So a keyed-columnar combined with route transposition has long been an accepted explanation of how K3 can be solved.  Is it the only way?  No, but it does involve the keyword “KRYPTOS” so it’s a very satisfying solution to the decryption problem.

Here’s a synopsis with pictures from Gary Phillip’s Realm of Twelve site.

Kryptos K3 Solution #2

Now I understand the temptation to leave the “?” with K4.  I wouldn’t say it’s an unassailable argument however.  Basically you have to leave off the “?” to get the grid to be 7×48 and for the whole thing to work.

Once again, the “Q” is in not an artifact of decryption.

It’s fascinating that it’s possible to pull the same message out of a set of letters with very different methods.  I don’t doubt there’s more.  I still wonder every now and then if you could develop an encryption system where you can pull two separate messages out of the same group of ciphertext.  I’m sure it’s possible but I can’t conceive of how to do that right now.

Kryptos K3 solution method confirmed.

Q and ? still intact.

K4 is still 97 letters long and starts with OBKR.