Lasgun-Shield Contact and Catastrophic Feedback: A Resonant-Coupling Model of Yield Variability Tested Against 38 Archival Incidents, 10080–10230 AG

1. Introduction

Few facts of Imperial field engineering are as settled as the consequence of firing a lasgun into an active shield: a release of nuclear magnitude that destroys the weapon, the generator and whatever stands near either. Because the lasgun cannot safely be turned on a shielded opponent, combat between shielded forces reverts to the blade. Deliberate use against people is widely held to violate the spirit of the Great Convention's ban on atomics, and the Landsraad tribunals whose proceedings we examine (series GC-7) treated it as within the Convention's reach. That classification is tribunal practice, not the Convention's wording.

Much less settled is the size of the release. Inquiry files record yields differing by more than a factor of twenty for apparently similar equipment, a spread commanders and armourers treat as intrinsic to the phenomenon. We ask instead whether the yield could depend sensitively on a deterministic variable that no one can measure when contact occurs.

We develop a resonant-coupling model in which the phase between a modulation of the lasgun beam and the shield's own field oscillation governs the fraction of field energy released. The model, its parameters and its frequencies are our reconstruction, offered as a hypothesis and not as established Holtzman theory. We write from an Imperial vantage of about 10240 AG, some decades after Paul's Jihad, and give all dates in the Guild calendar.

2. System Description

The Holtzman shield resists incursion in proportion to the rate at which an object attempts to cross it. A companion analysis described this rate-triggered boundary, through which slow incursions pass while fast ones are stopped. We adopt that description and add one property it did not require: a characteristic field oscillation, reconstructed from generator service spectra.

Treated as a damped oscillator, the field has a natural frequency f₀ averaging 8.2 GHz and a quality factor Q averaging 12 across the spectra examined. Here and below, ± values denote unit-to-unit standard deviations: 0.6 GHz for f₀ and 2 for Q. The resonant bandwidth B = f₀/Q is therefore about 683 MHz, and propagating the relative spreads of 7.3% and 16.7% gives about ±125 MHz. Values vary with manufacture and age.

The lasgun is a continuous-wave laser projector. We do not propose that the shield responds to its optical carrier. Ixian emitter specifications instead record a beating between longitudinal cavity modes that imposes an amplitude modulation at 8.3 ± 0.4 GHz across emitter series, and it is this envelope, close to the field's natural frequency, that we hypothesise couples to the shield. At about 2 kW, the beam acts only as a trigger; the energy released belongs to the field.

Both devices are largely Ixian manufactures. The inspection regime that sustains Ix's tolerated position certifies control architecture and leaves field-interaction behaviour outside its scope, one reason the resonant properties of paired equipment have never been specified jointly.

3. Analysis / Model

We write the yield Y of a contact event as the product of a scale term and three dimensionless factors: Y = Y₀ · r · c(φ) · D(Δf) · h(τ). Here Y₀ is the yield at full coupling, zero detuning and saturated duration for a reference generator, and r is the involved generator's rating relative to it. We do not model the microphysics of the release; Y₀ is an empirical scale in terajoules (TJ).

Phase enters through a coupling function, c(φ) = (1 + 3 sin²φ)/4, where φ is the offset between beam modulation and field oscillation when the beam first engages the field. The function runs from 0.25 at φ = 0° to 1 at φ = 90°, a ratio of exactly 1:4. Since φ and 180° − φ give identical coupling, only the range 0–90° is identifiable. Under uniform φ, c has mean 0.625 and standard deviation 0.265, a coefficient of variation of about 0.42 and a log-scale standard deviation of about 0.48.

Detuning enters as D(Δf) = 1/(1 + β|Δf|/B), with Δf the frequency difference. We estimated β at 0.10 (95% interval 0 to 0.31), so the data cannot distinguish a weak frequency dependence from none. We retain the term because a resonance model requires it; a small β would mean that resonance governs how the release is triggered while field collapse, once begun, proceeds on its own terms. Duration enters as a saturating term, h(τ) = 1 − exp(−τ/T), because a beam that remains engaged draws out more of the field before the generator fails.

A worked example fixes the magnitudes. For a reference generator with Δf = 0.1 GHz and a 60 ms contact, D = 0.986 and h = 0.865 at the fitted T of 30 ms. With Y₀ = 150 TJ, the predicted yield is about 128 TJ when φ = 90° and about 32 TJ when φ = 0°. Under the model, identical equipment and contact time can thus differ fourfold in yield through phase alone.

4. Validation Against Field Data

The incident record comes from inquiry files in the Salusa Secundus Military Archives and Landsraad Public Archives records of Great Convention proceedings, covering 10080–10230 AG. It comprises 38 accidental or illicit contact events: stray fire in mixed engagements, emitter discharges in armouries and transport holds, and a few deliberate uses brought before Landsraad tribunals. Several events come from the Arrakis conflicts of 10191 AG and the early Jihad. Because shielded commanders avoid the lasgun and deliberate use invites sanction, the corpus is small by construction.

Yield estimates survive for 26 of the 38 events (68%), reconstructed by inquiry assessors from damage radius and structural loss. For 19 of these 26, the generator's service spectrum and a contact duration are also recoverable, the latter from witness timing or instrumented records. We calibrated Y₀, β and T on the 7 cases with instrumented records, obtaining Y₀ = 150 TJ (95% CI 95–235), and held out the remaining 12 for validation. Nine events carry two independent assessor reconstructions; their disagreement implies an archival error of about 0.35 on the log scale (95% CI 0.24–0.64, 9 df).

For each documented event, Δf is the difference between the generator's measured natural frequency and the specified modulation of the emitter series named in the inquiry file; individual emitters were never measured, so within-series tolerance adds unquantified error. Across the 19 generators, Δf averaged 0.1 GHz (SD 0.45; range −0.8 to +1.1 GHz). That spread is narrower than the 0.72 GHz expected under independence because paired equipment usually came from the same Ixian production period, and the two frequencies correlate at about r = 0.65 in these events. Sixteen of the 19 (84%) lay within one bandwidth (|Δf| < B), close to the 86% a normal distribution with that mean and spread gives. The spectra place most pairings in the regime the model assumes; they do not test resonance itself.

Across all 26 yield estimates the range was about 9 to 205 TJ, with a median of 52 TJ and a log-scale standard deviation of 0.80. A Monte Carlo simulation drawing duration, rating and detuning from the 19 documented cases, with archival error added, gave 0.78 with uniform phase and 0.61 with phase fixed. With only 26 cases, a Kolmogorov–Smirnov comparison separated neither simulation from the observed distribution.

A sharper test uses the 12 held-out cases. After removing the modelled contribution of rating, detuning and duration, the residual log-yield standard deviation was 0.58 (95% CI 0.41–0.98, 11 df). Uncertainty in Y₀ shifts every held-out prediction together and is absorbed by the residual mean, but uncertainty in β and T varies between cases and adds about 0.15 on the log scale, so the no-phase expectation is about 0.38. Since the archival component is itself estimated from nine pairs, we compared the two estimated variances: F(11, 9) = 2.32, p = .11; against archival error alone, F(11, 9) = 2.75, p = .07. Adding uniform phase predicts about 0.61, inside the interval, and 11 of 12 held-out yields fell within the 90% predictive interval (10.8 expected). The excess variance is suggestive at most, and any unmeasured factor with a log-scale spread near 0.45, such as rating error, beam geometry or emitter tolerance, would produce the same pattern.

Duration behaved as the saturating term predicts. Contacts shorter than 30 ms (n = 8) yielded less than contacts of 30 ms or more (n = 11), with a geometric-mean ratio of 2.2 (95% CI 1.2–4.0; Welch t(15.2) = 2.83, p = .013). Durations in these 19 cases ran from 8 to 140 ms; none of the longer contacts reported in witness accounts could be timed reliably.

Inverting the model to estimate φ for each held-out case, folding each estimate onto 0–90°, gave implied coupling values outside the admissible range of 0.25 to 1 for five of the 12, as expected when archival error rivals the phase effect. We excluded these instead of clipping them, which would pile probability at 0° and 90°. The seven remaining angles gave no evidence against uniform phase (Kolmogorov–Smirnov D = 0.21, p = .86), a result of little weight.

Yields were reconstructed from damage long after the event, and the largest events may be under-recorded because they destroyed the witnesses and instruments. Events with surviving spectra may over-represent garrisons with complete service records. Because the corpus consists of inquiry files opened after destructive events, a contained discharge would be unlikely to enter it at all. The validation set is small, and the frequencies and coupling function are our reconstruction; another functional form could fit these data equally well.

5. Failure Modes

Could a shield be detuned so that lasgun contact became survivable? Our model gives no support to that hope. At 1.2 GHz of detuning, D falls to about 0.85, a reduction of roughly 15%, and even at the upper limit of β the reduction is about 35%; both figures extrapolate past the largest sampled offset of 1.1 GHz. A release of several tens of terajoules, cut by a third, is still nuclear in scale. That all 38 recorded events destroyed both devices agrees with this but adds little independent weight, since the corpus admits only destructive events; the absence of a bounded regime rests on the model and on the canonical account of the interaction.

Phase offers no better control. It is set by the instant the beam engages a field oscillating near 8 GHz, and measuring it would require synchronised sub-nanosecond instrumentation on both devices during a contact that is almost always unintended. The minimum coupling of 0.25 still leaves a quarter of the full-coupling yield. Duration, the one variable a crew might influence, saturates within tens of milliseconds, faster than any operator can release a trigger.

6. Conclusion

A model in which the lasgun triggers the release of Holtzman-field energy and phase governs its size is compatible with the spread of yields in the archival record. The held-out residuals exceed what archival and calibration error predict, but not significantly, and the detuning coefficient cannot be distinguished from zero, so phase remains one candidate among several unmeasured factors. If the hypothesis holds, the effect's reputed unpredictability would be the signature of a deterministic variable inaccessible at contact, a structure also proposed in foldspace engineering, where some field parameters take definite values only once the field forms. The model offers no detuning, pairing or procedural limit that bounds the release, which leaves the sanction Landsraad tribunals have attached to deliberate use as the only effective control.