Effects of nanoscale confinement on the distribution of glass transition temperatures (T(g)s) in free-standing polystyrene (PS) films are determined via a multilayer/self-referencing fluorescence method employing a pyrene dye label. Average film T(g)s yield a T-g-confinement effect in agreement with the molecular weight (MW) dependence reported by Forrest, Dalnoki-Veress, and Dutcher. Multi layer films, with one pyrene-labeled layer, reveal that a 14 nm thick free-surface layer in sufficiently thick films(>= similar to 56 nm) exhibits T-g=T-g,(bulk) -similar to 34 K, independent of film thickness and indicative of a strong T-g gradient near a surface. In sufficiently thin films (>= similar to 56 nm), a 14 nm thick free-surface layer reports T-g that decreases with decreasing film thickness and is equal to the T-g of a 14 nm thick middle layer and the average film T-g. Thus, the strongly perturbed T-g at the two surfaces affects T-g several tens of nanometers into and across the film, resulting in greater Tg reductions than observed in supported films. This study also tests de Gennes' "sliding motion mechanism", devised to explain the MW dependence of the T-g-confinement effect in free-standing films. No midlayer chain in a multilayer film forms loops or bridges reaching a surface. de Gennes' mechanism indicates that T-g reductions occur only at locations where segments are present from chains forming loops or bridges at a surface. Major T-g reductions (as large as similar to 54 K below T-g,(bulk)) are observed in midlayers of nanoconfined free-standing PS films, disproving a key premise of the mechanism