A rate through time. The same total can be delivered in different sequences.
A bounded observer / a real safety question
The total is the same.
The future can differ.
A single dose number cannot tell you when exposure arrived, what remained from before, or what another exposure would meet. A candidate continuous SNT model keeps those histories visible, turning “safe” into a question you can actually test.
D This chapter shows a conditional mathematical model and documented decisions. Its interactive values have no physical units and make no personal radiation-risk estimate.
01BIG → SMALL1.00
02SMALL → BIG1.00
SAME TOTAL · DIFFERENT STATE · SAME NEXT TEST
01 / The experimentReverse the history.
Reverse the history.
Keep the total fixed.
D Two equal-length exposure periods deliver 0.80 and 0.20 normalized units. Swap their order. Change the time between them or the recovery time. The model evolves throughout delivery and waiting.
CONTINUOUS SNT CANDIDATE / TWO HISTORIES0.80 + 0.20 = 0.20 + 0.80
EXPOSURE → RETAINED STATE → RESPONSETWO HISTORIES
Big then smallSmall then big
The total-only ledger1.000 = 1.000It cannot distinguish these histories.
Retained state after the pair— / —Warm / cool · normalized units
Model response after the pair— / —Warm / cool · unitless ledger, not a risk percentage
D Waiting changes the state that the second exposure meets. The colored paths are model outputs, not measured human outcomes.
P The sum-only reading is exactly equal by construction. In the quadratic setting, the two-period response can also agree while retained states differ. The identical third exposure then separates the modelled futures. See the mathematical checks.
H Whether a real biological endpoint follows this continuous rule must be tested for a specified population, preparation, exposure range, timing and outcome. A failed recovery prediction or challenge-curve translation would reject that proposed fit.
02 / What makes this continuousA curve is not a clock.
A curve is not a clock.
It needs a state.
D The candidate continuous SNT construction keeps the state that remains while dose arrives. Recovery and the acute response must be specified separately. The acute curve alone does not tell us how fast a body recovers.
Past input persists and recovery changes what later input meets. This demo sets R(x) = x / τ.
The candidate response potential h gives the increment at the current state.
P In the instantaneous-pulse limit, a dose d arriving at state x adds h(x + d) − h(x). This follows from integrating the displayed response rule during a pulse. For the fixed illustrative rules above, the numerical time-step check passes.
D The law is useful because it tells us what to measure next: an acute response curve, a separately measured recovery process, and matched future challenges. An order effect rejects a specified order-blind summary for that endpoint; it does not establish this mechanism by itself.
Same total. Same gap. Reverse the order.
P A reproducible difference in a declared endpoint rejects a total-only or otherwise order-blind prediction for that tested setting.
Measure recovery before the next dose.
H Do the later challenge curves follow the state transition predicted from a separate recovery measurement?
Make the future expose the missing state.
P A matching pair of endpoints need not mean matching states. The same later intervention can distinguish them.
H Continuous SNT is a proposed model, not a validated replacement for radiation-protection guidance. The 6 + 12 versus 12 + 6 Gy example in Murray’s order-reversal paper concerns specific mouse tumour and immune endpoints in radiotherapy; it cannot be transferred directly to low-dose public exposure.
03 / The knowledge horizonSafety sharpens
Safety sharpens
as the record deepens.
D Each extra fact can remove exposure histories compatible with a record. Under one fixed, correct model, the set of compatible predictions can only stay the same or shrink. This is a statement about information; it is not a promise of perfect certainty.
COMPATIBLE FUTURES / FIXED MODEL01 / 05
Only the total is known
Many dose-rate histories fit one number. Their predicted retained states and later responses need not agree.
Uncertainty remains: timing, recovery, person, endpoint, model fit.H More context can improve a conditional prediction, but unobserved facts, variation among people, uncertain model structure and future chance remain. “Infinitely sure” would require assumptions that real safety decisions cannot verify. A failed model prediction can widen the horizon again.
04 / One place, two clocksFukushima.
Fukushima.
Protecting life had two histories.
D In March 2011, decision makers had to limit radiation exposure while preserving transport, shelter, water, medicine and continuous care. Later evidence lets us see more of both risks. It does not give any one person a perfect counterfactual.
INSIDE THE DECISIOND A reconstruction of competing decision variables, not a transcript of any official’s private reasoning.
Leave, shelter, or move with care?
The radiological desk must act before individual future exposures are known. The hospital ward cannot treat a fragile patient as a dose number: transport and interruption of care have their own immediate hazards.
Radiation clockWhat exposure follows each route and delay?
Care clockCan oxygen, fluids, staffing and safe transport continue?
THE RETROSPECTIVE FRAME
Both risks were real.
P UNSCEAR modeled a Futaba scenario in which evacuation lowered first-year adult dose from about 43 mSv without evacuation to 0.35–2.5 mSv with evacuation. These are modeled group counterfactuals, not individual measurements. Dose table A-18.1 ↗
P A retrospective study of Futaba Hospital records found 39 of 338 patients died before evacuation was complete, with one missing. Care interruption and difficult transport were documented; the study does not isolate each death’s cause. Hospital study ↗
P A separate hospital study found higher mortality among medically vulnerable people who remained without adequate support. “Stay” alone was not a safe answer either. Shelter study ↗
D The defensible lesson is to compare exposure avoided against the care a protective action can actually sustain, in that setting and with those people.P UNSCEAR reports no documented resident health effects directly attributable to radiation and none expected to be detectable at population scale. That does not mean zero individual risk. UNSCEAR FAQ
D The continuous SNT interaction above has not been validated as a Fukushima risk calculator. This scene instead applies Gate 2’s simpler discipline: a useful safety state must preserve both exposure history and the conditions of care. OECD NEA retrospective
05 / The evidence behind the experienceA testable law.
A testable law.
A boundary on its claims.
The model and the historical case have different kinds of evidence. The links here keep their limits visible.
The state-recovery paper
D Murray derives the candidate continuous construction and the context-refinement result. Recovery and human transfer remain to be established.
Read Murray’s paper ↗Equal BED, unequal biology
D A precise order-blind null and an endpoint-specific radiotherapy example. It sets a test; it is not a population-safety result.
Read Murray’s paper ↗Jack Devanney’s Rate SNT
P Jack’s continuous SNT article presents a worked rate model and explicitly credits Daniel Murray for its mathematical development. Its numerical scenarios are model outputs.
Read Jack’s article ↗Fukushima and evacuation
D Dose reconstruction, hospital records and later care studies expose different parts of one safety decision.
UNSCEAR dose reconstruction ↗Futaba Hospital study ↗Unsupported shelter study ↗Methods, assumptions, source qualifications and loss conditions: radiation safety source ledger. Continue the observer story in Inside, or examine the claim register in the Atlas.