Eric J Korpela wrote: I see the page -- but where the results? That is the most interesting part. Measuring the dc bias of the random generator this way (with only 1000 samples) is pretty meaningless. Here goes for some machines I have at hand that aren't MS BASIC varients. All use the generic BASIC version listed on your page, with exceptions noted. Wang BASIC (1st generation 2200T CPU) Time: 2:46 Accuracy: 1.14320000E-05 Random: 13.68, 22.9, 10.0, 9.5, 2.04, 17.9, 15.6, ... Notes: all math is double precision BCD RND is particularly slow on this BASIC Wang BASIC-2 (2nd generation 2200VP CPU) Time: 4.94 seconds Accuracy: .000000076 Random: 2.16, 4.98, 7.22, 14.01, 27.2, 1.16, 21.7, ... Notes: all math is double precision BCD Processor Technology BASIC/5 Time: 3:52 Accuracy: .1021 Random: 30.2, 19.3, 6.5, 17.1, 4.4, 20.7, 2.1, ... Notes: must use RND(1) must use A=A*A instead of A=A^2 Processor Technology Extended Cassette BASIC Time: 3:17 Accuracy: .0379 Random: 9.79, 13.6, 8.0, 34.6, 23.4, 4.28, ... Notes: must use RND(0); A=A*A instead: time = 1:41, accuracy = .0009 IBM 5120 Time: 39 seconds Accuracy: 3.458354E-10 Random: .042, 0.89, 25.3, 11.4, 9.5, ... Notes: must use RND, and use RND(val) to set the seed only one statement per line allowed numerics are double precision binary TI CC40 Time: 5:42 Accuracy: .00000011 Random: 13.0, 11.7, 5.71, 6.9, 2.1, ... Notes: uses double precision numbers with base-100 encoding I have a few other machines I could run this on, but would require some rearranging to get them set up: HP-87, Heath H89 Benton Harbor BASIC, a few varieties of CP/M BASICs, a TI-74 handheld.