Previous studies of the dynamics of hydrated proteins had shown a feature resembling an exceptionally broad glass-softening endotherm. Its onset temperature, denoted as T(g), was indefinable in one calorimetric study of hydrated lysozyme and was in the 148-218 K range in another study, depending upon hydration. Other methods reported this T(g) as ~170 K. We argue that glass-formation of biomaterials should be studied by measuring a property on both the cooling and heating paths and it should be ascertained (i) that there is thermal hysteresis of the measured property, (ii) that the real and imaginary components of a dynamic property obey the Kramers-Kronig relations, and (iii) that there is an effect of annealing that is consistent with the glass phenomenology. We report the real and imaginary components of the dynamic specific heat, C(p)' and C(p)", of dry and two hydrated lysozyme samples on the cooling and the heating paths as well as the effects of annealing and changing the frequency. For the most hydrated (34.6 g water per 100 g lysozyme) sample, C(p,app) does not show thermal hysteresis in the 160-230 K range, C(p)' varies in a sigmoid-shape manner with T while C(p)" remains close to zero, and there is no effect of annealing. We interpret these findings in terms of continuous development of ice-like aggregates of immobile H2O as more H-bonds form on cooling, and continuous deterioration of the aggregates on heating. As the equilibrium constant between the aggregates and mobile H2O increases on cooling, configurational degrees of freedom of H2O molecules and lysozyme segments decrease. Consequently, the net change in enthalpy is small but the change in C(p) is large. Mobility of the lysozyme segments still depends upon the mobility of H2O molecules.