Is there a minimum polarizability principle in chemical reactions?

For complete fragmentations of the type A(m)B(n...) --> mA + nB + ... the change of the dipole polarizabilty Delta alpha = Sigma(i)nu(i)alpha(i) and its cube-root Delta alpha(CR) = Sigma(i)nu(i)alpha(i)(1/3) in as well as the atomization energy D-at are calculated from literature data (nu(i) is the stoichiometric coefficient). We have taken into account a large number of molecules containing the atoms H, C, N, O, S, P, F, Cl, Br, and I as well as the metals Fe and Os. The ranges of D-at and Sigma(i)nu(i)alpha(i) covered by the fragmentations are between 150 and 15 000 kJ mol(-1) and -6 and 170 x 10(-41) C-2 m(2) J(-1), respectively. In most cases Delta alpha > 0 is observed, whereas we always find Delta alpha(CR) > 0. Additionally, we observe a linear relationship D-at = A(mu) + B mu Sigma(i)nu(i)alpha(i)(u) between the atomization energy D-at and the sum of the dipole-polarizabilities of all chemical species taking part in the fragmentation. The linear relation is obtained for mu = 1 and 1/3. Our observation implies that the most stable isomer has the lowest polarizability and that in chemical reactions the most stable species (reactants or products) have the lowest sum of alpha(1/3).