A Week of One's Own Choosing: Occupation, Play and Rest in 1,043 Time-Use Diary Weeks Kept by 286 Residents of Chiark Orbital, 1790–1845 CE

1. Introduction

Everyone in the Culture may spend a week as they please. No one is obliged to take up an occupation, no one is rewarded for doing so, and nothing is held back from a person who takes up none. Contact tells visiting civilisations as much, and visitors from societies organised around obligatory labour tend to ask what then fills the days. The Culture's answer is that people fill them with projects they have chosen: building, tending, teaching, composing, studying, organising and a great deal else, pursued for as long as they remain worth pursuing (Haldrenne, 1668 CE).

Earlier studies of this question have relied on what residents say. Interview work records that citizens find the line between occupation and play hard to draw (Torvasque, 1731 CE), that young adults treat a first occupation as a trial and expect to leave it (Brimmelore, 1779 CE), and that people recall changing direction many times over a long life (Dathwane & Orrimund, 1812 CE). Recollection over decades is a poor instrument for amounts of time. A person's memory of a hundred years of work tells us what they thought mattered, not how many hours a week it took.

Self-kept diaries answer a different question, which is how a particular week was spent. Methodological work on the form shows that diaries kept at the time agree far better with an independent record than recollection does, provided entries are short and are written the same day (Quillamar, 1826 CE). This paper uses the most extensive diary series known to us, deposited with the Hub of Chiark Orbital by residents who chose to keep one (Chiark Orbital Hub, 1790–1845 CE).

We write in 1850 CE. We ask how weekly time in a chosen occupation varies with age across a life that commonly lasts 350 to 400 years, how that time is shared with play, company and rest, and how often a person's principal occupation changes. Our analyses of the diaries were not specified before we read them, and we say below which ones we settled on after looking at the data.

2. Methods

The diary series is the Hub's own project. From 1790 CE the Hub of Chiark Orbital invited a sample of residents drawn from its resident index, which records age and household for essentially every resident (Chiark Orbital Hub, 1846 CE), to keep a seven-day diary once a year for as long as they cared to. Participation was by invitation and by choice at every step. A diarist could decline any year, stop at any time, and withdraw a week after writing it. By long custom the Hub does not read minds or monitor private life on Chiark, and the diaries record only what their writers chose to set down. The series ends in 1845 CE, after 56 annual rounds, when the Hub closed it at the diarists' common request. The series name, structure and consent rules are the Hub's; the coding scheme and every analysis below are ours.

Invitations went to 640 residents aged 18 or over, stratified by age at invitation into seven bands of 18–49, 50–99, 100–149, 150–199, 200–249, 250–299 and 300 and over, with 120, 110, 100, 100, 90, 70 and 50 invitations respectively. Of these, 301 consented (47.0%). Consent was 50.0% in each of the first three bands, 49.0% in the fourth and 48.9% in the fifth, but 37.1% in the sixth and 34.0% in the seventh. Of the 301, 286 (95.0%) returned at least one usable week; 15 returned none. Many diarists began in a year after their invitation, which is why the youngest band holds 34 diarists although 60 invitees in that stratum consented (57 returned a usable week), and why the youngest diary age is 23.

Diarists returned 1,102 weeks in all. We excluded 59 because fewer than five of the seven days were complete, which leaves 1,043 weeks for analysis. Diarists contributed a median of 3 weeks (range 1 to 11, mean 3.6), spread irregularly over the 56 annual rounds, and 24 contributed only one. Ages at the time of the diary run from 23 to 387 years. A diarist's weeks fall into different age bands as they age, so the counts of diarists in Table 1 add to more than 286.

Each diary divides the waking day into dated entries. The series holds about 31,200 entries for the weeks analysed. Two of us coded each entry into one of five categories, defined before coding. A chosen occupation is sustained work on a project the person has taken up and could put down, whatever its kind. Play is games, sport and performance for its own sake. Company is time given to conversation, meals and gatherings. Rest is sleep outside the night, idleness and quiet. All else, chiefly travel, errands and care of the body, is Other. To assess reliability, both coded a random 10% of entries (3,120) independently. They agreed on 92.7% and Cohen's kappa was 0.91. Weekly hours in each category were summed from the entries, and waking hours recorded per week averaged 108.3.

The main outcome is weekly hours in a chosen occupation, which is continuous and roughly symmetric (mean 29.2, standard deviation 9.6). Weeks within a diarist are not independent: about 44% of the variance lies between diarists (one-way ANOVA estimator). We therefore fitted linear regressions by ordinary least squares with standard errors clustered by diarist (286 clusters), and report 95% confidence intervals and p values from the normal approximation.

Three models were fitted. The first compares the seven age bands with the youngest as reference, adjusting for diary year in decades. The second replaces the bands with age in centuries and its square. We chose the quadratic after inspecting the band means, so its p values are conditional on that choice. For the age at which the quadratic peaks we report a percentile interval from 2,000 bootstrap resamples of whole diarists. The third uses only diarists with two or more weeks (262 diarists, 1,019 weeks) and regresses hours on age in decades after removing each diarist's own mean, which isolates change within people. For the principal occupation, a diary pair is two successive diaries by one diarist, and a change is recorded when the diarist names a different principal occupation in the later one. We fitted a logistic regression of change on age at the earlier diary and on the interval between diaries, again clustered by diarist.

3. Results

Weekly hours in a chosen occupation rose from early adulthood to the second century of life and fell in the fourth. Table 1 gives the band means with the other categories. Mean hours were 24.2 (95% CI 20.5 to 27.8) at ages 18–49, 32.6 (30.9 to 34.4) at 150–199 and 20.7 (18.3 to 23.0) at 300 and over. The youngest band is small, with 53 weeks from 34 diarists, and its interval is wide.

[object Object]

In the band model, every band from 50–99 to 250–299 differed from the youngest. The differences were 5.00 hours (1.30 to 8.69, p = .008) at 50–99, 8.29 (4.19 to 12.38) at 100–149, 9.02 (4.89 to 13.15) at 150–199, 7.28 (3.26 to 11.29) at 200–249 and 5.80 (1.21 to 10.38, p = .013) at 250–299. The band of 300 and over did not differ detectably from the youngest, at −2.85 hours (−7.20 to 1.49, p = .20). The bands as a set differed strongly (Wald χ²(6) = 85.7, p < .001), and diary year had no detectable association with hours (−0.30 per decade, −0.65 to 0.06, p = .10).

The quadratic in age gave a linear term of 16.23 hours per century (12.49 to 19.97) and a squared term of −4.64 (−5.59 to −3.69), so hours rise and then fall. The curve peaks at about 175 years, with a bootstrap interval of 165 to 183 from 2,000 resamples of diarists, refitting the year-adjusted model each time. The peak sits inside the 150–199 band, where the band mean is highest. These intervals describe where this panel's curve turns, given the quadratic form, and do not locate a turning point in any individual.

Within diarists, we found no trend. Across 262 diarists with two or more weeks, hours changed by −0.04 per decade of age (95% CI −0.40 to 0.32, p = .84). Each diarist's weeks cover at most 54 years, a short stretch of the curve, and the interval excludes only trends larger than about four hours per century downward or three upward. The age contrast in Table 1 therefore comes mainly from differences between people of different ages.

Few weeks held almost no occupation. Twelve of the 1,043 (1.2%) recorded fewer than five hours, of which 4 fell among the 53 weeks in the youngest band (7.5%) and 6 among the 108 in the oldest (5.6%). At the other end, 127 weeks (12.2%) recorded 40 hours or more, with the largest counts at 100–149 (33) and 150–199 (32).

The other categories moved in other directions. Play fell from 24.3 hours at 18–49 to 20.5 at 200–249, then rose to 22.3 in the oldest band; its range was smaller than for any other category. Rest rose across the bands, from 18.8 hours at 18–49 to 30.1 at 300 and over, company rose in the oldest band to 31.2 hours, and Other fell from 15.0 to 4.5 hours. These means carry no intervals, and we draw no inference from them beyond the description.

Principal occupations changed in 354 of 757 successive diary pairs (46.8%), among 262 diarists. The share fell with the age at the earlier diary, as Table 2 shows. In the logistic model, the odds of a change fell by a factor of 0.62 per century of age (95% CI 0.52 to 0.75, p < .001), adjusting for the interval between diaries, and rose with that interval by a factor of 1.16 per year (1.13 to 1.19). Intervals ran from 1 to 46 years, with a median of 8, so a recorded change is a change between diaries, not a count of all changes.

[object Object]

4. Discussion

In this panel, time given to a chosen occupation was neither constant nor uniformly falling. It was low in early adulthood, highest in the second century and lowest after 300, where weekly hours fell back to about the level of the youngest band. A reading that treats occupation as something pressed on the young and shed with age does not match these diaries, since the youngest adults gave fewer hours than any band between 50 and 299, and about as few as the oldest. The Culture has no mechanism that would press anything on anyone, and in these diaries residents in the middle of a long life give more hours to occupation than residents at either end.

The same diaries show a loosening of commitment that is clearest before age 200. Changes of principal occupation grew rarer with age in this panel, with 64.0% of pairs changing from the youngest band, 45.9% at 150–199 and 35.5%, 39.2% and 36.2% in the three oldest bands. Adjusting for the interval between diaries leaves a clear decline per century. People in their first decades act as the interviews describe, taking occupations up and leaving them (Brimmelore, 1779 CE), while older diarists change direction less often, and do so in weeks that still hold many hours of occupation.

Rest and company both rose in the oldest band, which fits the idea that later life redistributes time among kinds of activity as well as reducing it. We cannot say more, because the category means have no intervals and the oldest band is small. Play varied least of the categories, which sits easily beside interview evidence that play and occupation shade into one another (Torvasque, 1731 CE).

Our within-person result limits all of this. The diaries do not show an individual's occupation hours rising and falling over a long life, because no individual is observed for more than 54 years and the estimated within-person trend is flat. The curve in Table 1 may describe how people at different ages differ, which could reflect the experience of different generations, the survival of different kinds of people to older ages, a matter of who chose to continue living, or who chose to take part, as well as ageing. A longer series would be needed to separate these.

5. Limitations

Consent fell to 37.1% and 34.0% in the two oldest invitation bands, against about 50% in the others. If the residents who declined differ in how they spend their weeks, the oldest bands are less representative than the younger ones. The Hub holds no record of how those who declined spend their weeks, and we cannot test this.

The series covers one Orbital over 56 annual rounds. Chiark's residents may spend their weeks differently from those of other Orbitals or of General Systems Vehicles. The curve and its peak should be read as a description of this panel.

Diaries record what writers chose to set down. We cannot rule out that some people recorded their weeks more fully when they were busy, or described rest as occupation and occupation as play. The coders agreed closely (κ = 0.91), but two coders working from the same text share what the text omits. The category definitions are ours, and the line between a chosen occupation and play is one that residents themselves find hard to draw (Torvasque, 1731 CE).

Weeks are nested within diarists, and diarists with many weeks contribute more to every estimate. We addressed this with clustered errors and a bootstrap by diarist, with 286 clusters, but we did not model diarist-level characteristics such as the nature of the occupation. The quadratic form was chosen after inspecting band means, so its p values are conditional on that choice, and the peak age depends on that form. A cubic or a piecewise form might place it elsewhere. The principal-occupation outcome depends on how diarists named their occupations, and diaries a median of 8 years apart will miss changes that began and ended in between.