Most pages about a mouse that double-clicks on one click tell you the same thing and cite nobody. This page is the opposite: a single, versioned reference for what double-click chatter testing actually shows, backed by an open dataset you can download, check, and cite. Where we have first-hand measurements, they are here with the numbers; where we do not yet, we say so.
The goal is simple — to be a source you can verify rather than trust. Everything below is either computed from the tool’s real code or captured from real browsers, and all of it is in the machine-readable double-click input behavior dataset (v1, JSON).
What chatter is (and what it is not)
“Chatter” is a worn or bouncing switch sending a second press your software never asked for, so one click lands as two. The catch: a deliberately fast double click produces the exact same short interval. A browser test can show you the timing; it cannot, on its own, tell chatter from technique. That is why every number here comes with its conditions, and why one flagged event is never a diagnosis.
The threshold decides what counts as “short”
The review threshold (40 / 80 / 120 ms) is the cutoff below which the gap between two presses is flagged. It is not a pass/fail line. The same clicks flag differently at different thresholds — including ordinary fast clicking flagging at a high cutoff. The full worked examples, computed with the real shortIntervals function, are in how the threshold changes what gets flagged and in the dataset.
How three browsers actually behaved
On one macOS machine we ran the same operations in Chrome 152, Safari 18.6 and Firefox 155. The results were consistent across engines:
| Check | Chrome 152 | Safari 18.6 | Firefox 155 |
|---|---|---|---|
| Double-click, 6 deliberate clicks @ 80 ms | 0 / 5 short | 0 / 5 short | 0 / 5 short |
| Scroll units (6 down events) | px: 1392 | px: 1458 | px: 1308 |
| Interrupt: window blur | “Window lost focus” — identical | ||
The full cross-browser and interruption data is in the same input in three browsers and why your test got interrupted. The honest headline: on this setup the engines agreed on units and codes — the differences people expect (for example a notched wheel reporting line units in Firefox) were not testable on a trackpad and are flagged below as pending.
The open dataset
Everything above is published as one file so it can be cited and re-checked: the double-click input behavior dataset (v1) carries the method version, the exact inputs and results, the environment, the reproduction steps, and an explicit list of what has not been captured yet. It is released under CC BY 4.0 so you can quote it with attribution.
Not yet captured (stated honestly)
- Notched-wheel (non-trackpad) scroll units per engine — to confirm whether any engine reports
lineunits instead of pixels. - Windows and Linux runs of the same operations.
- Additional browsers (Edge, Opera) and OS versions.
Version history
- v1.2.2 (2026-09-14) — removed tab-switch from the verified cross-engine results (no independent raw line) and listed it as not-independently-verified; corrected a dataset field reference.
- v1.2.1 (2026-09-14) — interruption evidence made precise: separated on-screen status from the saved report line, and noted tab-switch resolves to the same window-blur record.
- v1.2.0 (2026-09-14) — completed the raw reports (Safari/Firefox double-click and per-browser interruption lines with timestamps) so every summary row is traceable; added pinned snapshots.
- v1.1.0 (2026-09-14) — added cross-browser keyboard codes and verbatim raw reports for auditability; corrected the scroll wording to stop attributing the magnitude difference to the engine.
- v1.0.0 (2026-09-14) — initial: threshold sensitivity, cross-browser double-click, scroll units, interruption behavior.
Method browser-v1.0. The latest v1.x URL always serves the newest revision; for a citation that will not change under you, cite the immutable pinned snapshot v1.2.2. The full changelog is inside the dataset file.
First-hand, reproducible, and explicit about its limits. These are browser-observed events; they document how engines report input in the tested sessions and do not certify any physical device. Run your own in the double-click check.