We argue that a substantial increase in the cross section for Coulomb excitation in the region of the Double Giant Dipole Resonance should be expected from Coulomb excitation of excited states involved in the spreading of the one-phonon resonance, in a manifestation of the Brink-Axel phenomenon. This generates an additional fluctuating amplitude and a corresponding new term to be added incoherently to the usual cross-section. The appropriate extension of an applicable reaction calculation is considered in order to estimate this effect. [4][5][6]. The isoscalar double giant quadrupole resonance has also been observed in the proton emission spectrum from the collision of 40 Ca with 40 Ca at a laboratory energy of 44 A Mev [7].When the data on DGDR excitation for 136 Xe and 197 Au are compared with coupled-channel Coulomb excitation calculations [8], it is found that, in the harmonic approximation, the calculated cross sections are a factor of 2 to 3 smaller than the measured ones. A similar discrepancy, albeit somewhat smaller, is found for 208 Pb.Several effects that are not taken into account in the coupled-channel theory have been considered as possible explanations of this discrepancy. As examples, we mention the effect of anharmonicities [9,10] and the quenching of the 1 + DGDR state [11]. Here we will consider a new potentially important mechanism, which consists in the (one phonon) Coulomb excitation of backgrount states responsible for the large spreading width of the one-phonon GDR, as suggested long ago by Brink and Axel [12]. Due to the complicated background of intrinsic states, the amplitude for this process varies rapidly wth energy and possesses an average close to zero. Its contribution to the cross section can be sizable, however. In close analogy to this situation is the the well-known case of nucleon-nucleus elastic scattering. There, the cross section is the sum of the slowly-varying contribution of average optical scattering and of the fluctuating contribution compound nucleus formation and decay. In figure 1 we show a schematic picture of the couplings involved.We first sumarize our main result. The cross-section for Coulomb excitation to the DGDR energy region contains in fact two distinct components which peak at ∼ 2E GDR . However, while the usual component σ DGDR has a width which may be estimated as ∼ 2Γ GDR , the fluctuating Brink-Axel component has a width which is just ∼ Γ GDR . As a result of this, the bump observed in the two-phonon region has an effective width between these limits. The enhancement factor for the peak-value of the cross-section will be given roughly by (1 + Γ ↓ 1 /Γ 1 ), whereas the crosssection integrated over the peak is just about (1 + 1 2 Γ ↓ 1 /Γ 1 )σ DGDR due to the smaller width of the second component. For heavy nuclei Γ ↓ 1 ∼ Γ 1 , and we get enhancement factors of ∼ 2 and ∼ 3/2 for the peak and for the integrated crosssections respectively. Furthermore, these enhancement factors should be reduced and tend to unity as the collision time becomes shorter t...