Mobility as Doctrine: Thumper-Summoned Worm Mobility and Operational Radius in Forty-One Fremen–Harkonnen Engagements, 10110–10190 AG

1. Introduction

Writing in 10248 AG, with sietch registers and Harkonnen records both open, we ask why forces equipped with ornithopters, crawlers and supply columns so often failed to find, fix or pursue Fremen raiders. The period is the Harkonnen tenure before 10191 AG; the covert Sardaukar deployment of that year falls outside it.

Several Fremen terms of art carry the argument. A sietch is a concealed cave community, governed by a naib, a leader who traditionally holds office until a challenger kills him. A thumper is a spring-driven stake planted in open sand whose rhythmic beat summons a sandworm. Maker hooks are the goads with which a rider climbs a worm, opens the leading edge of a ring segment and steers the animal.

Earlier analysis of the conflict, summarised most fully by Kesteven (10236 AG), treated Fremen effectiveness as the product of fieldcraft and of the Harkonnen failure to count the enemy. That failure was real: Harkonnen census returns counted a small fraction of the population the sietch registers now show. Yet that literature modelled both sides with the same depot-radius geometry, and never asked what happens when one side can summon its transport from the terrain itself.

We argue that worm mobility, with the Fremen water economy, inverted the usual relation between terrain hostility and reach. We test three predictions: that the Fremen radius materially exceeded the Harkonnen one; that engagements clustered beyond Harkonnen reach and inside worm-mobile country; and that exchange rates favoured the Fremen there.

2. Methods

Harkonnen operational radius was reconstructed from the garrison and supply returns of the fief (Harkonnen Garrison Command, 10110–10190 AG), following the supplied-column method of Oszmar (10229 AG). For each element we halved the usable endurance after a 20% reserve to obtain an out-and-back radius; under air cover the smallest radius binds. Table 1 sets out the inputs. The ornithopter is sensitive to thermal and particulate loading, which explains the conservative fuel reserves.

Fremen reach required a different model, because a rider summons a worm with a thumper instead of waiting to meet one. Our own estimate of mature-maker density in the southern erg, one animal per 2,000–4,000 km² (spacing roughly 45–65 km), accords with basin-scale work on territorial spacing, and implies that a thumper in open sand usually falls within some animal's hearing. We therefore expressed worm availability as the probability that a call is answered within time t, modelled as 1 − exp(−λt), where λ is a regional response rate estimated from 212 recorded thumper calls in the Sietch Tabr testimonies (Sietch Tabr Oral History Collection, 10196–10240 AG). A 50 km survey cell was classed as worm-mobile when the modelled chance of an answer within two hours was at least 0.5. Because λ was estimated for two regions only, the southern erg and the northern margin, this classification reproduces the regional boundary and is a mapping device, not an independent finding.

The Fremen radius came from 47 recorded journeys with known endpoints, each made as a chain of ridden legs with dismounts between them. Journey length averaged 580 km (range 420–740 km). The testimonies treat these journeys as the practical limit of sustained riding and rider endurance before a party had to rest at a sietch or cache, so we take half of one journey as a conservative out-and-back radius: about 290 km (range 210–370 km). Water does not bind this figure, chiefly because raiding parties drew on hidden caches along their routes. Stillsuit losses, small at rest, become significant under sustained heavy exertion, and a raid is such exertion; the caches, not the suit alone, removed the need for a resupply column.

Forty-one engagements of 10110–10190 AG with usable locations and force sizes for both sides were drawn from Harkonnen field returns and checked against Fremen testimony and the sietch register (Fremen Planetological Survey, 10205–10240 AG). Each was coded by distance to the nearest major installation other than the target itself. Only major installations based ornithopter wings and relief columns; satellite outposts held small garrisons and no aircraft. The 128 km radius served as a cut-off fixed before coding, and each engagement was also coded by worm-mobile status. Location was tested with Fisher's exact test, since one expected cell count fell below five. Deaths on each side were modelled as counts with negative binomial regression, with the log of that side's committed force as an offset, so that the incidence-rate ratio (IRR) compares death rates per combatant. To test sensitivity, we refitted each model 1,000 times with every death count and every committed-force figure multiplied by independent factors drawn uniformly from 0.5 to 1.5, and report the 5th–95th percentile range of the IRR.

Installation coverage was estimated for the study theatre, an 800 × 600 km block running south from the Shield Wall margin into the open erg, by a Monte Carlo test on 60,000 random points. Relief coverage counted a point as covered if it lay within 128 km of a major installation, the same definition used for engagements; nominal coverage counted any installation, satellites included.

3. Results

On these inputs the Harkonnen combined-arms radius was about 128 km and the Fremen radius about 290 km, a ratio of roughly 2.3 (range 1.6–2.9 across the journey range). In the southern erg (148 calls) the modelled median wait was about 35 minutes, and about 91% of calls were answered within two hours. On the northern margin (64 calls) the median wait was near five hours and only about 24% were answered within two hours. The worm-mobile cells therefore coincide with the southern erg, some 55% of the theatre.

Engagement location followed the prediction (Table 2). Of 41 engagements, 29 (71%) lay more than 128 km from any major installation, and 26 of those 29 (90%) fell in worm-mobile cells. Among the 12 nearer engagements, only 3 (25%) did so. The association was strong (Fisher's exact test, p < 0.001; conditional odds ratio 23, 95% CI 3.5–220); the wide interval reflects the small cells.

In far engagements Fremen parties lost about 6% of those committed (roughly 250 of 3,900) against about 15% near installations (roughly 270 of 1,750). The negative binomial IRR for Fremen deaths in far engagements was 0.41 (95% CI 0.27–0.64, p < 0.001), and the perturbation range was 0.33–0.52. Harkonnen losses moved the other way: about 25% of committed troops died in far engagements (roughly 1,300 of 5,200) against about 7% near installations (roughly 160 of 2,200), an IRR of 3.4 (95% CI 2.3–5.3, p < 0.001; perturbation range 2.7–4.4).

Three engagements illustrate this. A raid the Harkonnen returns call the Red Sink action struck a harvesting operation 240 km from the nearest major installation, in worm-mobile country; about 180 Fremen lost 8 dead, the 120-strong detachment lost 47, and relief took seven days to assemble. A second raid concentrated about 220 Fremen against a satellite spice-storage outpost about 270 km from the nearest major installation, close to the mean Fremen radius, by staging through an intermediate sietch. Twelve Fremen died; of a 140-strong garrison, 89 were killed and 34 captured. The third case is the reverse. A siege of a garrison post 95 km from a major installation, outside worm-mobile country, lasted three weeks and cost about 220 Fremen 67 dead (about 30%), while the 300 defenders lost 21 and held.

Installation geometry changed over the fief. Between 10110–10140 and 10160–10190 AG the network grew from 25 installations (5 major, 20 satellite) to 43 (8 major, 35 satellite). The early network lay within about 200 km of the Shield Wall margin; later majors stayed within about 230 km, while new satellites were pushed up to about 400 km south. Nominal coverage within 128 km rose from about 45% to about 80% of the theatre, and the mean distance from a random point to the nearest installation of any kind fell from about 180 km to about 75 km. Relief coverage moved far less, from about 35% to about 45%, and the mean distance to the nearest major installation fell only from about 235 km to about 175 km. Inside the worm-mobile block, nominal coverage rose from about 4% to about 60%, but relief coverage only from nil to about 7%. Personnel grew only about 12%, so the average garrison per installation fell by about 35%. Engagements rose from 11 in the early period to 22 in the late one (the other 8 fall between), and 15 of the late engagements struck satellite outposts added after 10140 AG. Every sampled point, by contrast, lay within 290 km of one of the theatre's 18 registered sietches.

4. Discussion

Much of Fremen effectiveness under the fief came from a logistics asymmetry. Fremen parties could reach more than twice as far as Harkonnen columns under air cover, chose to fight where the difference was largest, and there lost fewer of their own and killed more of the enemy per combatant.

Worms alone do not explain the inversion. The sietch caches removed the water line that tethers any conventional force, the stillsuit cut routine loss, and sietch water discipline kept those reserves intact between raids. Worm-riding removed the fuel and vehicle line, leaving the raiding party no supply tail to cut.

Harkonnen forces could not simply follow. Rhythmic machinery draws worms, which is why harvesters depend on carryalls; a shield drives a worm into frenzy, so shielded infantry was a liability on open sand (Vaddeh, 10231 AG). Coriolis storms ground ornithopters, and the Guild's refusal of weather satellites, bought with Fremen spice, left the Harkonnen without orbital surveillance (Spacing Guild Arrakis Station, 10150–10191 AG). The same animal that opened the deep desert to the Fremen closed it to Harkonnen installations.

The late-fief dispersal was a specific doctrinal error. Adding satellite outposts brought most of the theatre within nominal reach, yet relief from the major installations still reached less than a tenth of the worm-mobile block. The new outposts sat in the band Fremen could reach at will, with thinner garrisons than before, and they became the most frequent targets. Denser spacing could not close the gap without major bases inside the erg, where no installation could be sustained. A doctrine fitted to Arrakis would have mapped worm territory as the enemy's road network and sized garrisons to the 290 km threat radius.

Fieldcraft still mattered, but inside a structural advantage the fief's planners never modelled.

5. Limitations

Concealment biases every source. Fremen hid worm-riding from outsiders, and Harkonnen field officers attributed Fremen reach to smugglers or unseen aircraft, so the Harkonnen series understates both Fremen numbers and reach. The oral testimonies were recorded after 10191 AG, when secrecy had lapsed, but decades after the events, and the thumper wait times rest on recollection. The response-rate model also assumes a constant rate within each of two regions, so it cannot resolve variation within the erg, and seasonal or storm-driven change remains untested.

Casualty and force figures are uncertain to perhaps ±50%. Harkonnen returns may inflate enemy dead, and testimony may omit costly defeats. The perturbation analysis shows that the direction and rough size of both IRRs survive errors of that order, but the point estimates should be read as approximate.

Small raids without Harkonnen witnesses are likely missing from the 41 engagements, which probably understates the far-range share. The findings describe one theatre in one period and should not be carried to the post-10191 AG campaigns without fresh analysis.