Removal or Suppression? A Field Model of Noship Stealth Tested Against 47 Known-Position Detection Trials, 14880–15190 AG

1. Introduction

Nearly every stealth system in the Guild's long-term review has eventually been beaten (Harrow, 15160 AG). Hull coatings, emission baffles and decoy screens reduce a vessel's signature, and a reduced signature is still a signature that better detectors eventually find. The noship appears to be an exception. A noship is a vessel enclosed by a no-field, an Ixian field device that hides whatever lies inside it from prescient search and, on the operational record, from ordinary instruments as well.

Our question is whether that apparent exception reflects a different engineering principle. We distinguish two. Under signature suppression, the vessel still interacts with its surroundings, only more weakly, and detectability should rise as instruments improve. Under interaction removal, the no-field eliminates the interaction a detector would register, and detectability should stay flat. The practical test is whether noship stealth has eroded across successive generations of detection technology.

This paper is written from a vantage of about 15200 AG, in the Scattering era after the Famine Times, when noships in their developed form were in use. Dates for the late operational record carry an uncertainty of several decades and are given as approximate.

A second question follows from the first. The same device that defeats instruments also defeats prescience. If it does so by removing a field interaction, prescience may itself depend on some such interaction with what it perceives. We treat that corollary as our own hypothesis and test it only as far as the trial record allows.

2. System Description

The lineage begins with the no-chamber, a shielded room built on Ix during the reign of Leto II, inside which a person could live unseen by the God Emperor's prescience. By the late period the principle had been scaled from rooms to hulls. Our account rests on the Ixian design series for no-field generators (Ixian Consortium, 14870–15150 AG), which records field geometry and power budgets without disclosing generator internals.

Operationally, noships are built on Ix and used by the Bene Gesserit, by the Tleilaxu and by returnees from the Scattering, chiefly to move people and cargo beyond the reach of prescient search (Bene Gesserit Chapter House, 15195 AG). That Ix could develop and sell such a device at all depends on the tolerance its field technology has long enjoyed under the Butlerian prohibition.

The device's effect on prescience is well attested in the Sisterhood's records; its effect on instruments rests on the trial registers. No released source establishes how the no-field works. Marn (15171 AG) proposed that it is a derivative of the Holtzman field family, the same physics that forms the fold in heighliner transit and whose safe use already ties Navigator prescience to the behaviour of a forming field. We adopt that derivation as our own working hypothesis; nothing in the analysis below depends on the exact generator design.

3. Analysis / Model

We express both principles as predictions of the per-attempt detection probability p in each of three detector generations: before 14950 AG, 14950–15080 AG and after 15080 AG. Oskarre (15142 AG) compiles a register of major advances in long-range detection and counts twelve over the span, three, four and five in the successive periods. We place each period at the cumulative number of advances at its midpoint, about 1.5, 5 and 9.5.

Model A, signature suppression, gives a baseline detection probability p₀ that grows exponentially with detector advances, p = p₀e^kc, where c counts advances. Neither parameter can be fitted from a record with no detections, because any trajectory with a small enough baseline reproduces zero successes. We therefore fix both from the erosion review of signature-suppression systems in Guild service (Harrow, 15160 AG), which counts advances on Oskarre's register and reports baselines near 2% per attempt and a doubling of detectability roughly every three advances (k ≈ 0.23 per advance). Under that calibration Model A predicts per-attempt detection of about 3%, 6% and 18% across the three periods. Because the advances concerned are improvements in instrument hardware, the model is a claim about instruments, and we score it on the instrumental trials alone.

Under Model B, interaction removal, detection probability does not change with detector advances. In its strong form the probability is zero. Its weak form allows a small constant rate from causes unrelated to signature, such as occasional leakage of crew emissions beyond the field boundary.

Each model is scored by the binomial likelihood of the observed instrumental outcomes in each period (4, 7 and 10 trials), and the two are compared by the likelihood ratio. The prescient trials enter separately, as an estimate of a constant rate. Scoring Model A on all 47 trials would require assuming that prescient detectability erodes on the instrumental schedule, so that pooled comparison is reported only as a sensitivity result. Because Model B's strong form sits at a boundary and the models are not nested, we report the ratio and twice its logarithm as a descriptive measure of support and attach no χ² significance level to it. We also report how the comparison changes as the assumed erosion rate varies.

4. Validation Against Field Data

Validation requires ground truth. An observer who never detects a noship cannot know that one was there, so routine transit logs are no evidence of stealth. We restricted the analysis to the Guild's detection-trial registers (Spacing Guild, Office of Operational Trials, 14880–15190 AG), in which Guild or Ixian crews placed a noship at a position recorded before the attempt and a separate team tried to find it. These registers hold 47 trials involving nineteen hulls. Twenty-one used the best instruments then available. The other twenty-six asked a Navigator to locate the noship with the limited prescience used for fold navigation, treating it as a hazard on a charted route. In five of these the Navigator was also told the sector in which the vessel lay.

Not one of the 47 trials produced a detection. Using the exact binomial method, zero successes in 21 instrumental trials bound the per-attempt detection rate at 16.1% (two-sided 95% upper limit), zero in 26 prescient trials bound it at 13.2%, and zero in all 47 bound it at 7.6%. The five cued Navigator attempts, taken alone, give a bound of 52.2%, which is too wide to show whether cueing helps.

With no events in any period, a test for trend has nothing to examine. The per-period bounds carry the evidence instead (Table 1). They overlap throughout, and for both the instrumental trials and the full set the tightest applies to the most recent period, when detectors were best.

Under the calibrated Model A, the probability that all 21 instrumental trials return no detection is about 7.8%. The strong form of Model B assigns this outcome a probability of one, so the record is about 13 times more probable under removal than under calibrated erosion (2 ln LR ≈ 5.1), which is moderate support. The weak form of Model B remains compatible with the record: a constant rate of 2% would leave the 21 instrumental trials clean with probability of about 0.65, and all 47 trials with probability of about 0.39.

Pooling changes the magnitude considerably. If prescient detectability is assumed to erode with each instrumental advance, the probability of a clean record across all 47 trials under calibrated Model A falls to about 0.35%, and the ratio rises to about 290 (2 ln LR ≈ 11.3). We know of no reason why Navigator prescience should improve with detector hardware, so this larger figure is conditional on an untested assumption.

Sensitivity to the erosion rate sets the limits of the result (Table 2). With detectability doubling every two advances, the instrumental record favours removal by a factor of about 6,000. With doubling every five advances, or with a baseline of 1%, the ratio falls to about 3, and against a trajectory as shallow as k = 0.015 per advance it is about 1.6, which barely discriminates between the models.

Across periods, the prescient trials were also clean (none of 5, 9 and 12 attempts), bounding a constant prescient detection rate at 13.2%. Harrow (15179 AG) found that Navigators routinely register ordinary hulls along charted routes as hazards, so the absence of any such response to noships is informative. The outcome fits the removal hypothesis if both detection routes depend on a shared interaction; read through the functional model of prescience as distributed computation over sensed inputs, it would place the no-field on the input side. The trials do not establish that placement, since a constant prescient rate of 2% would still leave all 26 attempts clean with probability of about 0.59.

5. Failure Modes

Several conditions could produce the same record without the mechanism we propose. The first concerns prescience: the descendants of Siona Atreides, bred during Leto II's reign, are invisible to prescient search by inheritance, with no device involved (Tavarre, 15183 AG). Earlier, in the reign of Paul Muad'Dib, a Guild Navigator's presence was enough to cloud another prescient's sight. Neither involves a no-field. Prescient blindness therefore has more than one cause, and the prescient trials alone cannot show that the no-field acts through a field interaction.

A second concern is the record itself. The trial registers are restricted, and those released to the Navigation Academy may omit trials the Guild preferred to keep secret. A detection might have been logged as an equipment fault, and Ixian crews designed some trials whose outcome concerned them. Both would bias the record toward non-detection, and none can be excluded from the released excerpt.

Third, the trials tested direct detection of the vessel. An absence can itself be conspicuous: a region where instruments return nothing, or where prescient vision shows a gap, may betray a noship to an analyst who searches for voids. No trial tested such negative-space inference, and a perfect removal mechanism would offer no protection against it.

Finally, the result depends on the erosion assumption. If signature-suppression stealth erodes far more slowly than the Guild's review indicates, twenty-one clean instrumental trials cannot tell the two principles apart.

6. Conclusion

Across roughly three centuries and three generations of detectors, 47 known-position trials produced no detection of a noship by instrument or by Navigator prescience. The record bounds per-attempt detectability below about 7.6% and shows no sign of the erosion that has overtaken other stealth systems. On the instrumental trials it favours interaction removal over calibrated signature suppression by a factor of about 13, more if erosion is fast and hardly at all if it is slow, while leaving open a small constant rate of detection. The larger pooled margin rests on an untested assumption.

A suppressed signature is a debt that improved detectors eventually collect; a removed interaction leaves no such debt, although it remains exposed to inference from the void it creates. We recommend that future trials test negative-space search directly, record cued and uncued prescient attempts separately, and use subjects of the Siona line as concealed targets, a device-free control for concealment from prescience.