# Decision Record ## Decisions - Treat the glass harp as a set of individual friction idiophones rather than a single fabricated structure. - Keep pitch and tuning claims out of the first packet until real glasses are measured. - Use water level as a future measured variable, not as a fabricated design table in this pass. - Record every artifact as `concept_only` or `pending_measurement`. - Do not create CAD, DXF, glass dimensions, fill-height tables, pitch maps, or tuning values in this pass. - Uplift readiness to `L2 V5 build-packet candidate` because the packet now separates glass inventory, water-fill workflow, tray/support planning, BOM, cut-list, and repeatability gates. - Add `cut-list.csv` as estimated tray, label, log, and storage families only; all sizes remain `TBD`. ## Alternatives Considered - **CAD-first tray design:** rejected because glass count, glass spacing, damping, and spill behavior are unknown. - **Pitch-map-first packet:** rejected until real dry and filled glass responses are measured. - **Permanent fill marks:** deferred until a repeatable water-level workflow is proven. - **Uniform glass set only:** kept as a possible branch, but not assumed; found or mixed glasses may be musically useful after measurement. - **Publishing estimated water volumes:** rejected; all fill values remain pending measurement. ## Open Questions - Which real glasses will define the first measured set? - How will each glass be labeled without interfering with play or cleanup? - What water-level recording method is repeatable enough for retuning? - What table, tray, or fixture keeps glasses stable while preserving resonance? - How will pitch drift be tracked during a playing session? - What safety protocol handles cracks, chips, and breakage? - What tray or pad material preserves sustain while preventing glass movement? - How many drain/refill cycles are needed before a fill record is trusted? - How should dry response and filled response be compared without implying a final scale? ## Promotion Gates - L2 now provides the review scaffold; promotion beyond L2 requires a selected glass set, measured dry and filled response logs, and a repeatable fill-recording method. - L3 requires reviewed setup authority and repeatable tuning/playability tests. - L4 requires empirical session data, retuning notes, and documented revisions.