#### dazzlepecs

Registered Senior Member
i was reading that if all 10^25 molecules in a brick just happened to oscillate in the same skywards direction it would jump upwards, but wouldnt jumping upwards reduce entropy which is impossible? But it IS possible (although improbable) for the molecules to randomly align this way?

(if im being painfully laymanish please set me straight)

Local variations in entropy are possible, as the brick is not a closed system.

When you're talking about the entropy of a system, normally you're talking about the entropy when it's in some sort of stable, unchanging equilibrium state. If you're considering a system like a box where all the molecules are moving towards one side of the box, clearly the system is not in equilibrium.

Entropy and equilibrium are related by Carnot's principle. Entropy is all the "energy" in a system from the time the system was created - at absolute zero no system can have any. In the Carnot engine ideally no entropy is lost, but this is physically impossible if you want energy to do something; it has to move around the system and losses occur.

Since all the systems that do have energy have always had some, all systems have entropy, or a "store of energy" which changes. Stat mech uses heat to define this, but heat is only part of the spectrum.