In a recent article concerning the calculation of molecular polarizabilities [ 11, steps towards the generalisation of the geometric approximation (GA) were indicated although no numerical studies were presented. To complement the previous investigation, we report here some polarkability calculations which employ one such generalisation. We use notation and methods as described in Ref. [l].To recap briefly, the polarkability computed by the GA a, =Ape;can be considered as a variational procedure which employs one scaling parameter A, [l, 21. However, the restriction to a single parameter is not necessary and one can envisage a family of such variational parameters. To take a specific example, appropriate to conjugated systems, in the functional J'", defined by Hirschfelder et al.[3]
5'2) = (+(o)~-E(z)~+(o))+(+(l)~H(l)-E(l)~+(o))
+(+(O)lH(l) -, ( 1 ) 1 + ( 1 ) ) + (+(l)lH(o) -E(o)l+(l)) (2) +(+(2)IH(O) -E(0)1+(0)) +($,(O)lH(o) -E(o)l+(z))one can choose to employ trial functions of the formThe spin orbitals i, j and a, b denote, respectively, m and (r orbitals; rn, n are used to denote both.