How Kanly Ends: Termination, Duration and Bystander Loss in 311 Declared Vendettas of the Landsraad Registry, 9719–10190 AG

1. Introduction

In 9719 AG the Landsraad High Council ordered that every declaration of kanly between Houses be copied to Kaitain, with the loss returns each party was required to file when the feud closed (Landsraad High Council, 9719 AG). The copies became series LR-K of the Imperial Archives. For most of the five centuries that followed, the series served heralds and Masters of Assassins checking whether a feud was still open, and nobody read it from end to end.

Kanly is the feud the Great Convention permits, a declared vendetta between Houses conducted under the strictest limits the Convention sets. Its violence belongs to the War of Assassins, the limited warfare whose rules prescribe the permitted weapons and whose stated purpose is to spare those outside the quarrel. Commentary on kanly has dwelt on its forms, the interval it imposes between affront and violence, the formal duel and the permitted instruments (Vesk, 10232 AG; Imhoff, 10236 AG; Tarvaine, 10204 AG). It has said little about what the feuds actually cost, or how they stopped.

This paper reads the registry for how declared kanly ended, how long it ran, and who outside the two Houses died while it ran.

We completed the reading in 10242 AG, using the copy registers for every declaration of 9719–10190 AG. We stop at 10190 AG because the Arrakis change of fief and the Jihad years that followed altered what a declaration meant, and because the registers for those years are not complete.

2. Sources and Method

Each feud in series LR-K has a declaration entry and a closing entry. The declaration entry names the declaring and the declared House, the herald, the date and the grievance in a fixed form. A closing entry, filed by each House, states how the feud ended and lists the House's losses under three heads, namely members of the House, its retainers and sworn agents, and persons belonging to neither (Landsraad Registry of Kanly and House Losses, 9719–10190 AG). The third head is our measure of bystander loss. We take it as the House reported it, and we return to that weakness in Section 7. The registers record 311 declarations in the window. By 10190 AG, 284 had closed. Of the remaining 27, 21 were still open and 6 (1.9% of declarations) had lost their closing entries to damage in the copy registers, close to the rate an earlier audit of the series found for its first three centuries (Oudevar, 10221 AG). The enmity of the Atreides and the Harkonnens is older than series LR-K and is not among the declarations counted. Analyses of ending, duration and loss use the 284 closed feuds.

The closing entries also record whether a Judge of the Change was seated at any point in the feud, since a Judge's commission is noted in the margin of the declaration it concerns. They record, too, any finding by a Judge or by the High Council that a party used a weapon or method outside the permitted list of the War of Assassins, which we call a finding of forbidden means. House rank, which would let us distinguish feuds between equals from feuds between unequal Houses, is entered for only 118 of the 284 pairs (41.5%). The first and second authors coded every closing entry independently and settled disagreements by joint reading of the original copy; disagreement arose on ending type in 14 entries, all of them lapses that one coder had read as settlements. Proportions are compared as risk ratios with 95% confidence intervals on the logarithmic scale and with exact tests for two-by-two tables. Durations are reported as medians and interquartile ranges, because a handful of very long feuds would distort any mean (Hollenby, 10230 AG).

3. How Kanly Closed

Settlement before a Judge of the Change ended 131 of the 284 closed feuds (46.1%). The Judge, appointed by the High Council and the Emperor, oversees a kanly negotiation and its formal battles; a settlement entered before a Judge carries a schedule of cessions and a date after which neither House may revive the grievance. Lapse or withdrawal ended 58 (20.4%), usually by a declaring House that let its declaration fall when the herald's renewal came due (Roquefeuil, 10228 AG). Loss of fief or line ended 41 (14.4%), among them House Garrandt, liquidated about 10179 AG after defeat in kanly with its creditors unpaid (Reyes-Okafor and Aldevash, 10241 AG). Rulings of the High Council ended 34 (12.0%), and formal combat between champions or principals ended 20 (7.0%).

The endings differ sharply in how long they took to arrive (Table 1). Formal combat was the quickest, with a median of 2.5 years from declaration, which suggests it served to close feuds that neither House wished to prolong. Feuds that ended in the loss of a House ran longest, with a median of 14 years. Shares in Table 1 are rounded and sum to 99.9.

4. Duration and the Long Tail

A closed feud ran a median of 5 years (interquartile range 3–11), and 212 of the 284 (74.6%) closed within a decade of declaration, though a small number ran very much longer. Nineteen feuds (6.7%) were still open fifty years after declaration, and the longest, ended by the extinction of the declared House, lasted 206 years. Nine of the nineteen ended in the loss of a fief or line and six by ruling of the High Council; only three reached settlement before a Judge. These long feuds were inherited. In each of the nineteen, the declaration outlived the head of House who made it, and in eleven the closing entry was filed by a grandchild or more remote heir of the declarant. The registry gives no sign that a long feud was conducted any less formally than a short one, since renewals were filed on time and heralds' notices kept their form, so formality, on this record, did not keep a feud short.

5. Loss Outside the Houses

Of the 284 closed feuds, 171 (60.2%) recorded no death outside the two Houses. The other 113 recorded 37,716 bystander deaths between them. The distribution is extremely uneven. The nineteen feuds that ran fifty years or more account for 35,110 of those deaths (93.1%), and every one of them recorded some. The 265 feuds that closed in under fifty years recorded 2,606 between them; among the 94 of those with any bystander death, the median was 15 and the largest 292.

Feuds in which a Judge of the Change was seated look different from those in which none was (Table 2). A Judge was seated in 162 of the 284 closed feuds (57.0%). Among them, 49 (30.2%) recorded any bystander death, against 64 of 122 (52.5%) with no Judge seated, a risk ratio of 0.58 (95% CI 0.43–0.77; exact test p < .001). Findings of forbidden means followed the same pattern, with 11 of 162 (6.8%) against 23 of 122 (18.9%), a risk ratio of 0.36 (95% CI 0.18–0.71; p = .003).

This comparison is crude by necessity. Settlement before a Judge requires a Judge, and the Judge-seated group therefore contains every feud that ended that way; a feud already moving toward settlement may also have been more likely to receive one. Rank, entered for fewer than half the pairs and perhaps not at random, cannot be held constant.

The association is not an artefact of the long tail, however. A Judge was seated in 7 of the 19 feuds of fifty years or more, all of which recorded bystander deaths. Among the 265 shorter feuds, 42 of 155 with a Judge (27.1%) and 52 of 110 without (47.3%) recorded any bystander death (risk ratio 0.57, 95% CI 0.41–0.79; p = .001).

6. Discussion

Declared kanly, as the registry records it, was a short and narrow institution. Most feuds ended within a few years, most by settlement or lapse, and most killed nobody outside the Houses that declared them. This agrees with the view that the Great Convention's forms, by fixing an interval between affront and violence and a list of permitted weapons, kept the feud within the Houses (Vesk, 10232 AG). The Convention's ban on atomics against persons set the outer ceiling on what a kanly could become (Vantanza et al., 10240 AG). The cost of kanly lay in its tail. Nineteen inherited feuds produced almost all of the recorded loss outside the Houses, and they did so without abandoning the forms. Any account of kanly that treats it as uniformly limited misdescribes the institution, whose limits held for most feuds and failed in a few that lasted for generations.

An association between a seated Judge and lower bystander loss is the finding most likely to matter, and the one this record supports least firmly. We attribute it to the Judge. A Judge oversaw the feud's formal battles and arbitrated between the Houses, and, on our reading of the closing entries, entered most findings of forbidden means; the association with bystander loss also holds once the long feuds are set aside. Findings of forbidden means are weaker evidence, since they are mostly entered by Judges and so observable chiefly where one sat. The plausible alternative is parity. A Judge may have been seated more readily in feuds between Houses of comparable standing, which would also have been less likely to spill outward, since neither party could hope to win by attrition. The Judge reading is the more likely of the two because a Judge's oversight bears directly on how a feud is fought, while parity would act only through the parties' calculations; but with rank entered for 118 pairs, the registry alone could not separate them. Doing so needs rank for the remaining pairs and the date on which each Judge was seated, both of which lie outside series LR-K.

7. Limitations of the Record

Bystander loss is self-reported by the Houses, each of which had reasons to understate deaths that a Judge or the High Council might treat as evidence of forbidden means. The undercount is unlikely to be uniform, because a House that lost its fief often filed no closing entry of its own, leaving only its opponent's return. The 21 open and 6 damaged entries are excluded, and feuds still open in 10190 AG are on average younger than the closed ones, so the duration figures describe feuds that closed, not all feuds declared. Kanly conducted without a registered declaration, which the forms forbid but cannot prevent, does not appear at all.

Series LR-K was kept so that heralds could answer one question: is this feud open? The present reading extends that herald's index into a record of how feuds closed and what they cost, and it inherits the index's design, in which the Judge appears only as a marginal note. The commission files of the Judges of the Change at Kaitain, which date each seating, and the Landsraad precedence rolls, which give rank for every House in every year, hold what the herald's index never carried. A reading that links them to series LR-K would extend the audit of the copy registers (Oudevar, 10221 AG) from the condition of the registers to the conduct of the feuds they record.