The Korean paper explicitly says that its study was not designed as a complete replication of NTP, while the Japanese paper says that its study was not intended as a direct replication. Both tested male Sprague Dawley rats using only 900 MHz CDMA modulation, at only one nominal whole-body SAR—4 W/kg. Neither study included GSM, a lower-SAR arm, a dose-response curve, female animals, or Ramazzini-range far-field exposures.

Nor did these studies find “no tumors.” In the Korean study, endocardial schwannomas occurred in 2 of 70 exposed rats and in none of either control group; one glioma occurred in each group. In the Japanese study, one endocardial schwannoma occurred in each group, and one glioma occurred among the 68 analyzed exposed rats. Those differences were not statistically significant, so they cannot be presented as evidence of an exposure effect—but they also cannot honestly be summarized as “there was no cancer.” The accurate statement is that no statistically significant exposure-related increase was detected under that particular protocol.

The Korean investigators themselves acknowledged that 70 males per group provided limited statistical power for rare tumors. They stated that an approximately 20% increase in cardiac schwannomas or malignant gliomas would not readily be detected with that group size.

Why the NTP response space remains substantially untested

NTP used both GSM and CDMA modulation at 0, 1.5, 3, and 6 W/kg. In male rats, the GSM malignant-glioma counts were:

0 → 3 → 3 → 2

The GSM malignant-heart-schwannoma counts were:

0 → 2 → 1 → 5

The CDMA malignant-heart-schwannoma counts were:

0 → 2 → 3 → 6

The CDMA malignant-glioma counts were:

0 → 0 → 0 → 3

The GSM adrenal-medulla pheochromocytoma pattern was also nonmonotonic:

11 → 24 → 28 → 14

Thus, the clearest mid-dose brain and adrenal patterns were not reproduced—or even tested—by placing one CDMA point at 4 W/kg.

Ramazzini occupied an entirely different region of the response surface: 1835 MHz GSM at estimated whole-body SARs of 0.001, 0.03, and 0.1 W/kg, with exposure beginning prenatally and continuing for nineteen hours per day until natural death. Its statistically significant male heart-schwannoma result occurred at 0.1 W/kg—forty times below the nominal 4 W/kg used by Japan and Korea.

Consequently:

A 4 W/kg CDMA null cannot adjudicate a 1.5–3 W/kg GSM brain signal or a 0.1 W/kg GSM far-field heart signal.

It tests neither the relevant modulation nor the relevant intensity window.

The boundary-layer interpretation

The term boundary-layer null is useful because 4 W/kg is not simply “more of the same exposure.” It was selected because it is the animal reference point historically used in constructing human RF limits. Separate W-CDMA work by Ohtani and colleagues found that 4 W/kg could elevate rat core temperature by approximately 1.5 °C and upregulate some heat-shock and heat-shock-factor genes after six-hour exposures, whereas comparable changes were not found at 0.4 W/kg or after the shorter 4 W/kg schedule. That demonstrates that, under at least some exposure systems, 4 W/kg can move the organism into a qualitatively different thermoregulatory and cellular-stress state.

That does not prove that heat-shock protection masked cancer in the 2026 experiments. The long-term studies did not measure HSP70, HSF1, Nrf2 activation, antioxidant enzymes, mitochondrial compensation, or ROS at sequential intervals throughout the two-year exposure. Body temperature was assessed only during the preliminary 28-day phase, not longitudinally in older and heavier animals. The Korean authors specifically listed that restriction and the absence of a dose-response analysis among their limitations.

The studies nevertheless showed that the exposed animals were not physiologically indistinguishable from controls. Korean animals had persistent food-consumption and body-weight differences, and the authors themselves speculated that periodic RF exposure might have provided thermal energy. In Japan, the exposed animals consumed less food, had lower body weights through much of the study, and had markedly higher terminal survival—64.7%, versus 34.3% for cage controls and 42.9% for sham controls. The Japanese authors recognized that the feeding, weight, and survival differences could alter the background occurrence of late-life tumors.

That makes the scientifically supportable boundary-layer statement:

At 4 W/kg, the animals may have occupied a metabolically and thermoregulatory distinct state in which stress-response, antioxidant, feeding, body-weight, survival, or repair dynamics differed from those operating at lower SARs. Because the necessary time-course biomarkers were not measured, the studies cannot determine whether compensation masked an earlier or lower-intensity perturbation.

Why the transient-response studies matter

Durdik and colleagues found an increase in ROS immediately after one hour of UMTS exposure that was no longer evident after three hours, with no persistent DNA damage or apoptosis under the tested conditions. Importantly, the authors noted that changing oxygen levels in the cell medium could have contributed to the time difference, so that experiment supports time dependence, but does not by itself prove biological adaptation.

The 2025 Jamaludin experiment gives more direct support to an adaptation interpretation. In that small study—six rats per group—MDA, sperm impairment, and testicular injury were generally greatest in the four-hour exposure group, followed by improvement at eight and twenty-four hours. The authors interpreted the pattern as possible activation of antioxidant and cellular-repair mechanisms. The study is too small and too different from the cancer bioassays to prove an NTP mechanism, but it demonstrates why a single terminal observation can miss a dynamic biological trajectory.

Together, these studies support the experimental principle:

The biological response must be measured as a time course—not merely as a terminal endpoint.

An early oxidative or calcium-signaling disturbance, subsequent compensation, and eventual terminal pathology are three different biological moments. A study that measures only the last one cannot tell us whether the first two occurred.

Replacement section for the manifesto

The 2026 Japan–Korea Studies: A Boundary-Condition Null, Not a Replication of the NTP Response Surface

The 2026 Japanese and Korean studies should not be described as negative replications of the National Toxicology Program bioassay. Both investigations were carefully performed GLP toxicology studies, but their designs tested only one narrowly defined coordinate: male Sprague Dawley rats exposed to 900 MHz CDMA-modulated RF at a nominal whole-body SAR of 4 W/kg. The Korean authors expressly stated that their experiment was not a complete NTP replication, and the Japanese investigators stated that theirs was not intended as a direct replication. Neither study tested GSM modulation, NTP’s 1.5 or 3 W/kg exposure groups, Ramazzini-range far-field intensities, female susceptibility, or a dose-response relationship.

Their finding was therefore narrow but legitimate: under that particular 4 W/kg CDMA protocol, no statistically significant exposure-related increase in tumor incidence or genotoxicity was detected. That is not the same as finding no cancer. The Korean exposed group contained two endocardial schwannomas where neither control group contained one, although the difference was not statistically significant. The Korean authors also acknowledged that seventy animals per group provided limited power for detecting rare tumors and would not readily detect an approximately 20% increase in heart schwannomas or malignant gliomas.

Most importantly, the experiments did not test the parameter regions in which the clearest nonlinear signals appeared. NTP’s male-rat GSM malignant-glioma counts were 0, 3, 3, and 2 across 0, 1.5, 3, and 6 W/kg. Its GSM adrenal-medulla tumor pattern also rose at the intermediate levels and declined at 6 W/kg. Ramazzini reported a significant increase in male cardiac schwannomas at an estimated whole-body SAR of only 0.1 W/kg under lifetime GSM exposure. The animal evidence therefore cannot be represented by a simple assumption that more average power must always produce more disease. The response appears dependent upon the biological endpoint, tissue, modulation, intensity, timing, and state of the receiver.

A null result at one point cannot falsify a nonlinear response surface. A CDMA-only result cannot falsify a GSM-sensitive hypothesis. A 4 W/kg result cannot resolve a 0.1, 1.5, or 3 W/kg window. And a terminal pathology assessment without longitudinal calcium, oxidative, mitochondrial, heat-shock, and repair measurements cannot determine whether an early biological disturbance occurred and was later compensated for.

The 4 W/kg condition may itself represent a biological boundary layer. Independent work has shown that 4 W/kg can engage measurable thermal and heat-shock responses under some RF protocols. The 2026 studies did not measure the necessary stress-response biomarkers over the two-year exposure period, yet they did report differences in food consumption, body weight, metabolic state, and—in the Japanese experiment—survival. Those physiological shifts demonstrate that 4 W/kg should not automatically be treated as a neutral extension of a lower nonthermal exposure regime.

The appropriate scientific classification is therefore:

A 4 W/kg CDMA boundary-condition null: no statistically significant carcinogenic or genotoxic effect detected under one high-SAR, single-modulation protocol, with lower-SAR, GSM-modulated, far-field, nonlinear, tissue-selective, and adaptation-dependent windows left unresolved.

Density Gating: Why the Recurring Target Tissues Matter

The animal tumor pattern is not simply a random elevation of all cancers in all tissues. The repeatedly implicated targets form a biologically coherent cluster: cardiac Schwann cells, brain glia, and the adrenal medulla. The Korean investigators themselves described the three NTP target tumors—glioma, cardiac schwannoma, and pheochromocytoma—as being linked to the nervous system. The 2025 WHO-related animal systematic review independently judged the certainty of evidence as high for increased gliomas and malignant heart schwannomas in male rats.

This is precisely where RF Safe’s density-gating concept becomes explanatory. The S4–Mito–Spin framework does not predict that every cell and tissue must respond equally. It predicts that vulnerability will rise where several conditions converge:

high densities of voltage-sensitive signaling structures;
intensive calcium handling;
strong mitochondrial or NADPH-oxidase amplification;
spin-sensitive redox chemistry;
high energetic demand; and
limited or state-dependent antioxidant and repair reserve.

RF Safe defines the S4–Mito–Spin chain as S4 voltage-sensor timing disturbance, mitochondrial and NOX redox amplification, and spin-sensitive radical or electron-transfer chemistry. Density gating adds the proposition that the local abundance, organization, and operating state of those systems determine whether a particular tissue crosses a response threshold.

A cardiac schwannoma is not a tumor of the contractile heart muscle. It is a tumor of Schwann cells associated with neural structures within the heart. That distinction strengthens rather than weakens the tissue-selectivity argument. The affected cell lineage is involved in maintaining electrically active nerves inside an organ characterized by continual electrical excitation, calcium cycling, high mitochondrial demand, and tightly controlled redox metabolism.

Likewise, glial cells do not merely occupy empty space around neurons. They maintain ion balance, metabolic support, calcium communication, repair, and the extracellular environment required for coherent neural activity. Adrenal-medullary chromaffin cells are excitable neuroendocrine cells in which membrane depolarization and voltage-gated calcium entry control catecholamine release. Voltage-gated calcium channels are established transducers that convert membrane-potential changes into intracellular calcium signals in neural, cardiovascular, and neuroendocrine systems.

Density gating can therefore be expressed conceptually as:

Biological response probability ≈ field-pattern coupling × susceptible-target density × mitochondrial/redox gain × current biological state ÷ compensatory and repair capacity

This is not yet a completed quantitative law. It is a falsifiable prediction about where effects should cluster and why the same nominal SAR may yield different outcomes in different tissues, at different intensities, under different modulation patterns, or at different points in the exposure timeline.

The resulting response curve need not be monotonic. Below one threshold, coupling may be too weak to produce a measurable change. Within an intermediate window, the field may disturb S4-dependent timing, calcium dynamics, or redox regulation without fully activating systemic defensive programs. At a higher boundary condition, thermoregulation, heat-shock signaling, antioxidant induction, altered feeding, metabolic adaptation, or repair may suppress or obscure the earlier endpoint. At still greater exposure, overt thermal injury may dominate.

That is not an excuse for every null result. It is a specific experimental model that can be tested by measuring both disturbance and compensation over time.

The two manifesto lines that should change

Instead of:

“Experimental animals have produced both serious positive findings and important negative replications.”

Use:

Experimental animals have produced serious positive findings alongside tightly bounded 4 W/kg CDMA nulls that do not test the lower-SAR, GSM-modulated, far-field, nonlinear, or adaptation-dependent response windows identified by NTP, Ramazzini, and subsequent mechanistic research.

And instead of:

“It is established that the 2026 Japanese and Korean studies did not reproduce those findings.”

Use:

It is established that the 2026 Japanese and Korean studies detected no statistically significant tumor increase under one 900 MHz CDMA protocol at 4 W/kg. Their designs did not reproduce the NTP dose-response surface and cannot adjudicate GSM-specific, lower-SAR, Ramazzini-range, tissue-density-gated, or time-dependent compensation hypotheses.

The decisive next experiment

To move the boundary-layer concept from a strong interpretation to a direct experimental test, the next bioassay should pre-register:

• both GSM and CDMA arms;
• multiple points below, within, and above the suspected window;
• Ramazzini-range far-field groups;
• both sexes and sufficient numbers for rare tumors;
• continuous temperature and metabolic monitoring;
• sequential calcium, mitochondrial, ROS, HSP70/HSF1, Nrf2, antioxidant-enzyme, lipid-peroxidation, and oxidative-DNA-damage measurements;
• prospective tissue mapping of S4-bearing channels, calcium-handling machinery, mitochondrial/NOX density, and repair capacity.

The central testable prediction would be:

Intermediate, modulation-sensitive exposures will produce stronger timing, calcium, or redox disruption before full compensatory signaling is engaged, while the 4 W/kg condition will show a different balance between primary perturbation and thermometabolic defense. The strongest persistent effects will cluster in tissues whose density-gated receiver architecture provides both efficient coupling and strong redox amplification.

That makes the RF Safe position more rigorous, not less forceful. The 2026 papers do not disprove S4–Mito–Spin. Properly interpreted, they help identify one boundary of the parameter space that the framework says must be mapped.