright cytoplasm returning on both sides, would overturn it. So would a demonstration that the mapping from the post's annotations onto the deposited originals points at structures other than the ones indicated.
  • A laboratory report supplied separately and not reproduced, using the terms "detachment" and "plaque", would close most of flag 1.
  • A mycoplasma or adventitious-agent screen, if performed and simply not reported, would close flag 5.
  • If the publications used as a positive control do report viability figures, flag 2's severity objection narrows to timing alone.

Epistemic Status

All five posts and the procedure description are observed: read in full. The contract laboratory's descriptions are observed as reproduced by the author; no independent copy exists, so whether the reproduction is complete is unresolved.

Of roughly 232 published images, a small number are observed: six micrographs read at full size, fifty-two acquisition footers read, four particles measured against scale bars. The remainder are not inspected.

Two of the deposited originals are observed: TEM_Path2_035_Mag_ 18500.tif and TEM_path1_043_Mag_ 11000.tif, retrieved from the deposit, verified against the checksums it publishes, and read at full resolution. The remaining files in the two transmission-electron-microscopy archives are not inspected.

The size measurements at flag 3 are this assessment's own, taken from the published scale bars with edges judged visually, and are good to roughly ±10–15%. The comparison figures it is measured against are observed for SARS-CoV-2 and for the CDC statement on measles, both read in the cited text. The membrane measurements are taken from the originals at 0.380 nm per pixel by intensity profile across the boundary, and the structures measured were located by mapping the post's annotations onto the original by the ratio of their dimensions, a mapping confirmed by resizing the original to the published width and comparing.

See Also