It is hypothesized that metabolic and mechanical changes in human loco
motion associated with changes in speed v are constrained by two attra
ctive strategies: Q(metab) = 1 and Delta Q/(metab)/Delta v = a positiv
e definite constant. Q(metab) = Delta E(k) s(-1)/ml O-2 s(-1) where De
lta E(k) s(-1) is the summed increments and decrements per unit time i
n the translational and rotational kinetic energies of the body's segm
ents and ml O-2 s(-1) is the rate at which chemical energy is dissipat
ed. The expected constancy of Delta Q(metab)/Delta v was derived from
an extension of Ehrenfest's adiabatic hypothesis by which transformati
ons (increases, decreases) in locomotion v can be considered as adiaba
tic, even though the biological conditions are nonconservative and non
-rate-limited. The expected significance of Q(metab) = 1 was derived f
rom stability considerations of the symmetry per stride of stored and
dissipated energy. An experimental evaluation was provided by collecti
ng metabolic and mechanical measures on walking (10 subjects) and runn
ing (9 subjects) at progressively greater treadmill speeds but within
the aerobic limit. Results revealed that walking was restricted to Q(m
etab) less than or equal to 1, with a nonlinear trajectory in v x Q(me
tab) coordinates shaped by Q(metab) = 1 (primarily) and the constancy
of Delta Q(metab)/Delta v. Running satisfied Q(metab) > 1, with a line
ar trajectory in v x Q(metab) coordinates conforming to Delta Q(metab)
/Delta v = a constant, with the constant predicted from invariants in
the mechanical space v x Delta E(k) s(-1). Results also suggested that
the metabolic costs of running might be predictable from measures mad
e in the v x Delta E(k) s(-1) space.