01
Record the two visible conditions precisely
Create one row for each supplied capture. At 390 pixels, record submit button covered by fixed panel. At 1280 pixels, record submit button visible. Attach A and B to their respective rows and retain version V7. Do not label the desktop form fully working merely because its button is visible. Likewise, the mobile observation identifies a visible obstruction without establishing every possible interaction outcome. The useful first record describes what the supplied images show, leaving the behaviour field unresolved until an appropriate interaction check occurs.
02
Leave the missing width and click outcomes unresolved
Add a 768-pixel row marked not observed. Do not interpolate a pass from the wider capture or assume the narrow obstruction persists unchanged. Layout rules can change between widths. For all three rows, keep submission outcome separate from button visibility. A proposed follow-up is to inspect the missing width and exercise the form in a controlled review environment. That is planned work, not completed evidence. If a later check produces a result, record its version, width and action rather than silently filling the old row with a new observation.
03
Use a coverage card with independent fields
The reusable card records page, release version, viewport, capture reference, visible condition, interaction performed, observed result and next check. It can report two supplied captures out of three requested widths while still showing zero supplied interaction results. Those counts describe the review pack, not a defect rate or a probability of failure. If the fixed panel is changed in V8, create new evidence for that version. Preserve V7 so reviewers can trace the concern without confusing an earlier observation with the current interface.