It is probably just a coincidence that “Kryptos” is a column transposition key for the rotations used in K3 – as I think Sanborn said. The number codes (0362514) for “Kryptos” happen to have an alternating pattern. To see the pattern, repeat the numbers:
036251403625140362514 : to make it easier to see, cut some the digits off the ends…
62514036251403 and then pull out every other number…
6 5 4 3 2 1 0 and …
2 1 0 6 5 4 3
So when the rotation used a column width divisible by three and then a column width with no odd roots, it apparently injected the “every other” phenomena that could also be obtained if a keyed transposition had been used with “Kryptos”.
This indicates that “Kryptos” was not used in any way for K3. Perhaps it is not used at all in K4 either. That messes with my notion of using the Vigerner table when trying ideas on K4.