Wartime radar has crossed more than 80 light-years. Its recoverable range, using the best current or committed receiver references in this study, ends before the closest star to us.
Between 1935 and 1945, Earth developed a rapidly expanding network of pulsed directional radar transmitters. By 2026, those emissions had travelled approximately 81 to 91 light-years. They passed Proxima Centauri, the closest star to the Earth, in the 1940s and continued forward through interstellar space.
The arrival of those emissions, however, is not by itself notable. Radio waves will propagate independently of a receiver can distinguish them from noise. The pressing question is as follows: what receiver would be required to recover the strongest defensible wartime radar signal at a real stellar distance?
The first release of The War Reaching the Stars answers that question.
With supported phase-insensitive processing, ten years of observation and the radar beam held continuously on the observer, the recovery horizon is only 0.10 light-years. That is 2.45% of the distance to Proxima Centauri.
Even the ideal upper bound does not reach the nearest star. If every pulse could be added perfectly in phase for ten years while the beam remained continuously on the observer, the strongest source-backed horizon would reach 2.42 light-years. Proxima Centauri is 4.24 light-years away. No surviving source establishes the pulse-to-pulse phase stability required by that ideal calculation.
At a 1% beam dwell, the most favorable corner of the declared assumptions requires a receiver 30.7 times more sensitive than the best modeled current or committed reference. The alternative is time, given that under the same ideal-coherence assumption, that receiver would need about 9,420 years of observation.
The conservative end of the identified range is 40,182 times the present receiver reference.
These figures demonstrate how quickly free-space spreading overcomes even powerful directional leakage.
From Experiments to Wartime Infrastructure
During the interwar period in the 1930s, radio detection was increasingly put into operational use. The United Kingdom, for example, built the Chain Home warning network. The German Reich developed the Freya and Würzburg radar families. The United States fielded Signal Corps, naval and airborne systems under SCR and later AN designations. The Soviet Union developed the RUS (РадиоУлавливатель Самолетов or Radio Aircraft Catcher) series. The Empire of Japan, Kingdom of Italy, France, Canada, Australia and other countries built, adapted or operated their own equipment.
The war accelerated that transition. Radar moved from fixed early-warning stations into anti-aircraft fire control, night fighters, patrol aircraft, surface ships, submarines, coastal batteries and navigation systems. It was used to find aircraft before visual contact, direct guns, guide interceptions and search through darkness or poor weather.
The technology also changed during the same decade. Early warning systems often worked at metre wavelengths with large antennas and broad coverage. Microwave development allowed shorter wavelengths, narrower beams and equipment compact enough for aircraft or mobile fire-control units. A single category called “radar” therefore covered transmitters separated by orders of magnitude in frequency, different pulse lengths, different repetition rates and radically different scan patterns.
Those differences matter at interstellar distance. A fixed warning station, a rotating naval set, a spiral-scanning airborne radar and an automatically tracking gun-laying radar do not illuminate the same direction for the same amount of time. Production totals do not reveal operating hours. Installed equipment does not imply continuous transmission. Peak transmitter power does not describe the average power radiated over an entire day.
The signal shell is therefore not a uniform expanding broadcast. It is the accumulated geometry of short pulses sent through moving beams. An observer in one direction would receive intermittent crossings shaped by antenna pattern, scan law, operating schedule, frequency and transmitter stability. Recovering that structure is much harder than establishing that electromagnetic energy passed the observer.
The current public calculation begins with a small set of United States systems because those technical records were sufficient to construct the first controlled receiver comparison. That starting point is useful but historically incomplete. Britain, Germany, the Soviet Union, Japan, Italy, France and other wartime programmes must eventually be represented through their own technical and operational records. Allied intelligence summaries can help check captured equipment, but they cannot substitute for national production records, deployment histories or operating logs.
First Comparison
The headline result is anchored by the SCR-584, an American gun-laying radar introduced during the Second World War. Its documented transmitter peak power lies between 250 and 300 kilowatts. A measured antenna gain has not yet been recovered, so the calculation derives a gain bracket from the documented four-degree beam. Combining the transmitter-power interval with that gain bracket gives a peak effective isotropic radiated power between approximately 322 and 773 megawatts.
The upper end of that range is deliberately favorable to detection. The receiver references are also represented by favorable scalar sensitivities across their modeled bands. The result does not fail because the calculation selected a weak radar, a poor receiver or a short campaign.
It fails after giving the link budget substantial advantages.
The historical comparison contains five selected United States radar systems. A sixth system, the SCR-540, remains in the evidence register to test waveform and gain uncertainty but is excluded from the EIRP ranking. A deliberately exaggerated nonhistorical comparison case tests the outer boundary of the calculation. It is never treated as a radar that actually existed.
This is a source-backed seed set, not a claim to have reconstructed the complete global wartime network.
Why the answer spans 30.7 to 40,182
The receiver requirement is reported as a non-probabilistic identified range. It is not a confidence interval and it is not a detection probability.
The favorable endpoint combines the highest SCR-584 EIRP bound, perfect pulse coherence, 1% beam dwell, full operating time, quiet solar conditions and a known-signal threshold. The conservative endpoint uses the lower EIRP bound, no pulse-to-pulse coherence, 0.1% beam dwell, 25% operating time, an active-Sun background and a survey-wide threshold.
Pulse coherence is the dominant unresolved factor. Holding the other favorable assumptions fixed, removing ideal phase coherence raises the receiver requirement by roughly 171 times. Gain, dwell, operating time and the detection threshold then compound that penalty.
The width of the range does not weaken the current-array conclusion. The entire source-backed range remains outside present receiver capability at Proxima. The uncertainty determines whether the next receiver frontier is difficult or extreme. It does not turn the present result into a possible detection.
The calculation has established that current receiver resources do not suffice. The surviving evidence does not yet determine whether closing the gap is principally a difficult receiver-engineering problem or an extreme-scaling problem. That is the uncertainty the next historical work must resolve.
A result that had to survive correction
The released calculation was not produced by polishing the project’s original interface. The inherited implementation contained 27 recorded defects, including unit errors, placeholder physics, unsupported predictions and interface elements that implied capabilities the code did not possess. It was archived rather than quietly reused. The useful residual link-budget work was retained as a legacy demonstrator, but no production result reads its predictions.
The replacement began with a deterministic analytic screen. It evaluates band compatibility, free-space loss, pulse bandwidth, receiver sensitivity, beam dwell, operating fraction, three coherence models, solar and sky backgrounds, campaign duration and detection thresholds. The complete matrix contains 131,472 rows.
The first completed screen still required a scientific correction pass.
The most important correction concerned aggregate radar energy. Emitters occupying the same spectral region can contribute power to the same detector statistic. Emitters in disjoint frequency bands cannot be added as though they were one co-channel source. The initial aggregate calculation did exactly that and produced an invalid network value near 145 at Proxima. The corrected implementation clusters emitters by overlapping spectral support, adds power only inside each cluster and combines independent clusters in quadrature. The corrected SKA-Mid extreme network bound is approximately 104 across four spectral clusters.
That aggregate value is intentionally narrow in meaning. It assumes deliberately generous unit counts, continuous boresight exposure, ten years and no background penalty. It represents recoverable aggregate energy. It does not preserve individual pulse identity and therefore does not establish recovery of the historical radar record.
A second correction concerned probability language. An early version described a 90% bound, but the available records do not support defensible probability distributions for coherence, beam dwell, uptime or antenna gain. Computing a percentile without those distributions would manufacture precision. The release therefore reports the minimum and maximum across the stated combinations of assumptions. No probability is attached to the interval.
A third correction made provenance enforceable. Every source assertion must resolve to a source record with a title, issuing body, retrieval location, section locator and archived checksum. Missing source records now stop the build. Derived bounds and deliberately generous assumptions remain visible as different classes rather than being allowed to resemble measurements.
The final correction turned an abstract sensitivity deficit into an observing consequence. The release added a campaign-time solver across every radar system, current receiver, distance anchor, coherence bound and dwell assumption. That calculation produced the 9,420-year figure at Proxima. It is the clearest expression of the scale mismatch: the favorable ideal case is not a slightly longer observing proposal. It exceeds recorded human history.
None of these corrections reversed the scientific conclusion. They made the conclusion narrower, reproducible and harder to misread.
What Does the Result Indicate?
The result leads us to make four conclusions.
Firstly, the wartime signal shell and the detection frontier are radically different objects. The shell now extends beyond 80 light-years. The supported single-emitter recovery horizon is one tenth of a light-year.
Second, perfect processing does not rescue the present receiver comparison. The source-backed ideal ceiling still ends 1.82 light-years short of Proxima even with continuous pointing for ten years.
Third, realistic beam exposure changes an already negative result into a severe resource requirement. At 1% dwell, the favorable ideal case demands a 30.7-fold receiver improvement or about 9,420 years with the best modeled reference.
Fourth, the main unresolved historical quantity is not whether the wave arrived. It is how the transmitters behaved from pulse to pulse and how often their beams occupied any one direction. Those details determine the scale of a future receiver, not the adequacy of receivers available now.
This release does not estimate whether another civilisation was present, whether it was listening or whether it would interpret the signal as war. It does not treat the region inside the historical shell as a detection volume. It does not turn an analytic threshold into an empirical detection probability.
It establishes the receiver resources required under the stated historical assumptions and the separately identified comparison case.
Receiver-Frontier Visualization
The public receiver-frontier page renders the released result files directly. It does not recompute the science in the browser. The same underlying tables feed the expert feasibility note and the interactive charts.
The calculation is covered by 27 unit and regression tests. These include inverse-square scaling, band rejection, directivity ownership, coherence ordering, aggregate-frequency behavior, campaign-time solutions and receiver-threshold cases. A locked environment reproduces the result from one command and repeated clean runs produce identical output files. The sources, assumptions, limitations, result tables and figures are included with the calculation.
That machinery matters because the conclusion is a null result at current receiver scale. A null result is only useful when the assumptions that produced it remain inspectable.
Open the original standalone receiver-frontier visual
What comes next
The immediate work returns to the historical record. Every value used in the feasibility calculation must be tied to the exact document, page, measurement or declared bound that supports it. Conflicting values must remain visible rather than being averaged into false certainty. Antenna gain, efficiency and coherence assumptions must stay attached to the radar system from which they came.
Three targeted searches come first because they address the uncertainties that most strongly control the receiver requirement:
- A measured SCR-584 antenna gain or radiation pattern to replace the present beam-derived gain bracket.
- Transmitter-architecture evidence that constrains pulse-to-pulse phase stability for the magnetron and amplifier-chain transmitters.
- Operational schedules and scan records that constrain uptime and beam dwell toward a fixed observer direction.
These records will show whether the gap is closer to tens of times present receiver sensitivity or many thousands of times. They are more valuable now than adding more systems to an uncertainty range whose dominant terms remain unresolved.
The next public result will then reconstruct a small source-backed section of the wartime network. It will connect radar specifications to antenna patterns, scan behavior, deployment evidence and operating schedules. The subset will be chosen because its documentation is strong enough to support an exact emission model, not because it can be presented as the whole global network.
That reconstruction will produce the first observer-conditioned record: what a receiver in one direction would encounter as the Earth rotated, beams scanned, units entered service and operating schedules changed. It is the necessary bridge between the receiver frontier reported here and the later historical question of whether the signal preserves a recognisable technological mobilisation.
Multinational expansion follows from evidence, not from a decorative list of countries. Each programme must be supported by original technical records, production evidence, deployment history and operating material where those records survive. Missing or destroyed archives will remain part of the result rather than being filled with undocumented assumptions.
Consequences
Earth’s wartime radar emissions did reach other star systems. For every documented historical system tested here, the required sensitivity at Proxima remains beyond the receiver references.
Supported processing reaches 0.10 light-years. Perfect but undocumented coherence reaches 2.42. A favorable 1% beam-dwell case at Proxima requires 30.7 times more receiver sensitivity or about 9,420 years of observation. The conservative requirement rises above 40,000 times.
The wavefront is large. The information frontier is not.
That is the first result of The War Reaching the Stars.