Several studies have analysed postoperative or spontaneous variations
in astigmatism. The methods of quantification of these variations, fre
quently used in these studies, are based on different ways of calculat
ing:subtraction, vectorial or polar methodes : Naylor, Jaffe and Claym
an, Cravy, Naeser, Russell et al. and recently, Holladay et al. Althou
gh these methods have the advantage of taking into account the axial c
omponent of astigmatism and are useful for a single patient, they have
many problems and they are more and more numerous (five in the last t
hree years). We have computerized these formulae on a personal compute
r and studied their performances when the cylinder power decreases wit
hout modification of the axes, when the axes are modified without modi
fication of the cylinder power and when the two axes change with the s
ame angular difference. These calculations demonstrate that these form
ulae are not linear and introduce statistical errors before the statis
tical decision tests. Moreover, their application is only possible for
evaluation of astigmatism axes between 0(0) to 180(0) and not for sem
i meridians and corneal topography.