The behavmur of the threshold anomaly for non-central potentials, wh,ch account for collectwe excltat,ons m heavy-~on colhs~ons. ~s investigated. It ~s shown thai the non-central potentmls should exhibit an energ5 dependence at energies m the vicinity of the Coulomb barrier Th~s energy dependence is, however, different from that of the elastkc optical potenual, occurring at lower energies It ff further shown thai there are correctmns to the tradmonal collectwc model such that. ff the trans~tmn potential ~s expressed as the derivative of the optical potential, the corresponding deformation length will be complex and energy-dependent Simple model calculauons are presented Durmg recent years, evidence of the energy-dependence of the real (the threshold anomaly) and imaginary parts of nucleus-nucleus optical potentials has been found by careful analyses of the elastxc scattering at energies m the vtcmlty of the Coulomb barrter [1][2][3][4][5] This energy dependence has been attributed to the couphng of non-clasttc channels to the elastic channel tn this energy region [6.7]. The threshold anomaly has also been related to the rapid mcrease of the surface imaginary potential as the energy ts mcreased above the Coulomb barrier and the consequent correction to the real potential through a dispersion relation [8][9][10]. The optical potential can thus be written aswhere go is an energy-independent potential (double-folded potential, for example), I,I'(E) ts the tmagmary potential and the dispersive real potenual AI'(E) ts defined aswhere P denotes a prmclpal value mtegral. Generahzlng this to the case of a set of strong-coupled channels, it was suggested by Satchler [ 1 1 ] that Permanent address SERC Daresbury Laborator3., Daresburs, Warrmgton WA4 4AD, UK each element of the potential matrix, which defines the coupled channels equation, should be expected to exhtbtt a threshold anomaly. In such a case, the energy dependence of the different diagonal and off-diagonal elements of the potenttal matrix will be attributed to couphngs to other channels not mcluded m the subset of coupled channels. However. a priori, there ts no reason for all the terms of the potential matrix to have the same energy dependence. Satchler [ 1 1 ] argued that, m the case of collective excitations, one would expect a similar energy dependence m the coupling mteractlon (transition potential) and the elastic optical potenttal, though one could expect a shift m the threshold of the coupling potential by approxtmatel,, half of the excitation energy. He further concludes that tfone uses an optical potential that ts consistent with the dispersion relatton but still needs an energy-dependent deformation length, then either the optical model itself or the nuclear structure model. or both, are madequateThe excitation functions of the first excited 3 -state of 2°8Pb by ~60 at energies around the Coulomb barrter have been measured and analysed [ 12,13] and more recently detailed angular dtstrtbutions for the same system were measured at energies from above t...