Just for one contribution to get through probably would have taken a couple of minutes, not weeks (see footnote), considering the state of technology back then.
[size=8pt]ARPANET was a packet-switched network (using TCP-IP - - look it up on Wikipedia) just like the "Internets" are today. Same technology, just implemented differently. At best there'd probably be a couple minutes' latency, mostly depending on how fast (or not) your modem was and how busy (or slow) the mainframe or box you were accessing was. Remember, almost all of it was done using the Telnet protocol (look it up on Wikipedia as well. Or, remember those old "character cell" terminals that used to be everywhere a couple generations ago? Think of them) which is nothing but text--text is much, much faster to transmit a screen at a time over a network than the graphics and other junk webpages are cluttered with these days (and all seem to want to load all at once.)
Packet switching has always been faster than, say, UUCP. If something like this were to have been done on the "old" Usenet (before the days of NNTP) it wouldn't have been unusual to encounter latencies of several hours up to a couple days. Although now that packet switching has also been supported in UUCP since about the mid-90s at best, this is kind of a moot point today.