[8] It's not always avoided, however; Gradus ad Parnassum (1725), probably the most famous species counterpoint instruction book, includes an example using crossed voices early in the text.
An early example of medieval voice-crossing can be found in what Richard Taruskin (2009, p. 394) calls “English twinsongs.” “These songs, among the earliest polyphonic vernacular settings to survive in any language, employ a more sophisticated sort of voice-leading, through contrary motion and voice crossings.”[16] Further examples of voice crossing can be found in music of the fifteenth century, where “the voices overlap constantly.”:[17] The early seventeenth century, as in this canon by Michael Praetorius: The eighteenth century, as in Bach’s Concerto for Two Violins: And the late nineteenth century , in the finale of Tchaikovsky's 6th Symphony “which begins with a composite melody that is shattered among the whole string section (no single instrumental group plays the tune you actually hear, an amazing, pre-modernist idea)”[18] As Tom Service points out, Tchaikovsky's approach to instrumentation here was indeed prophetic.
Some nineteen years after the première of the "Pathétique" symphony, Arnold Schoenberg was exploring a similar voice crossing technique involving flute, clarinet and violin in "Ein Blasse Wascherin", a movement from his seminal melodrama Pierrot Lunaire (1912).
It leads to ambiguity, as the ear interprets the step from B to C in one voice, and is fairly consistently avoided in contrapuntal writing.
[20] Voice overlaps are common in Bach chorales, but again are discouraged or forbidden by most theory texts.