Main Source: Jay Papers ID 1537 (encoding worksheet)
Jay Papers ID 598 is a clean copy, in which decoding with THE=332' (WE007) has been attempted for the first four codes ("SHI-IG-THEIR-EDGE") but abandoned.
Robert R. Livingston's letter of 28 November 1781 to John Jay, enciphered in THE=332' (WE007) and referred to as the final text below, was at first inadvertently enciphered with a code that was to be delivered to Jay by William Palfrey. However, Palfrey had been lost at sea in December 1780 and Livingston re-enciphered the letter and enclosed it in his letter of 13 December 1781. (See another article for details.)
While no detail is known about Palfrey's cipher, Jay Papers include a copy of this letter enciphered with an unidentified cipher (THE=227') different from THE=332' (WE007). It is probable that it was enciphered with Palfrey's cipher.
The encoding worksheet allows one to partially reconstruct Palfrey's cipher.
The text in the encoding worksheet is slightly different from the final text. Such differences are marked with underlines and strikethroughs below.
(plaintext lines omitted)Congress are occupied in taking measures for an early & active campaign, and they feel themselves satisfied with every thing both at home and abroad.
(plaintext lines omitted)
By comparing the plaintext and the code numbers in the encoding worksheet, one may infer what each number represents. In the following, readings in parenthesis ( ) are supported by at least two instances, those in brackets [ ] are guesses, and those in angle brackets < > are dubious guesses.
Probably, an "a"-like symbol under a code number represents an "e"-ending and that over a code number represents an "s"-ending.
In the above dump, blue color indicates where the clean copy (Jay Papers ID 598) differs from the encoding worksheet (Jay Papers ID 1537). Some differences are the same as the final text and may be attributed to Livingston's editing, while others appear to be simple errors/omissions in copying.
The following is a partially reconstructed list of code numbers of Palfrey's code. Dubious guesses are annotated with a remark.
9<h/t>*unclear manuscript; double reading; collision with 563
17<.?>*no other punctuation
257<r> *collision with 378; unclear manuscript; probably an enciphering error
340<sup>*unprobable entry; the clean copy renders this "517"
417(at) *unprobable entry
470<manne> *unprobable entry
577<humble> *unprobable entry; the clean copy has "572[hum] 299a[ble]", which looks more likely.
595<ons> *unprobable entry, unless a is missing
In the above, a remark "unprobable entry" indicates the word or syllable is not listed in the widely used 600-element printed template (see another article). If Palfrey's cipher was not based on the same template, entries such as "at" and "sup" may be acceptable.
It would seem clear that at least Palfrey's code had a vocabulary of 600 words and syllables and that it was a two-part code, i.e., the elements were assigned numbers in random order.
If the above code was indeed Palfrey's code, it would mean two-part code was introduced in December 1780 at the latest, i.e., before John Laurens' code (THE=278'/WE005) (see ibid.).