The simple (or 'anagram') cipher described on the previous pages does not really hide the original letters from the plaintext. If, for example, a message is transposed using a key that is 9 characters long, then the first nine letters of the ciphertext will contain the first nine letters of the plaintext, albeit in a different order.
The columnar transposition cipher, as will be shown below, does a better job of dispersing the original plaintext letters throughout the ciphertext, whilst again not replacing any letters from the plaintext with any new ones.
In this example, we will apply a columnar transposition to a message using a key that is 6 characters long. The first step is to write that message into a table with 6 columns, one for each character in the key:
So far, this is exactly as what would happen with a simple / anagram transposition cipher. The next step is the same too: rearrange the columns according to the characters in the key. In this example, we will use the key of '314250' and, when applied to the table above, we get the following:
To get the ciphertext using the simple / anagram method, we would now read from left-to-right and down the rows: LCUOMATARRA ... etc.
To get the ciphertext using the columnar transposition method, we read down each column, starting with the leftmost column and finishing with the one on the right:
Combining all the columns together and inseeting a space after every block of five characters, gives the ciphertext below:
To illustrate how effectively the transposition cipher scrambles the letters from the plaintext, I have underlined the first 3 letters from the plaintext, 'ACO' (from 'A COLUMNAR TRANSPOSITION ...'), in the ciphertext above. Whereas you can sometimes solve a simple / anagram transposition cipher by simply looking at it, the example above shows that this is much more difficult to do with a columnar transposition cipher and this is because letters that were contiguous in the plaintext can be a long way apart in the ciphertext.