Spectral distributions (SD) of the light scattered from three high molecular weight polystyrene standards in dilute solution over a wide range of scattering angles (5.7-154 degrees) were precisely determined from a series of carefully executed Light scattering experiments. At x << 1, the SD have only one narrow peak, where x = (qR(g))(2), with q being the scattering vector and R(g) the radius of gyration of the polymer coil. As expected, this peak is purely due to translational diffusion and has a line width Gamma(peak1) = Dq(2), with D being the diffusion coefficient of polymer. In the intermediate range of 1 < x less than or equal to 15, the SD have two peaks. We have identified that the normalized line width Gamma(peak2)/Dq(2)) of the second peak is associated with the internal motions and follows the plot of [1 + 2 Gamma(n)/(Dq(2))] versus x, wherein Gamma(n) is the line width associated to the nth-order internal motion with n = 1 in 1 < x < 3; n = 2 in 3 < x < 6; n 3 in 6 < x < 10; and n = 4 in 10 < x < 15. Thus, we can obtain Gamma(n) from Gamma(peak2)/(Dq(2)) = [1 + 2 Gamma(n)/Dq(2))]. These measured Gamma(n) values agree with the predicted values according to the non-free-draining bead-and-spring model. However, the same model also predicts that, for 3 < x < 15, the contribution from [1 + 2 Gamma(1)/(Dq(2))] Should be dominant in the SD. This contradiction between the theory and observation leads us to believe that some of the energy-favored internal motions are not observable in dynamic LLS at particular x, which is related to the observation length scale (1/q). For x > 15, the two peaks merge into one broad peak. The plot of [Gamma](2)/(q(3)k(B)T/eta(0)) versus x(1/2) of this peak approached a plateau value which is lower than a theoretically predicted value, where ([Gamma](z), k(B), T, and eta(0) are the z-average line width, the Boltzmann constant, the absolute temperature, and the solvent viscosity, respectively.