The molecular picture of dynamic tube dilation (DTD) was tested for viscoelastic and dielectric data for two series of entangled binary blends of linear cis-polyisoprenes (PI) having 10(-3) M-2 = 308 and 10(-3) M-1 = 94.0 or 21.4. The volume fraction v(2) of the high-M chains was varied from 0.005 to 0.5. The Struglinski-Graessley parameter r(SG) (= (M2MeM1-3)-M-2), specifying the constraint release (CR) contribution to the relaxation of high-M chains at small v(2), was quite different for the two series of blends (r(SG) = 0.0093 and 0.79). For large v(2) (= 0.5), the full-DTD picture treating the relaxed portions of the chains as a simple solvent held satisfactorily in the entire range of time t irrespective of the r(SG) value. However, on a decrease Of v(2), this picture began to fail at intermediate t and further at long t (in the terminal regime of the high-M chains), and the failure was more significant for the blend with small r(SG). This failure occurred when the number beta(CR)(t) Of the entanglement segments equilibrated through the Rouse-CR dynamics was smaller than the equilibration number assumed in the full-DTD picture, beta(f-DTD)(t) = {rho'(t)}(-d) with rho'(t) and d (= 1.3) being the dielectrically determined tube survival fraction and the dilation exponent, respectively. A test was made also for the partial-DTD picture incorporating the maximum number of the segments that can be CR-equilibrated under the given rho'(t) value. The behavior of the blends of various v(2), including the low-v(2) limit, was considerably well described by the partial-DTD picture, although nonnegligible differences still remained between this picture and experiments. Thus, the partial-DTD picture (including the full-DTD picture as an extreme case) was quite tempting, and a brief discussion was made for a possible improvement of existing molecular models on the basis of this molecular picture.