Sietch Water Reserves and the Fremen Water-Discipline Economy: A Compliance-Threshold Model of Twenty-Three Sietches, Validated Against Ledgers of 9880–10191 AG

1. Introduction

Arrakis has no rainfall in the period studied, and the sandtrout encapsulate free water at depth. The Fremen answer has two faces. One is hardware: windtraps, dew precipitators, sealed cisterns and the deathstill, which renders the water of the dead to the tribe. The other is custom: the ration, the water debt, the water rings that record water held in trust, and severe sanction for waste or theft.

Sealed caches are older than the greening. Sietch records later gathered into the Survey register show a standing tithe of collector yield to hidden stores from their earliest entries, generations before the Imperial Planetologist reached Arrakis. Our reading of the register is that Pardot Kynes did not create these hoards but gave them their present purpose, the greening of the planet, and that Liet-Kynes carried the plan on until his death in 10191 AG. Trajectory modelling in this journal places the plan's greatest fragility in its early windtrap build-out, when water is diverted from use into storage. Fremen society is therefore poor in the water it permits itself to use, whatever its total holdings.

Earlier Survey work modelled sietch reserves without separating the cache from the working store (Threll, 10224 AG), and comparative work on closed-resource systems has argued that compliance can substitute for storage (Straka, 10221 AG). Neither tested that substitution against sietch ledgers. A doctrinal study in these pages reads the naib's office as custody of the tribe's water, placing water discipline inside sietch constitutional order.

Writing from the Survey at Sietch Tabr in 10238 AG, we treat custom as the enforcement layer of the reserve system and locate the compliance level below which the median sietch's margin falls into the high-risk band.

2. System Description

The sietch holds water in two stores. The operating reserve comprises the cisterns that supply ration, cooking and ritual use. The cache, in sealed and often distant catchments, is closed to consumption by custom. The register records about 12% of collector yield passing to it as a tithe, and the recorded share does not change across the ledger period; Kynes redirected the tithe without, on the register's evidence, altering its size. Our model treats the cache as a sink that the operating reserve feeds but never draws on.

Inputs come almost entirely from windtraps and dew precipitators. Their yield varies with seasonal humidity and with Coriolis-storm damage, which can strip a collector field in one passage (Vashtel, 10212 AG; Qasmir, 10205 AG). Deathstill reclamation is hydraulically minor: at Tabr, with the mortality implied by Medical Corps returns, it yields a few hundred litres a year, well under 0.1% of inflow (Arrakis Medical Corps, 10195 AG).

The stillsuit does not feed the reserve; it returns the wearer's own water in a closed loop, and an engineering review in this journal finds that its losses stay near zero at rest and in moderate work, becoming significant only under sustained heavy exertion. We therefore model the stillsuit as reducing each person's net draw on the cistern, set at a ration of 2.0 L per person-day. For Tabr's census population of 1,120 this gives an annual requirement of about 818,000 L.

Discipline enters through two behaviours. A compliant household keeps to the ration and surrenders reclaimed water; a non-compliant one draws about 40% above ration and retains part of what it reclaims. Compliance, written c, is the share of households that behave in the first way.

3. Analysis and Model

The model steps monthly. Operating reserve rises with collector yield net of tithe and with deathstill returns, and falls with draw, seepage and evaporation. Population sizes come from Survey headcounts for the twenty-three sietches in the register (median 790, interquartile range 560–1,140).

Annual collector yield varies stochastically (standard deviation 9% of each sietch's mean yield), and major environmental shocks occur at random. The Tabr ledger records nine such shocks across 312 years, five humidity droughts and four storm-damage events, a Poisson rate of about one per 35 years. Three conflict disruptions in the same ledger are excluded from the shock distribution, since conflict acts on the system through enforcement and is the effect the validation tries to isolate. Each shock removes a deficit drawn from the ledger distribution (mean 17% of annual requirement, range 8–34%). A multi-year drought counts as one event, with its deficit spread evenly across its recorded duration (one to fourteen years among the five droughts). A moderate shock is defined throughout as a 15% deficit.

The operating margin is the reserve held at the end of the dry season, as a percentage of one year's requirement. Margins below 25% form the high-risk band, and a margin below 10% counts as collapse. Each sietch was run 10,000 times per compliance level over thirty years. Time to collapse was analysed with Kaplan–Meier estimates and a Weibull hazard model with a shared frailty term for sietch, with runs that did not collapse censored at thirty years. Simulation uncertainty intervals replace p-values, which simulated samples render meaningless.

4. Validation Against Field Data

Three parameters were calibrated on Tabr records alone: yield variability, and the shock rate and deficit distribution from the reserve ledger (Fremen Planetological Survey, 9880–10191 AG). Storage loss came from cistern surveys. Each sietch's mean yield came from its own collector-yield series, kept apart from its reserve ledger; compliance was fixed at 100%, never fitted. The Tabr fit is therefore in sample (r² 0.91, root-mean-square error 4.1 percentage points of margin). Two further registered sietches, S-7 (184 years) and S-12 (127 years), served as out-of-sample tests (Fremen Planetological Survey, 10008–10191 AG), with r² 0.89 and 0.87 and errors of 4.6 and 5.2 points. Run at 70% compliance with the same calibrated parameters, the model fitted Tabr worse (error 11.3 points), predicting shortfalls the ledger does not show.

Two disturbed periods give the sharpest test. During a humidity drought of roughly 10020–10033 AG the reserve fell to about 14–18% of annual requirement. During a period of disputed authority and weakened enforcement, roughly 10083–10100 AG, it fell to about 7–11%, briefly entering the collapse band before emergency transfers from neighbouring sietches arrived. Entries for several of those years are incomplete, so the minima are ranges.

We regressed the model's absolute error on indicators for the two periods, with Newey–West standard errors for autocorrelation. Drought years added 1.1 points of error (95% CI −0.6 to 2.8), indistinguishable from zero. Because this drought is among the five that informed the shock distribution, its absorption is partly built into the calibration and is the weaker of the two tests. Conflict years, whose shocks were kept out of the inputs, added 7.9 points (95% CI 5.2 to 10.6), with the model overpredicting the reserve. Refitted with compliance free, those years gave about 70–80%. The worst recorded crisis thus coincides with a failure of discipline.

Tabr's water judgments were delivered orally; Survey clerks transcribed 61 cases of water offence heard between 9920 and 10190 AG (Sietch Tabr Council of Elders, 9920–10190 AG). First offences (38 cases) were resolved by assessment of water debt in 34 (89.5%). Repeat offences (17 cases) brought suspension of ration rights in 13 (76.5%). Theft from the cache (6 cases) brought a capital sentence or expulsion to the open desert in 5 (83.3%). This graduated ladder is our reconstruction of one sietch's practice; the records show sanctions applied, not the compliance they produced.

5. Failure Modes

At full compliance the median margin was 31% (interquartile range 27–35%), and 21 of 23 sietches (91.3%) lay outside the high-risk band. The median margin crossed 25% at a compliance of 88% (95% uncertainty interval 82–93%). At 70% compliance the median margin was 18% (interquartile range 12–27%), and 15 of 23 sietches (65.2%) sat in the high-risk band. At 50% the median fell to 11% (interquartile range 7–16%), with 20 of 23 (87.0%) in the band.

Collapse followed the same gradient, though not at a constant rate. The thirty-year probability of collapse was 4% at full compliance, where the median time to collapse was not reached. At 70% compliance the median time to collapse was 14 years (interquartile range 9–22), and at 50% it was 5 years (3–8). These estimates pool heterogeneous sietches, and the robust few form a long upper tail that flattens the late pooled hazard. A single log-linear compliance term fitted poorly, so we estimated the Weibull model with a break at 70%. Between 100% and 70% compliance, each 10-point fall multiplied the hazard of collapse by about 3.3 (95% uncertainty interval 2.6–4.2); below 70% the multiplier was about 1.7 (1.3–2.3). At low compliance the margin already sits close to the collapse line, and ordinary yield variation drives most failures, so each further fall adds proportionally less hazard. Hazard thus rises fastest in the upper range, where Fremen practice appears normally to have sat. At 70% compliance a single moderate shock takes the median sietch from 18% to about 3%, below the collapse line within the same dry season.

The threshold responds most to collector yield: a 10% change in mean yield moves it by about 7 to 8 points either way. A 15% change in operating reserve capacity shifts it by about 3 points, and storage loss across its surveyed range by about one. Three counterfactuals isolate the cache. Suspending the tithe raises net yield to the operating reserve by about 13.6% and lowers the threshold to 78% (72–84%). A 50% larger operating reserve lowers it to 80% (74–86%). Opening the cache to consumption removes any threshold within the sweep: every sietch stays viable at 50% compliance.

Custom therefore guards two boundaries: frugal daily use keeps the operating reserve out of the high-risk band, and the ban on drawing from the cache, whose breach draws the gravest recorded sanction, keeps the greening on schedule.

6. Conclusion

Hardware alone cannot keep the median operating reserve out of the high-risk band while the tithe is paid and the cache is left closed. Compliance of about 88% is needed for that. Collapse risk is already 13–24% over thirty years at 85–90% compliance and reaches 78% at 70%. The close fit of the full-compliance model outside the conflict years implies that historical compliance at Tabr was usually high, and the transcribed judgments show the graduated sanction that sustained it. Water discipline is in this sense part of the infrastructure.

The model cannot say whether the coupling was designed or evolved. We favour co-evolution, in which the tithe and the severity of sanction ratcheted together, and offer it as a hypothesis. The ledgers cannot test it, because the recorded tithe share is constant from 9880 AG onward; any ratchet predates the written series. Older oral accounts of the first tithes and sanctions are the evidence that could separate a gradual tightening from a single founding act.

Three limits apply. The first is parameter quality: headcounts and ration returns are of uneven reliability. The second is coverage, since only three sietches have ledgers long enough for testing and others remain closed under sietch custom and Survey access rules. The third is the assumption that each sietch is closed; only one emergency transfer to Tabr in the conflict years has a recorded volume, about 4,000 L from S-12 (roughly 0.5% of annual requirement), so the network's share in Tabr's recovery cannot be estimated.