Not yet reviewed by Pith; the record is open.

We have measured temperature dependence of low-frequency Raman spectra in CuGeO$_3$, and have observed the quasi-elastic scattering in the $(c,c)$ polarization above the spin-Peierls transition temperature. We attribute it to the fluctuations of energy density in the spin system. The magnetic specific heat and an inverse of the magnetic correlation length can be derived from the quasi-elastic scattering. The inverse of the magnetic correlation length is proportional to $(T-T_{SP})^{1/2}$ at high temperatures. We compare the specific heat with a competing-$J$ model. This model cannot explain quantitatively both the specific heat and the magnetic susceptibility with the same parameters. The origin of this discrepancy is discussed.