The Drell-Hearn-Gerasimov and Bjorken sum rules are special examples of dis
persive sum rules for the spin-dependent structure function G(1)(nu, Q(2))
at Q(2) = 0 and infinity. We generalize these sum rules through studying th
e virtual-photon Compton amplitudes S-1(nu, Q(2)) and S-2(nu, Q(2)) At smal
l Q(2), we calculate the Compton amplitudes at leading order in chiral pert
urbation theory; the resulting sum rules will be ableto be tested against d
ata soon available from the Jefferson Laboratory. For Q(2) much greater tha
n Lambda (2)(QCD) the standard twist-expansion for the Compton amplitudes l
eads to the well known deep-inelastic sum rules. Although the situation is
still relatively unclear in a small intermediate-Q(2) window, we argue that
chiral perturbation theory and the twist-expansion alone already provide s
trong constraints on the Q(2)-evolution of the G(1)(nu, Q(2)) sum rule from
Q(2) = 0 to infinity.