On Dec 4, 2005, at 8:54 AM, Stephen Waits wrote:

> Numbers will certainly repeat within the period, since the period  
> is so much larger than our integer representation.  For example, a  
> 32 bit integer can only represent about 4e9.  So, the likelihood of  
> getting any specific integer, in the 32 bit case, is 1/(2^32).

I should add to this that you can approximate when you'll get  
repeats, and it's likely to be much sooner than you'd guess.  For  
more information, see the [Birthday Problem][1].

--Steve

[1]: http://mathworld.wolfram.com/BirthdayProblem.html