Measurement of visual acuity with logarithmic charts: comparison between methods of threshold estimation

The variability of the repetition of visual acuity measurements with a log MAR acuity chart with five optotypes in each row of acuity was studied in four groups of subjects. In three groups the threshold was calculated by probability of seeing curves, but the optotypes were read in different orders. For the fourth group the acuity was taken as the smallest row discriminated. Estimating the threshold by a probability of seeing curve did not improve the threshold's stability. Poorer acuity was detected when the optotypes were read vertically. It is suggested that clinical visual acuity charts are not suitable tools for generating data points to be fitted by probability of seeing curves.