Terminal access to computer systems has long been and continues to be a problem of major significance. We foresee an increasing demand for access to data processing and storage facilities from interactive terminals, point-of-sales terminals, real-time monitoring terminals, hand-held personal terminals, etc. What is it that distinguishes this problem from other data communication problems? It is simply that these terminals tend to generate demands at a very low duty cycle and are basically bursty sources of data; in addition, these terminals are often geographically distributed. In the computer-to-computer data transmission case, one often sees high utilization of the communication channels; this is just not the case with terminal traffic. Consequently, the cost of providing a dedicated channel to each terminal is often prohibitive. Instead, one seeks ways to merge the traffic from many terminal sources in a way which allows them to share the capacity of one or a few channels, thereby reducing the total cost. This cost savings comes about for two reasons: first, because of the economies of scale present in the communications tariff structure; and secondly, because of the averaging effect of large populations which permit one to provide a channel whose capacity is approximately equal to the sum of the average demands of the population, rather than equal to the sum of the peak demands (i.e., the law of large numbers). This merging of traffic and sharing of capacity has been accomplished in various ways such as: polling techniques, contention systems, multiplexing, concentrating, etc. Many of these are only weak solutions to the problem of gathering low data rate traffic from sources which are geographically dispersed.