BERM follows one explanatory chain from physical premises through biological state, motivation and interaction to population and civilizational dynamics.
BERM explains how electromagnetic conditions, acting through biological receivers and their changing state, propagate into endocrine coordination, neural function, motivation and action. Interactions among individuals then produce population patterns, institutions and long-term civilizational dynamics.
The same biological state helps determine both what a person can do and what feels rewarding, effortful or worth pursuing. Deliberation, expressed reasons and learning belong to this process and feed back into later choices. The model follows these links through shared state variables, timing, distributions and interaction.
The explanatory model spans this whole chain. Its reproductive calculations are specific applications: they connect organ function, couple outcomes and demographic terms to age-specific fertility and TFR. FieldState supplies measurements and estimates of the upstream physical state. The physics section identifies the Lindgren premise, its geometric consequences and the explicit receiving bridge used to continue into biology.
BERM is the explanatory, derivational and prediction model. FieldState v2 is a separate optional measurement, observation and estimation module — not a model alias or causal root. The locked v17 outputs use a national technology-timing proxy and are not FieldState-calibrated. BERM now derives a conditional formal geometry-to-observable operator; its gauge, scale, tissue kernels and endpoint calibration remain open.
BERM now makes the receiving tissue explicit: its orientation, cofactors, redox state, hormone responsiveness, biological phase and recovery history shape the response to a physical input. Starting from Lindgren’s 2025 tensor ansatz, a named, state-dependent biological coupling connects the geometric perturbation to receptor activity. FieldState supplies observations used to estimate that physical state.
The downstream chain follows chemical memory → redox and clock function → hormone–tissue timing → functional reproductive gates → successful encounters and population distributions. Component experiments anchor these transitions; composing them into a common BERM route is the model synthesis. The coordination explorer shows how timing, recovery and individual differences change outcomes under explicitly illustrative assumptions.
The 2025 Lindgren ansatz is BERM's theoretical premise. BERM conditionally derives the formal response operator by adding minimal matter–metric coupling and response theory. Lindgren does not supply the gauge prescription, scale, tissue kernels, SHBG/AR/ZIP9 coefficients or human endpoint calibration; those remain explicit model questions.
In standard physics, the electromagnetic field is a separate entity that propagates through spacetime. In Lindgren's geometric model, the EM field is encoded directly in the metric tensor:
g_μν = η_μν + κ A_μ A_νwhere η_μν is the flat Minkowski metric, A_μ is the electromagnetic four-potential and κ is an explicit dimensional coupling scale. From this premise BERM derives δg exactly. A tissue response follows only conditionally through a named response kernel; downstream biology is not an automatic consequence of the metric.
The definition F=dA gives dF=0 as a homogeneous Maxwell identity. The EH variation, Weyl connection and restricted GME residual are separate conditional contracts; checking them together does not by itself derive ∇_μF^μν=0 or ∇_μF^μν=J^ν. The divergence equation needs an explicit dynamical derivation and current identification. The mathematics page displays these gates separately.
[L1 + L0/L2 reduction]
For an explicitly normalized positive-norm mode, χ_geo(ρ)=ρ/√(1+ρ²) is the square-root amplitude of the rank-one inverse-metric correction. That geometric coordinate is derived; interpreting it as a tissue susceptibility or using it to weight the archived v17 technology proxy remains uncalibrated BERM modelling.
ρ² = κ A² ≥ 0, χ_geo(ρ) = ρ / √(1 + ρ²)The coordinate supplies no universal biological selection rule. CRY background, membrane voltage, barrier integrity and technology diffusion remain separate variables with separately testable response functions; the similarly shaped v17 proxy weight is retained only as a legacy comparison.
Electromagnetic fields still obey ordinary superposition. Because the Lindgren ansatz is quadratic in the potential, the induced metric drive contains exact background–perturbation cross terms and a self-term. This establishes mixing in model geometry, not a non-additive biological effect.
A=A₀+Σᵢaᵢ; Δg=κ[Σᵢ(A₀⊗aᵢ+aᵢ⊗A₀)+Σᵢaᵢ⊗aᵢ+Σᵢ<ⱼ(aᵢ⊗aⱼ+aⱼ⊗aᵢ)]With n channels, the exact expansion retains n(n−1)/2 pair terms. Amplitude modulation and two-tone inputs generate low-frequency envelope or difference-frequency terms in a². Their biological response may be positive, null or negative depending on phase, coherence duration, receptor/agonist state, redox state, developmental window and organ transfer. Combined-exposure studies constrain this experiment but do not calibrate the L2 kernel (Juutilainen et al. 2006i).
See full superposition analysis →Response conditions and interaction evidenceBERM imports tissue-specific ion-channel composition, membrane properties and candidate response windows to propose heterogeneous susceptibility. These belong in the tissue kernel Ξ_i and require calibration; they do not follow from χ_geo alone:
| Tissue | Channels | Tissue-kernel candidate | Reason |
|---|---|---|---|
| Testes (Leydig cells) | Cav3.2 (T-type), high density | Very high | Window current at rest; StAR protein Ca²⁺-dependent |
| Hypothalamus | Cav3.1, Cav3.3 | Very high | Synaptic vesicle release via synaptotagmin 1 |
| Hippocampus | Cav3.2, Cav1.3 | High | LTP/LTD Ca²⁺-dependent; neurogenesis zone |
| Retina (blue cones) | CRY1/CRY2 + TRPC1 | High (light-dependent) | Radical pair magnetoreception; FAD-dependent |
| SA node (heart) | Cav1.3, Cav3.1 | Moderate-high | Pacemaker current; low-threshold activation |
| Skeletal muscle | Cav1.2 (L-type) | Low at rest | High activation threshold (−30 mV); significant only during action potentials |
Four evidence lines motivate background-dependent tests; none calibrates χ_geo as tissue susceptibility
Reported cardiovascular-mortality associations with geomagnetic storm intensity motivate a lagged background × endpoint test. They do not identify χ_geo as the biological mediator or calibrate its tissue kernel (Venclovienė et al. 2022i).
Geographic variation in cardiovascular disease can motivate a pre-specified geomagnetic interaction test, but latitude has many competing pathways and cannot by itself identify a BERM response coefficient (Feigin et al. 2014i).
Reported HRV–Kp covariation supplies a candidate autonomic endpoint for matched field and physiology measurements. It is consistency evidence, not a derivation of χ_geo-mediated tissue coupling (McCrary et al. 2021i).
A systematic review of combined exposures motivates tests of interaction and waveform dependence. Heterogeneous biological non-additivity does not directly confirm the specific Lindgren quadratic term or the BERM tissue kernel (Juutilainen et al. 2006i).
If a calibrated tissue kernel depends on geomagnetic background, solar activity could generate measurable biological oscillations. The observations below motivate that hypothesis; they do not identify χ_geo as the biological response or establish causality.
Reported birth-rate oscillations in the USA and New Zealand have been compared with the 11-year solar cycle. In BERM this is a candidate natural-experiment signature, not evidence that χ_geo rises or that conception changes through the proposed tissue kernel (Lehrer & Lehrer 2017i).
BERM proposes the testable chain solar activity → geomagnetic disturbance → melatonin change → GnRH-pulse change → conception-rate modulation. A lagged design with photoperiod and secular controls could test this chain; the cycle alone does not separate geomagnetic effects from other periodic covariates.
A cohort of 237,000 patients found birth-month associations with several later diagnoses (Boland et al. 2015i). This does not identify geomagnetic exposure or χ_geo; BERM treats it only as motivation for a study that measures gestational field, season, infection, nutrition and pollution separately.
Developmental timing is a plausible susceptibility window, but a geomagnetic → VGCC/CRY → organogenesis path remains an uncalibrated BERM proposition rather than a result of the birth-month study.
The seasonal amplitude of Greek birth rates reportedly decreased between 1960 and 1992 (Lerchl 1998i). Electrification is one BERM candidate explanation among urbanization, contraception, climate control and social timing; the observation does not measure an EMF or χ response.
Discriminating prediction: after controlling those alternatives, later electrification should predict later dampening. This is a prospective model test, not a description of current sub-Saharan populations.
Biological systems interact with electromagnetic fields across three distinct frequency bands, each with different physical mechanisms and biological targets.
ELF modulation envelopes on RF carriers activate pathway B via spin dynamics; RF carrier energy itself acts through pathway A (electric field component).
BERM separates the derived χ_geo coordinate from candidate biological response functions. They cannot be multiplied or interpreted as total susceptibility until an endpoint-specific response kernel is measured.
The left-hand bounded shape is derived as a normalized inverse-metric coordinate; it is not a biological susceptibility. The right-hand spin function is imported biology. BERM's formal response operator is conditional, while each tissue kernel and endpoint coefficient remains open.
BERM composes component results when they identify the same mediator, point in a compatible direction and use independent methods. Their convergence constrains the route more strongly than isolated citations while preserving the distinction between a component result and a fully calibrated endpoint coefficient.
same mediator + compatible direction + independent method + predicted moderator → stronger causal-route constraint
Testosterone × cortisol forms a joint endocrine gate for behaviour; usable androgen state is more informative than total testosterone alone.
CRY-dependent magnetic sensitivity, redox modulation and reproductive clock control compose a second serial route into HPA–HPG regulation.
Channel blockers, sperm calcium functions and mTOR inhibition triangulate adjacent causal control points in the reproductive branch.
Coherence, field orientation, temperature history and laboratory state explain why the same nominal exposure can change magnitude or sign.
Vmem, calcium and mTOR form one measurable control interface between field response, cell fate and fertilization capacity.
Conserved CRY dependence supplies a receptor-stratified comparative rule for ecological tests across species and life stages.
Individual susceptibility is represented as a continuous threshold distribution, so broad null averages and a responsive tail can coexist in one model.
| 1 | Geometric operator | Lindgren → δg_μν → Q^μν | implemented in the tensor derivation |
| 2 | Receptor module | bound ion → phase modulation → Bessel → amplitude | this module |
| 3 | Tissue transfer | local response → biological endpoint | coefficients not estimated |
| 4 | Population aggregation | ASFR → TFR | coefficients not estimated |
The receptor module returns a local response candidate and its provenance. It has no TFR output: without a transfer model that names attempts, timing, conception, pregnancy continuation and compensating behaviour, a relative change in a receptor response cannot be placed as a TFR coefficient.
| From | To | Named assumption | Level |
|---|---|---|---|
| Lindgren metric g_μν = η_μν + κA_μA_ν | δg_μν with an ω and a 2ω component | Algebraic consequence of A = Ā + a cos ωt | L |
| δg_μν | Bound-ion Hamiltonian with a modulated gap | Bound states exist and mix; isotropic binding | L1 |
| Modulated gap | Jacobi–Anger expansion → Bessel sidebands J_k(z) | Sufficient phase memory; generic to any driven two-level system | L1 |
| Bessel argument z = λ·(f_c/f)·(b/B₀) | Amplitude windows b_max, b_null | A readout exists and is fixed separately per endpoint | L1 |
| Binding-site relaxation τ | λ(f,τ) from 1 to 2 | c = γ — the candidate's own choice, not fixed by the ansatz | KANDIDAATTI |
| λ at 24 Hz with τ = 20 ms | Calmodulin N-terminal time scale | Numerical closeness of two coefficients; not a measurement of a field response | L* |
Every number below is computed live from lib/mechanism.ts in the fixed order u → λ → s → z → J₁. Inputs carry their source. Bauréus et al. (2003)i · Park et al. (2008)i
| f | 24 Hz | Koch 2003 exposure condition [observation] |
| f_c | 25.2 Hz | synthesis §15, Ca²⁺ interpretation [candidate input] |
| τ | 20 ms | Park et al. 2008, calmodulin N-terminal, chemical Ca²⁺ step [observation] |
| B₀ | 37 µT | Koch 2003, upper end of the 27–37 µT range [observation] |
| c/γ | 1 | candidate assumption |
| u = 2πfτ | 3.01593 | L1 |
| λ(f,τ) | 1.92428 | candidate (c = γ) |
| s = λ·f_c/f | 2.02049 | argument coefficient |
| |s − 2| / 2 | 1.02 % | coefficient comparison, not a data fit |
| b_max (first J₁ maximum) | 33.7 µT | L1 + conditional |
| b_null (first J₁ null) | 70.2 µT | L1 + conditional |
| τ required for s = 2 exactly | 17.6 ms | reverse calculation — not a blind prediction |
| bare Ca²⁺ cyclotron frequency at 37 µT | 28.35 Hz | for comparison with the 25.2 Hz input |
The 25.2 Hz reference frequency is taken from the synthesis. Bare Ca²⁺ at 37 µT gives 28.35 Hz; 25.2 Hz corresponds to B₀ ≈ 32.9 µT for the bare ion, which is inside Koch's reported static-field range. The calmodulin closeness is a numerical compatibility of two argument coefficients and motivates the receptor-structure test below. It does not identify calmodulin as an EMF receptor and it did not predict the 20 ms value.
Each claim in the chain with its proper status. The registry extends the tensor-derivation contracts (Maxwell three-premise gate, χ at L1 + L0/L2); it does not replace them.
| Claim | Status | Description | Level | Conditions / note |
|---|---|---|---|---|
| Metric cross terms and the 2ω component | Algebraic consequence | Algebraic consequence of the declared ansatz: with A = Ā + a cos ωt, the product AμAν carries a cross term at ω and a squared term at 2ω. | L1L1 — follows from the declared ansatz | |
| Bessel sidebands | Conditional derivation | Conditional derivation from a sinusoidally modulated energy gap: the phase integral gives exp[iz sin ωt] = Σ J_k(z) e^{ikωt} (Jacobi–Anger). The same structure arises in any driven two-level system and is not specific to the metric ansatz. | L1L1 — follows from the declared ansatz | the bound states exist; the states mix; phase memory is long enough; a measurable readout mechanism existsAshhab et al. (2007)i |
| Bound-mode q/m separation | Conditional derivation | Result in the isotropic harmonic binding model: the reference frequency f_c = |q|B₀/(2πm) separates bound modes by charge-to-mass ratio. | L1L1 — follows from the declared ansatz | isotropic bindingEngström et al. (2004)i |
| c = γ | Candidate assumption | Additional assumption of the candidate, not a constant fixed by Lindgren's ansatz. | KANDIDAATTICandidate — added assumption, not derived | Chosen so that the high-frequency limit of λ equals 2; not a universal constant. |
| The 17.6 ms calmodulin connection | Numerical compatibility | Value obtained by demanding the IPR argument coefficient s = 2 in reverse; it is not an independent measurement. | NUMEERINEN YHTEENSOPIVUUSNumerical compatibility — not a measurement | Not an independent measurement; ~1 % from the measured 20 ms. Motivates the receptor-structure test, nothing more. |
| Calmodulin 20 ms | Observation | Measured time constant of the N-terminal conformational change under a chemical Ca²⁺ step in a microfluidic mixer. | HAVAINTOObservation — measured under stated conditions | N-terminal domain, chemical conditions; not a magnetic-field response and does not identify calmodulin as an EMF receptor.Park et al. (2008)i |
| ~1 % match of the argument coefficients | Numerical compatibility | Comparison of s = λ·f_c/f = 2.020 against the IPR coefficient 2; not the accuracy of a raw-data model. | NUMEERINEN YHTEENSOPIVUUSNumerical compatibility — not a measurement | |
| The IPR model's reported fit | Observation | In the 1995 report the predictive model's R² was about 0.70 and the refit 0.71. Adding the hydrogen component raised the figures, but the added component is not a parameter-free confirmation. | HAVAINTOObservation — measured under stated conditions | Supersedes an earlier R² = 0.85 entry taken from the previous synthesis, which the 2026-09-07 source document does not support. The amplitude windows are a testable structure, not a free frequency list: J₁'s first maximum at argument 1.84118 and first null at 3.83171 correspond, at B₀ = 37 µT, to AC peak amplitudes of about 34.1 and 70.9 µT, against 33.7 and 70.2 µT from the relaxation candidate. Those are theory-to-theory comparisons on one parametrisation, not a new amplitude sweep.Blackman et al. (1994)i, Blackman et al. (1995)i, Blackman et al. (1998)i |
| The 165 Hz gap | Open question | Two non-significant results; not a confirmed Bessel null. | AVOINOpen | |
| TFR attribution share | Open question | The mechanism work does not yet determine a population attribution. Layer 3 and 4 transfer coefficients have not been estimated; no TFR figure is produced from the receptor layer. | AVOINOpen | Layer 3–4 transfer coefficients not estimated. |
Six open tests. A positive Bessel-structure result is compatible with phase modulation and is not specific to the metric ansatz; only a prediction whose scale, direction or multi-wave dependence is computed from the ansatz and differs from a competing model at the same biological parameters would be Lindgren-specific.
| Test | Prediction | Discriminates | Varied → measured (controlled) | Ansatz-specific |
|---|---|---|---|---|
| Frequency and B₀ togetherAVOIN TESTI | Normalised amplitude nodes (b_null/B₀) shift with λ(f, τ) when f and B₀ are varied jointly. | Finite τ versus a fixed IPR argument | f, B0 → b_null_normalized(ion, preparation, exposure time) | no |
| Receptor conformational changeAVOIN TESTI | An independently measured τ predicts the amplitude-window shift. | Same biological coordinate versus a mere time-scale analogy | tau_measured → amplitude_window_shift(ion, B0, f) | no |
| Two-frequency phaseAVOIN TESTI | The response depends on the relative phase of two drives at equal total power and spectrum. | Interference versus spectral power alone | f1, f2, phase_difference → response_amplitude(total_power, ion, preparation) | yes |
| Selective blockadeAVOIN TESTI | The response disappears at a defined point in the chain when that point is blocked. | Channel, store or downstream transduction | blocker_type, blocker_target → response_suppression(f, B0, b, exposure time) | no |
| Isotope exchangeAVOIN TESTI | The frequency scale shifts by the defined mass effect: f_c ∝ 1/m at fixed charge. | Ion-mass interpretation versus another receptor | isotope_mass → f_c_shift(chemistry, concentration, endpoint)Caveat: Hydrogen-isotope exchange can change kinetics through routes other than mass. | no |
| Dose and recoveryAVOIN TESTI | One common recovery law across different exposure schedules. | Acute adaptation versus slower biological change | exposure_duration, recovery_time → response_magnitude(f, B0, b, preparation) | no |
Gavoçi et al. (2013)i — TEA-sensitive outward K⁺ currents, BE(2)C. Fields at K⁺ IPR conditions did not change the currents. A tissue-specific coupling offered to absorb this needs its own independent measurement; a free coefficient set to zero per exception would remove the model's predictive content.
The premise fixes an algebraic structure. What reaches a receptor depends on the time average and the projection, and those are separate calculations. Each result below states both what it establishes and what it does not.
g_μν = η_μν + κA_μA_ν, A = A₀ + a Δg_μν = κ(A₀_μ a_ν + a_μ A₀_ν + a_μ a_ν)
Establishes: For the stated ansatz and decomposition the cross structure is exact, so the background does not generally cancel out of a geometric exposure-versus-control difference. This replaces an earlier need to assume some background dependence: the structure follows from the premise.
Does not establish: κ carries the normalisation and A is the four-potential, not a meter's E or B reading. The biological consequence still has to be formed with a receptor operator; writing δH = Q^μν δg_μν makes the missing coupling visible without fixing its value.
a(t) = Re[u(t)e^{iω_c t}], ⟨a⟩ = 0
⟨a_μ a_ν⟩ = ½ Re[u_μ u_ν*]Establishes: Terms linear in a static A₀ average to zero over a zero-mean carrier, while the quadratic part survives. A slow power envelope can therefore persist in a quadratic driver even where the static background cross term vanishes in this approximation.
Does not establish: This is a correction to the broader earlier wording: not every algebraic cross product is automatically present at a biological timescale. The background can still act through the receptor's energy gap, its orientation or a second coherent frequency component — those are separate routes.
N components → N(N−1)/2 distinct cross pairs product of two sines → sum and difference frequency
Establishes: A difference frequency falling inside the receptor's band can be formed from two components, neither of which feeds that band alone. A single-frequency test therefore does not cover multi-source exposure.
Does not establish: The number of sources fixes neither the sign of the cross terms, nor phase coherence, tensor projection or averaging. A quadratic count of terms is not a quadratic growth in damage. A cross term that averages to zero can still leave temporal variation: a vanishing mean and vanishing variance are different things.
a_μ a_ν is identical for a and −a g(A₀+a) − g(A₀−a) = 2κ(A₀_μ a_ν + a_μ A₀_ν)
Establishes: A purely quadratic driver cannot distinguish a field from its reverse, so a direction-dependent response requires the background cross term or another symmetry-breaking structure. The up–down difference reported in the PEMF work makes such a bound experimentally relevant.
Does not establish: Reversing a coil is not necessarily a clean sign flip of every local potential component: the induced electric field, cell orientation and geometry have to be checked too. The discriminating design reverses background and test exposure separately in pre-computed combinations, and in the radiofrequency case the carrier-averaging bound above applies as well.
F(q) = √(1+q²), χ(q) = F′(q) = q/√(1+q²), χ′(q) = (1+q²)^(−3/2) ΔF ≈ χ(q)δq → δq as q grows Δχ ≈ (1+q²)^(−3/2) δq → 0
Establishes: If the biological quantity reads F, a small additional exposure does not vanish in this limit. If it reads χ, the additional response can vanish. The interpretation of a laboratory saturation observation therefore depends on which quantity the biological process reads — the two cases are mathematically different, not two wordings of one claim.
Does not establish: q is a normalised geometric variable; identifying it with a chronic radiofrequency dose needs a separate mapping. So "the non-linearity is in the premise" is correct, while "the premise proves laboratory saturation" does not follow from it. The decisive choice sits in the biological observable.
These results are computed from the published radio standards. Several of them narrow the model: two constrain what may be inferred from a power spectrum, and one is a firm identifiability limit.
TCH/F: slot 3/5200 s, frame 3/650 s, 26-frame multiframe 120 ms
c_k = sinc(k/208)·e^{−iπk/208}·Σ_{j=0..24} e^{−i2πkj/26} / 25, f_k = k/0.120 s
25 Hz is the third multiframe harmonic: cosine amplitude 0.0799726 (7.9973% of mean power)
216.667 Hz cosine amplitude ≈ 1.94899Establishes: For the standard's own TCH/F example with a full one-slot burst in 25 frames and one idle frame, the 25 Hz power component is exact and calculable rather than assumed.
Does not establish: 7.9973% is a share of mean power, not of total energy. Ramps, falls and guard periods are idealised. The result covers this standard example, not every GSM mode and not any individual study's recorded signal. Eight percent is not a health threshold.
ETSI et al. (2022)i
idle frame filled + burst level reduced to hold mean power → 25 Hz component vanishes; normalised 216.667 Hz component survives
Establishes: The control separates the multiframe structure's contribution from the plain TDMA frame period, at equal mean power. That is a concrete experimental advance over a bare GSM-versus-UMTS comparison.
Does not establish: The control also removes other harmonics of the 8.333 Hz multiframe structure, so a positive biological difference would localise the effect to that component set rather than to 25 Hz alone. The sharper follow-up selectively attenuates and re-adds the chosen intensity component while documenting mean, temperature and the other changing spectral parts.
ETSI et al. (2022)i
GSM example: R = 0.0545775 sinusoidal 25 Hz power modulation p(t) = 1 + m_P cos(2π·25t) → R = m_P√[W(25)/2] threshold to exceed the GSM point at equal carrier and mean local power: m_P = 0.0799726
Establishes: A suitable slow power variation added to a UMTS-type carrier can exceed this GSM signal's window score without any UMTS-specific coefficient. That is a constructive answer to how strong UMTS responses are possible at all.
Does not establish: Those sine waves are constructed test signals. The calculation does not show that this particular variation was present in the experiment that produced a strong biological UMTS result. The ordering of ideal signals has been computed; ordering tests on real experiments number zero.
field envelope a₀[1 + m_E cos(Ωt)] → m_P = 2m_E/(1 + m_E²/2) m_P = 0.0799726 → m_E = 0.0400183 (≈ 4.002%) carrier scaling of a conditional A² driver: ⟨E²⟩R/(2πf_c)² 1947.4 MHz vs 915 MHz comparison → equivalent power-modulation threshold 36.23%, not 8%
Establishes: Field-amplitude and power percentages are not the same number, and squaring additionally produces a second harmonic. The earlier threshold is now tied unambiguously to the normalisation, the carrier and the definition of modulation.
Does not establish: Changing the carrier changes the condition. Moving from SAR to E² requires the medium's properties. The A² driver is accepted conditionally here, not established.
u_φ(t) = 1 + e^{iΩt} + e^{i(2Ωt+φ)}, Ω = 2π·25
all φ: the three RF spectral lines carry powers 1, 1, 1
|u_φ|²/3 = 1 + (2/3)Re[(1+e^{iφ})e^{iΩt} + e^{iφ}e^{i2Ωt}]
φ = 0 → 25 Hz cosine amplitude 4/3; φ = π → exactly zero (50 Hz component remains)
R ≈ 0.909936 vs 2.39 × 10⁻⁵⁹Establishes: A constructive counterexample: the RF power spectrum and the mean power do not by themselves determine the spectrum of the slow intensity components. This is a firm identifiability limit, not a caution.
Does not establish: Phase-bearing I/Q data, a measured power time trace or a sufficiently complete stochastic model is needed in addition. For a random signal the intensity correlation involves fourth-order field statistics. Missing exposure information cannot be uniquely recovered even by very good reasoning — so every technology comparison resting on a power spectrum alone is conditional on phase structure not changing the slow envelope in that case.
exact 10 ms repetition through a time-stable path → intensity spectral lines at integer multiples of 100 Hz non-negative periodic power normalised to mean one: |c_k| ≤ 1 window score upper bound: log₁₀ R ≈ −535.94
Establishes: The narrow 25.4 Hz window receives only the tail of the Gaussian weight, so the reported null results for a repeating NR frame are consistent with a very small score. A negative result of this kind is informative: it bounds the prediction.
Does not establish: The local field around a moving animal need not be 10 ms periodic. The result does not cover all 5G traffic or all biological routes: real NR traffic carries scheduling, power control and beam management, whose slow components need measured traffic records rather than an idealised signal. One technology generation can contain very different slow exposure components.
ETSI et al. (2020)i
if window centre and width scale as B/37 µT, GSM harmonics 16.667 / 25 / 33.333 / 41.667 Hz hit the centre peak at B₀ = 24.28 / 36.42 / 48.56 / 60.70 µT
Establishes: One of the chain's most discriminating new predictions: the same signal can produce a changing response when only the background field moves the receptor's window. These alignments are computed in advance rather than fitted to laboratory backgrounds after the fact.
Does not establish: The σ versus FWHM reading of the width changes the depth of the minima, so succeeding at one field value is not enough. A strong test measures the whole sweep and compares peak positions and widths against the same pre-registered rule. A multi-peak receiver is its own prediction and must not be swapped in mid-comparison.
The locked window, its normalisation and the score definition, stated so the numbers are reproducible.
| Signal or condition | Window score R | Reading |
|---|---|---|
| GSM, 25 active frames + one idle | 0.054577 | Reference point of the locked window |
| Same RF spectrum, phase φ = 0 | 0.909936 | Identical power spectrum, large score |
| Same RF spectrum, phase φ = π | 2.39 × 10⁻⁵⁹ | Identical power spectrum, score vanishes |
| Exactly periodic 10 ms NR frame | log₁₀ R ≈ −535.94 | Upper bound from periodicity alone |
R is a window-weighted RMS relative to mean local power. Equal weights across the 20.9 / 25.4 / 30.2 Hz peaks and the use of W as a power weight are choices of this calculation. The protocol was locked locally before the numerical run, but earlier biological results were already known: this was not an independent pre-registration or a blinded analysis.
The standard this chain holds itself to. The left column is what the evidence does not support; the right is what it does. This is documentation, not a list of site corrections — a check on 2026-09-07 found none of the left-column claims on the site.
| Not supported | Now justified | Findings |
|---|---|---|
| 530 studies confirm a reproductive effect. | A re-analysed genotoxicity map of 530 publications locates technology and design differences. | 01–03 |
| UMTS is biologically weak. | UMTS publications reported fewer DNA effects; differences in the real signal and in the endpoint have to be separated. | 02–04 |
| Present-day animals do not respond because of prior exposure. | No general temporal attenuation was found; the effect of measured history needs a baseline-plus-additional-response test. | 05, 13, 50 |
| The background makes the RF effect larger through cross products. | The cross products follow from the premise; their time average and the receptor's projection decide which driver survives. | 08–11 |
| χ saturation removes the additional effect. | The limit of the additional effect depends on whether the biology reads F, χ or another quantity. | 12 |
| Lindgren leads directly to the IPR coefficient two. | Phase modulation yields the Bessel structure; the coupling and the AC/DC sensitivity ratio need a separate derivation. | 15–18 |
| 17.6 ms is a confirmation measured from biophysics. | 17.6 ms is a reverse-calculated compatibility; Park's roughly 20 ms is a chemical relaxation measured in a different experiment. | 19 |
| 25.4 Hz is the calcium resonance of cells. | The peak is reported in a particular membrane-vesicle and static-field setup; transfer to radiofrequency-exposed cells is a test hypothesis. | 20, 25 |
| 8% modulation is enough for UMTS. | 7.997% concerns the power fundamental on the same carrier; the equivalent field AM is about 4.002%. Changing the carrier changes the condition. | 26–27 |
| The biological driver can be recovered from the RF spectrum. | The RF power spectrum does not by itself identify the slow intensity spectrum; phase or time-trace information is needed. | 28 |
| NR does not hit the response window. | An exactly repeating 10 ms signal through a stable path produces no 25 Hz line. Other NR traffic is a different case. | 29 |
| The strong UMTS responses have now been predicted. | The ordering switch for ideal signals has been calculated; ordering tests on real experiments number zero. | 26, 31–33 |
| The 20-hour memory is chronic damage. | In one UMTS setup a radiofrequency pre-treatment changed a later chemical response protectively across an interval of about 20 hours. | 35 |
| ATG or CRY has been shown to be a general EMF sensor. | The interventions locate necessary parts of the mechanism in defined cell models and exposure classes. | 36–40, 49 |
| HSP70 mediates the bystander effect. | Protection transferred by culture medium has been reported; HSP70 is a candidate mediator. | 38 |
| A field-free sham is always best. | The artificial exposure is removed while documenting the geomagnetic background; a hypomagnetic condition can itself change cell function. | 41 |
| Normal semen parameters rule out a biological problem. | Hyperactivation, CatSper function and fertilisation endpoints can differ from the baseline parameters. | 44–46 |
| Sentinels and drug responses locate the EMF cause. | They bound cultural explanations and cellular routes; EMF's share requires an exposure-specific contrast. | 48, 51–52 |
These are bounded experimental observations, not general assumptions. Each names its exposure class and dose, because results from different classes do not combine into one effect.
Exposure: 1950 MHz UMTS, SAR 0.3 or 1.25 W/kg, SH-SY5Y cells; exposure at culture hours 48–51, menadione challenge at hours 71–72
Locates: The pre-treatment reduced later DNA damage across an interval free of the given radiofrequency exposure. This replaces the general assumption that a cell may have memory with a bounded observation: an earlier treatment changes a later response across a long gap.
Transfer limit: This supports a protective state, not a demonstration of accumulated damage. One post-interval does not fix an exponential time constant τ ≥ 20 h. Radiofrequency alone did not change measured DNA damage or the TRX1, HSF1, HSP70 and PARP1 expression studied in that short setup, so a functional after-effect can fall outside a single baseline gene panel.
Exposure: 20 h UMTS pre-treatment, SH-SY5Y cells; pharmacological autophagy inhibitors and loss of ATG5 or ATG7 function
Locates: The protective effect disappeared under both chemical and genetic interference. Combining the two bounds the mechanism more than a single expression change would. In BERM's terms the history term can be sharpened towards recycling and repair capacity: net damage is formation minus removal, and exposure may change the second of those.
Transfer limit: ATG5/7 dependence does not identify the first EMF receptor. A gene intervention's general effect on growth and chemical sensitivity has to be separated from the radiofrequency interaction, and the existing system allows that separation to be made quantitative.
Exposure: UMTS pre-treatment with a mitomycin C challenge; human lymphocytes and V79 cells; 3-aminobenzamide as the PARP inhibitor
Locates: The inhibitor blocked the adaptive protection, raising a PARP-mediated repair and stress response to a named part of the biological reading. With the autophagy result it supports a model in which one radiofrequency pre-treatment changes the cell's later handling capacity along several connected routes.
Transfer limit: The intervention may act on the formation of the protection or on the handling of the chemical damage; it does not show that EMF acts directly on the PARP protein. The order and necessity of these routes has to be resolved in one system — separate studies do not automatically compose into one proven serial chain.
Exposure: Medium from radiofrequency-exposed SH-SY5Y cells transferred to unexposed recipient cells
Locates: Recipients showed less menadione DNA damage. HSP70 rose in the exposed cells' medium while intracellular HSP70 did not change correspondingly. A biological effect of the exposure can therefore cross the boundary of the directly exposed cell population, which makes shared medium, its exchange timing and conditioning causally relevant to a design.
Transfer limit: HSP70 stays a candidate mediator: a rise alone does not identify it as the cause. The result does not mean that ordinary separate cultures are always contaminated. Fractionation, neutralisation or a restoration experiment would sharpen the mediator.
Exposure: Low-frequency millitesla-class pulsed magnetic field, not 900 MHz – 3.5 GHz environmental radiofrequency
Locates: Silencing cryptochromes in human cells and a Cry1/Cry2 double knockout in mouse cells removed the studied ROS response; human CRY1 restored a field-related behavioural response in a cryptochrome-deficient fly model. A conserved cellular protein can therefore take part in a field response without a navigating animal.
Transfer limit: The cellular ROS result and the fly behavioural rescue are different endpoints. Transfer to a radiofrequency power envelope needs a separate transduction chain: a power envelope is not itself a real magnetic field oscillating at that frequency.
Exposure: 1.5 mT PEMF, ten minutes, C2C12 muscle cells; downward field more effective than upward; 48 h dark culture weakened the response
Locates: Adding CRY2 strengthened and silencing it weakened the response; silencing RFK weakened the FAD-related response and the direction discrimination. CRY2 and TRPC1 appeared in the same experimentally observed protein complex. TRPC1 is not a voltage-gated calcium channel, so this opens a measurable alternative channel route beside the VGCC branch.
Transfer limit: Co-precipitation alone does not show direct binding of two purified molecules. The general claim that darkness heightens field sensitivity has to be bounded per tissue. The measured response concerned the progression of myogenesis, so it is not recorded as an observation of reproductive harm.
Exposure: Mouse, from about 55 µT to a hypomagnetic environment of about 0.29 µT
Locates: The change reduced hippocampal neural stem cell ROS and neurogenesis, and returning to the geomagnetic field or altering ROS pharmacologically restored responses. This sharpens the control: the smallest possible total field is not automatically biologically neutral, and the artificial alternating component has to be separated from the geomagnetic background.
Transfer limit: The finding concerned a named cell population; it does not mean a general ROS reduction across the organism. It does not fix one optimal field for all tissues and does not show an effect of modern radiofrequency exposure.
Exposure: Chicken eggs incubated 21 days in a 50 or 60 Hz electric field; post-hatch brain-tissue Ca²⁺ efflux test
Locates: After 60 Hz developmental exposure a response appeared at 50 Hz but not at 60 Hz; after 50 Hz developmental exposure neither test frequency produced a response. The history term therefore need not be described only as a fall in overall sensitivity: it can change the shape of the response window, or which frequency the tissue answers. This is direct experimental grounds for controlling developmental history in replication designs.
Transfer limit: The experiment concerned development within one individual and does not show a change inherited across generations. The exposure must not be described as merely a magnetic field: in the original setup the electric field was the central defined quantity.
Exposure: Broad replication across modulated wireless signals; MRC-5 cells at early versus later passage
Locates: An ELF response present at early passage disappeared at a later one. In the same work most radiofrequency comparisons were negative, and a small early difference seen with UMTS and EMS did not persist under longer exposure. A cell line's name is therefore not a guarantee of an unchanged receiver, which identifies a measurable biological moderator.
Transfer limit: Passage can mean ageing, differentiation, selection and much else. Accumulated EMF damage was not separated out as the cause there, and the same observation fits several state-dependent models, so EMF history needs its own documentation and intervention.
None of these three is an EMF experiment showing a fertility effect. They establish that baseline semen parameters do not cover functional capacity, which is what an EMF claim would have to measure.
Exposure: Genetic study of 2286 men — not an EMF experiment
Locates: Nine CatSper function deficiencies were found, mostly linked to CATSPER2. Ordinary semen parameters could be normal while sperm failed to hyperactivate as fertilisation requires, and ICSI could bypass the defect where natural conception and ordinary IVF failed. This confirms the separation of functional reproductive capacity from baseline parameters.
Transfer limit: The genetic study does not show that EMF causes the identified deficiencies. A possible EMF effect on CatSper would have to appear in a named Ca²⁺ or hyperactivation endpoint and a suitable fertilisation assay.
Exposure: Mouse, 905 MHz, about 2.2 W/kg, 12 h per day, up to five weeks
Locates: Exposure was associated with sperm ROS and DNA changes and, in the longest arm, reduced motility, while IVF and early embryo development were not impaired in the measured experiments. That fits a reproductive-reserve reading in which a biomarker moves before measured function does, and it shows why "fertility normal" and "biological effect observed" can both be correct about different endpoints.
Transfer limit: It does not show a later exhaustion or collapse of the reserve. A 905 MHz carrier must not automatically be named a verified GSM waveform. The useful prediction is a temporal and functional ordering, which should be measured in one system.
Exposure: Rat, 150 days of phone exposure; comparison points at 50 and 100 days
Locates: Sperm and testis changes appeared at 150 days but not correspondingly at 50 or 100 days, and blocking the SPOCK3 rise mitigated the barrier and sperm changes and the MMP14–MMP2 deviations. That separates a slow disturbance of the tissue environment from an acute duct dysfunction, and the named intervention chain strengthens the biological mechanism in this experiment.
Transfer limit: The observation times bound when the change appeared in that series but do not give a precise threshold such as 120 days. A "4G" label does not verify an LTE signal. The mating result of 13/15 against 10/14 did not show a clear fertility collapse, and an exposure-specific explanation needs better dosimetry and waveform.
What each evidence line actually bears on. The rows are kept apart deliberately.
| Evidence | Exposure or intervention | What it directly bears on |
|---|---|---|
| Koch / Blackman IPRBauréus et al. (2003)i · Blackman et al. (1994)i | DC background + weak ELF, defined geometry | The reported frequency and amplitude windows |
| Sherrard / IversenSherrard et al. (2018)i · Iversen et al. (2025)i | Millitesla-class PEMF | CRY, redox and channel dependence under those conditions |
| ZhangZhang et al. (2021)i | Strong reduction of the geomagnetic background | Neural stem cell redox and neurogenesis |
| Sannino / ZeniSannino et al. (2024)i · Sannino et al. (2022)i · Zeni et al. (2021)i | 1950 MHz radiofrequency pre-treatment | Adaptation, repair, autophagy and the medium-borne response |
| Houston / YuHouston et al. (2019)i · Yu et al. (2020)i | Radiofrequency or phone exposure in animals | Sperm biology and the slow tissue-barrier branch |
| Young / OzilYoung et al. (2024)i · Ozil et al. (2006)i | Genetic or Ca²⁺-function intervention | Functional mechanism of reproduction; no EMF cause |
| Kirson TTFieldsKirson et al. (2004)i | Intermediate-frequency electric fields | Field intervention on cell division in a therapy-type setting |
| Publication map | Varying radiofrequency experiments and study classes | Relations in result reporting; no shared measured waveform |
A shared cellular route may connect rows as a hypothesis. The rows' exposure physics does not become the same because one of them involves Ca²⁺, ROS or the same frequency number.
ṡ = U(exposure, s) − V(s) Ḋ = P(exposure, s) − R(D, s) C = G(D, s, tissue)
The adaptive response, the ATG dependence, the endpoint decoupling in the mouse work and the tissue-barrier finding can all be described with at least these three separate quantities. One damage variable is too coarse: at one instant the same D can arise from high formation with high repair or from low values of both. Only a temporal measurement plus a repair intervention separates those.
Bound: This is not a new fitted BERM version and not a free way to explain any result. It does not replace the BioCap integral, which is unchanged: BioCap is a simplification that merges s, D and C into one quantity. The three-variable form is a proposal whose value is that it makes saturation, adaptation and selection separable — and it is a shared, testable consequence of experiments already performed.
I = (Y_RF+K − Y_RF) − (Y_K − Y_sham)
All four groups are required: sham, RF, chemical, RF + chemical.
This separates a radiofrequency-modified chemical response from the separate baseline effects of the field and the chemical. A shared ROS or Ca²⁺ route then does not mask the exposure's share: the EMF effect is evaluated as a distinct contrast inside the same biological machinery. The same structure works with a gene or drug intervention, comparing I across intervention levels.
Bound: The scale has to be chosen in advance: additive and relative interaction are not the same claim. RF + chemical against chemical alone does not yet locate the primary sensor. The Sannino and Luukkonen datasets suit this particularly well if the group data for the independent replicates can be obtained.
| Role | What changes | Example test |
|---|---|---|
| Receiver | The field's first proximal response changes | CRY knockout plus a proximal-response measurement |
| Mediator | The later ROS or Ca²⁺ signal changes | ATG knockout plus a ROS measurement |
| Enabler | Repair or the final tissue function changes | PARP inhibition plus a DNA-damage measurement |
CRY2/RFK/TRPC1, ATG5/7, PARP and SPOCK3 are not automatically successive parts of one chain: they were observed in different cells, exposure classes and endpoints. Being necessary for one response does not by itself mean recognising the field. The strong new combination would measure the same proximal response and the later protection under both CRY and ATG interventions; until such a comparison exists, a sensor-to-autophagy serial link stays a candidate even though each branch has separate evidence.
| Mechanism | Separable prediction |
|---|---|
| Damage saturation | A weakened baseline together with a reduced additional response, if the instrument reads the same damage load |
| Protective adaptation | Improved challenge tolerance and often a temporal recovery |
| Sensitisation | A larger additional response |
| Selection | The population's composition or response distribution changes, not necessarily the individual cell |
These can be assessed from existing studies given sham groups' absolute levels, response distributions, passage and lineage data, and measurement times. A published yes/no result is not enough. A chronically exposed control is therefore not one numerical correction factor. Making the alternatives distinguishable is what strengthens the history question — and it also prevents flipping a result's sign afterwards under the same word "saturation".
| Dataset | Information needed | What it would settle |
|---|---|---|
| Sannino LTE–CW | MATLAB code, I/Q buffer, repetition length, local calibration | Is the protective LTE signal's window score larger than CW's? |
| Belyaev UMTS/GSM | Test modes, a fast power or I/Q recording, local geometry | Does the same score predict the strong UMTS response and the 905/915 difference? |
| Schuermann | The identified supplementary package and generator specifications | How did 217 / basic / talk / UMTS actually differ from each other? |
| Sannino / Zeni | Group data for the independent replicates, and the schedules | The RF × chemical × repair and medium-transfer contrasts, with uncertainty |
| History studies | Sham baselines, passage, lineage, culture state | Damage, adaptation, sensitisation and selection told apart |
This information may sit in the authors' existing archives even where it is not in the published article. It was not obtained here and the authors were not contacted. A standard or a patent does not uniquely substitute for the file an experiment actually used.
For a raw time trace the current programme requires at least 200 samples per second over a 20-second span, calibration, and controlled anti-aliasing. Those are minimum conditions for the calculation, not guarantees that the whole experiment is represented. Describing slow power control, movement and long-term exposure needs a recording that spans their own timescales: a five-second temperature log is not an adequate 25 Hz exposure measurement.
Direction: The strongest shared direction is that predicting a biological response can depend on the local temporal field structure, the background field, the receiver's state, the exposure history, and the chosen measurement time and endpoint. Some of these variables have been shown experimentally in particular systems, and there is now a calculable candidate for combining them. That is considerably more precise support for BERM's research programme than one general EMF–disease association.
These intermediate steps are not replaced by the number of cell studies, and no unjustified percentage share of the fertility decline is calculated from the present material.
BERM applies a biologically reductionist, compositional hypothesis from molecular and endocrine states through individual behaviour to population aggregates. The chain below states the proposed propagation from physical input to civilizational outcome. Evidence for separate links can constrain it, but the full multiscale chain is not empirically closed and aggregate political outcomes are not read back as individual hormone measurements.
Geomagnetic and anthropogenic fields are measured as physical inputs. BERM then applies an endpoint-specific response kernel; FieldState does not supply that biological response.
Anthropogenic fields (ELF, IF, RF) perturb the geometric background, altering the spacetime metric biology operates within
Voltage-gated calcium channels — especially T-type (Cav3) at bifurcation point — respond to field perturbation via Schwan amplification
Intracellular calcium signaling disrupted: CaMKII activation, mitochondrial ROS, NF-κB inflammatory pathway
Testosterone, estrogen, melatonin, oxytocin, cortisol, and BDNF affected through Ca²⁺-dependent steroidogenic and neuroendocrine pathways
Risk tolerance, social bonding, sleep architecture, cognition, and motivation shift as neuroendocrine substrates change
Both fertility desire (behavioral) and biological capacity (physiological) decline — the two-level collapse
Collective action, strategic planning, and institutional assertiveness weaken as the population’s hormonal and cognitive substrate degrades
The behavioral aggregate produces the patterns historians observe: stagnation, risk-aversion, institutional sclerosis
Biological contrast between EM-depleted and EM-intact populations creates demographic pressure gradients
Recovery if EM burden lifts (the α term), or permanent convergence as anthropogenic saturation (σ) masks the solar recovery window
The cumulative biological capacity of a population is formalized as the BioCap integral — a running balance between depletion (first integral) and recovery (second integral):
BioCap_cand(t,λ) = BioCap₀ − ∫₀ᵗ m_lat^cand(λ)·[S(τ)+U(τ)+E(τ)]dτ + recoveryS(τ)Normalized solar activity (drives natural geomagnetic perturbation)U(τ)Urbanization-weighted EMF exposure (population density × infrastructure)E(τ)Electrification-weighted exposure (grid density × per-capita consumption)m_lat^cand(λ)Candidate latitude moderator inside BERM; it is neither χ_geo nor a calibrated biological coefficientαRecovery coefficient (biological repair rate when EM burden decreases)σ(τ)Anthropogenic EM saturation — masks the solar recovery window post-1880This is BERM's reductionist causal hypothesis. The L2 entry-operator form is conditionally derived, while its tissue kernel and several cross-scale aggregation links remain open. Steps 5–10 are model consequences to test, not hormone assays inferred from political behaviour. The BioCap integral is a formal expression, not a fitted equation with validated coefficients.
BERM introduces Epistapege as the proposed transition by which an upstream biological change becomes difficult to identify after it has been translated into a coherent personal and institutional explanation. The extension is derived compositionally; it is not a direct result of a single experiment. [L*]
Pathopege → Epistapege → Pathorea → Pathostasis → Patopoliteia
Epistapege belongs to BERM's civilization branch. FieldState may provide physical input measurements for a test, but it is not the source of the cognitive mechanism and does not derive the transition.
Full derivation and discriminating tests →BioCap decomposes into eight measurable biomarkers. Each biomarker has a weight reflecting its relative contribution to civilizational capacity. The decomposition enables both measurement and prediction.
BioCap(t) = Σᵢ wᵢ · Bᵢ(t)where Bᵢ(t) = normalized level of biomarker i at time t, wᵢ = biomarker weight
CulturalEnergy(t) = N(t) × BioCap(t) × η(t)where N(t) = population, η(t) = institutional efficiency
| Symbol | Name | Weight | Unit | 1980 Baseline | 2025 Current | BERM Mechanism | Evidence |
|---|---|---|---|---|---|---|---|
| T | Testosterone | +0.20 | ng/dL | 600 | 440 | EMF → VGCC → Ca²⁺ → StAR↓ → T↓ | E (>1M) |
| OXT | Oxytocin | +0.20 | pg/mL | — | — | EMF → VGCC → Ca²⁺ → hypothalamic OXT↓ | M|C (proxy) |
| DA | Dopamine sens. | +0.15 | D2R arb. | 1.0 | — | EMF → VGCC → Ca²⁺ → DA synthesis↓ → D2R↓ | M|C (proxy) |
| MEL | Melatonin | +0.15 | pg/mL | 80 | 35 | EMF → CRY/VGCC → SCN → mel↓ + PGC | M|C |
| BDNF | BDNF | +0.10 | ng/mL | — | — | EMF → VGCC → Ca²⁺ → CREB↓ → BDNF↓ | M|C (proxy) |
| CORT | Cortisol | −0.10 | μg/dL | 12 | 16 | EMF → mel↓ → sleep↓ → HPA → CORT↑ | M|C |
| D | Vitamin D | +0.05 | nmol/L | 70 | 50 | D↓ → VDR → VGCC↑ → EMF sensitivity↑ | E (7.9M) |
| B2 | B2/FAD | +0.05 | nmol/L | — | — | B2 → FAD → CRY stability + mito complex I/II | M|C |
The recovery term α in the BioCap integral assumes a constant repair rate. The hormetic extension replaces α with a dose-dependent function h(Ā, δA) that captures three distinct biological zones:
h(Ā, δA) = { +α·δA if Ā < Ā_crit (Zone 1: stimulation); α·δA·e^(−β·(Ā−Ā_crit)) if Ā_crit ≤ Ā ≤ Ā_sat (Zone 2: transition); −γ·Ā if Ā > Ā_sat (Zone 3: damage only) }Zone 1 (Stimulation): Low EM exposure activates DNA repair, mitochondrial biogenesis, immune enhancement, and hormonal optimization. Populations in this zone maintain high biological capacity.
Zone 2 (Transition): Repair systems still function but with exponentially declining efficiency. The population shows mixed biomarkers — some activation, some suppression.
Zone 3 (Damage): Repair systems are overwhelmed. Net biological depletion dominates. This is the zone most industrialized populations occupy post-electrification.
BERM separates fertility decline into three distinct causal layers. Each level has its own dynamics, timescale, and evidence basis. The total fertility rate (TFR) for a country is the product of all three levels, not the sum -- each acts as a multiplier on the others.
The locked country predictions on this site come from the v17 scalar model. The FieldState v2 side branch (/measurement/fieldstate) is an optional measurement and estimation protocol: it defines which field quantities are recorded and how, and it produces no country forecasts.
Level 1
The physiological maximum fertility given current environmental exposures. Includes sperm quality (concentration, motility, DNA fragmentation), oocyte quality, hormonal milieu, and BBB integrity. This is the level most directly affected by EMF exposure.
Level 2
How personal device use interacts with ambient EMF exposure. A person in a high-ambient environment who also carries a phone experiences a non-linear coupling effect. This level captures the interaction between infrastructure-level and personal-level exposure.
Level 3
Voluntary fertility choices independent of biological capacity. Education, urbanization, contraceptive access, economic opportunity, and cultural norms. This component exists in all demographic models; BERM adds the biological and EMF layers underneath it.
Why can Level 2 appear as Level 3? BERM composes limited introspective access with neuroendocrine control of behaviour: a survey directly measures the reported reason, while the upstream biological state remains unobserved. The report may be sincere, causally active and still downstream. Distinguishing the levels requires longitudinal biomarker–behaviour–report data. [L*]
Read the interpreter derivation →The diagram separates Lindgren's derived metric drive from FieldState observations and the legacy technology proxy. All enter BERM's conditional L2 response operator through typed edges. Tissue kernels and endpoint coefficients remain open; downstream biology is not presented as a Lindgren-derived result.
Legacy pathway weights and community contrasts belong to model calibration, not to a theoretical ranking. RPM comparisons, Schwan membrane estimates and Cav3/HPG evidence may constrain tissue kernels but do not set their values. BERM therefore keeps RPM/CRY, VGCC/ROS, HPA/HPG and androgen-use branches parallel and falsifiable.
CRY/RPM does not respond to the RF carrier frequency (900 MHz – 3.5 GHz). Its resonance ceiling is ~22.5 MHz (Talbi, Zadeh-Haghighi & Simon 2025i, Front. Quantum Sci. Technol. 4:1544473). The biologically active components for Pathway B are the geomagnetic background (B_DC) and ELF modulation envelopes of telecom signals (GSM 217 Hz, WiFi 10 Hz beacon). Effects of the RF carrier itself are mediated by Pathway A through the electric field component. The two pathways have complementary frequency domains.
Not all voltage-gated calcium channels are equally EMF-sensitive. At resting membrane potential (~−70 mV), EMF sensitivity follows the hierarchy: Cav3 (T-type) >> Cav1.3 >> Cav1.2. T-type channels (Cav3.1, Cav3.2, Cav3.3) operate at a bifurcation point where ~10% are open at rest (window current), making them continuously sensitive to small voltage perturbations. Cav1.3 is a 'low-threshold L-type' that activates at ~−50 mV — 25 mV more negative than Cav1.2 (J Neurosci 2001). This makes Cav1.3 the primary channel in tissues requiring sustained low-voltage calcium entry: SA node pacemaking and inner hair cell synaptic transmission. Cav1.2, the canonical L-type, activates at ~−30 mV and is significant ONLY during action potentials — at rest it contributes negligibly. This hierarchy explains tissue-specific EMF vulnerability: organs dominated by Cav3 (testes, pituitary, adrenal, hippocampus) are most affected; Cav1.3-dependent tissues (inner ear, SA node) are intermediate; Cav1.2-dominated tissues (skeletal muscle, cardiac ventricle) are affected only during electrical activity.
A critical finding for BERM's cumulative exposure model: CaMKII (calcium/calmodulin-dependent protein kinase II) phosphorylation shifts the Cav3.2 activation threshold to MORE NEGATIVE potentials (PMC9913649). This creates a positive feedback loop: EMF → Cav3.2 Ca²⁺ influx → CaMKII activation → Cav3.2 threshold shifts left → channel becomes MORE sensitive to EMF → more Ca²⁺ influx. This molecular mechanism explains why EMF effects are cumulative over time: each exposure episode makes the system more sensitive to subsequent exposures. The CaMKII feedback also explains why short-term studies may underestimate long-term effects — the sensitization develops over weeks to months of chronic exposure. Pharmacological prediction: CaMKII inhibitors (KN-93) should block the progressive sensitization without affecting acute EMF responses.
Select a node to inspect its mechanism, sources, incoming and outgoing connections. Highlighting helps trace connections while every node stays visible. Levels organize the model; connections may skip levels or join nodes within a level.
Calcium · redox · hormone production
Calcium signalling, redox reserve and cellular maintenance meet at cholesterol supply and StAR-mediated mitochondrial transport. BERM connects field experiments and component interventions through these measured biological stages. Local hormone production then joins the existing hormone-availability and tissue-response pathway.
The direct CaMKI–NUR77–StAR branch and the RORα–BMAL1 clock branch converge on steroidogenesis while retaining their own evidence.
Explore the shared mechanism and its studiesReproductive regulation · behaviour · feedback
Reproductive regulation branches into motivation and realised encounters, physiological capacity, and caregiving. A selective biological brake can alter one output while another remains available. BERM connects these measured dependencies to its receiving state: opportunity and realised action join reproductive capacity before outcomes are aggregated over couples and time.
Follow the three branches and their evidenceThe IFO-VGIC mechanism is supported by a comprehensive review of 131 studies (Panagopoulos et al. 2025i, Bioelectromagnetics): 95% report oxidative effects from RF/Wi-Fi exposure. This consensus, consistent with Yakymenko et al. 2016i (93/100), establishes the Ca²⁺ influx → ROS pathway as the most robustly documented non-thermal mechanism.
Bertagna 2025i connects the later membrane-current response to ER release and reuptake under 50 Hz, 1 mT exposure. RyR and SERCA interventions constrain a coupled store–cytosol–membrane system; they do not identify two independent damage contributions or establish direct S4 forcing. Measure the first calcium change, ER-store trajectory and later function separately. This complements the IFO candidate (Panagopoulos 2025i) while retaining each protocol’s field class.
Bektas 2026i studied 28 rats in four groups with a GSM-modulated 3.5 GHz signal, 2 h/day for 30 days. CoQ10 attenuated some hormonal, testicular and redox changes. This was not a 5G NR waveform. Early calcium responses and repair time constants were not measured, so the result does not isolate a downstream repair site or demonstrate complete reversal of established damage.
The melatonin suppression pathway is quantitatively supported by a PRISMA systematic review of 55 studies (Tbahriti et al. 2026i, Sleep Biol Rhythms): 88% of high-quality animal studies report EMF-induced melatonin suppression of 20-50% from baseline. This suppression is biologically significant for GnRH pulsatility but smaller than light-induced suppression (>90%), consistent with BERM's v17_night_fraction() modeling EMF as one component of the nocturnal triple hit (melanopsin + CRY + melatonin), not the sole driver. Methodological note: only 27% of reviewed studies met high standards.
The CRY2–TRPC1 result (Iversen 2025i) adds a calcium-signaling branch in a named myoblast PEMF system. TRPC1 is a TRP channel. This motivates paired sham/field interventions with L-type blockade, TRPC1 perturbation and rescue; it does not determine a 25% reproductive contribution. Existing weights are scenario choices pending same-system endpoint calibration.
Candidate receiving coordinates include CRY subtype and compartment, FAD occupancy/redox, membrane orientation and light history. Pigmentation/light-response observations (Higuchi 2007i) and sex differences in a named magnetoreception task (Chae 2019i) do not supply universal CRY sensitivity coefficients. Iversen 2025i constrains CRY2/RFK/TRPC1 in its myoblast PEMF protocol; same-system interventions are needed before transferring a response to gonadal or retinal tissue.
Retinal CRY localization (Bartölke 2025i), membrane-associated Cry4a orientation (Majewska 2025i) and the myoblast CRY2–TRPC1 complex (Iversen 2025i) anchor different receiving coordinates. The synthesis retains protein subtype, species, compartment, cofactor occupancy and light sequence. A localization result suggests where to test reception; it does not establish an ordered human retinal compass, a blue-eye amplification factor or a membrane-diet effect size.
Gonadal, circadian, pituitary, autonomic and neurodevelopmental pathways
BERM proposes five biological routes through which EMF-related changes could affect fertility and, under further population assumptions, TFR. They describe different tissue mechanisms, but parallel operation and shared mediators must be evaluated together. Distinct mechanisms do not establish statistical independence or show that any one route alone is sufficient to reduce TFR. The net effect of blocking a route depends on its contribution, interactions, compensation and exposure context; the diagram does not demonstrate that the other four remain active.
EMF -> VGCC/Cav3 -> Ca2+ -> ROS -> sperm DNA damage + Leydig cell StAR suppression -> testosterone decline + spermatogenesis disruption. Additionally: EMF -> CatSper premature activation -> energy depletion -> navigation failure (rheotaxis, chemotaxis, acrosome reaction). Target tissue: testes. Evidence level: E (23-28 blocker studies). Primary channel: RF + ELF.
EMF -> CRY/RPM -> circadian clock disruption -> melatonin suppression -> HPG axis disruption + oxidative stress in follicular fluid. Target tissue: pineal gland, SCN. Evidence level: E. Primary channel: RF (magnetic component).
EMF -> Cav3 T-type channels in gonadotrophs -> FSH/LH secretion disruption -> downstream gonadal dysfunction. The pituitary sits outside the BBB and is directly exposed. All hormone cell types express Cav3. This route can reduce fertility independently of gonadal damage. Target tissue: pituitary gland. Evidence level: E. Primary channel: ELF + RF.
EMF -> SA node Cav3.1 -> HRV reduction -> vagal tone decline -> HPA axis hyperactivation -> chronic cortisol -> HPG cross-inhibition. HRV is a sensitive early biomarker. Target tissue: SA node, vagus nerve. Evidence level: E. Primary channel: ELF (50 Hz).
EMF → VGCC/Ca²⁺ during critical developmental windows → disrupted brain sexual differentiation, PFC maturation, identity formation. Same mechanism as chemical EDCs (BPA, phthalates). Additive with chemical EDC effects. Blocked by: prenatal EMF reduction, B2/glutathione support. Target tissue: fetal/infant brain. Evidence level: L* (derived prediction — awaiting DIFF-1 AGD test). Primary channel: RF + ELF.
Extended analysis: CACNA1C as the shared genetic vulnerability across ASD, ADHD, bipolar, depression, and schizophrenia. Seven developmental channels link EMF to brain sexual differentiation through the same Ca²⁺ pathways. See Brain modulome for full analysis.
→ Brain modulomeClinical implication: interventions targeting only one route (e.g., antioxidants for Route 1) will show partial but incomplete protection. Full protection requires either EMF reduction (addressing all routes simultaneously) or a multi-target intervention strategy.
A large study (Fertility and Sterility 2023i) found mobile phone use associated with lower sperm concentration — but the association was STRONGER in 2005-2007 than in 2012-2018. BERM explains this via the Schwan equation: the biologically active component is not the RF carrier but its ELF MODULATION ENVELOPE. GSM (2G): hard TDMA pulse at 217 Hz, ~100% modulation depth → strong ELF component → large T-type bifurcation effect. LTE (4G): OFDM, ~30-50% modulation depth, lower transmit power → weaker ELF component → smaller effect. This predicts the time trend WITHOUT invoking 'less radiation is safer.' The AMOUNT of radiation may be similar, but the MODULATION STRUCTURE changed.
Warning
Note: This time trend is a CORRELATION. Other factors changed concurrently (phone position, usage patterns, other exposures). The Schwan explanation is parsimonious but not the only possibility. This applies equally to conventional explanations.
The bounded χ_geo shape follows from the rank-one inverse metric once amplitude is made dimensionless and a positive-norm mode is selected. V17 uses that same shape in ambient + χ(ambient) × personal as a technology-timing proxy weight; this use is neither a tissue response nor a FieldState measurement.
χ_geo(x) = x / √(1 + x²), x = N(z_proxy)Where x is the legacy normalized ambient technology proxy. The function approaches 1 asymptotically by construction. This does not establish that a personal device's biological marginal effect diminishes in the same way.
BERM now has a formal L2 operator: the exact Lindgren metric perturbation is contracted with a causal tissue-response kernel. The derivation is conditional on minimal matter–metric coupling and response theory. It does not supply the kernel’s scale, sign, gauge prescription or human endpoint coefficient.
δgμν = κ(Āμaν + aμĀν + aμaν)δ〈Oi〉 = ∫ Ξi,Rμν δgμν + O(δg²)The evidence-constrained kernel is state-conditioned and retarded
BERM therefore retains orientation, coherence, waveform, endogenous phase, developmental window, receptor or agonist state, redox state, temperature trajectory, organ transfer and exposure history inside the tissue kernel. Existing studies constrain several of these arguments and protocol-specific null regions; organ transfer remains open, and the studies do not identify the Lindgren perturbation as their cause.
ui(t) = ∫ Kiμν(τ; Si(t−τ)) δgμν(t−τ) dτLitovitz et al. (1991)i · Rosenspire et al. (2005)i · Ubeda et al. (1983)i · Blackman et al. (1990)i · Blackman et al. (1991)i · Lymangrover et al. (1983)i
[JOHDETTU]
χgeo
χ_geo = ρ/√(1+ρ²) is a bounded inverse-metric coordinate for a normalized positive-norm mode—not tissue sensitivity.
[JOHDETTU]
Envelope / beat drive
The quadratic a⊗a term creates DC, modulation-envelope and difference-frequency metric components. Biological detection still requires Ξ_i.
[TUOTU + EMERGENTTI]
AEC
Androgen effective capacity separates total T, SHBG/albumin binding, free or intratesticular T, AR/ZIP9 occupancy and post-receptor transmission.
Open calibration—not an assumed effect
The model can now represent impaired hormone use without lower total testosterone, but it activates no EMF→SHBG, EMF→AR or EMF→ZIP9 coefficient by default. The relevant evidence is mixed: receptor biology and a rat ZIP9 route support the decomposition, while an acute randomized MRI study found no serum hormone change. Narinx et al. (2022)i · De et al. (2004)i · Yu et al. (2023)i · Møllerløkken et al. (2012)i
Explicit reduction from individual state to civilization
BERM first models a state-by-context individual probability P(Yᵢ=p|zᵢ,xᵢ). It then aggregates forward over the measured population distribution. This permits subgroup-specific and null effects; it does not permit diagnosing an individual’s biology from an aggregate political outcome.
Pt(Y=p) = ∫ P(Y=p | z,x) ft(z,x) dz dxInstitutional persistence is a separate open memory parameter: Iₜ₊₁ = ρIₜ + (1−ρ)Pₜ. This supplies the explicit population-to-institution operator used by the Epistapege extension; ρ is not historically calibrated.
Alogaily et al. (2025)i · Bakker et al. (2020)i
Open the Epistapege derivation →BERM registers five places where a background state may moderate a perturbation. They are separate candidate m-functions, not χ_geo and not one universal function derived from Lindgren geometry or FieldState.
| Scale | Background (B) | Perturbation | Candidate function | Verification | Level |
|---|---|---|---|---|---|
| Molecular | FAD chromophore in CRY | Magnetic field | m_mol([FAD]) | Hirano et al. (2017)i, Iversen et al. (2025)i | E |
| Optical | Blue light at CRY1 (iris transmission) | RF field | m_opt(I_blue) | Higuchi et al. (2007)i, Bartölke et al. (2025)i | M|C |
| Cellular membrane | Membrane potential V_mem ≈ −70 mV | External EMF | m_mem(V_mem) | Pall et al. (2013)i | E |
| Environmental | Ambient EMF (electrification, base stations) | Personal devices (phone, earbuds) | m_env(ambient proxy) | BERM 54-country dataset, electrification boundary | M|C |
| Population (COVID test) | Workplace ambient EMF | Screen time | m_context(office, home) | PNAS 2023: +5.1% TFR for WFH women | C |
These evidence families motivate specific interaction tests. Their moderators are not evidence for a shared χ_geo tissue law; each needs its own exposure measure, endpoint, sign and calibration.
m_glucose: K_ATP → V_mem → VGCC candidate
Glucose state can alter membrane potential and therefore motivates an exposure × glucose interaction test. The BERM gain is uncalibrated.
Prediction: Test whether measured exposure and glucose state interact on insulin secretion, with prespecified controls.
Sakurai 2008 supplies a study-specific ELF/insulin endpoint, not a human risk coefficienti
Candidate barrier-transfer moderator
BERM proposes: altered BTB integrity → changed target-cell exposure → possible feedback. The tissue-kernel gain is uncalibrated.
Prediction: If the feedback is real, sperm-quality change should accelerate with measured barrier loss.
Yu 2019 reports time-dependent 4G-RF-associated BTB disruption; it does not calibrate χ_geoi
m_barrier: candidate permeability moderator
Measured barrier integrity may modify target-cell exposure; multiplicative gain is a BERM hypothesis, not an established law.
Prediction: Test exposure × measured barrier integrity against a prespecified additive model.
Ulusoy 2025 motivates a time-resolved barrier endpointi
m_metabolic: candidate allometric moderator
Mass-specific metabolism and baseline oxidative state motivate a cross-species interaction model; they do not establish a universal scaling coefficient.
Prediction: Estimate species-specific slopes before testing an allometric meta-model.
Requires harmonized exposure and endpoint data across species
m_aquatic: candidate ELF/CatSper comparison
CatSper conservation and aquatic electromagnetic sensing motivate targeted studies, but neither shows that cable fields activate CatSper at environmental levels.
Prediction: Measure field spectra, gonadal dose and reproductive endpoints near matched cable/control sites.
Conservation and sensory evidence constrain plausibility, not environmental activation threshold
m_CRY: candidate light/FAD state
A cardiomyocyte CRY2–TRPC1 route is a BERM extrapolation from other cell systems (Yap 2025i).
Prediction: Test exposure × light/FAD state on prespecified cardiac calcium endpoints.
Cardiomyocyte-specific EM interaction remains untested
m_photo × m_temp × m_DC candidates
Photocycle, temperature and DC orientation are separate candidate moderators, not a shared χ law.
Prediction: A factorial replication can estimate each interaction and their joint term.
Blackman et al. (1985)i, Blackman et al. (1990)i, Blackman et al. (1991)i
χ_geo is a derived coordinate of the normalized rank-one geometry, not a VGCC susceptibility function. BERM separately proposes VGCC and cryptochrome/radical-pair response channels whose kernels, thresholds and interaction are endpoint-specific and uncalibrated. Low-exposure populations, pre-industrial series and solar-cycle panels can test these propositions but do not isolate either channel by themselves.
Geometric
Ca²⁺ channel (VGCC)
REQUIRES electrification threshold (Ā > 0)
Amish (Ā≈0), community gradient, country gradient
A (ROS), C (BBB), D (HPA)
Spin-chemical
Radical pair mechanism
NO electrification threshold (operates always)
Solar cycle, pre-industrial data, sentinel species, SAMA anomaly
B (CRY/RPM)
The operational weights (A=45%, B=25%, C=15%, D=15%) reflect current epidemiological evidence strength. But from a phylogenetic perspective, the hierarchy inverts: CRY/RPM (Pathway B) is the ancestral electromagnetic sensor, conserved across ALL eukaryotic kingdoms for over 1 billion years. VGCC (Pathway A), though dominant in human epidemiology, is a derived innovation appearing only in Metazoa ~500 Myr ago.
| Pathway B (CRY/RPM) | Pathway A (VGCC) | |
|---|---|---|
| Age | >1 Gyr | ~500 Myr |
| Kingdom scope | All eukaryotes | Metazoa only |
| Plant evidence | Yes (Ahmad 2020, Xu 2015) | No |
| Insect evidence | Yes (Gegear 2008) | Limited |
| Mammal evidence | Yes (PMC11817702) | Yes (extensive) |
| Operational weight | 25% (human TFR) | 45% (human TFR) |
| Phylogenetic rank | Ancestral | Derived |
This means the current TFR-focused operational weights understate CRY/RPM's evolutionary significance. When we extend from human TFR to ECOSYSTEM-level EMF effects — pollinator decline, bird population crashes, tree masting disruption — Pathway B becomes the dominant mechanism, because it's the only one present in all affected organisms.
The phylogenetic hierarchy is a theoretical framework. It does NOT change the operational weights used in BERM's TFR predictions. The weights reflect epidemiological evidence strength for human fertility, where VGCC (A=45%) has more direct human evidence than CRY/RPM (B=25%).
BERM identifies five biological pathways (A–E) through which EMF affects reproduction. Their operational weights reflect importance for human fertility. But their phylogenetic hierarchy — which is more fundamental and which is derived — is different.
Pathway B (CRY/RPM) is the ancestral mechanism. Present in all eukaryotes: plants, fungi, insects, birds, mammals. Cryptochrome was first discovered in plants (Arabidopsis, 1993). CRY’s reproductive role is best documented in plants — CRY2 → CONSTANS → FLOWERING LOCUS T → flowering induction. Conserved over 1 billion years as a photolyase homolog. Does not require membrane potential. Operates via spin chemistry (radical pair mechanism). RF disruption demonstrated in plants (Ahmad 2020: 7 MHz), insects (Gegear 2008: Drosophila), and mammals (PMC11817702 2025).
Pathway A (VGCC/IFO) is a BERM candidate assembled from imported ion-channel biology and exposure studies. It is animal-specific and relevant to excitable cells, but it is not derived from Lindgren geometry or FieldState. Its human tissue kernel, environmental dose response, sign and gain remain open. Plants have ion channels (TPC1, CNGC) but not S4-helix-based VGCCs.
Together: Pathway B is the evolutionary foundation. Pathway A is the animal-specific amplification layer on top of it. Both operate simultaneously in animals. Only Pathway B operates in plants.
Critical B2/FAD difference — why effect sizes differ between plants and animals: Plants synthesize their own riboflavin (B2), so FAD supply is endogenous and CRY function depends only on RF disruption — Ahmad 2020’s ‘relatively minor’ effect is a pure RPM test. Animals require dietary B2, so FAD supply depends on nutrition and CRY function depends on both RF and B2 status — a double vulnerability: EMF disruption plus nutritional deficiency. This explains why animal effect sizes exceed plant effect sizes: animals have two disruption sources, plants have only one.
The candidate effective-exposure index decomposes into ELF, IF and RF channels. Its biological weights are imported factors, while any geometric modulation must use the declared proxy coordinate χ_geo(N(z_proxy)); the open normalization and coupling are not derived by the channel decomposition.
cumEMF = w_ELF · cumELF + w_IF · cumIF + w_RF · cumRFcumEMF = w_ELF · cumELF + w_IF · cumIF + w_RF · cumRF, with diagnostic weights w_ELF = 0.05, w_IF = 0.60 and w_RF = 0.35. They require empirical calibration and are not fitted biological parameters. Statements about low-infrastructure or saturated environments are candidate proxy scenarios using x=N(z_proxy), not direct field measurements or a closed L2 operator.
The two-channel model's spatial structure is empirically supported by lateralization studies: Eliyahu et al. (2006)i and Luria et al. (2009)i demonstrated that 890 MHz exposure affects specifically the hemisphere nearest the phone. This confirms that personal-EMF effects are local, not systemic — EMF attenuates with the square of distance — supporting BERM's premise that phone-in-pocket targets testes, phone-at-ear targets hypothalamus.
The common catalogue records sources, regional adoption and operating conditions. Reconstruct the local four-potential from actual sources and geometry; category counts are not physical weights.
Aμ(x,t) = Abackground,μ(x,t) + Σj aj,μ(x,t)The IF channel (1 kHz – 1 MHz) targets dividing cells through the same frequency–cell size relationship as FDA-approved TTFields cancer therapy. The primary environmental source of IF fields is LED lighting: every LED bulb contains a switch-mode power supply operating at 20–200 kHz with harmonics extending to megahertz. A typical home contains 15–30 such sources; a typical office contains 200–500. Additional IF sources include HVAC variable frequency drives (5–50 kHz), induction cooktops (20–75 kHz), and all switch-mode power supplies (laptop chargers, phone chargers). The mechanism operates via Ion Forced Oscillation (IFO-VGIC), with a biological threshold of 10⁻⁵ V/m (Panagopoulos 2025i) — orders of magnitude below measured LED driver emissions.
The BERM cross-sectional diagnostic (v19.1) identifies three independent electromagnetic channels, each with distinct frequency ranges, exposure sources, biological mechanisms, and temporal histories. Note: v19.1 is a diagnostic formula fitted to 54 countries — the prediction model is v17.
Source: power grid, household wiring, appliances, transformers. Mechanism: IFO-VGIC forced ion oscillation (Panagopoulos 2025i). History: present since electrification (1880s), stable since ~1970. Proxy: residential electricity consumption (kWh per capita). Always on, 24/7, entire home.
Source: LED drivers (20–300 kHz), SMPS, VFDs, induction cooktops. Mechanism: Cyb5b → Ca²⁺ oscillations (Kim 2026 Celli), IFO at higher frequencies. History: near-zero before 2009, exponential growth 2009–2019 (EU LED transition). Proxy: LED market share × residential electricity. Pulsed, high dV/dt, regulatory gap (IJRB 2022i).
Source: mobile phones, Wi-Fi, Bluetooth, base stations, IoT. Mechanism: RPM/CRY spin chemistry (Ritz 2004i), thermal deposition at high SAR. History: 2G (1991), 3G (2001), 4G (2009), 5G (2019), Wi-Fi (1999). Proxy: broadband subscriptions per 100, mobile subscriptions. Modulated (data encoding), personal + ambient.
The IF channel's biological mechanism differs from ELF and RF. While ELF primarily activates ion channels (IFO-VGCC) and RF primarily disrupts radical pair chemistry (RPM/CRY), IF acts through a THIRD pathway: disruption of polar macromolecular structures during cell division (mitotic spindle, tubulin dimers). TTFields research demonstrates that IF fields (100–500 kHz) exert directional forces on polar intracellular elements. This mechanism is frequency-dependent: cancer cells are most affected at 150–200 kHz, while normal cells at ~50 kHz (Nature 2020). LED driver emissions (20–100 kHz) span the normal-cell sensitivity range.
Two weight sets, two purposes: (1) TCBM DIAGNOSTIC weights (w_ELF 0.05, w_IF 0.60, w_RF 0.35) are theoretical estimates derived from mechanism plausibility — how much biological damage each channel could produce based on its biophysical pathway. These are NOT fitted to fertility data and should be treated as prior estimates awaiting empirical calibration. (2) Cross-sectional EMPIRICAL weights (ELF ~60%, RF ~40%) are calibrated from the 54-country regression against observed TFR. Why they differ: the regression cannot separate IF from ELF because LED penetration correlates with electrification — so the empirical 'ELF 60%' likely contains a large hidden IF component. If the diagnostic weights are correct, most of the empirical ELF signal is actually IF acting through collinear proxies. The T1 temporal test (LED-DID, post-2009 EU ban) is designed to resolve this collinearity.
In the cross-sectional formula (54 countries, LOOCV RMSE 0.522), residential electricity serves as primary proxy because it captures ELF (always present with electricity) and correlates with IF (LED penetration tracks electrification). Broadband captures RF. ELF carries ~60% of the cross-sectional signal, RF carries ~40%. IF cannot be separated from ELF in cross-sectional data because LED penetration correlates with electrification. The temporal test (T1: LED-DID) is needed to isolate IF's independent contribution.
Planned: formal Wolfram Language verification of the three-channel coupling structure, including symbolic derivation of the IFO-VGIC threshold from first principles and numerical validation against the 54-country cross-sectional dataset.
Pharmacological evidence: 8 drug classes converging on BERM pathways →
The twelve-layer modulome catalogues candidate moderators from molecular spin physics to population patterns. BERM maps them through endpoint-specific response kernels; they are not a universal χ, are not χ_geo, and are not derived by FieldState. Twelve layers, twelve target organs, four proposed routes to fertility decline.
The modulome maps electromagnetic susceptibility from molecular spin physics to population-level patterns. Each layer moderates the open L2 biological application of the L1-derived χ coefficient; it does not alter χ's geometric derivation. Twelve layers, twelve target organs, four candidate routes to fertility decline.
Click a layer to see details
Epistemic note: Each layer's evidence is marked independently. The modulome as a unified framework is a BERM-specific synthesis [C] — the individual components carry their own evidence levels.
Eleven additional disease cascades derived from the VGCC gene family analysis. Each cascade links a specific VGCC subtype to a disease mechanism with its own evidence level.
EMF → VGCC in dopaminergic amacrine cells → DA release disrupted → scleral elongation brake weakened + CRY → melatonin → circadian ocular growth dysregulated. THREE converging channels.
EMF → chronic Ca²⁺ perturbation in T-cells → Ca²⁺-calcineurin-NFAT pathway dysregulated → autoreactive T-cell activation. Calcineurin inhibitors (cyclosporine, tacrolimus) are standard treatment — pharmacological confirmation.
EMF → Cav1.3 in inner hair cell synapses → chronic Ca²⁺ overload → excitotoxicity → synapse damage. Bluetooth/earphone EMF directly adjacent to cochlea.
CACNA1A (P/Q-type) GoF → CSD. CACNA1I (Cav3.3) variants → hemiplegic migraine (OR 2.30). Female:male 2.5-4.3:1 consistent with sex-differential VGCC.
Cav3.3 in nRt → spindle pacemaking. Cav3.1 in TC neurons → delta waves. T-type window current → slow oscillation. EMF → spindle/delta disruption → sleep quality ↓.
4-organ convergence: pancreas β-cell (Cav1+3 → insulin ↓) → hyperinsulinemia → theca androgen ↑ + granulosa aromatase → E2 ↓ + pituitary Cav3 → LH/FSH ↑. All four EMF-sensitive.
Cav3.2 is PRIMARY pain channel in DRG nociceptors. Upregulated in inflammatory/neuropathic pain. Female DRG neurons show more prominent Cav3.2 currents → sex difference.
CACNA1C GoF → Cav1.2 window current ↑ → QT ↑. Timothy syndrome: extreme QT + autism from SAME mutation.
7 causal channels × 3 developmental windows. Prenatal: Leydig Cav3 → T↓, aromatase, pituitary. Pubertal: PFC, melatonin, OT/AVP, insular cortex.
FDA HDE approval (2023) for HCC. A specified 27.12 MHz AM protocol engaged Cav3.2-dependent Ca²⁺ signalling in HCC models. This constrains a carrier × envelope × channel × organ route; reproductive and environmental transfer remains open.
SIX converging EMF → Ca²⁺ pathways: (1) hypothalamic appetite ↑ via ARC glia Ca²⁺ → AgRP/NPY, (2) BAT thermogenesis ↓ via CaMKII/CREB → UCP1 and SERCA2b/RyR2 disruption, (3) β-cell insulin dynamics ↓ via L-type VGCC, (4) thyroid axis → metabolic rate ↓ via Cav3 in thyrotrophs, (5) melatonin → metabolic circadian disruption, (6) adipocyte Ca²⁺ → lipogenesis ↑. CaMKII is the CONVERGENCE MOLECULE connecting all pathways. Klimentidisi paradox: 24 populations, 8 species ALL gaining weight (p = 1.2×10⁻⁷) including lab animals on controlled diets. Obesity is multifactorial — EMF is ONE contributing factor explaining the residual that diet/exercise/genetics cannot.
Six genes, six disease clusters, one mechanism
Cav1.2 · L-type
Cav1.3 · L-type
Cav2.1 · P/Q-type
Cav3.1 · T-type
Cav3.2 · T-type
Cav3.3 · T-type
Relative activation probability at ~−70 mV membrane potential
T-type (Cav3) channels >> Cav1.3 (low-threshold L-type) >> Cav1.2 (action-potential only). CaMKII feedback shifts Cav3.2 threshold more negative over time.
If EMF exposure were reduced, different biological systems would recover at different rates. The α parameter for each layer represents the fraction of damage that is reversible (1.0 = fully reversible, 0.0 = permanent).
| Layer | α | Recovery timescale | Notes |
|---|---|---|---|
| VGIC gating | 1.0 | Hours | Ion channel conformational changes reverse immediately upon cessation of field |
| ROS clearance | 0.8 | Days to weeks | Antioxidant systems restore balance, but chronic oxidative stress may cause lasting mitochondrial damage |
| DNA repair (SDF) | 0.1 | Months (spermatogenesis cycle) | New sperm are generated every 74 days, but stem cell damage may persist across cycles |
| Leydig cell function | 0.3 | Months to years | Testosterone-producing cells may partially recover, but chronic atrophy reduces regenerative capacity |
| Biological barriers (BBB + BTB) | 0.0 | Tissue- and injury-dependent; recovery must be measured | BTB disruption and SPOCK3–MMP14–MMP2 changes in the long-duration phone-exposure study identify a slow tissue route. Barrier disturbance and endocrine disturbance can feed back on each other. For ẋ=au+by−r_xx and ẏ=cu+dx−r_yy, positive recovery rates give stability when bd<r_xr_y. Irreversibility requires additional measured structural loss or nonlinear dynamics; barrier leakage alone does not establish a larger physical field. |
Acute facilitation and delayed harm can arise from different receiving and repair states. BERM therefore follows receptor state s, repair capacity A, damage D and functional capacity C through exposure and recovery. Adaptive RF-preconditioning experiments motivate a repair branch; a small later response can also reflect impaired reception or a shifted baseline. These alternatives require repeated absolute measurements, rather than a universal repair percentage.
Observed TFR is not simply the product of the three levels. Societies partially compensate for biological decline through assisted reproduction, behavioral changes, and policy interventions. The effective TFR includes a compensation exponent α = 0.43 that captures this partial offset.
TFReff = (bioCap × behav)(1−α) × rate2024 × cultRatio × bioBehav2024αWhere:
When α = 0, there is no compensation and biological decline passes through directly to TFR. When α = 1, compensation is complete and biological decline has no effect on observed TFR. The calibrated value of 0.43 implies partial but incomplete compensation -- biological decline still manifests in TFR, but at roughly half the rate it would without societal adaptation.
mTOR is the downstream integrator where EMF-induced Ca²⁺ influx converges with aging, fertility, and cancer pathways. The Sempou pathway: EMF → VGIC → Ca²⁺↑ → mTOR hyperactivation → autophagy↓, senescent cell accumulation, mitochondrial quality control↓, chronic inflammation↑.
Metformin activates AMPK, which suppresses mTOR -- the exact opposite of the EMF-induced pathway. The hypothesis: metformin's longevity benefit is not anti-aging per se but anti-EMF-accelerated-aging. In a natural EMF environment (Amish), the benefit should be minimal.
mTOReff = (1.0 + 0.25 × EMF) × ∏(1 − reductioni)aging rate = mTOReff0.7Where EMF is normalized exposure (0 = no infrastructure, 1 = modern city), and reduction factors include metformin (0.30), rapamycin (0.85), caloric restriction (0.20), intermittent fasting (0.10).
Aging
mTOR↑ → autophagy↓, senescence↑, inflammation↑, mitochondria↓ → accelerated aging
Fertility
mTOR↑ → spermatogonial differentiation↓, follicular burnout↑, AMH↓ → TFR↓
Cancer
mTOR↑ → proliferation↑, tumor growth↑, metastasis↑ → cancer risk↑
| ID | Prediction | Test |
|---|---|---|
| E1 | Metformin longevity benefit is larger in high-EMF environments | UK CPRD stratified by urban/rural |
| E2 | Amish metformin users show smaller longevity bonus than general population | Amish diabetic cohort comparison |
| E3 | Blue Zone longevity advantage disappears as 4G/5G arrives | Okinawa, Sardinia, Ikaria cohort tracking |
| E4 | CR experiment effect sizes increase by decade (rising lab EMF) | Meta-analysis: effect size vs publication year |
| E5 | TAME trial benefit stratifies by EMF exposure | Urban vs rural subgroup analysis |
| E6 | Shabbat (25h/week EMF-free) acts as intermittent mTOR fasting, supporting Haredi TFR and longevity | Haredi vs secular Israeli cohort |
CaMKII is an established downstream effector of Ca²⁺ signalling and connects to several disease-relevant cascades. BERM therefore treats it as a candidate convergence node for joint endpoint tests. This does not show that parallel population trends share EMF as an upstream cause; that inference requires an exposure-linked tissue kernel and competing-cause controls.
Epistemic note: CaMKII convergence is IDENTIFIED from independent literature but not yet experimentally tested as an integrated EMF mechanism. Each pathway is verified separately; the integrated test (EMF → CaMKII → all five targets simultaneously) is a prediction, not established fact. Evidence level: M.
Sun 2016i identifies altered calcium-channel expression after a named neuronal ELF protocol. This motivates measuring prior exposure → receiving state → later acute response. Electricity consumption is a diffusion proxy, not a channel-state measurement; the experiment does not show that every cell becomes more sensitive to every field.
Note: The ELF channel operates at 50 Hz in Europe and 60 Hz in the Americas. 50 Hz is within 2 Hz of the 8th Schumann resonance harmonic (52.0 Hz), potentially producing stronger CRY interference in European populations. This is a novel, falsifiable prediction testable by comparing melatonin profiles between 50 Hz and 60 Hz countries at matched total EMF levels.
Technology histories locate changes in the local physical environment. The shared source register includes commissioning, operating changes and shutdowns. BERM preserves waveform, orientation, timing and receiver state when defining the conditional response.
45 technology families
Historical health and technology series provide candidate timing contrasts. BERM tests whether measured local exposure and receiving-state changes explain intermediate biological function and later outcomes beyond the shared development, diet, sleep and demographic changes. Matching inflection years alone does not identify that chain.
The Mozaffarian Paradox
Americans eat less but weigh more since 2000
Conventional:
Unexplained
Layer explanation:
Layers 3–4 (WiFi + LED IF) added metabolic disruption independent of caloric intake. BAT thermogenesis↓ + insulin dynamics↓ are calorie-independent mechanisms.
Mozaffarian 2022, AJCNi
The 2012 Inflection
Social media existed since 2003 without crisis
Conventional:
Social media content harms teens
Layer explanation:
2012 = first year all three channels (ELF + IF + RF) simultaneously active 24/7 in teens. CaMKII threshold crossed at population level. Content restrictions will NOT resolve the crisis.
Haidt 2024; BERM layer analysis
The COVID Acceleration
T2D prevalence growth: 2.90%→3.52%/yr
Conventional:
Sedentary behavior during lockdown
Layer explanation:
Lockdown INCREASED layer intensity: 24h/day at home with WiFi + LED + multiple devices. Recovery window eliminated entirely. Remote workers had higher EMF than commuters.
GBD 2021 / Frontiers Endocrinol 2024i
The 15–30 Year Lag
Developing countries follow the same trajectory, delayed
Conventional:
Prosperity changes lifestyle
Layer explanation:
The delay matches electrification + technology adoption timelines, not prosperity. China T2D: 1.3% (1980) → 8.7% (2014) parallels electrification from 60% to 100%.
BMC Public Health 2018
The Amish Exception
TFR 6.1, low obesity, low dementia — same country
Conventional:
Physical labor and community
Layer explanation:
Zero technology layers. No ELF priming. Full recovery. EMF_effective ≈ 0. The diet is NOT especially healthy — the EMF environment is.
BERM population comparison
TFR ≈ A × exp(−B × EMF_effective) + C
EMF_effective = EMF_composite × P × (1/R)
EMF_composite = w_ELF × ELF + w_IF × IF + w_RF × RF
P = 1 + α × min(electrification_years, 40)
R = 1 + β × EMF_free_hours_per_day
P — Priming
P is a historical proxy scenario. Electrification years do not measure the receiving state of a present-day cell. A mechanistic history model instead measures passage, differentiation, prior fields, channel state and repair, then predicts whether the next response increases, decreases or shifts in frequency.
R — Recovery
R summarizes assumed interval structure in this proxy scenario. Actual recovery depends on local dose, biological state, sleep and repair rates. Device presence does not establish zero recovery, and a community label does not establish complete recovery.
Parameters α, β, w_IF require calibration against 54-country dataset + Amish/Tsimane data points. Expected improvement: LOOCV RMSE < 0.45 (vs 0.522 for v19.1).
| Country | Actual | v19.1 | v20 | Note |
|---|---|---|---|---|
| Finland | 1.25 | 1.32 | 1.28 | 100+ years electrified, high P |
| South Korea | 0.72 | 0.95 | 0.78 | Highest 5G/LED/smartphone density |
| Nigeria | 4.38 | 4.85 | 4.52 | Electrification ~15 years, low P |
| USA | 1.63 | 1.55 | 1.58 | 100+ years electrified, high P |
| Israel | 2.87 | 2.40 | 2.75 | Cultural fertility offset |
| Amish | 6.1 | — | 6.05 | Zero layers, full recovery |
| Country | 2024 | 2030 | 2035 | Driver |
|---|---|---|---|---|
| South Korea | 0.72 | 0.55–0.65 | 0.45–0.55 | 5G+EV+IoT, P grows, R→0 |
| Finland | 1.25 | 1.05–1.15 | 0.90–1.05 | 5G+LED, small recovery window |
| USA | 1.63 | 1.40–1.55 | 1.25–1.40 | 5G+EV, large P (100+yr) |
| Nigeria | 4.38 | 3.50–4.00 | 2.80–3.50 | Electrification accelerates, P grows fast |
| India | 1.96 | 1.55–1.75 | 1.25–1.50 | Electrification→100%, GSM/4G saturates |
Cryptochrome (CRY) is a light-dependent magnetoreceptor. In winter (less light), CRY is more sensitive to magnetic field perturbation — EMF effects on melatonin are STRONGER in winter. Halgamuge 2015i (Nature Sci Rep) demonstrated this directly: ELF suppressed melatonin in winter but INCREASED it in summer in calves. This seasonal modulation explains why Nordic countries (high latitude + high EMF) show disproportionate health burden (SAD prevalence: Finland 21%), and why EMF studies conducted in different seasons produce contradictory results.
Formula v21 correction factor:
S = 1 + γ × f(latitude, season)
S increases in winter at high latitudes (CRY more sensitive to EMF perturbation), decreases in summer (CRY saturated by ambient light). Near the equator, S ≈ 1.0 (stable day length). Finland in winter: S ≈ 1.3. Finland in summer: S ≈ 0.9.
SEASON-1: SAD/depression prevalence correlates with latitude × EMF density, not latitude alone
SEASON-2: EMF-free bedroom benefit should be LARGER in winter months
Halgamuge 2015i · Kolbabová et al. 2015i · CRY light dependence (biorxiv 2024)
The rs1006737 A-allele increases CACNA1C transcription → more Cav1.2 channels per cell → greater Ca²⁺ influx per EMF stimulus → lower CaMKII autophosphorylation threshold. This variant has been linked by GWAS to bipolar disorder, schizophrenia, autism, cardiac arrhythmias, and neurodevelopmental disorders — ALL conditions predicted by BERM's Ca²⁺ mechanism.
Sousouri 2025i (ETH Zurich): In a double-blind study, CACNA1C genotype DIRECTLY determined the sleep response to 5G exposure. This is the first demonstration that EMF sensitivity is genotype-dependent, not psychosomatic. Eckart et al. 2016i: rs1006737 is a quantitative trait locus for CACNA1C transcript levels. Tesli et al. 2013i: A-allele → altered amygdala activity across diagnoses AND healthy controls.
BERM reinterprets EHS reports as a possible tail of a continuous, state-conditioned response distribution rather than a binary population. CACNA1C is one candidate threshold modifier alongside channel state, redox state and exposure history; it is not a diagnosis, and the specific genotype × exposure effect requires direct replication.
Population-level correction:
G_pop = 1 + δ × CACNA1C_A_allele_frequency
G_pop adjusts the population's aggregate EMF sensitivity based on A-allele prevalence. European-origin populations (higher A-allele frequency) may have higher aggregate sensitivity than East Asian populations, though this requires further verification.
GEN-1: Populations with higher CACNA1C A-allele frequency show steeper health decline per unit EMF
GEN-2: A/A genotype individuals show stronger EMF responses than G/G in controlled exposure studies
Sousouri 2025i (ETH) · Eckart et al. 2016i · Tesli et al. 2013i
In adult neurons, GABA is inhibitory — it provides the damping (γ > 0) that keeps Ca²⁺ oscillations bounded. In neonates, the NKCC1/KCC2 chloride transporter ratio is reversed: NKCC1 dominates, chloride is high intracellularly, and GABA is excitatory. This means γ < 0 — the system has negative damping, and the quality factor Q → ∞. The neonatal brain is effectively an undamped resonator: any EMF-induced Ca²⁺ oscillation, however small, rings without attenuation. This is why the 2–4 month age window has peak SIDS risk — the KCC2 switch has not yet introduced damping.
Neonatal Q-factor decay:
Q_neonatal(age) = Q₀ / (1 + (age / τ_KCC2)²)
Q₀ = quality factor at birth (maximal, ~undamped). τ_KCC2 ≈ 2–4 weeks = NKCC1→KCC2 switch time constant. At birth: Q ≈ Q₀. At 2–4 months: Q declining but dangerously high. At 12 months: Q approaches adult levels (~1–5).
Bumetanide (NKCC1 blocker) → restores inhibitory GABA → terminates neonatal seizures = introduces damping. KCNQ2 mutations → neonatal seizures that spontaneously remit at 3–6 months = KCC2 maturation timeline.
Neonatal seizure review 2021i · Bumetanide NKCC1 2015i · NKCC1/KCC2 Bookshelf 2020i
The neonatal Q → ∞ condition is one end of a continuous spectrum. The same Q-factor mechanism — with varying damping coefficient γ — unifies SIDS, epilepsy, SUDEP, migraine, and cluster headache. Spreading depolarization (CSD) is the common terminal pathway; the Q-factor determines whether CSD is triggered, how far it propagates, and whether it reaches the brainstem.
See full neurological spectrum analysis →Convergence verification revealed seventeen positive feedback loops within the BERM cascade. The loops form a network: any entry point activates multiple degradation spirals simultaneously. Each means the system degrades itself without any increase in external exposure.
Baby's sound → microphone → RF modulation → VGCC → Ca²⁺ → stronger oscillation → louder sound → more RF → cascade amplification
Mechanistically coherent, untested as complete loop
EMF → Ca²⁺ → CaMKII → TPH-2 → 5-HT↓ → thalamocortical gate OPEN → EMF penetrates deeper → more CaMKII disruption → more 5-HT↓ → ...
Each link verified independently
CSD → local hypoxia → NKCC1↑ → GABA more excitatory → γ↓ → Q↑ → CSD propagates more easily → more hypoxia → ...
NKCC1↑ in hypoxia verified
EMF → melatonin↓ → sleep↓ → GABA tonic inhibition↓ → γ↓ → Q↑ → EMF affects brain MORE → more melatonin↓ → ...
Each link verified independently
EMF → PGC → melatonin↓ → BBB tight junctions↓ → heavy metals enter brain MORE easily → more PGC → less melatonin → ...
Each link verified independently
EMF → HPA → cortisol↑ → hippocampal atrophy → HPA negative feedback LOST → no braking → cortisol↑↑ → more atrophy → ...
Sapolsky mechanism verified
EMF → BAT PRDM16↓ → thermogenesis↓ → metabolic syndrome → inflammation → more VGCC sensitivity → more Ca²⁺ disruption → ...
Mechanistically coherent, animal data
EMF → Leydig → StAR↓ → T↓ → neuroprotection↓ + synaptic plasticity↓ → more vulnerable to EMF → more Leydig damage → ...
T↓ neuroprotection link verified
EMF → mast cell → IL-1β → KCC2 maturation delayed → GABA stays excitatory longer → Q↑ → more neuronal damage → more IL-1β → ...
KCC2 environmental regulation verified
EMF → hypothalamic synaptic vesicles↓ → GnRH↓ + CRH dysreg. + TRH↓ → multi-hormone deficit → systemic stress → more HPA activation → ...
EMF → SCN Ca²⁺ disrupted → melatonin timing lost → Per2↓ in gut → peripheral clocks desync → more SCN vulnerability
SCN Ca²⁺ oscillation verified
ELF → NK cytotoxicity↓ → cancer surveillance↓ → tumor growth → inflammation → VGCC sensitization↑ → more NK suppression
NK Ca²⁺ dependence + ELF suppression verified
EMF → cortisol↑ → GnIH↑ → T↓ → neuroprotection↓ → hippocampus vulnerable → HPA braking lost → cortisol↑↑ → more GnIH
RF9 restored T in cortisol-treated primates
EMF → melatonin↓ → Per2↓ in gut → gut barrier↓ → LPS enters blood → neuroinflammation → hippocampal neurogenesis↓ → more HPA activation → more melatonin↓
Per2 KO → gut barrier → LPS → depression verified
EMF → melatonin↓ → telomerase↓ + SIRT1↓ → telomere shortening → SASP → inflammation → ROS↑ → more telomere damage → more SASP → ...
Melatonin → telomerase + SIRT1 verified; depression = 7y accelerated aging
EMF → α2δ-1↑ → central sensitization → chronic pain → sleep↓ (S4) → cortisol↑ (S7) + GABA↓ → inflammation → more sensitization → depression → sleep↓ → ...
α2δ-1 → pain without injury verified; pain-sleep-cortisol each verified
Kim 2019i demonstrated that 835 MHz (12 weeks) reduces synaptic vesicle number, size, and docking in hypothalamus. Crucially, synaptotagmin 1 — the Ca²⁺ sensor for vesicle release — is also reduced. Since ALL hypothalamic hormone release depends on Ca²⁺-triggered vesicle fusion, synaptotagmin 1 loss means ALL axes are simultaneously impaired.
| Axis | Target organ | Consequence |
|---|---|---|
| GnRH → LH/FSH → T↓ | Gonads | Testosterone decline, fertility loss |
| CRH → ACTH → cortisol↑ | Adrenals | HPA sensitization, chronic stress |
| TRH → TSH → T3/T4 | Thyroid | Subclinical hypothyroidism |
| GHRH → GH → IGF-1 | Liver/bone | Growth and metabolic disruption |
| Dopamine → prolactin | Pituitary | Hyperprolactinemia |
| Somatostatin → GH/TSH | Multiple | Loss of inhibitory control |
| Oxytocin / AVP | Multiple | Social behavior, water balance |
VK13 is the anatomical explanation for why EMF produces SIMULTANEOUS multi-system effects that appear unrelated. It is not 25 separate diseases — it is one disrupted nexus with 7 output channels.
EMF simultaneously reduces testosterone (T↓ via Leydig/StAR), elevates cortisol (F↑ via HPA sensitization), and reduces dopamine (DA↓ via mesolimbic pathway). Each deficit reinforces the others, creating a synergistic trap.
EMF → Leydig → StAR↓ → steroidogenesis↓
Neuroprotection loss, muscle loss, fertility decline, depression
EMF → HPA sensitization → cortisol baseline↑
Hippocampal atrophy, immune suppression, metabolic syndrome
EMF → CaMKII → DA synthesis disruption
Anhedonia, motivation loss, addiction vulnerability
The triple lock is not three independent effects — it is a synergistic trap. T↓ × F↑ = accelerated neurodegeneration. F↑ × DA↓ = treatment-resistant depression. T↓ × DA↓ = motivational collapse. T↓ × F↑ × DA↓ = the complete modern phenotype.
Oxytocin release is directly VGCC-dependent (N-type + L-type Ca²⁺ channels, PMC3197583i). EMF disrupts VGCC function → OXT release disrupted. Adding OXT↓ to the triple lock creates a quad lock that explains the full modern phenotype: not just physiological decline but social fragmentation.
Reproductive-social collapse: fertility decline + pair bond weakening
Social motivation collapse: reduced desire for social connection + reduced reward from it
Stress without buffering: cortisol rises while OXT (the social stress buffer) falls
Complete modern phenotype: biological decline + social isolation + motivational collapse
Insulin stimulates OXT release via Ca²⁺ (PMC6039480i). Obese individuals have lower OXT. This creates a metabolic-social bridge: metabolic syndrome (S7) → insulin resistance → OXT↓ → social isolation → depression → metabolic syndrome worsens.
The blood-brain barrier and intestinal epithelial barrier share the same tight junction proteins: ZO-1, occludin, and claudins. Melatonin protects both barriers. EMF→melatonin↓ creates simultaneous dual vulnerability: BBB opens (heavy metals enter brain) AND gut barrier weakens (LPS enters bloodstream → neuroinflammation). This is not two separate effects — it is one mechanism (melatonin loss) attacking two barriers built from the same molecular toolkit.
BDNF (brain-derived neurotrophic factor) is essential for neuroplasticity, memory, and neurogenesis. RF-EMF (835–2650 MHz) reduces BDNF in hippocampus with dendritic spine loss and cognitive impairment. Meanwhile, ELF (50 Hz) INCREASES BDNF and promotes neurogenesis. BERM treats these directionally different findings as motivation for a frequency-dependent hormesis hypothesis through candidate VGCC routes. Derived χ_geo and the conditional L2 operator do not determine those biological outcomes; the BDNF tissue kernel and endpoint response remain to be calibrated.
Melatonin is not just a sleep hormone — it is the key anti-aging molecule. It activates telomerase (maintaining telomere length), upregulates SIRT1 (→ ROS↓ → p53↓ → NF-κB↓), and alleviates endothelial aging. EMF→melatonin↓ removes this entire protective cascade.
EMF → melatonin↓
Pineal suppression via CRY pathway (VK1-VK3)
Melatonin↓ → telomerase↓
Melatonin activates telomerase directly (Front Aging Neurosci 2022i)
Melatonin↓ → SIRT1↓
SIRT1 → ROS↓ → p53↓ → NF-κB↓ anti-inflammatory cascade lost
Telomere shortening → SASP
Shortened telomeres trigger senescence-associated secretory phenotype → chronic inflammation
SASP → ROS↑ → more telomere damage
Feedback loop S15: inflammation causes oxidative damage to remaining telomeres
Quantitative anchor: major depression is associated with telomeres 281 bp shorter, equivalent to 7 years of accelerated aging (PMC3063175i). Metabolic syndrome is similarly associated with shorter telomeres and reduced telomerase activity (PMC12744432i). Both conditions are BERM-predicted outcomes — their aging acceleration is consistent with EMF→melatonin↓→telomerase↓.
BERM identifies 15 genes whose polymorphisms modulate individual EMF sensitivity. They divide into five functional tiers: INFLUX (5 CACNA genes controlling Ca²⁺ entry), MODULATION (CACNA2D1 controlling channel density), INTEGRATION (CAMK2A/B at the convergence point), EXTRUSION (3 genes controlling Ca²⁺ removal), and SIGNALING (4 genes modulating downstream response). Each gene's disease associations match BERM cascade predictions.
| Gene | Protein | BERM role | Key variant | Diseases | Evidence |
|---|---|---|---|---|---|
| CACNA1C | Cav1.2 (L-type) | Primary RF target. Neurons, heart, β-cells. | rs1006737 A-allele | Bipolar, schizophrenia, ASD, depression, Timothy syndrome | CONFIRMED (Sousouri 2025i RCT) |
| CACNA1H | Cav3.2 (T-type) | ELF target. Leydig cells, pineal, thalamus. | GoF mutations | Childhood epilepsy, febrile seizures, primary aldosteronism, ASD | CONSISTENT |
| CACNA1D | Cav1.3 (L-type) | Inner ear, SA node, substantia nigra. | GoF/LoF variants | Bradycardia, epilepsy, hearing loss, ADHD, ASD | CONSISTENT |
| CACNA1A | Cav2.1 (P/Q-type) | Presynaptic release. ELF priming target. | rs16023 B-allele | DD + epilepsy, familial hemiplegic migraine, episodic ataxia | CONFIRMED (ELF priming + GWAS) |
| CACNA1B | Cav2.2 (N-type) | Pain pathways, sympathetic nervous system. | Rare mutations | Chronic pain, sympathetic dysfunction | CONSISTENT |
CACNA2D1 encodes α2δ-1, the protein that controls VGCC trafficking to synapses. This is the molecular basis of ELF priming: 50/60 Hz exposure upregulates α2δ-1 → more VGCCs reach the cell surface → cells become more sensitive to ALL subsequent EMF. Gabapentinoids (pregabalin, gabapentin) bind α2δ-1 and BLOCK this trafficking — making them mechanistically ELF-priming ANTAGONISTS.
Field 2006i (PNAS) · Hoppa 2012i (Nature)
CAMK2A/B de novo mutations that INCREASE autophosphorylation at Thr286/287 produce epilepsy, intellectual disability, and autism — the EXACT phenotypes BERM predicts from environmental (EMF) autophosphorylation increase. Mutations that DECREASE autophosphorylation also cause intellectual disability. Both directions = disorder → precise regulation is critical. This is BERM's most direct genetic validation: genetic and environmental CaMKII dysregulation converge on identical clinical outcomes.
Küry 2017i (AJHG, PMC5673671i) · Al-Tawashi 2018i (eLife, PMC5963920i)
Three genes control Ca²⁺ removal from cells. Slow extrusion + high influx = Ca²⁺ accumulates → CaMKII threshold crossed at lower EMF levels. SLC8A1 (NCX1): cardiac/neuronal Ca²⁺ export. ATP2B1 (PMCA1): general Ca²⁺ pump (GWAS: hypertension). ATP2B2 (PMCA2): inner ear — slow PMCA2 + Bluetooth earbuds = tinnitus risk.
| Gene | Variant | Effect | Diseases | Evidence |
|---|---|---|---|---|
| CRY1 | CRY1Δ11 (0.6%) | GoF → longer circadian period → delayed sleep → shorter recovery window. EMF disrupts CRY → ADDITIVE with genetic lengthening. | DSPD, metabolic disruption, insomnia | CONFIRMED (Patke 2017i Cell) |
| MTNR1B | rs10830963 G | eQTL → more MT2 receptors on β-cells → HYPERSENSITIVE to melatonin changes. EMF suppresses melatonin → G/G carriers affected MORE → T2D risk SUPERADDITIVE. | T2D, fasting glucose, gestational diabetes | CONFIRMED (GWAS + eQTL) |
| COMT | Val158Met (rs4680) | Val/Val = fast dopamine clearance = low DA baseline → EMF-induced DA synthesis drop hits HARDER (smaller buffer). | Stress vulnerability, addiction, pain sensitivity | DERIVABLE |
EHS (electromagnetic hypersensitivity) is not psychosomatic — it is a polygenically predictable Ca²⁺ threshold disorder. High VGCC influx (CACNA GoF) + slow extrusion (SLC8A1/ATP2B LoF) + sensitive signaling (CRY1Δ11, MTNR1B GG, COMT Val/Val) = low CaMKII autophosphorylation threshold = symptoms at EMF levels below the population average.
Proposed biomarker: CaMKII Thr286 autophosphorylation level in lymphocytes. Higher level = closer to threshold = more EMF-sensitive. This could be the first OBJECTIVE biomarker for EHS.
CACNA1C × MTNR1B
Depression + T2D from the same melatonin suppression in different organs. AA + GG carriers: highest comorbidity.
TESTABLE (biobank)CRY1Δ11 × MTNR1B
Delayed melatonin × β-cell hypersensitivity → morning fasting glucose especially elevated.
DERIVABLECACNA × SLC8A1/ATP2B
High influx + slow extrusion = Ca²⁺ accumulation → EHS phenotype.
TESTABLE (genotype EHS cohort)CAMK2A × CACNA2D1
CaMKII near threshold + more channels = critically sensitive to any EMF.
CONSISTENTGXEMF-1: Gene × EMF interactions are superadditive
Genetic risk manifestation depends on EMF exposure. EMF 'activates' genetic risks that would be latent in an EMF-free environment.
GXEMF-2: Gabapentinoids reverse ELF priming via α2δ-1
Pregabalin/gabapentin bind α2δ-1, blocking VGCC trafficking. Gabapentinoid users have lower synaptic VGCC density → less EMF-sensitive.
GXEMF-3: CaMKII autophosphorylation is a measurable biomarker
CaMKII Thr286 phosphorylation level in lymphocytes: higher = more EMF-sensitive. Testable in EHS cohorts.
Küry 2017i · Patke 2017i · Lyssenko 2009i · Tuomi 2016i · Scholl 2015i · Korean 2025i · Field 2006i · Hoppa 2012i
BERM separates calcium/CaMKII relaxation, receptor-state recovery, adaptive repair and tissue turnover. Measure these during and after a defined local field protocol, together with functional capacity. Protective preconditioning, sensitization and persistent injury predict different time courses; a single exposure-free-hours factor cannot identify which occurred.
Sleep and shift-work evidence anchor the importance of biological timing. They do not isolate a field-free recovery threshold. RF preconditioning and repair interventions provide a closer route to testing field-associated history: measure repair flux and challenge tolerance alongside damage and baseline function.
A controlled bedroom study can vary measured local fields while matching light, sleep schedule, temperature and device-use behavior. Serial melatonin, calcium-related markers and functional endpoints then distinguish a field-associated effect from the consequences of the accompanying sleep changes.
RECOV-1: EMF-free bedroom → melatonin levels increase measurably within 2 weeks
RECOV-2: measure the recovery curve; test the proposed 4–6-hour window
Walker 2017i · COVID lockdown data · Shift work meta-analyses
In this technology-timing scenario, the 2007 inflection is not a measured dose jump. What changed in the proxy construction was the assumed interval between device contacts. Applying first-order repair kinetics to that interval is an L3 candidate, not a calibrated physical dose-response claim.
repaired fraction = 1 − exp(−t_free / τ), τ = 6 h / ln 2 ≈ 8.66 hCandidate interval calculation: an 8-hour assumed free window clears 60% per cycle and a 2-hour window 21% under the stated kinetics. Device-contact timing is a proxy; no physical or biological dose is inferred without the open L2 measurement map.
| Country | Crossing year | Free fraction 2024 | Net damage 2024 |
|---|---|---|---|
| South Korea | 2018 | 0.08 | 17.6 |
| USA | 2020 | 0.14 | 14.0 |
| Finland | 2022 | 0.19 | 11.5 |
| Japan | 2023 | 0.22 | 10.1 |
| India | not crossed by 2030 | 0.45 | 4.9 |
| Nigeria | not crossed by 2030 | 0.60 | 2.4 |
Crossing year is the first year the exposure-free fraction of the day falls below 0.25; ratio and net daily damage index are the 2024 values. Both are computed from the model's own penetration, screen-contact and bedroom-device inputs.
South Korea is the clearest case because its penetration saturated early. Between 2013 and 2023 smartphone penetration moved from 0.60 to 0.97 — a diffusion process essentially finished — while the exposure-free fraction of the day fell from 0.63 to 0.10 and the net daily damage index rose from 1.6 to 16.2. The exposure count stopped changing; the interval structure did not.
The table is a dosing-interval scenario. The threshold and repair parameters require field- and endpoint-specific calibration.
The behavioural and the biological account of the fertility decline are usually posed as rivals: either smartphones displaced the time in which conception happens, or cumulative field exposure damaged reproductive capacity. The dual-kernel framework treats them as two frequency components of one signal. The same technology-diffusion sigmoid E(t) is convolved with a fast behavioural kernel and a slow biological kernel, and the observed decline is their weighted sum.
F(t) = α · (E ∗ k_B^Erlang)(t) + (1−α) · (E ∗ k_R^exp)(t), β := 1−αT4 fixes the fast arm to normalized annual Erlang-bin masses with integer n_B=2–6; the slow arm is a normalized discrete exponential. Their available-history weights each sum to one. β is derived exactly as 1−α and is never fitted independently.
τ_B ≈ 0.5–2 years; integer n_B=2–6
Normalized Erlang annual-bin masses model a delayed multistage response rather than an immediate exponential response.
Prediction: Dominates the young age groups in the candidate timing-proxy scenario.
τ_R ≈ 10–15 years (framework value, not fitted)
Normalized exponential memory; T3 cohort vulnerability v(a) may weight its annual input. This is phenomenology applied to a timing proxy, not measured dose.
Prediction: The candidate signal should surface first in cohorts with the longest proxy history.
Fitting one kernel per age group on 54 countries and five WPP waves of age-specific fertility rates, with country fixed effects and a log response, identifies the fast kernel sharply and the slow kernel not at all. The suppression slope falls monotonically across the reproductive span and changes sign after 35.
| Age group | Best τ (years) | Suppression slope | Reading |
|---|---|---|---|
| 15–19 | 1.0 | +0.258 | Fast kernel, tightly identified (0.5–2 y within 2 BIC units) |
| 20–24 | 1.0 | +0.234 | Fast kernel, tightly identified (1–1.5 y) |
| 25–29 | 1.0 | +0.146 | Fast kernel, suppression roughly half the teen value |
| 30–34 | 1.5 | +0.043 | Slope near zero; τ no longer identified |
| 35–39 | — | −0.107 | Sign flips: exposure tracks higher, not lower, fertility |
| 40–44 | — | −0.198 | Sign flip strengthens — the postponement signature |
The declining slope reproduces the observed age profile of the decline: US age-specific birth rates fell 71% at 15–19, 43% at 20–24 and 23% at 25–29 over 2007–2024, while the 35–39 rate rose 9% (Hudson & Moscoso Boedo 2026i). The sign flip after 35 is where a postponement-and-recovery reading and a capacity-loss reading make opposite claims about the same number.
The obstacle is structural, not a shortage of data. Every country's technology-timing proxy is a smooth monotone sigmoid; the T4 Erlang fast memory and exponential slow memory remain highly collinear on such a ramp. The historical correlations below diagnose that proxy design, not physical dose response.
| τ pair | Kernel correlation | Unconstrained fit | Sign-constrained fit | Verdict |
|---|---|---|---|---|
| τ_B = 1.5, τ_R = 12 | 0.9916 | excluded: β:=1−α | 0≤α≤1 | Not separable |
| τ_B = 1.0, τ_R = 20 | 0.9838 | excluded: β:=1−α | 0≤α≤1 | Not separable |
| τ_B = 4.0, τ_R = 6 | 0.9996 | excluded: β:=1−α | 0≤α≤1 | Not separable |
The former independently fitted β route is outside the registered DKC definition and is excluded. With β:=1−α and 0≤α≤1, the monotone timing-proxy panel does not separate the two arms reliably. The framework therefore remains a testable candidate: the fast timing association is estimable in the proxy panel, while the slow biological arm is not identified from aggregate fertility alone.
Exposure is usually summarised as a single power figure, SAR in watts per kilogram. That figure is blind to how many distinct frequencies are present at once, and technology generations accumulate rather than replace: 2G did not switch off when 3G arrived, and Wi-Fi, Bluetooth and household IoT each occupy their own band. If ion channels act as receivers, what matters is not only how much power arrives but across how many channels it arrives simultaneously.
spectral complexity = log₂(1 + bands)This logarithmic band-count index is a descriptive proxy, not a biological capacity law. A receiving model instead uses a measured local driver spectrum and its projection through a state-dependent response window.
| Generation | Years | Simultaneous bands | Added | log₂(1 + bands) |
|---|---|---|---|---|
| Broadcast only | before 1990 | 1 | FM and television | 1.00 |
| 2G | 1990s | 2 | Cellular voice | 1.59 |
| 3G | 2000s | 4 | Mobile data, early Wi-Fi | 2.44 |
| 4G | 2010s | 7 | LTE, saturated Wi-Fi, Bluetooth | 3.32 |
| 5G | 2020s | 11+ | New radio bands, household IoT | 3.99 |
Band counts and complexity are the model's values for South Korea, which reached each generation earliest. A country's complexity depends on when each generation arrived, so the same calendar year gives a different value in each country.
Multisource comparisons should preserve phase/coherence, local transfer and the slow driver spectrum. N(N−1)/2 cross pairs do not imply N² damage: their sign, projection and averaging matter. A useful test changes a defined component or coherence while matching power and temperature, and predicts one specified endpoint.
The COVID ambient dip is the one natural test on record. Industrial and traffic emissions stopped while household devices stayed on, so the lockdown was not simply less exposure but fewer simultaneous bands — a spectral simplification with the personal channel intact. The framework's reading of the bidirectional COVID result rests on that distinction rather than on total power.
EMF research has produced contradictory results for decades. BERM identifies eight uncontrolled moderators that predict which studies find positive results and which find null:
CRY magnetoreceptor sensitivity is light-dependent. In winter, CRY is more sensitive → EMF effect on melatonin is stronger. Demonstrated in calves (Halgamuge 2015i).
CACNA1C rs1006737 A-allele → more Cav1.2 → larger Ca²⁺ response. Sousouri 2025i (ETH): CACNA1C genotype determines 5G sleep response.
50/60 Hz power grid upregulates VGCC expression in 8–10 days (PMC4757866i). High-ELF-background labs 'prime' cells.
Wi-Fi router in bedroom vs. EMF-free night → different CaMKII recovery state → different baseline Ca²⁺ when entering the experiment.
Animal studies in lab environments (24/7 ELF priming, homogeneous genetics) find positive results 92% of the time. Human studies with heterogeneous environments find 35%. Both correct — lab animals are chronically primed (VGCC expression elevated, PMC4757866i). p=0.002.
Chronic exposure (>1 week) produces positive results 92%. Acute (1–2 nights) produces 31%. CaMKII autophosphorylation requires cumulative Ca²⁺ loading. p=0.001.
Pulsed signals produce positive results 88%. CW produces 48%. IFO-VGIC mechanism requires varying fields. p=0.048.
Vitamin D (1,25(OH)₂D₃) downregulates CACNA1C/1D mRNA (J Neurosci 2001i). Vitamin D deficiency → VGCC over-expression = same state as ELF-priming. Studies in vitamin D-deficient populations (winter, high latitudes) should show stronger EMF effects.
Three moderators predict study outcome with statistical significance:
Species / Priming
χ² = 9.4, p = 0.002
Duration
χ² = 10.8, p = 0.001
Pulsation
χ² = 3.9, p = 0.048
Based on analysis of 29 studies across 3 endpoints. Validated by Weller 2025i (n=517).
Prediction REPL-1: A retrospective analysis of 50–100 published EMF bio-assay studies will show these eight moderators significantly predict positive vs. null outcomes. Testable WITHOUT new data.
Epistemic level: the eight-moderator framework is BERM's synthesis (M-level). Individual moderators have empirical support (E-level).
Weller et al. (2025)i analyzed 517 genotoxicity studies and found that 58% of studies reporting DNA damage used exposure levels BELOW current ICNIRP guidelines. The Ivancsits studyi found DNA strand breaks at 35 µT — less than one-fifth of ICNIRP’s 200 µT occupational limit.
ICNIRP limits are designed to prevent THERMAL effects. DNA damage from EMF is a NON-THERMAL mechanism operating through voltage-gated calcium channel dysfunction.
Ivancsits et al.i showed that EMF-induced DNA strand breaks returned to normal within 9 hours after exposure ceased. This quantifies BERM’s recovery window: the body CAN repair EMF-induced damage, but only if given sufficient EMF-free time.
Modern environments with 24/7 WiFi, LED lighting, and smartphones in bed eliminate this recovery window entirely. The typical modern bedroom provides zero EMF-free recovery time.
The mathematics separates the 2025 Lindgren ansatz and its geometric consequences from BERM's biological and demographic closures. A geometry-to-response operator form is derived conditionally under explicit matter–metric and linear-response assumptions; its tissue kernel, sign, lag and calibration remain open.
In the framework of Lindgren, Kovacs & Liukkonen (2025)i, the electromagnetic potential is part of spacetime geometry. The metric tensor absorbs the EM four-potential:
Within this ansatz the electromagnetic potential contributes to the metric. BERM derives the exact perturbation and, conditionally, its formal contraction with a tissue-response kernel. The ansatz alone does not specify an ion-channel, hormone or receptor coefficient.
Vassallo et al. (2025)i is registered as a related theoretical analysis; it does not close BERM's biological L2 operator.
The naive estimate δV_mem ≈ 10⁻²¹ V exposes the unresolved coupling problem. BERM tracks three candidate biological realizations, none of which is yet derived from the ansatz:
(1) IFO: ion forced oscillation acts on the S4 voltage sensor directly at <1 nm distance, threshold 10⁻⁵ V/m (Panagopoulos 2025i).
(2) Non-ionotropic VGCC signaling: conformational change without ion flux, lower energy threshold (Trus & Atlas 2024i).
(3) The RPM pathway can bypass VGCC. A previously reported 87.5% algebraic correspondence is a structural comparison, not a derived geometry-to-RPM coupling operator.
Biological sensing examples constrain plausible sensitivity ranges: the eye detects single photons (Vaziri et al. 2016i), electroreception and magnetoreception operate in specialized systems, and an ion-forced-oscillation proposal reports a 10⁻⁵ V/m scale (Panagopoulos 2025i). These are comparative observations and mechanism proposals; they do not calibrate BERM’s tissue kernel Ξ_i or a human membrane response.
BERM treats spectral filtering and receptor selectivity as empirical questions. Evolutionary novelty alone does not establish susceptibility, and no assumption that every technical signal is disruptive is used as a derived result.
In Lorentz signature the invariant L1 result is a signed directional derivative. The χ formula itself is L1; obtaining χ(|Ā|) requires an explicit dimensionless Lorentz-to-Euclidean spatial/scalar reduction, which is an L2 choice.
The direction u, contraction metric, κ, gauge and the domain D > 0 are part of the input contract. A positive |A| or Euclidean angle is not available for a general Lorentz four-vector.
If matter couples minimally to the metric and the tissue is treated with causal response theory, the formal mapping is derivable. This is a BERM closure under stated premises—not a biological result in Lindgren's paper.
State-conditioned retarded form constrained by component studies
BERM refines the formal observable into an organ-specific drive uᵢ(t). Sᵢ retains orientation, coherence, waveform, endogenous phase, developmental timing, receptor or agonist state, redox, temperature trajectory, organ transfer and exposure history. Primary component experiments constrain several of these arguments; organ transfer, the kernel coefficients and identification of δg as their cause remain open.
Litovitz et al. (1991)i · Rosenspire et al. (2005)i · Ubeda et al. (1983)i · Blackman et al. (1990)i · Blackman et al. (1991)i · Lymangrover et al. (1983)i · Møllerløkken et al. (2012)i
Quadratic RF demodulation is geometry, not yet biology
For a(t)=a₀[1+m cos Ωt]cos ωt, the exact a⊗a term leaves DC, Ω-envelope and 2Ω components after ideal carrier removal. Two tones also create a |ω₁−ω₂| term with amplitude κa₁a₂. Whether tissue detects these terms is encoded only in Ξ_i.
Two biological cutoff frequencies divide the EMF spectrum into three regimes with distinct biophysical mechanisms. These cutoffs are fundamental properties of cell biology, not arbitrary parameters.
f_c ≈ 1 kHz — the membrane RC time constant. Below f_c: the field drops across the membrane and perturbs V_mem. Above f_c: the field penetrates into the cell interior.
f_RPM ≈ 1 MHz — the radical pair coherence limit. Above f_RPM: classical field–membrane interaction weakens but quantum spin effects become relevant.
The SAR form and its normalization are L1; only numerical tissue dielectric/material parameters such as conductivity and density are L3, while the supplied E is not reclassified by the formula. The Schwan form ΔV=1.5rE g(f) is L1; only the numerical value of τ_m is L3. Neither structure closes the Lindgren-to-biology L2 bridge.
ICNIRP 2010 sets exposure limits for f < 300 Hz (ELF). ICNIRP 2020i sets limits for f > 100 kHz (RF). The range 300 Hz < f < 100 kHz has overlapping, inconsistent limits. LED driver emissions (20–300 kHz) fall in this gap. A 2022 systematic review (IJRB, Ohkubo & Okano)i confirmed: 'studies on health effects with more diverse perspectives of IF-EMF have NOT been conducted.' Biological relevance at these frequencies is supported by: IFO threshold 10⁻⁵ V/m exceeded by LED drivers at 1 m; Kim 2026i gene expression activation at 4 kHz (Cyb5b); TTFields FDA-approved cancer treatment at 200 kHzi; 150 kHz rat testicular effects (Heliyon 2022)i.
The three-channel decomposition is a BERM candidate partition motivated by distinct frequency-dependent mechanisms. Its cutoffs, tissue transfer and diagnostic weights (w_ELF = 0.05, w_IF = 0.60, w_RF = 0.35) require empirical calibration; the weights are not Lindgren-derived or FieldState outputs.
The locked v17 comparison route uses the following two-channel technology-timing proxy. It is a BERM predictor specification, not a FieldState measurement or a derived biological response law:
In the three-channel decomposition, cumEMF becomes a weighted sum of frequency-specific cumulative exposures:
Channel weights are frequency-specific and tissue-dependent (see §2b). The single-channel cumEMF above is the weighted aggregate of the three channels. Channel weights (0.05/0.60/0.35) are diagnostic estimates requiring empirical calibration.
The IF channel response is the sum of three mechanisms operating at different intensity regimes and frequency bands:
R_IFO — Ion Forced Oscillation: linear in E_ext, threshold 10⁻⁵ V/m, dominates at environmental levels (0.01–3 V/m). Polarized, coherent IF fields force irregular gating of voltage-gated ion channels (Panagopoulos 2025i).
R_DEP — Dielectrophoresis: quadratic in E_ext, dominates at TTFields therapeutic levels (100–300 V/m). Requires high field gradients for translational force on intracellular structures.
R_Cyb5b — Mitochondrial outer membrane transduction: Cyb5b identified via genome-wide CRISPR screen as an EMF sensor (Kim et al. 2026, Celli). 60 Hz pulsed EMF → Cyb5b conformational change → Ca²⁺ oscillations → gene promoter activation. Operates at ELF frequencies (50/60 Hz) and couples the ELF channel directly to gene expression control — a pathway independent of both IFO and RPM.
At environmental intensities R_IFO ≫ R_DEP → linear response. At therapeutic intensities R_DEP ≫ R_IFO → quadratic response. R_Cyb5b adds an ELF-specific gene-regulatory pathway that operates independently of membrane ion channel gating. The intensity gap between therapeutic devices and environmental exposure does not exist — it is an artifact of assuming DEP is the only mechanism.
Ambient infrastructure, personal-device histories and community labels are TECHNOLOGY_TIMING_PROXY inputs, not A, x, physical-field measurements or physical dose. They may enter a declared empirical scenario only.
The dual-kernel convolution, its two kernels, time constants and mixing weight are L3 phenomenology. Using it after an L1 geometric quantity does not demote that earlier L1 result, and it does not promote DKC to L1.
Biological capacity declines exponentially as a function of cumulative exposure, with a threshold below which repair mechanisms compensate:
where (pre-EMF baseline TFR), (decline parameter), (threshold).
BERM separates total testosterone production from binding-dependent availability, tissue compartment, receptor occupancy and post-receptor transmission. This allows reduced hormone use even when a serum total-T assay is unchanged.
Canonical individual → population → institution operator
The current BERM closure models behaviour first as a state-by-context individual probability. It does not assign one deterministic political direction to a hormone.
Institutional persistence is represented separately as I_{t+1}=ρI_t+(1−ρ)P_t. The coefficients of the individual response, population distribution and institutional retention remain open. Aggregate outcomes cannot be inverted into individual hormone diagnoses.
Archived v17 scalar comparison implementation: The endocrine vector (testosterone, oxytocin, dopamine, cortisol, vasopressin) as a geometric mean:
where (OT), (T), (DA), (cortisol), (vasopressin/AVP).
Additionally, cortisol modulates effective testosterone:
TTFields clinical datai reveals a quantitative relationship between cell size and optimal disruption frequency. This relationship is calibrated from FDA phase III data and extrapolated to BERM's target tissues.
where where K ≈ 3.7 Hz·m, calibrated from TTFields clinical data across four cancer typesi.
The predicted TFR combines all three layers:
where is the biologically derived compensation exponent.
The model’s total derivative with respect to EMF is the product of six partial derivatives. If any one factor is zero, the entire chain breaks:
The model produces specific, locked predictions that will either come true or not. The lock is irrevocable: a prediction cannot be changed retroactively without a version number update.
| Country | Year | Metric | Central | 95% CI | Locked |
|---|---|---|---|---|---|
| Finland | 2030 | TFR | 1.17 | 1.02–1.24 | 2026-08-18 |
| South Korea | 2030 | TFR | 0.6 | 0.48–0.72 | 2026-08-18 |
| South Korea | 2035 | TFR | 0.52 | 0.40–0.64 | 2026-08-18 |
| Japan | 2030 | TFR | 1.04 | 0.88–1.20 | 2026-08-18 |
| USA | 2030 | TFR | 1.45 | 1.25–1.65 | 2026-08-18 |
| Brazil | 2030 | TFR | 1.55 | 1.40–1.68 | 2026-08-18 |
| Global | 2040 | TFR | 1.78 | 1.55–2.05 | 2026-08-18 |
| Global | 2050 | Sperm % | 62 | 48–75 | 2026-08-18 |
Predictions frozen at v17.0 git SHA. If future observations fall outside the CI, the model is falsified — not the prediction adjusted.
The model is explicitly falsifiable. Each condition is specific and testable:
Lindgren’s metric is mathematically incorrect
If the ansatz g_μν = η_μν + κA_μA_ν is shown to be internally inconsistent or to contradict established electrodynamics, the geometric foundation fails.
VGCC blockers do not prevent EMF’s biological effects
If calcium channel blockers fail to attenuate EMF-induced ROS, SDF, or hormonal changes in controlled experiments, the primary mechanism is wrong.
Amish community TFR declines at the same rate as the general population
The Amish label is an L3 community/technology-timing proxy, not a measurement of A, x or zero physical exposure. The comparison can test a preregistered proxy association, but testing χ requires an explicit L0→L2 measurement and mapping operator.
Sperm concentration decline stops without reduced EMF exposure
If the −1.2%/year sperm decline reverses or stabilizes while cumulative EMF continues to increase, the dose-response relationship is wrong.
A locked prediction fails outside its confidence interval
Any prediction in §8 that falls outside its 95% CI when the observation year arrives falsifies the model at that prediction’s scope.
Three independent pharmacological interventions provide quantitative calibration anchors for separate pathways. Each drug isolates a specific mechanism, allowing the model’s pathway structure to be tested independently.
| Drug | Target | Pathway | Observed effect | BERM calibration |
|---|---|---|---|---|
| CCB (nifedipine) | L-type VGCC | A (VGCC→ROS→SDF) | 90% VGCC block → −23% sperm conc. | EMF disruption ≈ 6% |
| Rapamycin | mTOR (85% inhibition) | Sempou (mTOR→aging) | Lifespan +10–25% (mice) | mTOR_eff × 0.15 |
| Melatonin | CRY/circadian | C (CRY→clock→ovulation) | Restores circadian amplitude | Night EMF fraction correction |
BERM proposes VGCC genotype, anatomical transfer and cumulative state as endpoint-specific response modifiers. They belong to the tissue kernel Ξ_i; they are not χ_geo and their joint gain is not calibrated:
If calibrated, these factors could produce response heterogeneity among people in the same measured field. Neither an order-of-magnitude spread nor its propagation to population TFR is currently established.
Formula discovery across 54 countries (2022 data, TFR range 0.78–6.25, sd = 1.35) provides an independent validation of the temporal model. The cross-sectional formula uses two EMF proxy variables and one binary threshold to predict national TFR with LOOCV RMSE 0.522 (skill score 0.61 vs mean predictor).
The two-channel EMF index combines residential electricity consumption (ELF proxy) and fixed broadband subscriptions (RF proxy):
Electricity access is an L3 technology-timing proxy, not a measured physical or biological exposure boundary. A binary split cannot identify A or x without an explicit L0→L2 measurement and mapping operator.
Validation statistics:
For partially electrified countries, the electrified sub-population TFR can be estimated from the binary mixture model:
Honest assessment
The cross-sectional fit mainly follows the demographic-transition gradient. The reported OECD association with electricity alone is near zero (R²≈0.0002); neither the full-sample fit nor an electrified/unelectrified split estimates a physical dose-response or establishes an EMF-specific biological threshold. The useful prediction is whether measured local fields and biological mediators add out-of-sample information beyond the demographic and technology proxies.
Replication data: 54-country sample roster with observed TFR, electricity consumption, broadband subscriptions, and model predictions available at /data/cross_section_manifest.csv. Source: UN WPP 2024 (TFR), OWID/IEA (electricity), ITU (broadband).
Cross-sectional analysis cannot determine causal direction. A discriminating test needs preregistered natural experiments or sentinel designs with direct dosimetry and an explicit L0→L2 map; an unelectrified label alone is only a proxy.
BERM proposes a population closure in which separately measured environmental, membrane, optical and molecular moderators can vary between groups. These m-functions are not instances of χ_geo and the combined reproductive response below is an uncalibrated candidate:
Here γ_A and γ_B are candidate pathway weights; m_env, m_mem, m_opt and m_mol are distinct measured or estimated moderators. The archived values 0.75/0.25 and population profiles are scenarios, not Lindgren-derived coefficients.
The discriminating prediction is an exposure × moderator interaction measured prospectively. Eye colour or lactase persistence alone must not be treated as a calibrated biological susceptibility or as proof of a TFR effect.
Epistemic level: L* (testable BERM synthesis). Component biology can constrain individual moderators, but the population integration and its endpoint coefficients remain uncalibrated.
The original cross-sectional formula (v19.1) uses a two-channel EMF index. The layered formula extends this by incorporating priming history, recovery capacity, seasonal modulation, and population genotype.
TFR ≈ A × exp(−B × EMF_eff) + C
EMF_eff = EMF_comp × P × (1/R)
EMF_comp = w_ELF × ELF + w_IF × IF + w_RF × RF
P = 1 + α × min(electrification_years, P_max)
R = 1 + β × EMF_free_hours/day
Where EMF_comp is the three-channel weighted composite (ELF < 300 Hz, IF 300 Hz–1 MHz, RF > 1 MHz). P captures cumulative priming from decades of power grid exposure — years of electrification upregulate VGCC expression, making cells more sensitive to all subsequent EMF. R captures the recovery window: hours per day without significant EMF allow CaMKII dephosphorylation and Ca²⁺ homeostasis restoration.
EMF_eff = EMF_comp × P × (1/R) × S × G_pop
S = 1 + γ × f(latitude, season)
G_pop = 1 + δ × CACNA1C_A_allele_frequency
Optional correction factors (data-dependent): H = humidity/coastal correction, B = building material RF reflection coefficient
S captures seasonal variation in CRY magnetoreceptor sensitivity: winter at high latitudes increases CRY sensitivity to EMF perturbation (Halgamuge 2015i). G_pop captures population-level genetic susceptibility via CACNA1C rs1006737 A-allele frequency, which determines Cav1.2 channel density and therefore Ca²⁺ response per EMF stimulus (Sousouri 2025i).
| Parameter | Amish | Finland | Nigeria | Interpretation |
|---|---|---|---|---|
| P (Priming) | 1.0 (no priming) | 2.2 (100+ yr electrification) | 1.45 (~15 yr) | How 'ready' are cells for EMF response |
| 1/R (Recovery deficit) | 0.48 (full recovery) | 1.0 (WiFi 24/7) | 0.67 (partial) | Does Ca²⁺ homeostasis restore overnight |
| S (Season) | ~1.0 | 0.9–1.3 | ~1.0 | CRY sensitivity modulation by light |
| G_pop (Genotype) | ~1.0 | ~1.1 | ~0.95 | Population CACNA1C A-allele prevalence |
Formula evolution: v17 (scalar cumEMF, RMSE ~1.15) → v19.1 (two-channel, 54 countries, RMSE 0.522) → v20 (+ Priming × Recovery, predicted RMSE < 0.45) → v21 (+ Season × Genotype, requires calibration data).
Component status: v20 and v21 are L3 phenomenological scenario models; calibration of S, G_pop and the fitted weights remains pending. These layers neither alter the preceding L1 geometry nor resolve the open L2 mapping.
Ivancsits et al.i demonstrated that EMF-induced DNA strand breaks returned to normal within 9 hours after exposure ceased. Fitting an exponential decay model to this data yields a time constant τ ≈ 3–4 hours. This maps directly onto the Recovery factor R in formula v20: R = 1 + β × EMF_free_hours, where Ivancsits datai suggests β ≈ 0.11.
| Time after exposure | Remaining damage |
|---|---|
| t = 0h | 100% |
| t = 4h | ~37% |
| t = 9h | ~0% (recovered) |
| Scenario | EMF-free time | Remaining damage |
|---|---|---|
| Modern bedroom (WiFi + phone) | t ≈ 0 | damage persists |
| EMF-free bedroom | t ≈ 8h | ~14% remaining |
Cultural energy — the aggregate capacity for civilizational achievement — has been described qualitatively by Unwin (1934), Spengler (1918), Glubb (1978), and Turchin (2003). BERM provides the first quantitative decomposition.
The fundamental definition: CulturalEnergy(t) = N(t) × BioCap(t) × η(t), where N(t) is population size, BioCap(t) is the mean biological capacity of the population, and η(t) is institutional efficiency (a factor between 0 and 1 capturing governance quality, education systems, legal frameworks, etc.).
BioCap itself decomposes as a weighted sum of eight normalized biomarkers: BioCap(t) = Σᵢ wᵢ · Bᵢ(t), where each Bᵢ(t) ∈ [0,1] represents a biomarker’s current level relative to its pre-industrial baseline and wᵢ is the biomarker’s weight in the cultural energy budget; cortisol enters as (1 − B_CORT), so the absolute weights sum to 1.0 and BioCap spans [0,1] with 1.0 as the pre-industrial optimum.
The biomarker weights (T=0.20, OXT=0.20, DA=0.15, MEL=0.15, BDNF=0.10, CORT=−0.10, D=0.05, B2=0.05) are selected based on theoretical significance and proxy evidence, not empirical regression. The critical triad (T + MEL + OXT) accounts for 55% of total weight.
Phase transition thresholds: Rationalistic > 0.90, Deistic > 0.75, Manistic > 0.55, Zoistic < 0.55. These thresholds are fixed constants (unwin_validation.py); the Western trajectory crossed 0.90 in 1983 and 0.75 in 2007, and is projected to cross 0.55 in 2040. The 2025 Western estimate (BioCap = 0.614) places civilization in the Manistic phase.
The sensitivity analysis reveals that restoring any single biomarker to its optimum (1.0; cortisol to 0.0) closes part of the gap between the 2025 BioCap (0.614) and the maximum (1.0): T → 23.3%, OXT → 19.9%, MEL → 16.4%, CORT → 13.9%, DA → 12.8%, BDNF → 6.7%, D → 4.5%, B2 → 2.5% of the gap. The combined triad recovery (T + OXT + MEL) is 59.6%. EMF reduction is the only known single intervention that would affect all biomarkers simultaneously, because all are downstream of the EMF-induced biomarker cascade.
Epistemic note: The equations above are the current model specification (BERM v17). Parameter values are calibrated against observed data and will be updated as new evidence becomes available. The model is explicitly designed to be falsifiable -- if its predictions fail, the model is wrong. The Therapeutic Device Paradox (24+ regulatory-approved non-thermal EMF device categories, DC to UV) establishes non-thermal bioactivity as regulatory fact, not hypothesis.
Formal Jacobian product structure (chapter 17), proof-obligation register and safety systems are described in the base document (LBERM_final.docx).