I started out by looking at the different ciphers to get a general understanding of how they worked. After this I focused on what patterns of letter frequencies the ciphers exhibited and what general indices of coincidence they had. For a summary of the results and flow chart of the process of cipher determination click here. Or check out my power point presentation here. (This requires Power Point. If you do not have Power Point you can veiw the slides as .jpeg images here)

    The data I found in books and articles did not present an easy way to distinguish between the Vigenere and Playfair ciphers. They both had a fairly flat frequency distribution and published index of coincidence of ~.38. Using CryptoTool I did further testing of the indices of coincidence and found that the value varies based on length of the keyword. It turns out that there is a distinct and different range of values for the two ciphers that does not overlap as long as the keyword is long enough. The testing seems to suggest that the keyeword has to be atleast 4 characters. This information was encouraging for application to the Expert System and it seems reasonable, although not optimal to limit the keyword to length 4 or greater. These test results are shown in the table below. The data also shows that the index of coincidence varies based on the length of the keyword. This information might be used later to find the length of the keyword in order to be able to decipher the message.