18 June 2026

This morning I told myself an old story. I have told it to groups before — participants sitting in a circle on the floor at the end of a personal development week, a candle burning in the middle, the instruction to focus and extinguish it through collective intention. The candle never went out. One of us would walk to the centre, lean down, and blow it out. The point being made: it doesn't matter what you know, what you've experienced, or what you intend. If you don't get up and do something, the candle stays lit.

I told myself that story on a morning when I needed it. Fourteen or sixteen days of low motivation, sore feet, sore legs, sore neck. Quick to sit, quick to scroll, slow to move. I recorded a voice note before I started — an honest account of where I was, not where I wanted to be. Then I blew out the candle. Fifty-two minutes and fifty-four seconds of Taijiquan, following David-Dorian Ross's TaijiFit Flow programme. I was sweating by the end. My legs were shaking.

The data confirmed what the body already knew.

The Experiment That Didn't Quite Work

I had intended to run a continuous verbal commentary while moving — recording observations in real time alongside the session. The idea was to produce a layered document: physiological data from the Watch, narrative from the voice recording, both timestamped and mappable against each other.

It didn't work. Not because the idea was wrong, but because talking and breathing are competing demands when the practice is Taijiquan. The breath is not incidental to the form — it is the form. Disrupt it to narrate it and you've undermined the thing you're trying to describe. The heart rate data showed this clearly: the sections of highest physiological demand — the single-leg kicks between minutes forty and forty-nine, where HR peaked at 141 bpm — were exactly the sections where the commentary almost disappeared. The narration stopped where the real work began.

This is useful information. Where the words run out is where the data starts to matter most.

The decision was made to separate the two. The .fit file produced by my Apple Watch carries the objective record. Analysis happens after the session. I respond to specific anchors from memory. The voice note, where it exists, is a pre- or post-session reflection rather than a live commentary. This is, as it turns out, methodologically sounder — closer to how experienced ethnographers have always worked. Geertz did not write his thick descriptions standing in the middle of the cockfight.

What the Watch Saw

Session duration: 52 minutes 52 seconds. Average heart rate: 107 bpm. Peak: 141 bpm. Calories: 331. The session divided into eleven five-minute laps via auto-lap, with a short final segment.

The HR profile showed a gradual rise through the opening sequences — crane and flying imagery, cloud hands, the window and curtain work — settling into a sustained band between 100 and 120 bpm for the middle third of the session. This is where approximately half the session time was spent. The final push, the kicks and single-leg balance work, drove the highest sustained load of the session.

I noted during the kicks that standing on my right leg was genuinely hard. The hamstring was tight. The right leg buckled inward at one point earlier in the session. I used my stick for the balance section. I kept going.

These observations — the right leg asymmetry, the stick, the shaking at the finish — are as much primary data as the heart rate figures. They are recorded here because they are consistent findings that need tracking across the 28-day streak beginning tomorrow.

Building the Recording System

Part of today's work was practical: designing the data capture system for the streak itself. This involved a longer conversation than expected about devices, apps, and the reliability of different metrics — particularly HRV in the context of permanent atrial fibrillation.

The short version: standard HRV methodology assumes sinus rhythm. In permanent AFib, beat-to-beat variation is produced partly by the irregular rhythm itself rather than purely by autonomic tone. This makes HRV readings higher and more variable than equivalent sinus rhythm readings, and makes day-to-day comparison less clean. The numbers are not useless — they form one data stream among several — but they are not the primary metric they might be for someone without AF. The .fit file data is more reliable as a daily comparator for this project.

A consistent morning protocol has been established: two minutes with the Mindfulness app before getting up, lying still. This controls the measurement conditions in a way the passive overnight capture does not. The reading goes to Apple Health automatically.

Walking training will begin shortly and will be logged identically — the Watch records it, the .fit file is exported, the same analytical process follows. A weekly ECG reading has been added as an optional field in the log, not for diagnostic purposes but as a timestamped record of rhythm at that point in the streak. If nothing changes, it confirms the baseline. If something does change, there is documentation of when.

The 28-day log document has been built. Each daily entry takes a readiness score and HRV reading before the session, populates session metrics from the .fit file, and includes a brief qualitative response to specific anchors from the analysis. A summary grid at the back will show all 28 days across a single table as the streak builds. The trajectory is what matters, not any single entry.

A Note on Method and AI

This post was drafted by Claude from the conversation that produced the decisions described above. That conversation is primary research data. The thinking in it — the decision to abandon the live commentary, the HRV reliability question, the recognition that designing the recording system is itself autoethnographic — emerged from the session and the discussion of it. The AI contributed information and analytical structure. The intellectual direction, and the fifty-two minutes on the mat, were mine.

The use of AI in this project has been addressed elsewhere on Here be Dragons. The division of labour here is practical and transparent: I live it and move it; the machine reads the data and keeps the log. This gives me the mental space to write about what the movement actually feels like, which is what Here Be Dragons is for.

The process of working out which data to collect, which tools to trust, and which metrics are meaningful for a body with this particular history — that process is not preamble to the research. It is the research. The instrument is being built in real time, by the person it is built to measure.

Tomorrow the streak begins. Nineteen June 2026.

Renegotiating Gravity.

#methodolgy