An ACT-R model was developed for the first experiment of Johnston, Hawley, and Elliott (1991). Subjects in their experiment studied 96 4 or 5 letter words at the rate of 1 word per 2 seconds. The first and last 4 words were buffers but the middle 88 were critical. Subjects were then tested with 206 words which consisted of a buffer of 30 words followed by the 88 critical words mixed in with 88 foils. The words were presented for recognition camouflaged in dots which either disappeared at a slow rate or fast rate. The subjects were to read the word as fast as they could. In the ACT-R simulation the reading of the word was governed by the Read-Word production above. The actual timing consisted of three components. There was the time to encode to the letters (estimated at the default of 200 msec. during study, 750 msec. during test in the fast uncovering condition, and 950 msec. during test in the slow uncovering condition), the time to retrieve the word in the Read-Word production, and the time to say the word (given a standard estimate of 500 msec.). After reading the word, ACT-R recognized the word using the same productions as used in the Recognition Memory Section 7.2 to model the length-strength effect. ACT-R simulated 32 subjects, the same number as in the Johnston et al. experiment. The non-default parameter settings were the encoding times (set longer than usual to reflect the difficult encoding conditions), the latency scale parameter F set at 1.30 sec., the activation noise parameter s set at .65, and the activation threshold parameter t set at 0.90.