Never double the leading tone

The leading tone is never doubled. It is the scale’s most determined tone — a half step below the tonic with one place to go — and two voices carrying it would owe the same resolution: satisfy both and there are parallel octaves, deny one and a leading tone hangs unresolved. The same logic protects every strong tendency tone: the chordal seventh, and any chromatically altered tone, are likewise left single.

The ban is really a diagnostic: whenever a progression seems to force parallel octaves at a cadence, the cause is almost always a doubled leading tone two chords earlier. It generalizes cleanly — before doubling any chromatic or tendency tone, ask what both copies would have to do next — and it explains the standard voicings: the dominant chord doubles its root or fifth precisely because its third is untouchable.

• Doubling the leading tone in the dominant chord and discovering the parallel octaves only at the resolution.
• Doubling it accidentally between an inner voice and a figurated line that touches the same degree.
• Extending a sequence’s doubling pattern onto a chord whose third is the leading tone.
• Doubling other tendency tones — the chordal seventh, a chromatic alteration — as if only the leading tone were protected.

• Ebenezer Prout, Harmony: Its Theory and Practice (1889), The opening chapters on harmony in four parts — The doubling tables’ firmest line: the leading note is not doubled.
• Luigi Cherubini, Cours de contrepoint et de fugue (1835), The rules for counterpoint in four parts — The strict-style form of the same ban.