Q & A on Mixing wiring schemes Q: I'm a college student from the Zhejiang University, Hangzhou China. I use your book "Computer Networks and Internets" for self-teaching ,for I am very fond of computer and the allied subject.I have finished the first half of the book-- Computer Networks. I think your book was very well written ,it's brief and clear for the beginners. However, in the course of my studying , I still have some problem that could not be solved by myself. first, in your book ,on page 94,the last fifth and forth line, you mentioned that "for example ,it is possible to have some computers connected to a network using Thicknet, while others are connected to the same network using Thinnet." However, in your later summary on page 99. Beginning from line 8 you summarized that "Although a given networks technology can support more than one wiring scheme ,all computers attached to a particular network must use the same scheme. " I just wonder whether I misunderstood your book,or there is actually a contradiction .Is it the difference between theory and reality? A: You are quite observant. In general, one cannot mix wiring schemes on a single network. There are two minor exceptions that were invented as ways to migrate from older wiring to a newer scheme. First, because Thick and Thin Ethernet use the same electrical signals on the cable, it is possible to join a (short) segment of thick cable to a (short) segment of thin cable, making part of the network use thick wiring and part use thin. Of course, the length restriction still applies to the overall network. Second, because all types of Ethernet use the same frame format and addressing scheme, it is possible to obtain a repeater (level-1 interconnect) that connects between a newer 10base-T hub and an older wiring scheme. Second, for an Ethernet with CSMA/CD, how to make access for each computer attached to the network fair? Can you give me some details which you neglected in your book? The fairness is probabilistic: When a collision occurs, each station that was involved in the collision chooses a random delay, d, before retrying. The station that chooses the smallest delay will proceed to transmit. Assuming the random number selection is fair, all stations have an equal probability of being allowed to proceed.