A cyclic, multi-channel, analog encoding scheme is proposed as a way of representing numbers in optical computers. It can be represented as an n-tuple of complex numbers or phasors. It is similar to residue arithmetic but applies to non-integers. With this scheme the operation of addition is mapped into multiplication. Thus summation of many terms must be carried out by sequential modulation rather than superposition. Though this is a disadvantage for some optical architectures, it seems well suited to certain integrated optic structures and, may be appropriate for certain systolic architectures.