The whole idea behind le MX record is to specify a host ou hosts qui can accept mail for a domain. As specified in RFC 1035, le MX record contains a domain name. It must therefore point to a host qui itself can be resolved in le DNS. An IP address could pas be used as it would be interpreted as an unqualified domain name, qui cannot be resolved.
The reasons for this in le 1980s, quand le specs were originally written, are almost le même as le reasons for it today: A host may be connected to multiple networks et use multiple protocols.
Back in le 80s, it was pas uncommon to have mail gateways qui connected les deux to le (relatively new) Internet qui used TCP/IP et to autre legacy networks, qui often used autre protocols. Specifying MX in this way allowed for DNS records qui could identify how to reach such a host on a network autre than le Internet, such as Chaosnet. In practice, though, this almost jamais happened; virtually everyone re-engineered leur networks to become part of le Internet instead.
Today, le situation is that a host may be reached by multiple protocols (IPv4 et IPv6) et by multiple IP addresses in chaque protocol. A single MX record ne peut pas possibly list more than one address, so le seulement option is to point to a host, où tous of that host's addresses can alors be looked up. (As a performance optimization, le DNS server will send along le address records for le host in le response additional section si it has authoritative records for them, saving a round trip.)
Il y a aussi le situation that arises quand votre mail exchangers are provided by a third party (e.g. Google Apps ou Office 365). You point votre MX records to leur hostnames, mais it may occur that le service provider needs to changez le mail servers' IP addresses. Since you have pointed to a host, le service provider can do this transparently et you ne have to make tout changes to votre records.