Living on the Tree: Limb Deflection, Bark Damage and Girth Response under Wookiee Dwellings of Different Occupation Age in Five Kashyyyk Tree-Cities, 28–24 BBY
Abstract
Wookiee tree-cities hang dwellings and walkways from the limbs of living giants, some for centuries, and no measurement has said whether such limbs adapt to the load or slowly fail. I surveyed 90 trees in five Kashyyyk settlements, 18 in each of five occupation-age classes from under 25 to over 300 standard years, in five annual rounds from 28 to 24 BBY. I recorded limb droop, droop creep, bark damage, girth increment against unloaded limbs, and the way keepers share the load. Mean droop was higher in the older classes, 40.2 against 20.8 mm per metre of span in the oldest and youngest, creep was lower, 0.17 against 0.54 mm per metre per standard year, and bark damage was greater, 19.3% against 4.2% of girth. Girth response to loading peaked in the middle classes at 13.6 percentage points and was weak in the oldest class (3.2, 95% CI −0.9 to 7.3). Older dwellings carried lighter and more evenly spread loads. Measured creep exceeded the rate implied by the class sequence by a factor of about three in the first three intervals, so occupation age is a poor stand-in for time on a single limb; rework, stiffening, changes in building practice or the loss of fast-drooping limbs could each account for part of the gap, and the data cannot yet separate them. The study cannot yet say whether long-occupied limbs adapt or slowly fail, and it names the re-surveys, cores and registers that could.
1. Introduction
On the 212th day of 26 BBY the survey party at settlement K-3 found the landing of a main walkway 11 centimetres below the mark it had held at the previous year's round. The walkway hung from a rhoka limb with a horizontal span of 9.2 metres, on a tree whose first load had been placed about 260 standard years before. The keepers of the adjoining dwellings said that most of the drop had come in a single night, when the limb made a long low sound through the evening and the walkway was cleared before dark, so that nobody was on it when it settled.
The keepers eased two cradles the next morning and moved part of the weight onto a second limb higher on the trunk. The party measured the landing again before it left the settlement that evening.
Wookiees hang dwellings from living wood that goes on growing beneath them. A limb that has carried a household for three centuries has been tuned to the load or is being undone by it, and no measurement says which. The reconnaissance of the great-tree stands recorded girth classes and platform counts and said nothing of what platforms do to limbs (Outer Rim Planetary Survey, 28 BBY). Work on unburdened limbs of emergent giants shows slow creep under their own weight (Tavreck, 29 BBY). Living wood lays down reaction tissue where it is stressed (Ulmane, 27 BBY), and bark under a point load cracks, crushes or heals according to how the load is applied (Staveth, 26 BBY). None of these was tested on a limb that carries a dwelling.
This paper reports a four-year field study from the Kashyyyk Institute of Arboreal Ecology, written in 24 BBY, which is the vantage of every statement below. It measured limb deflection, bark damage and growth response under dwellings and walkways in five tree-cities, with trees grouped by occupation age. It asks whether long-occupied trees adapt to their load or slowly fail. A four-year study cannot watch a process that runs for centuries, so occupation age stands in for time, and the paper tests what that substitution costs.
2. Methods
Settlements and trees. The Institute visited five tree-cities, labelled K-1 to K-5. Two species carry nearly every dwelling in them: the rhoka, a broad-limbed giant with thick fissured bark, and the ghuurn, a narrower tree with smooth bark and steeper limbs. I took 18 trees from each of five occupation classes, which gave 90 trees (22, 17, 19, 14 and 18 in K-1 to K-5). A tree qualified if it carried a dwelling or walkway on at least two limbs, its keepers consented, a climbing party could reach every loaded limb, and no dismantling was planned. Ghuurn made up 3, 5, 6, 7 and 8 of the 18 in classes A to E. The 90 trees carried 252 loaded limbs, at spans of 4.1 to 13.8 metres.
Measurements. Five rounds were held, one in each year from 28 to 24 BBY, each within the same fifty-day window. For every loaded limb I recorded the drop of its upper surface at the main load point below a level line sighted from the trunk collar, and divided it by the horizontal span to that point. This is the droop ratio, in millimetres per metre of span, and it includes whatever inclination the limb had before it carried anything. The class means of droop, bark damage and the two load indices are first-round values (28 BBY); creep uses all five rounds. Creep was the least-squares slope of a limb's droop ratio against standard years, averaged over the limbs of a tree. Bark damage was the share of girth, taped at the worst contact section of each limb, where bark was cracked, crushed or sloughed. Girth increment was taped midway along the loaded limb and on an unloaded limb of the same tree matched in girth to within 10%, as a percentage of starting girth over four years. The outcome is loaded minus unloaded, in percentage points; 73 trees had a matching limb.
Load and its spread. Dwelling and walkway masses were static estimates, good to about 15%, from an inventory of floor area, timber and fittings and the occupants stated by the keepers, and were assigned to limbs by an attachment plan. The heaviest-limb share is the fraction of a tree's supported mass carried by its most heavily loaded limb. Load intensity is the supported mass of a limb divided by its cross-sectional area at the collar, in kilograms per 100 cm2, averaged over the tree. Any rework by the keepers during the survey was recorded with its date, and creep for a limb was fitted only to rounds before the rework; in every reworked tree at least two rounds preceded it. A limb failed if it cracked through or dropped far enough to bring its load to rest on another support.
Occupation age. The age of a tree is the number of standard years from its first recorded load to the first round, in 28 BBY, taken from household registers (Zelmar, 24 BBY). Age belongs to the tree and not to the limb, so the age of the load on a limb loaded later, including by rework, is overstated by an unknown amount. For the two oldest classes the registers give a count of generations and not a date, and I converted generations at 35 standard years each. No tree gave both, so the constant could not be calibrated on this sample. Classes A to E are bounded at 25, 75, 150 and 300 years. The design treats a class as a stage in the life of one limb. That holds only if the classes come from one population of trees and practice, if limbs that fail leave no systematic gap, and if load histories follow a common path (Yorane, 25 BBY). I test one of these in the Results by comparing the difference in droop between neighbouring classes with the creep measured directly in the same classes.
Analysis. The tree is the unit, because limbs on one tree share species, soil, load history and keeper (Wessaly, 24 BBY), and limb values were averaged within trees. Trends across the five ordered classes use a linear and a quadratic contrast with a pooled variance, reported per class step. The oldest class is compared with the youngest by an interval for the difference in means with unequal variances. Failure counts use exact binomial intervals and an exact test. All other comparisons are descriptive. Twelve trend contrasts were examined, the linear and the quadratic for each of six outcomes, so I treat p below .004 (.05 divided by 12) as clearing correction for multiplicity and say so where a result falls between that bound and .05.
3. Results
Of the 90 trees, 88 contributed to creep, because one tree each in classes D and E was dismantled early; 73 contributed to girth, and all 90 to the other outcomes. Twenty trees (22%) had load reworked by their keepers during the survey: none in class A, 2 in B, 4 in C, 6 in D and 8 in E. Seven limbs failed over the four years, 2.8% of the 252 (exact 95% interval 1.1–5.6%), and they were 0, 1, 1, 2 and 3 limbs in classes A to E; class E (3 of 50 limbs) against class A (0 of 50) gave an exact p of .24. Counting each limb for four years, the failures came to 0.69 per 100 limb-years over 1,008 limb-years; classes A to C had 2 in 612 (0.33 per 100) and classes D and E 5 in 396 (1.26 per 100).
Droop was higher in each older class (Table 1). The mean droop ratio was 20.8 mm per metre in class A and 40.2 in class E, a difference between the classes of 19.4 (95% CI 14.2–24.6). The linear difference was 5.2 mm per metre per class step (Table 3), and the sequence bent only slightly (quadratic coefficient −0.73, −1.67 to 0.21; p = .13). Most of the difference between classes A and E was already present in class D.
Creep, the one droop measure followed within limbs, was lower in the older classes. A limb in class A gained 0.54 mm per metre of span each standard year (0.46–0.62), and a limb in class E gained 0.17 (0.12–0.22); the difference was −0.37 (−0.47 to −0.27). Creep was lower in each class through C and then levelled, as the positive quadratic coefficient shows (0.028, 0.012–0.044; p = .001).
Bark damage was greater in each older class, 4.2% of girth at contacts in class A and 19.3% in class E (difference 15.1 points, 10.6–19.6). Girth response followed another course (Table 2). Loaded limbs outgrew their unloaded partners by 4.1 percentage points in class A (1.0–7.2), by 13.6 in class C (10.0–17.2) and by 3.2 in class E (−0.9 to 7.3). The linear difference was nil, and the bend was marked (quadratic coefficient −2.30, −3.18 to −1.42; p < .001). The response peaked in the middle classes, and class C exceeded class E by 10.4 points (5.2–15.6), while the interval for class E includes zero.
Older dwellings carried lighter and more evenly spread loads. The heaviest-limb share was 61.0% in class A and 30.9% in class E (difference −30.1 points, −35.9 to −24.3), and load intensity was 3.12 kg per 100 cm2 in class A and 1.78 in class E (−1.34, −1.78 to −0.90). Creep was 69% lower in class E than in class A, and creep per unit of load intensity (0.17 in class A, 0.10 in class E) 45% lower. If creep were proportional to load, lighter loading would account for about half of the difference in logarithmic terms (49%). These are ratios of class means, and proportionality was not tested.
If a limb moved through the classes in turn, the difference in droop between neighbouring classes divided by the difference in median ages should equal the creep measured directly, and it did not. Between classes A and B the class means imply 0.17 mm per metre per standard year, against a measured 0.48; between B and C, 0.12 against 0.32; between C and D, 0.05 against 0.20; and between D and E, 0.007 against 0.17. The measured creep exceeded the implied rate by factors of 2.9, 2.6, 3.8 and 22. The last factor is unstable, because the droop difference between D and E was 1.3 mm per metre (95% CI −4.5 to 7.1).
Without the reworked trees, mean creep was 0.19 mm per metre per year in class E (9 trees; 95% CI 0.14–0.24) and 0.17 in class D (11 trees; 0.12–0.22), against 0.17 and 0.16 with them. Rework does not explain the levelling.
4. Discussion
Within the sample, four statements hold, each a difference between classes; only the creep rates were measured within limbs. Limbs in the older occupation classes droop more, and most of the difference between classes is reached by class D (150–299 years). Creep is lower in the older classes. Bark damage at contacts is greater in each older class, with no sign of levelling. Older dwellings carry lighter and more evenly spread loads. None of these says whether a limb is tuning itself to its load, and I do not think any of them can.
Girth response bears most directly on adaptation. A limb that thickens under load changes its stiffness in a way that droop alone does not record, and loaded limbs in the middle classes outgrew their unloaded partners by up to 13.6 percentage points. A weak response in class A may mean only that reaction tissue has not yet formed. A weak response in class E could mean that the limbs have finished adjusting, or that growth cannot repair a limb whose bark is damaged over nearly a fifth of its girth, and the data do not choose between these.
Measured creep exceeded the rate implied by the class sequence by a factor of about three in each of the first three intervals, and the check cannot be set aside. Limbs cannot be drooping that much faster than the sequence accumulates droop unless droop is being undone along the way, the older classes are missing the limbs that drooped fastest, today's young dwellings load their limbs harder than older ones did when new, these four years were unusually fast, or the age given to a limb is greater than the age of its load. Undoing could come through rework, stiffening or re-hanging. Failure offers another path, in which a limb leaves the population, though seven events cannot separate it from the rest. At the rate seen in classes A to C, 0.33 failures per 100 limb-years, about 28% of limbs would be lost in a first century, and at the pooled rate of 0.69 about half; both are illustrations and not estimates. Closing a threefold gap by loss alone would need the losses to be strongly concentrated among the fastest-drooping limbs. Between classes A and B, a 32-year step, the A-C rate removes only about 10% of limbs.
On building practice the data show keepers reworking dwellings in every class except the youngest, while the heaviest-limb share and the load intensity were lower. Whether keepers lighten a dwelling as it ages or older dwellings were lighter when built, the data cannot say, because load was not followed within trees. If creep is proportional to load, lighter loading leaves the rest of the difference in creep to tuning, to the loss of fast-drooping limbs, or to both.
5. Limitations
Occupation age stands in for time at a cost that the Results measure only in part. The two oldest classes rest on generation counts converted at an assumed 35 years. Wookiees are long-lived, so the constant may be short, and an error scales the ages of classes D and E together. Holding each tree in its class and rescaling only the medians, at 50 years per generation the median ages of D and E would be 300 and 550 instead of 210 and 385, the implied rate between them would fall from 0.0074 to 0.0052, and the factor between measured and implied creep from class C to class D would rise from 3.8 to 7.1. A longer generation therefore widens the gap in the check and does not close it. No within-tree change in bark damage or load is reported, so whether bark damage accumulates on a given limb is untested here.
The sample is of survivors that a climbing party could reach. Trees that were abandoned, felled by storm or given up after a limb failed do not appear in the registers I used, so class E is what remains of a larger set. Trees whose limbs could not be climbed safely were excluded, and those may be the weakest.
Age and load are entangled, since load intensity is lower in the older classes and no tree was chosen to cross the two. Species are unevenly spread, with ghuurn rising from 3 trees in class A to 8 in class E, and were not modelled. The droop ratio also carries the inclination a limb had before loading, and the steeper limbs of ghuurn would by themselves lower droop in the older classes and flatten the step from D to E on which the check leans. No result falls between .004 and .05.
Four years is short, and a linear slope fitted to five rounds averages events that arrive in steps, as the drop at K-3 did. Rework cut some series short, and the 20 trees affected were mostly the oldest. The 73 trees with a matched unloaded limb were not a random subset of the 90.
Return to K-3, where the landing fell 110 millimetres over 9.2 metres, 12.0 mm per metre, in a night, about 75 years of mean class D creep (0.16 mm per metre per year), and did not move in the rounds of 25 and 24 BBY. If long-occupied limbs are tuned to their load, the drop may be the last adjustment of a limb whose wood had caught up with the dwelling. If limbs fail slowly, it may be an early step in a sequence that rework postpones. The record holds one such drop in one limb and cannot say which. Four things could decide the matter. A re-survey of the same 252 limbs a decade or two on would follow each limb over decades and not across classes. Increment cores from the collars would read the growth rings directly. Continuous recording on a few limbs would catch step drops as they happen. The keepers' registers of limbs given up, read against Zelmar (24 BBY), would give failure rates over generations and not over four years.
References
- Outer Rim Planetary Survey (28 BBY). Reconnaissance of the great-tree stands of Kashyyyk, trunk girth classes and platform counts. Outer Rim Planetary Survey Reports, Report 119.
- Tavreck, M. (29 BBY). Creep under self-weight in the lateral limbs of emergent forest giants. Comparative Xenobiology Review, 11(4), 201–224.
- Ulmane, K. (27 BBY). Reaction tissue and the timing of girth response in stressed limbs of closed-canopy trees. Comparative Xenobiology Review, 13(1), 33–58.
- Staveth, R. (26 BBY). Crush cracking and callus closure in thick bark under point loads. Comparative Xenobiology Review, 14(2), 101–124.
- Yorane, T. (25 BBY). Space-for-time substitution in age-ordered stands, survivorship and cohort bias. Comparative Xenobiology Review, 15(3), 88–110.
- Wessaly, J. (24 BBY). Tree-level units and limb-level repetition in measurements on long-lived plants. Proceedings of Applied Speculative Statistics, 15(1), 61–84.
- Zelmar, P. (24 BBY). Household occupation registers of Wookiee tree-cities and the dating of first loads. Coruscant Journal of Galactic History, 13(1), 120–146.
Open in Uncited Press →