Here is my theory how the VM was actually enciphered — Mark I. It’s actually a fairly simple method, it would have been feasible in the time period considered, it is moderately labour-intensive, and it is able to statisfy some of the properties of the VM text. (Though not all of them — which is why I later went on to Mark II.)
Anyway, here’s the deal:
How it works:
Suppose you want to encipher the word “abdomen”:
The first thing you do is you decompose the letters which make up the word into the individual penstrokes you have used to write it down, from left to right:
As you note, we have used a surprisingly small number of different strokes, namely only six. Let’s write them down orderly, and assign a number to each of them:
Now we can replace each of the strokes from our ciphertext sequence with the number we have assigned to it in the list above:
Thus we have replaced the plaintext word “abdomen” with a string of digits, “12344342551625”. If we had used little obscure glyphs instead of the well-known arab digits, we would have arrived at a string of Voynich characters like “qochhchoiiqnoi”.
That’s all there really is to it.
Read more on the following pages:
Thoughts about the Voynich Manuscript 