Not quite. Arcnet was not a CSMA/CD but token passing, more in common with Token Ring rather than Ethernet. Arcnet used a bus, but a token was passed along to each node and that node sent a packet only when it got the token. Whenever a new node came online the rest of the nodes would reconfigure and start the token over again. Collisions only occured when new nodes came on line the first time. IIRC it was Datapoint that first came up with Arcnet for their multi-user terminals. For a small number of static nodes the Arcnet throughput was actually quite good. Whereas ethernet is probabilistic for access, token passing is deterministic in that you are guaranteed a slice of bandwidth to every node. At a minimum you always had roughly 1/n of available bandwidth, where n was the number of nodes. Arcnet degraded in a linear fashion as nodes were added, but Ethernet degradation can approach geometric rates, as anyone knows when they see that collision light on the UTP hub go on solid. Also, Arcnet was 2.5Mbps. The original version used coax (not Ethernet thinwire! I think it was RG-92?) into hubs, later versions went to twisted pair. There was a 20Mbps follow-up from SMC (the primary chip maker for Arcnet) but it never caught on. Arcnet was a common networking standard for S-100 systems in the early to mid 80s, using MP/M and CP/NET from Digital Research. I think it was one of the first network transports supported by NetWare too. Jack Peacock