Rahn-Otto: Mathematician's review (Step 2)

Project: Otto Rahn, Invisible Radiations of Organisms (Protoplasma-Monographien IX, 1936). Reviewer role: PhD mathematician. I check arithmetic, units, statistics, notation and hidden assumptions. Physics and biology verdicts belong chiefly to the physicist and contrarian; where I give a verdict it is on the mathematical content, and I say so.

Tags: [read] from the source text; [read: page image p.N] from a rendered page (PDF page numbers; book page = PDF page minus 14); [calc] my own arithmetic (numpy/Python, scratch only); [web] with URL; [memory] with confidence. Source gap: none (text available, 230 pages), but the deep read covers PDF pp.44-67 only. For claims 4-7 I read the passages myself in the extracted text and, for two figures, in page images; items I did not check are said to be unchecked.

Reader note: "probable error" (PE) in 1930s papers is 0.6745 standard deviations. "Effect = 5.3 times the probable error" therefore means about 3.6 standard deviations.

---

Corrections to Step 1

1. Nerve speed is 30 +/- 3 m/s, not 30.3 m/s. facts.json (summary, Chapter V) writes "30.3 m/s in sciatic nerve"; claims.md correctly says 30 +/- 3. The text layer reads "303 meters per second" and "30.3" is a misreading of the printed "30+3" sign (OCR dropped the plus-minus). I read the page image: "the rate of conduction in the nerve was found to be 30+-3 meters per second" [read: page image p.126]. The +/-3 matters: it is a 10% error bar, not a 0.3% one. The facts.json entry should say 30 +/- 3.
2. Counter sensitivity does not convert as printed. Barnes states the counter limit as "about 10^-9 ergs/cm2 sec, or about 500 quanta/cm2 sec" [read: p.~1684 of the text, Chapter I]. A 2020 A photon is 9.8e-12 erg, so 10^-9 erg/cm2/s is about 100 quanta/cm2/s [calc]. 500 quanta per 10^-9 erg means 2e-12 erg per quantum, about 1.25 eV, a wavelength near 1 micrometre (infrared) [calc]. The two numbers describe different photons. For ultraviolet the figure should read about 100 (at 2000 A) to about 150 (at 2500 A) quanta/cm2/s. Step 1 repeated "about 500 quanta/cm2/s" without noticing. Consequence: the strongest claimed senders (100-1000 quanta/cm2/s, Chapter VI) sit at, or within a factor of 10 of, the counter's detection limit. That helps explain why physical counters gave marginal and disputed results.
3. "43 times" is right; the book's "one-fiftieth" is rounded. facts.json and claims.md already say 43 (correct). I confirm: 6.1e23/342 x 9.8e-12 erg = 1.75e10 erg = 418-420 cal/g; 420/9.7 = 43 [calc]. Rahn's text says "one-fiftieth" [read: p.55]. No correction needed beyond noting the text's own rounding.
4. Barnes and Rahn 1933 may not be the physicist Barnes. The Investigator's lead (Macht cites "Rahn, O., Barnes, M.N., Diss.") is plausible. I did not verify. Step 1 should not assume Sidney W. Barnes did the oxycholesterol work. Not decisive for any verdict.
5. Table 10 "interval" and +/- values: Step 1 / the deep read treated the counts as "10-40 per interval". The values behave like counts per minute averaged over the stated interval, with error = square root of (rate / minutes) [calc]. Details in claim 2. This changes how the table should be read: the +/- are pure counting (Poisson) errors.
6. Typography/OCR slips that are not Rahn's errors (so they should not be attributed to him): "23 out of 20 experiments" reads "23 out of 26" in the image [read: page image p.157]. The deep read's note that the permanganate equation is unbalanced stands (typesetting omission of coefficients), I did not re-check it.
7. Step 1 says the book "mostly reports Gurwitsch's school." I agree. Add that Table 45 (see claim 6) and Table 10 contain internal numerical inconsistencies I found; they are listed below, not in Step 1.

What I confirmed from Step 1 [calc]: E(2020 A) = 6.14 eV = 141.5 kcal/mol; Rubner's 9.7 cal/g x 342 = 3.3 kcal/mol; Fricke's numbers (1.5e-19 cal per molecule, 3000 A, 10^-43 Boltzmann factor) are consistent; probability of a yeast cell (6 x 7 micrometres, area 4.2e-7 cm2) being hit at 1000 quanta/cm2/s is 4.2e-4 per second, 0.0252 per minute, 40 minutes for one expected hit (Rahn's arithmetic is right); the rotating-disk numbers (below) are arithmetically right; the onion-root and nerve numbers in claim 4 are as stated.

---

Claim 1. Living cells emit very weak short UV ("mitogenetic radiation") that speeds division in other cells

a. Mathematically consistent? The quantitative parts are consistent where checkable: wavelength/energy conversion, the single-quantum hit probability (4.2e-4/s at 1000 quanta/cm2/s) and the 40-minute figure [calc]. One hidden assumption: Rahn treats "probability of at least one hit" as the expected number of hits (valid only while the probability is small; at 40 minutes the expectation is 1 and the probability of at least one hit is 1 - e^-1 = 63%, not "likely every cell"). It matters little. A bigger issue: the book's central intensity estimate (100-1000 quanta/cm2/s) is "very inaccurate" by Rahn's own words and is derived biologically, not by a calibrated physical measurement. Using the corrected counter limit (correction 2), those intensities are at the edge of physical detectability, so the physical and biological detectors cannot be cross-calibrated. Statistics: Schwemmle's analysis of about 200 onion-root experiments (Fig. 30) puts the method's error at +/-10% for 500 mitoses [read: p.72]. That is what counting statistics predict: 250 mitoses per side gives sqrt(2/250) = 9% on a ratio [calc]. So the error band is a pure Poisson band; it contains no allowance for root-to-root biological variation or for observer bias in scoring mitoses. Rahn reports that all positive results lay outside the band [read: p.72] and that Reiter and Gabor's error is +/-20%. Julius (1935), chick fibroblasts: ratio 1.92 +/- 0.15 (56 quartz/glass pairs) against 1.01 +/- 0.05 (48 unirradiated pairs); the quoted "0.91 +/- 0.17 = 5.3 times the probable error" [read: p.97]. I get a combined error of 0.16 [calc] and 0.91/0.17 = 5.35, so internally consistent; but the "effect" is taken as a difference of ratios (1.92 - 1.01), where the natural measure is the ratio 1.90, and the quoted "5.3 PE" is about 3.6 sigma. Still a respectable signal if the pairs were independent and unbiased; the text does not say whether scoring was blinded. Fermentation test (Gesenius): of 102 experiments, 68 increased, 4 decreased... the digest/text layer shows "68 increase, 4 decrease, 3 within error" (garbled); the numbers as printed do not sum to 102 [read: p.84, text layer garbled; unchecked in image].

b. Consistent with established experiments? I defer to the physicist. Mathematically: modern work measures ultra-weak photon emission of order 1 to 1000 photons/cm2/s from tissues, chiefly visible/near-UV, from radical reactions such as lipid peroxidation [web] https://ar5iv.arxiv.org/html/1502.07316 (summary of a review; medium-weight check). So the intensity scale Rahn states is the same order as modern UPE; the contested part is the UV (190-260 nm) component and the division-accelerating effect.

c. Tested? Yes, many times, with mixed results. In the book: Gurwitsch school (onion root), Baron/Tuthill-Rahn yeast bud method, Wolff and Ras bacteria, Julius (fibroblasts), and by Rahn's own group (Ferguson and Rahn 1933; Tuthill and Rahn 1933/34). Against: Hollaender and Claus, NRC Bulletin 100 (1937): "all the attempts have yielded negative results", biological and physical detectors, two years' work [web, via Investigator] https://www.nature.com/articles/1401007c0. Authors in the Gurwitsch tradition (Volodyaev and Beloussov 2015, who are interested parties) argue those conditions were "new, and never checked before" [web, via Investigator, a model summary; not re-fetched] https://www.frontiersin.org/journals/physiology/articles/10.3389/fphys.2015.00241/full. Quickenden et al. 1989 (attempt to stimulate yeast division with weak UV) did not reproduce the effect; Wainwright et al. 1997 report "partial evidence", not reproducible [web, via Investigator: PubMed 33711853]. Rahn himself recorded that his own work had long stretches of complete negative results [read: Ch. IV]. No preregistered, blinded, multi-lab replication is on record in any source I have.

d. If true, what must change? Mathematically little: a weak UV flux of 100-1000 quanta/cm2/s needs no new law. What must change is the biology of signalling: single-photon triggering of division means an amplifier in the cell (Rahn's chain-reaction analogy, quantum yields up to 1e5 in Na2SO3 photo-oxidation [read: p.61]). That is not forbidden by physics.

Verdict (split). Weak emission of photons by living tissue: Established in the visible/near-UV (modern ultra-weak emission), Plausible, untested for the deep-UV band. Division-accelerating effect (strong form, as a reliable laboratory phenomenon): Never properly tested (no neutral, blinded, multi-lab protocol), with the two best-known controlled negative attempts (Hollaender and Claus 1937; Quickenden 1989) Tested, not supported; the proponents' objection is that the tests used untested conditions.

---

Claim 2. Ordinary biochemistry emits the light, each process with a narrow-band "fingerprint" read by yeast/bacteria (and weakly by counters)

a. Internally consistent? Three separate points. (i) Table 10 (counter data) [read: page image p.48]. The +/- values are consistent with Poisson errors of rates in counts per minute averaged over the interval: for the nine rows, rate/(error^2) is 10.4, 9.5, 10.2, 6.2, 4.0, 6.2, 9.0, 16.0, 3.7 against intervals of 10, 8, 10, 6, 4, 6, 9, 15, 4 minutes [calc]. So I read the table as counts/min, with error = sqrt(rate/T). On that reading, excess over control in standard errors is 3.2, 4.5, 1.5, 2.2, 1.6, 2.7, 3.6, 6.1, 0.4 [calc]. Seven rows are above 1.5, only four above 3; the dark FeSO4 + K2Cr2O7 row at 15 min is strongest (6.1); the last row (0.4) is null. This is an inference from the pattern, not stated by Rahn. What the table omits: counter dead time, background drift (a few counts per minute differences are well within the drift of 1930s counters), whether control and exposed runs were interleaved, and the counter's quantum efficiency (so no photon flux can be derived). Barth's proteolysis result "24 of 25 higher than control, 7 above 3x error" is a sign test with p about 1e-6 [calc], which is strong if the runs were independent and control and exposed order randomised; the text does not say. (ii) The strip spectra. Spectra were read as positive/negative strips of 10 or 50 A with a biological detector. Blood glycolysis: 5 positive strips of 10 A out of about 60 (1900-2500 A) [read: Ch. II]. There is no stated false-positive rate per strip. If each strip had a 5% chance of a spurious positive, five or more of 60 would occur by chance 18% of the time; at 10% it is 73% [calc]. So a "fingerprint" of 5 of 60 strips is only meaningful if the reproducibility of the same strips across independent runs is documented; Rahn says "two identical strips" disagreements occur and calls the method "rather crude" at 50 A [read: Ch. II, pp.~51-53]. Hidden assumption: that a positive strip is a photon band of the reaction, not a detector-state or secondary-radiation artefact. The text itself says nutrient media that have held microorganisms emit secondary radiation when exposed, so "no proportionality" and spectral purity cannot be assumed [read: Ch. VI]. (iii) Energy per event. See claim 3: for neutralisation (13 kcal/mol about 0.6 eV) and sucrose hydrolysis (3.3 kcal/mol, 0.14 eV) a single reaction event cannot make a 5-6 eV photon [calc; heat of neutralisation [memory, high]]. Oxidations (pyrogallol + O2, dichromate + Fe(II)) release several eV per event and are chemically plausible sources: effective excitation needs 1-6.6 eV supplied within about 1e-11 s [web] https://www.materialsscience.pwr.edu.pl/bi/vol24no4/articles/IMIM_14.pdf. So the claim's "even neutralisation and dissolving NaCl emit" is the arithmetically weakest part, "oxidation emits" the strongest. Also, Table 45 and the experimental tables contain inconsistent entries; see claim 6.

b. Consistent with experiment? Chemiluminescence of oxidations in the UV-visible is established [memory, high]; that of hydrolysis and neutralisation in the 190-250 nm range is not established independently of the biological detector. Frank and Rodionow, Gerlach (quartz transmits, glass does not), Audubert and van Doormal, Barth are cited for physical detection [read: p.48; Investigator lists their later work]. Audubert (1938) on emission from reactions is a lead for further reading [web, via Investigator].

c. Tested? Physical counters: positive reports by the cited authors, negative or contested by Lorenz (three error sources: water vapour, charge on the quartz, muscle stimulation by the induction coil) and in Hollaender and Claus. I have not read these papers.

d. If true? For oxidations, nothing. For hydrolysis/neutralisation, quanta of 6 eV from 0.1-0.6 eV events would need concentration of many events' energy, which is not described by standard kinetics (this is the content of the Lorenz objection).

Verdict (split). Oxidation-driven weak UV (counter evidence): Plausible, untested in a neutral controlled replication (the counts are marginal and the key negative studies disputed). Narrow-band spectral "fingerprints" read by a biological detector: Never properly tested (no false-positive rate, no independent replication of the same spectra on record in this file). Emission from neutralisation/NaCl dissolving/hydrolysis as one photon per reaction event: Refuted by calculation (per-event energy 0.04-0.6 eV against 5-6 eV photons); the weak form (rare excitations, quantum yield far below 1) is not refuted by this calculation.

---

Claim 3. The energy accounting is unresolved: one 2020 A quantum per sucrose molecule = 420 cal/g, 43 times Rubner's 9.7 cal/g

a. Mathematically consistent? The arithmetic is correct [calc, above]. The reasoning has three flaws. 1. The premise is a choice, not a law. "Each reacting molecule must emit at least one quantum of the shortest wavelength" [read: p.55] fixes a quantum yield of at least 1. A measured signal of a few counts per minute (claim 2) implies a quantum yield very far below 1. With yield f, the radiated energy is 420 x f cal/g; f = 1e-6 gives 4e-4 cal/g, 1e-5 of the heat. The "43 times" paradox disappears for any f below 2%. Modern ultra-weak emission has yields far below 1 [memory, medium; consistent with the 1-1000 photons/cm2/s scale from the source above]. Rahn's two alternatives (side reaction, or immediate reabsorption) omit this third option. 2. The real constraint is per event: 141.5 kcal/mol photon against 3.3 kcal/mol released, a factor of 43 per event (or 5.0 eV against 0.14 eV). A rare event can still need 43 molecules' heat, which kinetics gives no mechanism for. 3. The proposed fix fails on Rahn's own numbers. With Fricke's "usual" activation energy of 100 kcal/mol plus 3.3 kcal/mol heat, a leaked quantum has at most 103 kcal/mol, wavelength 2770 A [calc: 4.48 eV, 12398/4.48 = 2767 A], not 2020 A, and 2770 A is outside the range where the detector responds (stimulation only below 2600-2700 A [read: Fig. 28]). If the spectra "measure the energy of activation" [read: p.56], 2020 A means 141.5 kcal/mol, contradicting the assumed 100. Measured activation energies are lower still: acid-catalysed sucrose hydrolysis about 109 kJ/mol (26 kcal/mol), invertase about 7 kcal/mol [web] https://pmc.ncbi.nlm.nih.gov/articles/PMC2664835 and search results listed at https://explore.cnr.it/resource/item/195591 (search snippets; confirms the order of magnitude of my [memory, medium] figures). 26 kcal/mol corresponds to a photon of about 11,000 A, infrared, not ultraviolet. Also, activation energy is borrowed and returned, not net energy; a photon leaking out terminates the chain unless resupplied. Units and notation: "heat of hydrolysis" and "heat of activation" are used loosely [read: p.56]; Avogadro's number is 6.1e23 (immaterial); "one-fiftieth" is really 1/43.

b. Consistent with experiment? Rubner's 9.7 cal/g is close to the modern value of about 10 cal/g [memory, medium: roughly 14-15 kJ/mol]. The conflict is with the premise, not the calorimetry.

c. Tested? Not as an energy balance; no one measured the photon yield of the sucrose-invertase reaction in this record. Rahn says the origin "is by no means clear" [read: p.55], which is accurate.

d. If true? With low yield nothing changes. With a yield near 1 (what the 420 cal/g calculation assumes), thermodynamics would need rewriting; no one claims that.

Verdict (split). Rahn's arithmetic: correct. The apparent paradox (420 vs 9.7 cal/g, one quantum per molecule): Refuted by calculation as a real conflict (it exists only if the quantum yield is 1; a yield below about 2% removes it). The reabsorption chain as the resolution: Internally inconsistent on its own numbers (2770 A against 2020 A). Per-event energy for hydrolysis photons: unresolved, no mechanism given: Never properly tested.

---

Claim 4. Secondary radiation spreads along onion roots at about 30 m/s and along frog sciatic nerve at 30 +/- 3 m/s, so it may take part in nerve conduction

a. Mathematically consistent? I rechecked Chapter V [read: page image p.126]. Rotating disk at 3000 rpm = 50 rev/s, so 1 degree = 5.56e-5 s [calc]. Positive angles for 2.5 cm of root: 25-50 degrees (time 0.0022 s, corresponds to 39.6 degrees [calc]); for 5 cm: 40-70 degrees. Rahn says "an average increase of 15 degrees ... corresponds to 0.00083 s" so 2.5 cm in 0.00083 s = 30 m/s. Arithmetic is right [calc]. Problems: - The window midpoints are 37.5 and 55 degrees, a difference of 17.5 degrees (0.00097 s, 26 m/s) [calc]; the lower edges differ by 15, the upper by 20. The speed is 22-30 m/s depending on which edge, with a +/-12 degree window width (+/-0.0007 s) comparable to the quantity measured (0.0008-0.0010 s). So 30 m/s from the onion root is good to roughly +/-30% at best, and is a difference of two coarse windows, not a time-of-flight. - The "0.0014 s for local processes" is obtained by assuming the same speed through both halves and subtracting [read: p.112]; it is not independently measured. - Light itself travels at 3e8 m/s; 30 m/s means the signal is not propagating as light but as a relay (absorption, reaction, re-emission) with the dominant delay in the relay steps. The measurement does not distinguish "light relay" from a chemical or electrical wave carried along the root. - Nerve: Latmanisowa used a narrower window (3 degrees primary, 1 degree secondary), giving 30 +/- 3 m/s "in good agreement with physiological measurements" [read: page image p.126]. 30 m/s is within the range of frog A-fibre conduction at room temperature [memory, medium-high: about 20-40 m/s]. Agreement of one number is weak evidence: the conduction velocity of a nerve depends on fibre and temperature; the nerve length, temperature and fibre type are not stated in the passage. - Amplification: six cuvettes in series gave "27 times" shorter exposure time (10 s against 4.5 min) [read: Ch. II, p.~58; deep read]. 270 s/10 s = 27 [calc]. Rahn warns this is measured biologically only and that reciprocity and polarisation leave it unclear that physical intensity rose. A passive medium cannot raise energy; the 27 is a time to effect, not a gain in photons, so the word "amplifier" is a notational hazard.

b. Consistent with experiment? The established mechanism of nerve conduction is the ionic action potential (Hodgkin and Huxley 1952) [memory, high], which reproduces conduction speed from membrane physics and needs no light. This does not exclude a minor optical side channel; no one in this record measured one.

c. Tested? Anna Gurwitsch (1931, root), Latmanisowa (1932, nerve), Frank (muscle). Replication by others in this file: none. Later sources named by the Investigator (Frank and Rodionow, Rajewsky, Siebert and Seffert 1933-34) were physical-counter searches that did not show a clear signal in the book's account [read: Ch. IV].

d. If true? An optical channel in nerves would add a second conduction mode. Nothing in electromagnetism changes.

Verdict (split). Measured delays and the arithmetic: Plausible, untested (internally consistent, resolution poor). "Light relay at 30 m/s in root/nerve" as the explanation: Never properly tested (not distinguished from other relays). Role in nerve conduction: Historically superseded but instructive (the ionic action-potential model explains conduction without it).

---

Claim 5. Healthy blood radiates, cancer blood does not, tumours radiate, so blood radiation is a cancer test and cancer may be a radiation disorder

a. Internally consistent? The book's own statements conflict in places. (1) Gesenius lists patients with sarcoma, hypernephroma, glioma and myoma as showing "normal radiation" [read: Ch. VII, p.~180], with Rahn's footnote "This seems strange since the sarcoma tissue shows the same strong radiation as carcinoma tissue." So the rule "tumour radiates, blood does not" fails by the book's own account for sarcomas. Rahn's glosses "benign" for sarcoma, hypernephroma and glioma are medically wrong [memory, high: these are generally malignant]; the listed exceptions are thus largely malignant tumours, which weakens the diagnostic rule. (2) "Occasional exceptions" are admitted both ways: blood radiation "found infrequently connected with severe carcinoma, and has rarely been absent in patients where autopsy revealed absence of carcinoma" [read: p.~181]. No counts are given, so no sensitivity or specificity can be computed. A diagnostic test needs true/false positive and negative counts; this book gives none for Gesenius's three years of cases ("very encouraging"). Without them the "test" cannot be evaluated: with, say, 70% sensitivity and 90% specificity at 1% cancer prevalence in the screened group, positive predictive value is about 7% [calc, illustration only, not Rahn's numbers]. (3) The explanation of origin is withdrawn within the text: "this simple assumption does not agree with our explanation of these radiations as originating from biochemical reactions", since sugar and glycolysis are little altered in cancer blood [read: p.~181-182]. Braunstein and Heyfetz: rats showed decreased radiation from day 5 after inoculation while glycolysis and blood sugar were unchanged [read: p.~180]. (4) The "cancer quencher" explanation (mixing cancer blood with normal blood suppresses radiation; Lydia Gurwitsch and Salkind 1920) is a different mechanism from "absence of radiation": a measurable inhibitor, not a missing source [read: p.~180]. Roffo: 700 rats, 52% cancer in 10-11 months under ultraviolet, none under tungsten lamp [read: Ch. VII; I did not re-read the passage]; no group size, controls per arm or test statistics checked. I did not verify these numbers beyond the digest (unchecked).

b. Consistent with experiment? No modern blood UV-chemiluminescence cancer test exists in routine use [memory, medium]. The Investigator's lead: Soviet continuation to the 1950s (Gurwitsch's "cancer quencher", Pesochensky 1942, Avchina 1950, Sneshko 1955) and Naumova et al. 2018, who report "very promising results" and a peptide quencher but that research was "interrupted and almost abandoned" [web, via Investigator] https://jbpe.ssau.ru/index.php/JBPE/article/view/3308. Those authors work in the Gurwitsch tradition. I have not read the Soviet papers; they should not be dismissed, and I cannot confirm them.

c. Tested? Clinically in Berlin university clinics (Gesenius, 3 years), Heinemann, Klenitzky, Protti (cytophotolysis); Soviet clinics through 1955. No blinded, prospectively registered trial in this file's sources. Replications by outside groups: not found.

d. If true? Oncology would gain an assay; "cancer as radiation disorder" would add a signalling hypothesis, no change to physics.

Verdict (split). Blood-radiation absence as a cancer test: Never properly tested (no sensitivity/specificity on record, no blinded trial). "Cancer is a radiation disorder": Not testable as stated (the book itself withdraws the simple version). Roffo's UV-cancer link: Never properly tested here (single reported series, not checked).

---

Claim 6. The radiation shapes development and health (wound healing, larvae, yeast forms, crowding effect, harmful menstrual/fingertip radiation killing yeast through quartz)

a. Internally consistent? I found internal numerical problems. - Table 45 (sea-urchin eggs as senders, onion-root mitoses) [read: page image p.143]. Percent effect should be (exposed - control)/control. Recomputed [calc]: 167 to 173 = +3.6 (printed +0.3); 299 to 299 = 0 (0); 139 to 203 = +46.0 (+46.0); 215 to 301 = +40.0 (+40.0); 67 to 109 = +62.7 (+62.0); 214 to 226 = +5.6 (+0.5); 182 to 186 = +2.2 (+0.2); 92 to 86 = -6.5 (-0.6); 144 to 157 = +9.0 (+0.9); 423 to 568 = +34.3 (+34.0); 252 to 340 = +34.9 (+35.0). The small values are printed at one tenth of what the counts give; the large ones are right. Either a printing convention changed in the table or these are errors; as printed the table cannot be reproduced. The computed small values (2-9%) are within the counting noise of about 10% for 170-210 mitoses [calc], so they are null results either way. The pattern is bimodal: the effects are either near zero or 34-62%. Noise around a graded response would not look like that; it fits a threshold response ("induced / not induced") or selection of results, which the text does not rule out. Two rows have exactly equal control and exposed counts (299/299 and 268/268); for Poisson counts near 270-300 an exact tie has about 1.6% chance per row [calc], so two ties in ten rows is unusual (about 1% chance). I flag this as a curiosity for the reader, not as evidence of anything. - Yeast bud method: if radiation speeded only the first stage, buds would fall from 26 of 35 (74.3%) to 18 of 27 (66.7%) [read: p.~64; printed 74.4% and 66.8%, rounding slip, immaterial; calc]. - Oxycholesterol (Barnes and Rahn 1933): "Out of 5 batches ... 4 killed Mycoderma during the first day ... but not on the second day"; the fifth batch "had no effect" and is explained by a delayed start and loss of potency over a day [read: page image/p.~196]. So the claim is 4/5 positive plus a post-hoc account of the fifth, and the potency window (hours) is itself an untested free parameter. With a free parameter (decay time) fitted after the fact, the data cannot refute the claim; that is a falsifiability concern, not an arithmetic error. - The tadpole wound results (Blacher, 26-33% effects, double maximum) and the frog parthenogenesis ("3 of a few hundred eggs") are in the digests; I did not check them in the text (unchecked). - Sea-urchin eggs: furrowing increased in "23 out of 26 experiments" (printed "20" in the text layer, an OCR slip); a sign test on 23 of 26 gives p about 1e-4 [calc], as a count of experiments, if each was independent and unbiased.

b. Consistent with experiment? Modern studies of cell-to-cell photon effects (Albrecht-Buehler 1992, Beloussov and Volodyaev 2009-2015, Cifra and Pospisil) are listed by the Investigator as related work [web, via Investigator; not read by me]. Menstrual toxin: Macht (1949) and others cited as "Rahn and his co-workers claimed that oxycholesterol emits destructive ultraviolet rays" [web, via Investigator] https://psycnet.apa.org; this is a report of the claim, not a confirmation. Rahn's own 1932 public account (NYT 25 June 1932; Science Service via Papers Past): fingertip, eye and nose radiation killed yeast "in five minutes", right hand usually stronger, strength varying between people and over time [web, via Investigator] https://www.nytimes.com/1932/06/25/archives/evil-eye-is-found-to-be-based-on-fact-radiations-from-human-body.html. These are newspaper paraphrases of a talk, flagged by the paper itself as preliminary.

c. Tested? By Rahn's group and the Gurwitsch school; no neutral replication in this file. Volodyaev and Beloussov (interested parties) note the reproducibility problem in the field.

d. If true? Developmental biology would get a physical signalling channel; no change in physics.

Verdict (split). Cell-cell photon signalling effects on division/development: Never properly tested (positive series only from proponents; bimodal and inconsistent tables). Harmful human/oxycholesterol radiation killing yeast through quartz: Never properly tested (small series, post-hoc potency window; the possibility of non-radiative volatile effects is controlled by quartz/glass pairs in some designs [read: Ch. IV, Ferguson]). Wound healing effect (Blacher): Never properly tested (not checked in primary text).

---

Claim 7. Failed replications do not disprove the effect (the detector is only responsive at a particular physiological stage; failures are unpredictable; biological error is large)

a. Internally consistent? The numbers: Gurwitsch laboratory, 54 failures in 373 blood experiments = 14.5% [read: p.~92; calc 54/373]. Rahn's source (Golischewa 1933) says the 54 failed "on account of a poor quality of the yeast culture", evident because "all other associates using the same culture on the same day obtained negative results" [read]. So failures were assigned by an internal criterion (all positive controls in the lab fail the same day) rather than by an external one; that is defensible only if the criterion is fixed in advance. As statistics: excluding failed runs after seeing the results is acceptable only if the exclusion rule does not depend on the outcome for the experiment in question. It is a hidden degree of freedom here, since the "failures" are recognised by negative results. There is a logical structure worth stating: "the detector works only when physiologically ready and fails unpredictably" makes the hypothesis unfalsifiable by any single negative run unless a positive control with known readiness is run alongside. The honest form (Gurwitsch school practice) is to run known senders and discard only runs where the control also fails. Whether that was done in Hollaender and Claus or Quickenden is a factual question I cannot settle; the proponents say conditions were new (culture age, medium, detector layer thickness, induction time) [web, via Investigator, a model summary of Volodyaev and Beloussov 2015]. Reading the book itself: Rahn states the physical counters negative results (Rajewsky; Frank and Rodionow; Siebert and Seffert) and Lorenz's three error sources [read: Ch. IV], so the claim as stated in claims.md ("Rahn also points out the physical counter evidence") is correct. The Investigator's point on absence of evidence: the Western work almost stops after 1937 (Hollaender and Claus; about 600 papers published by then, per Nature [web, via Investigator]); Soviet work continues to the 1950s (dissertations 1942-55; Gurwitsch books 1947, 1948, 1968, 1974) and photon-counting studies resume in the West in the 1950s-70s (Colli and Facchini 1954; Quickenden and Que Hee 1974). Explanations to weigh (not decided here): (1) the effect was not real and interest faded with the negative bulletin; (2) the effect is real but fragile, and the Western literature stopped before a robust protocol existed; (3) institutional factors (the Hollaender and Claus bulletin as a decisive negative; war-time disruption; the Soviet context of Gurwitsch). Records that would decide: Rahn's reply to Hollaender and Claus (Tabulae Biologicae 1937), the full NRC Bulletin 100 protocols, and original lab notebooks. I have not read any of these.

b. Consistent with experiment? The claim concerns method, not a physical measurement. Rahn's candour about failure is itself data: the book records that Heinemann, Siebert, Wolff and Ras, and Rahn himself had periods of total negatives [read: Ch. IV].

c. Tested? The method-sensitivity claim was never tested by a pre-registered protocol that states in advance when a run counts as valid.

d. If true? No change to physics, but it would require biology to accept stage-dependent detector response as a standard control.

Verdict. Not testable as stated in its general form ("failures do not disprove"), because a run-by-run exclusion based on outcome is unfalsifiable; Never properly tested in its specific form (stage-dependent response with a pre-declared validity control).

---

Cross-cutting mathematical points for the reader

1. Poisson errors are the floor, not the ceiling. Every error band quoted (Schwemmle +/-10% at 500 mitoses, Table 10 +/-) matches pure counting statistics [calc]. Biological variance (between roots, cultures, days) is additional and the book states it is large.
2. Probable error versus standard deviation: PE = 0.6745 sigma. "5.3 PE" is 3.6 sigma.
3. Multiple comparisons: 60 spectral strips, many reactions and many conditions give many chances for a positive. Only independent, repeated spectra count.
4. Quantum yield: the energy paradox is a statement about one number (the fraction of reacting molecules that emit). Compute it before arguing about energy.
5. Worked check the reader can repeat: E = hc/lambda with hc = 12398 eV A: 2020 A gives 6.14 eV; 2770 A gives 4.48 eV; 3000 A gives 4.13 eV. Compare with 0.14 eV for sucrose hydrolysis.
6. Rahn also wrote Mathematics in Bacteriology (1939) [web, via Investigator: HathiTrust catalogue], which may show how he treated these error questions later; I did not obtain it.

Web verification log (6 calls budget)

• Sucrose activation energies [web] acid about 109 kJ/mol, invertase about 7 kcal/mol, see links in claim 3: confirmed the order of magnitude (the same search also returned a textbook-exercise value, 62.2 kJ/mol, so published values vary with conditions).
• Ultra-weak photon emission intensity and excitation energies [web] https://ar5iv.arxiv.org/html/1502.07316 and https://www.materialsscience.pwr.edu.pl/bi/vol24no4/articles/IMIM_14.pdf: snippets only, pages not opened.
• Rahn 1934 J. Bacteriol. 28:153 (PMC533662): fetch blocked by reCAPTCHA; not read. The Investigator's description stands as a lead.
• Facts not checked on the web: nerve speed range and heat of neutralisation (high-confidence [memory], used only as context), glioma/sarcoma classification [memory, high], Hodgkin and Huxley [memory, high].

Summary of items the Investigator found that I keep in the record

Hollaender and Claus 1937 (negative, NRC Bulletin 100) and the Gurwitsch-tradition objection to it (Volodyaev and Beloussov 2015, interested parties); Quickenden 1989 and Wainwright 1997; Rahn's CSH Symposium paper (1934) and his own statement that the radiation "is not of mysterious, but of physicochemical, origin" [web] https://symposium.cshlp.org/content/2/226.extract; Rahn's 1932 AAAS talk as reported by NYT and Science Service; the Soviet continuation of blood-quencher work to 1955 and Naumova et al. 2018; later photon-counting work (Colli and Facchini 1954; Quickenden and Que Hee 1974; Popp's biophotons from 1976); Stempell's 1932 book with almost Rahn's title; the Barnes/M. N. Barnes name question; the 1944 HathiTrust record of the book. None was downloaded (research/files/ is empty), so all are known to me only through the Investigator's notes.

Verdict table