The Famine Times as a Food-System Collapse: Inter-Planetary Grain and Protein Movements, Planet-Level Trajectories and the Cascade from Sareer to the Scattering, 13725–14300 AG

1. Introduction

In the two hundred years after Sareer the Imperium's inter-planetary grain movement fell by about 91% from its late-Tyrant baseline, and 71 of the 416 worlds for which food and population records survive lost more than half their inhabitants. The Famine Times have been treated as a political period and as a demographic one; this paper treats them as a food system in failure, with grain, protein and population as the quantities, and asks why some worlds crossed the mortality threshold and others did not.

We write c. 15230 AG, two generations into the Heretics era. The reconstruction is possible now because the Landsraad Public Archives released the post-collapse survey of planet-level food records in 15190 AG, three centuries after the compilation began, and because the CHOAM Directorate opened its partial delivery series for the 13700–14300 AG window in 15205 AG. The two series together cover 416 of the ~2,100 worlds in the pre-collapse Imperium register. The remainder either have no legible records for the window or lie in regions that fell out of the known universe during the Scattering.

The paper's contribution is a planet-level reconstruction and a trajectory-based classification, not a political history. The Tyrant's monopoly, the Fish Speaker dissolution and the Honored Matres' return are the surround and are not reargued; interested readers may consult Orrande et al. (13770 AG) and the companion paper on the Scattering's genetic consequences (Terrault-Voss et al., 15230 AG). What we do argue is that the food-system failure can be read off three structural variables set before Sareer, and that this reading accommodates the record better than any single-driver account.

2. Sources and Method

Three record series supply the reconstruction. The Landsraad Public Archives post-collapse food-balance survey (compiled 14780–15180 AG from locally retained records) carries for each of 416 worlds a yearly or decadal entry on principal staple output, livestock headcount, net imports, population estimate and famine-year flags. The CHOAM Directorate inter-planetary delivery series (13620–14300 AG) carries route-level grain and protein movements, in principle; its coverage falls from 94% of attested routes before 13725 AG to 31% by 13900 AG and recovers to 58% by 14300 AG. The Spacing Guild Operational Archives (restricted) hold noship departure manifests for the Scattering outflow, 1,350 entries covering 13725–14300 AG, of which 1,083 record a staff count and 212 record a crop or livestock payload.

The three series overlap in 227 of the 416 worlds. Where they do, the Landsraad survey's net-import figure agrees with the CHOAM delivery total to 7.4% at the pre-collapse decade and to 14.1% at the trough decade, the discrepancy rising with the share of unaccounted routes. We take the Landsraad figure as the planet-level entry and use CHOAM to apportion to routes. For 189 worlds without CHOAM coverage in the trough, net imports are the Landsraad figure alone.

The analysis is in three stages. First, we fit a food-balance accounting at each world for each decade: calories produced plus calories imported minus calories exported minus calories carried over equals calories consumed; and the same for protein. Second, we trace trajectories across 13725–14300 AG and classify each world by population loss within 200 years of Sareer: catastrophic (>50%), severe (20–50%), milder (<20%), or recovered (output restored by 14300 AG). Third, we fit a planet-level failure model to the catastrophic–severe split, with as inputs the pre-collapse staple structure (number of principal grains and share of each), the carryover-stock-to-consumption ratio for the two years before 13725 AG, and the counterparty concentration index of pre-collapse imports (share of calories routed through the top two trade counterparties). The failure model is a log-odds fit on the planet level, with 95% intervals from resampling worlds with replacement over 2,000 draws. Agreement between the two coders who extracted the Landsraad figures was summarised by a chance-corrected coefficient for ordinal readings.

We do not use the Brian-Herbert-tier material (cymeks, Omnius, Serena Butler): none of it bears on the food system between 13725 and 14300 AG.

3. The Grain Movement Record

Across the 416 worlds, inter-planetary grain movements averaged 2.44 GT per year in the century before Sareer (13625–13725 AG). The series falls almost immediately on Sareer, by about 18% within the first decade, as the Fish Speaker household and Guild transport commitments to Arrakis unwound; by 13800 AG the figure is 0.98 GT per year. It continues to fall across the next century to a trough of 0.21 GT per year across 13900–13920 AG, and recovers to 0.68 GT per year by 14300 AG. The slope is steepest between 13800 AG and 13870 AG.

Movement did not fall uniformly. Routes routed through the three principal pre-collapse hubs (Giedi Prime, Caladan and Richese) retained 61%, 58% and 39% respectively of their pre-collapse volume at the trough. Routes between worlds that had been second-tier under the Tyrant's monopoly fell more sharply: of 282 such routes carrying grain, 184 recorded no deliveries at all for at least one of the 13850–13920 AG decades. The surviving route pattern at the trough is a sparse network around four surviving hubs. Of the hubs, only Caladan maintained a comprehensively open outbound route list through the trough.

Protein movements, where separately recorded (218 of 416 worlds), fell faster and recovered later. Protein-GT-equivalents at pre-collapse were 0.72 GT per year across the 218; by 13920 AG the figure is 0.03 GT, a 96% fall. Livestock-dependent destination worlds (62 of 218) are overrepresented in the catastrophic class.

4. Planet-Level Trajectories and the Threshold Structure

By 13925 AG, 71 of the 416 worlds had lost more than half their inhabitants, 148 had lost between a fifth and a half, and 197 had lost less than a fifth. By 14300 AG, nine worlds had recovered pre-collapse grain output. The catastrophic worlds share three structural features before Sareer: a principal-staple share greater than 70% of calories; a carryover-stock ratio under 0.4 of annual consumption averaged across the two preceding harvests; and a counterparty-concentration index above 0.60.

The planet-level failure fit confirms these three. Carrying all three features raised the log odds of catastrophic loss by an estimated 4.1 (95% interval 3.2 to 5.0) over the baseline rate, and carrying none of them held it to 0.19 of the baseline (0.11 to 0.33). Only four of the catastrophic 71 carried none of the three features, and three of the recovered nine carried any one of them. Melange supply loss, measured as the drop in planet-level melange ration delivery across the window, adds information at the margin (log odds +0.42, 0.11 to 0.78) but does not substitute for the structural three.

Caladan is the clearest counter-example: a diverse marine protein base, an on-planet carryover ratio of 1.4 at pre-collapse (unusual for the Imperium), and no counterparty exposure above 0.25 to any single partner, including Arrakis after Paul. The planet records a 7.3% population drop across 13725–13925 AG, the mildest among surveyed Great-House seat worlds. Giedi Prime, by contrast, carried heavy industrial labour populations with on-world staple output under 30% of their consumption; it lost 54% of its population across the same window despite retaining its hub status.

5. The Scattering Outflow and the Condition for Recovery

The Guild manifest total of 1,350 attested noship departures across 13725–14300 AG understates the Scattering outflow: 227 of the 416 worlds record outbound movements with no CHOAM-registered destination inside the pre-collapse register, representing an unaccounted population movement of at least 11 to 14 trillion persons by Landsraad estimate. We treat this as the pre-collapse low bound and do not improve it.

What matters for recovery is less the magnitude of outflow than its selectivity. The nine recovered worlds lost between 3.1% and 11.2% of their pre-collapse populations and recorded outbound movements concentrated in agricultural and technical trades, suggesting a return re-import of practice rather than headcount. The 71 catastrophic worlds record outflows that were not selective: proportional losses by trade broadly match their populations, and no clear return-flow is visible before 14300 AG.

Recovery as recorded through 14300 AG proceeded by substitution within surviving hub networks, not by re-establishment of the pre-collapse route list. By 14300 AG, 71% of inter-planetary grain movements followed routes that had not been in the top 500 by volume before Sareer. The Imperium's food system was not reconstructed; a smaller, more locally resilient system replaced it.

6. Discussion

The food system that failed between 13725 and 14300 AG had three structural weaknesses before Sareer: concentration of calories in a single staple on most worlds, chronically thin carryover stocks under the Tyrant's monopoly, and dependence on a short list of counterparty worlds for the import share that remained. These three held across the Imperium's political variation and set which worlds crossed the mortality threshold once the monopoly dissolved. The melange withdrawal that is often named as the trigger was material only where the three structural features had already left a world one bad harvest from catastrophe.

The food-system reading sits comfortably with the political record. The Tyrant's monopoly suppressed the development of planet-level reserves because reserves were an inefficiency the monopoly did not price. His death removed both the suppression and the apparatus that had made inter-planetary trade safe; neither the Fish Speakers in dissolution nor the surviving Great Houses had the carrying capacity to replace it. What the record shows is not an orderly return to pre-Tyrant conditions but a cascading failure in which planet-level starvation set the pace of institutional dissolution rather than following it.

For the Scattering, the implication is more specific than the standard account. Outflow was heaviest from catastrophic worlds, and selective outflow from mild and severe worlds carried practice as well as persons outward. The scattered populations whose later return the Heretics-era record registers do not come from the catastrophic class alone; they come from the severe and mild classes, which retained enough social capacity to send organised fractions. The genetic reconstruction of the Scattering (Terrault-Voss et al., 15230 AG) is consistent with this reading, since the founder cohorts it identifies are not drawn disproportionately from the catastrophic planets.

7. Limitations of the Record

416 worlds are 20% of the pre-collapse register. The 1,684 worlds without legible records may differ systematically: Landsraad survey teams reached them later and, where they did, found no retained archives. The 416 are therefore better documented than the Imperium mean, and may be relatively less catastrophic than the whole. The 55% catastrophic threshold would be reached on the Landsraad's own working estimate that the proportion of catastrophic worlds among the unreached is at least as high as among the 416.

The CHOAM delivery series undercounts the trough. Route-level entries fall out of the record where recorder stations closed, and the share of attested routes falls from 94% to 31%. The planet-level food-balance accounting is therefore more trustworthy than the route-level accounting at the trough, and the hub-level pattern is the weakest of our three findings, in part for this reason.

The Guild manifest total excludes departures the Guild did not log; the Scattering outflow is a lower bound. The population estimates on which the loss classification rests come from Landsraad survey reconstructions of local records, and carry errors we do not quantify at the planet level. Carryover-stock ratios before 13725 AG are the weakest of the three structural variables (local accounting varied in whether a reserve was reported). The planet-level failure fit therefore identifies a pattern, not a mechanism; the mechanism is read off the pattern plus the known logistics of the pre-collapse system, and that step rests on inference. The identification of staple concentration and counterparty concentration as decisive is robust to every version of carryover ratio we tried, including the version that drops it as a variable altogether.