The speed of reading unrelated words as a function of luminance, size, and contrast, was measured with an eye movement monitor for fifteen young adults. Subjects read up to 5,000 words in a test session, with the exact number depending upon their acuity. The size of the smallest legible print at a given luminance and contrast for these subjects was found to fit well to the Blackwell-Taylor detection threshold data above about 1 minute of arc. At lower sizes inclusion of a resolution size term provided an excellent fit.

Reading speed was fit to a number of visual performance models. It was found that for most subjects that a ratio of the print size to an estimate of the threshold print size (a VLsize) gave the best fits to the data. The threshold size was computed with a fit to the Blackwell-Taylor detection threshold data, modified to include a resolution size term as above. For the sole remaining subject a slightly better fit was obtained with a VLcontrast model, where again the thresholds were modified by a limiting size term. The implication of these results for visual performance modeling is discussed.

The reading speed for all subjects varied rapidly with size near the acuity limit, but became almost independent of visibility parameters as long as size is two times the acuity limit. These results show that size is a powerful determinant of reading speed, and suggest that minification of about 1/2 power could be used as a field test for adequate visibility.