Abstract
The Landau-Zener formula for the transition probability at a crossing of potential energy curves is discussed. It is pointed out that the model used in the derivation has defects which make the range of validity of the formula much more restricted than is commonly supposed.