This clearly written paper is addressed to experimenters in search of a piece of Pascal code. It also covers a few relatively well-known aspects of the theory underlying multiplicative linear congruential generators. Most of the interesting material is on how to implement the generators. Methods are described and their efficiencies are evaluated experimentally. The authors propose Pascal programs for long-period generators for 32-bit machines.
The authors deal attentively with the possibilities of jumping ahead or back to new random substreams; I wonder whether this discussion is fully justified. They indicate that such possibilities avoid some programming burden, but this argument seems weak. More attention could have been paid to generating random numbers based on a given arbitrary seed and to function-like structures of the form random:=function(seed), where the seed can be any positive integer.