Abstract: In a large time-space-time division switching network, particularly one employing a multistage space-division switch, one time-division elementary time is too brief to enable transmission of addresses of crosspoints to be completed through all the stages of the multistage switch, and also provide for transmission of the speech sample. To cope with this situation, two successive network elementary times are used, each elementary time being divied into two half-times, i.e., a first half-time and a second half-time. First half-times are used for transmitting addresses of crosspoints to be closed immediately. Second half-times are used for storing data concerning further crosspoint addresses plus a speech sample. Addresses are propagated through a speech sample path according to a "staggered operation", thereby reducing the requirements for memory dedicated to crosspoint addresses.