In support of the view that entropy is lost on glass formation, the C'(p) relaxation spectra were regarded as experimental proof of time- and temperature-dependent loss of both ergodicity and entropy, and confirmation of isothermal glass transition. Also, both C(p) -> 0 and S -> 0 in the limits of t(obs) -> 0s, and T -> 0 K were cited as further proof. We found no experimental evidence in support of the view. A critical examination shows that the C'(p) and C ''(p) spectra used to reach these conclusions indicate phase-lag between the response and perturbation, not loss of ergodicity or loss of configurational degrees of freedom, and C(p) -> 0 asT -> 0 K is not evidence for S -> 0 as T -> 0 K.The notions of partial ergodicity and entropy and their dependence on t(obs) are inconsistent with the properties measured during cooling, heating and isothermal annealing, and thermodynamic consequences of the apparent proof are untenable. The premise that glass formation is a process of continuously breaking ergodicity with entropy loss does not merit serious consideration. (C) 2011 Elsevier B.V. All rights reserved.