“…These cross-contributions vanish always for the molecular S-model, the only non-zero one being the purely fluctuational contribution I A 1 I2 ) [6,8,11] ] usually taken into account in the spectral theories [15][16][17][18][19][20]. But it should be kept in mind that contributions of the same order as I A 12 > give non-zero cross-terms [7,8,11,13] ( Aa A2 + Ai Ao ) in the S-model. It is noteworthy that, in the molecular S-model, the factor causing the Lorentz-Lorenz function to be density-dependent is precisely the second-order change in molecular polarizability A 2 ) [4,5], whereas the first-approximation variation A 1 ) gives non-zero contributions but in the G-model [5, 23, Above, a(i)t] is the linear polarizability tensor of the i-th isolated molecule.…”