Statistical analysis of 136 natural tourmaline compositions from the literature reveals the presence and extent of coupled substitutions involving several cations and structural sites. In schorls and dravites these are a dehydroxylation type substitution (1) (OH)−+R2+ = R3++O2− and an alkali-defect type substitution (2) R++R2+ = R3++□, Al3+ being the predominant R3+ action. Substitution (1) which represents solid solution towards a proton-deficient end-member, R+ R 3 3+ R 6 3+ (BO3)3 Si6O18O3(OH), accounts for three times as much of the observed compositional variability as does (2) which represents substitution toward a hypothetical alkali-free end-member, □(R 2 2+ R3+) R 6 3+ (BO3)3Si6O18(OH)4. The occurrence of both of these substituions produces intermediates between end-member schorl/ dravite, R+ R 3 2+ R 6 3+ (BO3)3Si6O18(OH)4, and a new series within the tourmaline group, R 1−x + R 3 3+ R 6 3+ (BO3)3Si6O18O3−x (OH)1+x. In addition to dehydroxylation type, 2(OH)−+Li+ = R3++202−, and possibly alkali-defect type, 2R++Li+ = R3++2□, substitutions, a third type Li++O2− = (OH)−+□, occurs in the elbaites giving rise to Li-poor, proton-rich species. All three substitutions serve to reduce the Li-content of natural elbaite which, as a result, does not attain the composition of the ideal end-member, Na(Li1.5Al1.5)Al6(BO3)3Si6O18(OH)4. Substitution from elbaite and schorl/dravite toward R 1−x + R 3 3+ R 6 3+ (BO3)3Si6O18O3−x(OH)1+x is very extensive and may be complete. Substitution toward R 1−x + R 3 3+ R 6 3+ (BO3)3Si6O18O3−x(OH)1+x results in improved local charge balance. The mean deviation $${\Delta \zeta \left( {\text{O}} \right)}$$ from oxygen charge saturation $$\left( {\zeta \left( {\text{O}} \right) = 2.0} \right)$$ is at a maximum in end-member schorl, dravite and elbaite. Substitutions (1) and (2) progressively decrease $${\Delta \zeta \left( {\text{O}} \right)}$$ but substitution (1) does so more effectively, which may explain its predominance in nature. However, alkali-defective end-members appear to be unstable regardless of $${\Delta \zeta \left( {\text{O}} \right)}$$ . Substitution (3) in the elbaites cannot be discussed on the basis of charge balance considerations at present due to the lack of structural information on proton-rich species.