On the relationship between microcanonical and canonical Gibbs distributions

It is established that the agreement between the use of microcanonical and canonical Gibbs distributions is possible only if the temperature of a closed equilibrium system and its specific energy in a thermostat are mutually inverse functions. The analysis shows that the derivation of the canonical Gibbs distribution from the microcanoncial does not agree with the physical condition for separating a macroscopic part of a closed equilibrium system.