I tried for a while but am having trouble with the offsets (and possibly the correct edition of the DOI). The smartest way to get the offsets correct would be to copy the numbers from the image and cross reference them with the decrypted text. But who's got time for that? :)

Some Python code if anyone wants to play with it:

     words = doi.replace(',', ' ').replace('\n', ' ').replace('  ', ' ').replace(':', ' ').replace('.', ' ').replace("'", ' ').replace('"', ' ').replace(';', ' ').split(' ')
     words = filter(None, words)

     def getcode(w, i, offset):
          print w[i-offset][0],

     for code in [89, 263, 201, 500, 337, 480]:
          getcode(words, code, 0)

I used http://www.constitution.org/usdeclar.txt and deleted the header and title.