It is commonly assumed that orthographical lexical access in visual word recognition takes place in parallel, with all letters activated at the same time. In contrast, in the SERIOL model of letter-position encoding, letters fire sequentially (Whitney, 2001). I present further support for such seriality on several fronts. (1) The reasons that led to the rejection of serial encoding are shown to be invalid, and the virtues of a serial encoding are discussed. (2) The SERIOL model's serial mechanisms provide a natural account of counterintuitive letter-perceptibility patterns, and correctly predict the temporal evolution of these patterns. (3) Via simulations, I show that serial lexical access accounts for conflicting data which indicate both a presence and absence of positional effects at the lexical level. In contrast, the experimental results at the letter and word levels are difficult to explain under the assumption of parallel lexical access. (Contains 2 tables, 10 figures and 4 footnotes.)