Thu, 10 Jul 2003 14:37:37 -0700 (PDT) John McCarthy <jmc@steam.Stanford.EDU> Does the following work? Choose the bits of the 27 by 27 array to give the desired picture - whatever it may be. Jiggle the bits systematically so as not to change the picture too much and check the modified numbers for primality. Presumably log n changes will ordinarily yield a prime. Hmm. I may be off the mark here, but from my naive perspective, this "ordinarily yield a prime" seems optimistic for such a sparse and structured picture. There are only 25 "dots" in the picture. If you let each jiggle by about 2 spaces in all directions that gives you around 25 possibilities for each dot. Presumably the density of primes is about 1 in 27^2; so it seems, if there are any primes with such a property, that there is on the order of only one that will look reasonably close to your desired picture. On the other hand, if you had more dots (or even better, a grey-scale photograph), or less structure (so you could move things around more), it seems much easier. I bet it would be easier to take a 15pixelx15pixel 4-bit grey-scale photograph (this is so small it would be an icon) and encode it as a prime number.