The SERIOL model of orthographic analysis proposed mechanisms for converting visual input into a serial encoding of letter order, which involved hemisphere-specific processing at the retinotopic level. As a test of SERIOL predictions, we conducted a consonant trigram-identification experiment, where the trigrams were briefly presented at various retinal locations. The accuracy data exactly matched the SERIOL predictions. To further test the SERIOL model, we conducted the trigram experiment in the Hebrew alphabet (read from right to left) with native Hebrew speakers. However, the SERIOL predictions were not fully confirmed for Hebrew. Therefore we revised the SERIOL model, resulting in the SERIOL2 model presented here. SERIOL2 re-specifies how the retinotopic representation is transformed into a serial encoding of letter order. We present spiking-neuron simulations to illustrate how SERIOL2 accounts for the trigram data. Parameter values are provided in an appendix. (Contains 2 tables, 14 figures, and 1 footnote.)