My bad, the page layout was confusing. Page 13, last paragraph, before formula (6) (which looked like a page number instead).
Ok, now I see this funny formula. And I'm still sure that if I have a million kangaroos, best average jump is about sqrt(range) instead of 250000*sqrt(range).
You can confirm it easily even with JLP software which uses millions of kangaroos but correctly calculates jumps table based on the range only, ignoring the number of kangaroos. Try to multiply jumps distances by a million and you will get worse results.
Why? Probably because when we have a lot of kangaroos they start using birthday paradox and this formula doesn't work anymore.