In known optical distribution systems, a signal intended for the subscribers is distributed from a transmitting point to all network terminations (point-to-multipoint transmission). The receivers, which are connected to a passive optical network, are adapted to a common bit rate. If a need for a higher data rate arises at a network termination, this need can only be satisfied in the prior art by converting all receivers. This is not possible without interrupting the service. Furthermore, the conversion entails great expense, since the receivers of these network terminations where the need for information is unchanged have to be converted as well. By a time-division-multiplexing method, a time-division multiplex signal is generated which has a frame whose duration is equal to one bit period (T) of a digital signal, and which is divided into k time slots (ZS). At least two time slots (ZS) are used for one digital signal, and one respective time slot (ZS) is used for each of the remaining digital signals.