From Mourning to Questioning: How Citizens of Four Orbitals Have Kept the Memory of the Idiran War, from 349 Announced Observances, 1376–1949 CE

1. Introduction

The Idiran War ran from 1327 to 1375 CE and remains the largest war the Culture has fought. Nobody on any Orbital is obliged to remember it. The Culture keeps no calendar of commemoration that anyone could enforce, no office whose holder tends one and no law that could require attendance at anything. An occasion for remembering the war exists because some citizens choose to hold it and others choose to come, and it lasts only while it goes on attracting invitations and company.

Historians have written at length about the war. Histories of it composed on Orbitals have been listed by one of us (Asphodene, 1598 CE), and the grief of the first gatherings was described for a single Orbital over its first decades (Marrowell, 1634 CE). Remembrance as a practice has had no study across centuries or across Orbitals. General accounts of how the Culture keeps shared occasions stress invitation and reputation over obligation (Dunthrey, 1725 CE), and an earlier survey observed that citizens, who live long, gather readily on round anniversaries of events within their own lifetimes (Tressaly, 1647 CE). Neither asks whether a war that has passed out of the lifetime of nearly every citizen keeps the same hold.

We write in 1950 CE. Our sources close at the end of 1949 CE, and nothing here describes remembrance after that date or the future of the practice. We record how the war was remembered, and make no claim about the war beyond its dates.

Three questions guide the paper. How often have citizens held announced observances of the war, and in what forms? Who has taken part, and on what terms of invitation? And what have the gatherings said they were for?

2. Sources and Method

The Hub of an Orbital keeps a register of gatherings that organisers choose to announce, as a convenience to residents who wish to find an occasion. The Hub lists an event on request. It does not organise events, and by custom it does not read the minds of residents, so a listing holds only what its organiser chose to say. We use the registers of four Orbitals, each continuous from 1376 to 1949 CE, which is 574 years and 2,296 Orbital-years in all. The Chiark register is held in the Chiark Orbital Hub Records (Chiark Orbital Hub Records, 1376–1949 CE). The Hubs of the other three Orbitals, which we do not name at their request, deposited copies there by invitation (Hubs of Orbitals B, C and D, 1376–1949 CE). The third author is the custodian of the Chiark series; the practice of its custody has been described elsewhere (Ghastrey, 1884 CE).

From the four registers we drew every listing that names the Idiran War. There were 404. Twenty-nine were re-announcements of an occasion already listed, and 26 named the war only in passing, as one item in a gathering held for another purpose. Both were removed, leaving 349 observances: 102 on Chiark, 98 on the second Orbital, 91 on the third and 58 on the fourth. No listing carries a count of those lost in the war, and we record none. We also hold no service records of any individual and study no one's part in the war.

Two of us, Asphodene and Farrasthe, coded each observance independently from the full text of its announcement. The form of the gathering was placed in one of five classes: silent assembly, in which a silence of set length is the centre of the occasion; reading aloud; shared meal; music or performance; and walk. Where an announcement names several, we coded the first it names. The stated theme took one of four values: loss; the decision to fight; inquiry into the Idirans; and peace and the Culture's way, meaning the question of what the war showed about how the Culture should live. Coders agreed on form in 320 of 349 observances (91.7%, Cohen's κ = 0.89) and on theme in 313 (89.7%, κ = 0.84). Disagreements were settled by discussion, and the consensus codes are used throughout.

Five other variables were read directly from the listing. These were the year and the attendance, where the organiser reported one. We also noted whether the invitation was open to all residents or to a named circle. We noted whether a Mind took part, by invitation, and whether the announcement names a speaker who had been alive during the war. The forms, themes and invitation categories are our own constructions. Six periods were fixed before coding: 1376–1449 CE (74 years), then five centuries from 1450 to 1949 CE.

Rates are counts per Orbital-century, with exact Poisson 95% intervals. Counts in the 24 Orbital-by-period cells were more variable than a Poisson process allows, and the dispersion of 1.69 estimated from them (Pearson χ² = 25.29 on 15 degrees of freedom) is used to scale the likelihood-ratio test of a period effect after allowing for Orbital, which we report as an F statistic. The intervals in Table 1 take no account of this dispersion, and should be about 30% wider. Trends in shares across the ordered periods use the Cochran–Armitage test. Attendance is analysed as log₁₀ of the reported figure in a linear model with Orbital as a factor, because observances cluster within Orbitals. Anniversary clustering is tested by an exact binomial test on a single interval, 25 years counted from the war's end in 1375 CE, which we fixed before examining the data and did not vary.

3. How Often, and in What Form

Announced observances were most frequent in the generation after the war and thinned quickly. The first period holds 125 of the 349, a rate of 42.2 per Orbital-century (Table 1). The rate halves in the next century and settles between 8.0 and 10.5 from 1550 CE, with no clear recovery. After allowing for differences between Orbitals, the period effect is large (F(5,15) = 17.50, p < .001). The fourth century (1650–1749 CE) and the last (1850–1949 CE) have the lowest rate, 8.0 per Orbital-century each, about one-fifth of the first period's.

The Orbitals differ less than the periods do. Over the whole span the rates were 17.8 per Orbital-century on Chiark, 17.1 on the second Orbital, 15.9 on the third and 10.1 on the fourth, against 15.2 overall, and all four registers show the same fall.

Form changed alongside frequency (Table 2). In the first period silent assemblies and readings aloud made up two-thirds of observances (83 of 125, 66.4%). The announcements of that time describe long evening silences in halls that the Hub had opened on request, and readings by people who had lived through the war and chose to speak of it. By the last period silent assemblies had almost gone (1 of 32, 3.1%), and shared meals, readings and music or performance made up 26 of 32. Shared meals were the commonest form from the third period to the fifth, and reached 17 of 32 (53.1%) in the fourth. Music and performance, rare at first (1 of 125), accounted for 28 observances overall, 6 of them in the last period.

Anniversaries of the war's end drew gatherings. Forty-five observances (12.9%, Wilson 95% CI 9.8–16.8%) fell in the 22 years that are whole multiples of 25 years from the war's end, from 1400 to 1925 CE, which together make up 3.8% of the 574 years (exact binomial p < .001). The share was 24 of 243 (9.9%) in the first three periods and 21 of 106 (19.8%) in the last three, and it rose across the six periods (Cochran–Armitage z = 2.88, p = .004, from the per-period counts in Table 1). Fewer observances were held as time passed, but a larger share of those that were held were timed to an anniversary.

4. Who Took Part

Terms of invitation moved steadily toward openness. In the first period 40.0% of observances (50 of 125) were open to all residents, and the rest were for a circle named in the announcement, usually a household, a workplace or a group of friends. In the last period 81.3% were open (26 of 32). Over the six periods the open share rose without reversal (Cochran–Armitage z = 5.01, p < .001), and 187 of the 349 observances (53.6%, 95% CI 48.3–58.7%) were open overall.

Attendance was reported for 224 observances (64.2%), and the share rose from 52.0% in the first period to 87.5% in the last. Reported attendance had a median of 64 (interquartile range 33–120). The median by period ranged from 48 to 90, with no consistent direction, and after allowing for Orbital the period differences are small beside the scatter (F(5,215) = 1.66, p = .15). Form accounted for far more (F(4,211) = 22.41, p < .001). Medians were 150 for music and performance (20 observances), 105 for silent assembly (61), 71 for walks (19), 61 for readings (65) and 28 for shared meals (59). Open observances drew more than closed ones of the same form, period and Orbital: the open median was 76 (129 observances) against 46 (95), and the adjusted ratio of geometric means was 1.58 (95% CI 1.27–1.98, F(1,210) = 16.37, p < .001). Open observances were also somewhat more likely to report attendance at all (69.0% against 58.6%), which this ratio does not remove.

The presence of those who had lived through the war faded early. An announcement named a speaker who had been alive during it in 134 of the 349 observances (38.4%), but in 78 of 125 in the first period (62.4%) and in 42 of 78 in the second (53.8%). Only 11 of 40 in the third period and 3 of 32 in the fourth named one, and the last such announcement is dated 1731 CE. Some who spoke may have done so without the announcement saying so, and the registers record no ages.

Minds came more often as time passed. A Mind took part by invitation in 122 observances (35.0%, 95% CI 30.1–40.1%), whether the Orbital's Hub or another Mind, and the share rose from 28.0% in the first period to 46.9% in the last (Cochran–Armitage z = 2.91, p = .004). Announcements describe Minds attending, answering questions, or contributing readings and music. Hubs organised none of the occasions.

5. What Was Remembered

Loss was the stated theme of 182 observances (52.1%). The decision to fight accounted for 64 (18.3%), inquiry into the Idirans for 55 (15.8%), and peace and the Culture's way for 48 (13.8%). The weight of these shifted markedly. Loss fell from 75.2% of observances in the first period to 47.5% in the third and 21.9% in the last (Cochran–Armitage z = −8.34, p < .001), with a low of 11.9% (5 of 42) in 1750–1849 CE. The three other themes together made up 24.8% of the first period and 78.1% of the last. Loss is the commonest single theme in each of the first four periods. In 1750–1849 CE inquiry into the Idirans leads, and in the last period the decision to fight does.

Gatherings about the decision to fight ask why the Culture chose war. Their announcements describe readings from the debate over entering it, and discussions at which anyone could speak for or against the choice. Such gatherings accounted for 15.0% of the third period (6 of 40) and 25.0% of the fourth (8 of 32), and they remained between 21.4% and 31.3% of every period after that.

Inquiry into the Idirans is the least anticipated theme. In 23 of the 55 observances of this kind (41.8%) the announcement names its readings as translations from captured Idiran devotional writing, material held in the Idiran Captured Scripture Collection (Idiran Captured Scripture Collection, 1327–1375 CE) and since studied in its own right (see the study of captured texts in this journal). Others were talks on the Idirans' way of life. This theme was rare in the first period (9 of 125, 7.2%), and it peaked at 18 of 42 observances (42.9%) in 1750–1849 CE, before falling to 6 of 32 (18.8%) in the last period. We treat that single peak as a feature of one century and draw no trend from it.

Observances on peace and the Culture's way were never a majority. They rose from 6 of 125 (4.8%) in the first period to 9 of 32 (28.1%) in the last. The announcements ask what the war taught about the limits of a society that fights seldom, and what an Orbital owes to the neighbours it will never meet.

6. Discussion

The registers show a practice that survived by being kept up. Announced observances became far rarer after the first generation, and they neither vanished nor recovered: for four centuries after 1550 CE they held at between 8.0 and 10.5 per Orbital-century. Nothing compelled anyone to keep them going, and no listing records an obligation. They continued on all four Orbitals, with a growing share open to all comers.

Content changed as well as number. Mourning dominated at the start and fell below half of observances from the third period, while questions about the war's causes and meaning took its place. Two changes coincide with this: the speakers who had lived through the war stop being named after 1731 CE, and the forms move away from silent assembly, which almost disappears, toward a mix of readings, meals, music and walks. The data cannot say which of these drives the others. Loss was already below half of observances in the third period (47.5%), when speakers who had lived through the war were still named in more than a quarter (27.5%), so first-hand memory is not the only factor, and from the registers alone we cannot say how much it matters. A change in fashion across all four Orbitals would look the same.

Anniversary years drew a larger share of a smaller number of gatherings. That is what one would see if the war's memory came to depend more on occasions marked in advance, but the registers cannot show spontaneity directly, and we claim only the concentration.

Attendance at each observance gives no sign of decline: where reported, it differed by form and by openness, not by period, though the report rate itself rose over time.

7. Limitations of the Record

Everything here concerns announced gatherings. A household that remembered the war at its own table without telling the Hub is not in the registers, and observances may have moved over time between announced and unannounced forms. The decline in rate could partly reflect that movement, and the registers cannot tell us. The same applies to the movement toward open invitations, since an organiser who invites a wide public has more reason to list an occasion.

Four Orbitals are not a sample of the Culture. Three are known to us through copies that their Hubs chose to deposit, and Hubs that deposit may differ from those that do not. Every date is a year, and the period boundaries and the 25-year interval are our own choices.

Counts across Orbitals and periods are more variable than a Poisson process allows, and the F test corrects for that with only 15 residual degrees of freedom, so the interval estimates in Table 1 are too narrow. Attendance is missing for 125 observances. The missing figures are not random, and the analysis of attendance should be read with that in mind. Coding of form and theme was done by two of us; the third author supplied the Chiark series and answered questions about it, and took no part in coding. The Commission, whose members include the first two authors, has published histories of the war that some observances may have used as readings; we did not measure how far.