Actually I'm doing an online course and in the Address Resolution Protocol video lecture, this statement is made in the following context.
And here's what the instructor says:
It needs to because of the concept of a netmask. For example, look at this hypothetical setup. The gateway, in the middle, has a single IP address: 192.168.0.1. It has two network cards, one connecting it to the destination 184.108.40.206, one connecting it to the source, 192.168.0.5. The address 192.168.0.1 can really only be in one of these networks, the source network. The netmask needed for 192.168.0.1 to be in the same network as 220.127.116.11 is 18.104.22.168, or just one bit of netmask! But it can’t be that all IP addresses whose first bit is 1 are in the same network as 22.214.171.124 -- 192.168.0.5, for example, needs to be reached through the gateway. So instead we often see setups like this, where the gateway or router has multiple interfaces, each with their own link layer address to identify the card, and also each with their own network layer address to identify the host within the network that card is part of. For the gateway, the left interface has IP address 192.168.0.1, while the right interface has IP address 126.96.36.199.
I specially don't understand the bold statements. Why can't 192.168.0.1 be in both networks?