Consider, for example, a source data string whose alphabet comprises symbols a0 to am, having probabilities of occurrence equal to p(0) to p(m), respectively. If the source data string is a0a5a3 . . . then the first symbol a0 will be encoded within the sub-interval (0,p(0)). This represents a first subinterval within the original unit interval whose width A1 is equal to p(0) corresponding simply to the probability of occurrence of symbol a0. In order to encode the second symbol a5 of the source data string, its probability of occurrence conditional on the probability of symbol a0 occurring must be determined. Furthermore, the cumulative probability S(5) associated with the second symbol a5 must be calculated. Thus, the sub-interval corresponding to the second symbol a5 is a second sub-interval within the first sub-interval corresponding to a0. Mathematically, the width A2 of the second sub-interval is equal to p(0)*p(5), i.e. the product of the probabilities of occurrence of both symbols a0 and a5. The starting point of the second sub-interval within the unit interval depends on the width A1 of the first sub-interval and the cumulative probability S(5) associated with the second symbol a5, being equal to their product A1*S(5).