Gwella — data notes, 6–17 July
For a fortnight, the same wood, mostly the same loop, walked at whatever hour the day allowed. Nine walking sessions logged on the Coed Cymerau circuit and its variants, three separate Qigong sessions alongside them, five rest days worked in on a five-on-two-off pattern. The point of holding the ground still was never the ground itself — it was to have one fixed thing to measure everything else against while footwear, pace, and load all changed underneath it.
What follows is the record as the watch and the app kept it, not as memory kept it. Where the two disagree, that disagreement is data too.
| Date | Session | Distance | Timer time | Avg HR | Max HR | Ascent | Pauses ≥15s |
|---|---|---|---|---|---|---|---|
| 6 Jul, pm | Circuit 1 | 1.50 km | 31:02 | 92 | 126 | 34 m | 0 |
| 7 Jul, am | Circuit 1 | 1.51 km | 30:53 | 118 | 131 | 32 m | 1 |
| 9 Jul, am | Circuit 1 (extended) | 2.58 km | 1:12:56 | 110 | 147 | 109 m | 14 |
| 10 Jul, am | Circuit 1 | 1.56 km | 31:48 | 87 | 127 | 34 m | 0 |
| 13 Jul, am | Circuit 1-2 (1) | 2.90 km | 1:20:01 | 111 | 139 | 116 m | 11 (10)* |
| 14 Jul, am | Circuit 1-2 (2) | 1.66 km | 42:36 | 102 | 127 | 35 m | 9 |
| 15 Jul, am | Circuit 1-2 (3) | 1.54 km | 41:58 | 101 | 126 | 31 m | 3 |
| 16 Jul, am | Alt 1 | 2.94 km | 1:28:36 | 107 | 139 | 69 m | 21 (20)* |
| 17 Jul, midday | Coed Cae Fali † | 1.44 km | 43:25 | 123 | 144 | 74 m | 8 (7)* |
* Figure in brackets excludes a start-of-session "dead start" — the app registering stationary time before recording begins properly, not a genuine pause. Confirmed against the record stream on three sessions (13, 16, 17 Jul); treated as an artefact throughout, not counted against the walker.
† Different site to the rest of the table — steps, not the flat circuit — included for the fortnight's completeness, not for direct comparison.
Green = the fixed baseline holding. Zero pauses on the 6th and the 10th, one on the 7th — a walker moving through familiar ground without needing to stop for it.
Amber and rust = the baseline fraying. The 9th is the first crack, nine days before the circuit was formally abandoned — a longer route, double the pauses of any prior session, the highest heart rate of the fortnight. The 13th onward is a different project in slow motion: the naming shifts from "Circuit 1" to "Circuit 1-2," the distance and the pause count both climb, and by the 16th the walk has left the loop entirely for a waterfall and an old blacksmith's track. The rust rows aren't failures against the baseline. They're the baseline being outgrown in real time, which is a different thing.
The 1.2km mark
The circuit's informal marker — the point where "mountain legs," the automatic, unthought quality of movement, either arrives or doesn't. Heart rate and pace in the 1.1–1.3km band, same measure applied across sessions of very different character:
| Date | Avg HR | Avg speed |
|---|---|---|
| 6 Jul | 112 | 0.81 m/s |
| 7 Jul | 117 | 0.89 m/s |
| 9 Jul | 107 | 0.67 m/s |
| 10 Jul | 92 | 0.84 m/s |
| 13 Jul | 94 | 0.48 m/s |
| 14 Jul | 102 | 0.73 m/s |
| 15 Jul | 110 | 0.87 m/s |
| 16 Jul | 122 | 0.54 m/s |
Lowest heart rate at the mark and a strong pace together, on the 10th — the single cleanest read of the fortnight. The worst combination — high heart rate, slowest pace — on the 16th, which was also the day the route left the baseline behind and the pauses reached their peak. No clean trend line across two weeks; the mark reads differently depending on what kind of walk it sits inside. That itself is worth stating plainly rather than smoothing into a slope that isn't there.
A methodological note
Three sessions — 13, 16, and 17 July — show a stationary period sitting exactly at the recorded start time, before any real movement: 161 seconds, 23 seconds, and 80 seconds respectively. This is not hesitation at the threshold. It is the app taking a moment to begin recording properly after the button is pressed, confirmed by checking the record stream directly rather than assuming. Worth logging here in case it recurs, and worth stating for anyone else building a fixed-baseline method on this app: the first pause in any session should be checked against the session start time before it's counted as anything meaningful.
Nine walking sessions, three Qigong sessions, one circuit held still long enough to watch it stop being useful. What that stopping means is the next piece.