Historically, which network topology does classic Ethernet use at the logical level?

Introduction / Context:
Ethernet is one of the most widely used LAN technologies. Although modern Ethernet networks are often built with switches in a star shaped wiring layout, the original Ethernet design and its logical operation are based on a bus topology. Understanding the difference between physical cabling layouts and logical topologies helps clarify how access methods like CSMA/CD work and how collisions occur.

Given Data / Assumptions:

We are considering classic shared medium Ethernet, such as 10BASE5 and 10BASE2, and the logical behavior of early hubs and repeaters.
All stations on the Ethernet share the same collision domain and can hear each other's transmissions.
The question is focused on the topology used by Ethernet, not on the physical type of cable such as twisted pair or coaxial cable.

Concept / Approach:
In a bus topology, all nodes are attached to a single shared communication medium. Any station that transmits places signals on this common bus, and all other stations see those signals. Classic Ethernet with coaxial cable was implemented exactly this way: hosts tapped into a single coax cable that stretched along the network. Even when hubs were introduced, from the medium's point of view the network still behaved as one shared bus and used CSMA/CD to arbitrate access. Logical Ethernet segments formed one multiaccess bus, irrespective of the wiring pattern in the wiring closet.

Step-by-Step Solution:
Step 1: Recall that classic Ethernet uses CSMA/CD (carrier sense multiple access with collision detection), which assumes a shared medium that all stations can sense. Step 2: In early Ethernet implementations, multiple devices were connected to a single coaxial cable. When one host transmitted, the signal propagated along the cable to every other host. Step 3: This shared cable is the defining characteristic of a bus topology, where every node attaches to the same backbone. Step 4: Ring topologies, such as those used in token ring networks, and mesh topologies with many direct links, do not match this behavior. Step 5: Modern switched Ethernet often uses a physical star, but each switch port forms a separate point to point link, and logically the shared bus is replaced by many links, which is different from classic shared bus Ethernet.

Verification / Alternative check:
Diagrams in classic Ethernet standards and textbooks show stations attached to a single coaxial bus. Collisions occur because signals overlap on this bus. Hubs, when used, simply extend the collision domain so that from a logical perspective the medium remains a shared bus. These descriptions confirm that historically Ethernet's logical topology is a bus.

Why Other Options Are Wrong:
A pure star topology with switching is more accurate for modern full duplex switched Ethernet, but that is not how classic shared medium Ethernet operated logically.
Ring topology with tokens describes token ring or FDDI networks, not Ethernet.
Full mesh with direct links between every pair of hosts is rare and not associated with Ethernet LAN design.

Common Pitfalls:
Learners often see star shaped Ethernet cabling in offices and assume Ethernet is inherently a star topology. It is important to distinguish physical layout from logical topology. Historically, Ethernet started as a bus, and even with hubs, it behaved logically as a bus because all nodes shared one collision domain.

Final Answer:
Classic Ethernet uses a bus topology, in which all stations share a common transmission medium and form one collision domain.