The research reported here comprises an empirical investigation of the phenomenon of typographic allusion. In a preliminary study, subjects rated the perceptual qualities possessed by different typefaces (e.g. heavy—light, fast—slow). The results indicated that subjects agreed as to typeface characteristics and that typefaces were distinguished by such qualities. In Expt 1, subjects undertook a speeded decision task in which they responded according to which one of four adjectives appeared tachistoscopically. Each word appeared in a typeface whose qualities were either consistent or inconsistent with its meaning. Reaction times for inconsistent stimuli were significantly slower than those for consistent trials. In preparation for Expt 2, the same rating procedure was used to elicit subjects' judgements of the attributes of different animals. Subjects agreed as to animals' qualities and such qualities reliably distinguished between the animals. The names of these animals were then used as targets in a binary decision task and each one appeared in a typeface possessing qualities which were either congruent or incongruent with those of the animal. Subjects responded according to whether they considered the animal presented on each trial to be heavy or light, or fast or slow moving. Responses on trials in which the animal and typeface possessed conflicting attributes were significantly slower than responses when animal and typeface qualities were congruent. These results are discussed in relation to current views regarding the processing of written English. We argue that typographic features of words are able to access a semantic code and that this code can interact with the derivation of a linguistic code specifying a word's meaning and/or with post-lexical access decision processes.