It has lately been discussed different ways to construct pure random number generators by means of radiactive decay. I must admit that this is a very good way to produce such numbers, but for a number of reasons it is impractical to use such a device. (High radiation levels are needed too produce a significant amount of data.) What I have been thinking of is a way to reduce the amount pure random data needed to produce good random sequences. The way to do this would too have a weak random number generator like: R(n+1)=(17*R(n)+1)mod256, and then inject the pure random into this function giving the result R(n+1)=(17*R(n)+1+PR(t))mod256 or something of the like. Would this constitute a sufficiently secure stream of data for use in one time pads? Has this been thought of before ? THE REST OF THIS MESSAGE CAN BE SAFELY INGNORED: A friend of mine now serving in the air force gave me a report of the cryptographic devices used for communicating on permanent lines. (Twisted pair or something dug down). The machines used till now are huge totaly outdated mostly mechanical monsters. While not transmitting data the machines produce random data to hinder traffic analysys (like how many messages are sent), these data are produced by cookiebox sized thing, and I strongly suspect it to be of a very mechanical nature. Thats how a came up with the idea of a very inexpensive way to produce random data by mechanical means: Use a contaner filled with marbles some made of conducting, and some of non- conducting material. Then mix them in some way or another (this is bound to make a lot of noise) and in the container there should be a few closely placed points of contact that would be able to send a small current through a steel ball if it where to pass over the pair of contacts, and what do we have: Random Data ! (I know this is not very practical at all)