K1 and K2 are straightforward Vigenere ciphers that use the right hand side of the Kryptos Copperplate and the keywords “palimpsest” and “abscissa” respectively.

But what about K3?

Here’s one method and a site that will make it easy to see for yourself.

http://rumkin.com/tools/cipher/skip.php

Take K3:

ENDyaHrOHNLSRHEOCPTEOIBIDYSHNAIA
CHTNREYULDSLLSLLNOHSNOSMRWXMNE
TPRNGATIHNRARPESLNNELEBLPIIACAE
WMTWNDITEENRAHCTENEUDRETNHAEOE
TFOLSEDTIWENHAEIOYTEYQHEENCTAYCR
EIFTBRSPAMHNEWENATAMATEGYEERLB
TEEFOASFIOTUETUAEOTOARMAEERTNRTI
BSEDDNIAAHTTMSTEWPIEROAGRIEWFEB
AECTDDHILCEIHSITEGOEAOSDDRYDLORIT
RKLMLEHAGTDHARDPNEOHMGFMFEUHE
ECDMRIPFEIMEHNLSSTTRTVDOHW?

and count off every 192nd letter (starting on the 192nd letter)

SLOWLYDESPARATLYSLOWLYTHEREMAINS
OFPASSAGEDEBRISTHATENCUMBEREDT
HELOWERPArTOFTHEDOORWAYWASREMOV
EDWITHTREMBLINGHANDSIMADEATINY
BREACNINTHEUPPERLEFTHANDCORNERAN
DTHENWIDENINGTHEHOLEALITTLEIIN
SERTEDTHECANDLEANDPEEREDINTHEHOT
AIRESCAPINGFROMTHECHAMBERCaUSED
THEFLAMETOFLICKERBUTPRESENTLYDETA
ILSOFTHEROOMWITHINEMERGEDFROM
THEMISTXCANYOUSEEANyTHINGQ?

It’s interesting to note that the Q is not added but internal to K3 and the “?” is  included in the transposition.  I checked against the CIA’s page and Elonka’s and the Q is definitely part of the ciphertext.

So opinions aside, I’m not sure how anyone could argue that the “?” is part of K4 or that K3 ends anywhere but on that “?”.  The crazy thing is that by this method, the plaintext ends at the only non-letter  symbol which happens to be a ? which happens to end a question at the end of the section.

It seems to be a clear indication that K3 definitively ends there and that a new cipher awaits the analyst, especially as K4 only has 97 letters.

Kryptosfan