In my opinion, the article is

*definitely* too sensational.

The existence and the nature of negative temperatures - in the

*statistical mechanics* sense - is well known.

There is a reasonably extensive

wiki article on the subject, the subject was treated theoretically as far back as

1951 (at least), and is discussed in Kittel and Kroemer's classic

introductory text book on thermal physics.

This may be the first

*experimental* realization of negative temperatures though.

diablo_04, on 07 January 2013 - 01:57 PM, said:

I believe they measure with equations, In science lot of things can be measured with equations.

Actually you are sort of correct (despite what other posters have said). The authors of this paper are using the

*statistical physics* definition of temperature, and it is - to some extent - reasonable to claim that the ``negative temperatures'' measured by the authors is just a mathematical trick.

You can see in the wiki article I linked to above, that if the total energy of the simple model system described therein is positive (i.e. more than half of the particles are in the excited state) the

*statistical* temperature of the system will be negative.

But there can be differences between the

*statistical* temperatures of carefully controlled ensembles and the

*thermodynamic* temperatures of macroscopic objects.

After all, temperature is a property of macroscopic objects which determines the direction heat energy will flow. Can temperature be properly defined for only 100 000 atoms? Or even more importantly, is it accurate to define an ensemble of 100 000 atoms that start at extremely low temperatures with

Maxwell-Boltzmann statistics?

The original scientific paper published in Science is

here, and a draft of the paper is freely available on arXiv

here. Skimming the paper, it seems (to me, anyway) what the authors have actually

*done* is achieved a

population inversion in kinetic degrees of freedom.

The authors describe the statistics of the ensemble with the Maxwell-Boltzmann distribution, and derive the

*statistical* temperature from that; and the result is a negative temperature.

This is definitely a legitimate description of the system, but don't be confused between

*statistical* and

*thermodynamic* temperatures. The

*thermodynamic* temperature of their ensemble would still be very cold.

Mr Right Wing, on 08 January 2013 - 12:21 AM, said:

The atoms they were experimenting with had an average temperature just above zero. Average means some are less some are more. This means some drop below zero.

No, it depends on the type of statistics used. In Gaussian statistics, yes. In Poisson statistics (i.e. counting statistics), nothing can drop below zero.