At 12:43 PM 2/11/2003, Edwin Clark wrote:

Do you still have the sequence up to n = 5 million? If so I would
be interested in knowing whether or not the frequency of i such
that u(i) = u(i-1) + 2 continues to hold up. Gaps between successive
such i are quite small in my limited data (up to n = 2000).


No I don't still have the data but I do have the program that did the test, and now I have a faster computer (although it gets slower and slower as n increases).  I think the frequency you mention holds up, because I seen to remember that the density holds pretty constant.  I'll dust off my old program and let you know.