Whole notes have no stem, so voice identity cannot be read from stem direction alone

Whole notes are never stemmed. On a shared stave where stem direction is normally the reader's voice cue, a whole note removes that cue — engravers compensate with horizontal notehead offset rather than any stem-based device.

A whole note in one of two shared-stave voices rendered at the exact same x-position as a simultaneous note in the other voice, with no offset signaling two independent voices.

Gould, Behind Bars, Stems chapter (whole-note exception is universal and uncontested across Read, Ross, Stone); the horizontal-offset compensation is standard multi-voice practice (Gould, voice-pairing section).

This rule needs the notated content and the rendered result together — the score says what should happen, the page says whether it did.