These days, we take it for granted that it's very simple, and relatively costless, to generate random numbers. Really, these are "pseudo-random numbers", and we should pay careful attention to the quality of the algorithms that are used to generate them, as I've noted previously (and also here) on this blog.
Putting these qualification to one side, however, we've certainly come a long way from having to rely on that famous tome, A Million Random Digits With 100,000 Normal Deviates. Published by the Rand Corporation in 1955, this classic provided (lots of!) random digits that could be used as an aid to simple random sampling, as well random deviates generated from the standard normal distribution. The latter could be used in Monte Carlo simulations, for example.
The following is an extract from a handout that I use in my introductory statistical inference course: