Abstract
Szilard pointed out in 1925 that it is possible to base the foundations of statistical thermodynamics upon ``phenomenological'' principles, analogous to those of the non‐statistical ``classical'' theory. This approach is discussed and developed.