The probability method of analysis of astigmatism variations respects
both astigmatism components, quantity and axis. The two random variabl
e model that we introduce here analyses cylinder variations and axial
variations. After describing the contingency table and the probability
table, Expectation of each component and its confidence interval are
calculated together with the variance and the standard deviation. Cova
riance and correlation are calculated. If we want to compare different
groups we test cylinders with cylinders and axes with axes. This meth
od does not distort reality and is applicable to the semi meridians fr
om 0(0) to 360(0). Only large samples of patients can be studied by th
is method.