If the visual mechanisms underlying form perception scale similarly with eccentricity, then performance in regions of different eccentricity should be characterised by a single function of the form f( s/ e) where s is a spatial variable, like size or spatial frequency, and e is a parameter that represents the local scale at that eccentricity. This formulation implies that performance at a given size s0 and eccentricity should be identical to that at a size s1= s0 e1/ e0, at a different eccentricity, where e0 and e1 are the local scale parameters for the two eccentricities. We refer to this as the equivalent-size hypothesis. We tested the equivalent-size hypothesis by measuring contrast thresholds for detection and identification of four mirror symmetric letters (b, p, d, q) for a series of sizes at each of three eccentricities (2, 4, and 8 deg). Psychometric functions were obtained for each size and eccentricity with the use of a spatial, 4-alternative forced-choice, double-judgment technique. First, observers specified at which of four positions around the fovea the stimulus appeared. Then they responded with the letter name. At each eccentricity, contrast thresholds for detection and identification as a function of size were described well by a power function. A single power function scaled for eccentricity was able to account for either the detection or the identification behaviour, and a single scaling parameter for the two tasks would not suffice.