Abstract

Negative temperature is based on the Kelvin scale and the condition dU>0 and dS<0. Conversely, there is also negative temperature for dU<0 and dS>0. When the derivation of negative temperature is examined, it necessarily requires a decrease in entropy. The concept of negative temperature contradicts the usual meaning of "temperature", as well as contradicting some basic concepts of physics and mathematics. Negative temperatures can be shown to occur only in a system which is not in thermodynamic equilibrium. It is here proposed that the decrease of entropy is possible due to magnified fluctuations and internal interactions in some isolated systems. Further, a necessary and sufficient condition for the decrease of entropy is discussed quantitatively. In addition, some possible tests for the decrease of entropy in isolated systems are proposed. They should be confirmed by many stable states in nature.