[seqfan] Definition of A045916, pi(n) and the log-integral approximation

[Richard Mathar]

In case someone has access to Clawson's "Mathematical Mysteries" cited in A045916:

It would be nice to have an explanation put into A045916 why the numbers
differ much from A057752. The difference is not just a small effect from
changing the lower limit as seen in A106313 or using floor(), round, ceil()
in some mix. These explain only differences of the order +-1.
Should the formula in A045916 be kept or be changed to li(10^n)-pi(10^n)?

http://oeis.org/classic/?q=id%3AA045916|id%3AA057752|id%3AA106313