# Meridian instrument bench · events-v1

Two independently authored detector references introduce event lists, exposure-aware counts and pulse-height distributions. These are fictional Geiger and muon-detector examples, not ELDÆON recordings, calibrated radiation measurements or particle-identification results.

The bench has its own authored session: 15 August 2026, 12:00:00–12:02:00 UTC. It is not synchronized to the main field study. CSV is the supported canonical table representation here; no provider-native NDJSON decoder is claimed.

## Sources and calculation

Each detector has a retained `*-events.csv`, `*-coverage.csv`, and derived `*-counts.csv`. Event IDs and row positions remain inspectable. Coverage is recorded at [0,40), [60,85) and [90,120) seconds. Other intervals were not recorded. The recorded [20,30) window deliberately has zero events. The [80,90) window has five seconds of exposure.

All intervals are half-open. Geiger counts include all authored pulses; the muon reference includes only rows flagged as coincident. Rejected rows remain available in the source, with no independent physical classification claimed. Counts divided by actual recorded exposure give the displayed rate. No dead-time, efficiency, energy or dose correction is applied. Zero exposure produces an absent count/rate, not a zero. `methods.json` retains these rules.

The histogram uses eight 512-ADC-unit intervals from 0 to 4096. It includes only accepted events in the selected time window. Histogram filtering changes the event list, not the window's overall count/rate. Exact canonical `table_row` selectors identify contributing event rows, coverage rows and derived count rows. CSV row indices exclude the header.

## Standard contracts and reproduction

`observation.json`, `history.json`, `acquisition.json`, `bindings.json`, and `source-selection.json` use the existing V2 standard. Count products have processing records pointing to their event and coverage inputs, generator and methods. Source-selection validation checks every row. Structural and semantic acquisition checks pass, while the acquisition profile stays incomplete because physical calibration and clock characterization are absent.

Use Node 22 and canonical source commit `88bcbfcc8b17af30837313aff596cec8e1bde43a`, with its schema dependencies built. From this directory:

```sh
node build.mjs /path/to/disclosureos
```

This regenerates the exact sources and evaluates the canonical packet. The downloaded viewer includes the same event reader and renderer. The histogram is a transient viewer calculation, not an independent observation. The source authoring files are CC0-1.0; code follows the included MIT license. No npm publication is required.
