James Lovell, a leading member of the Committee of Foreign Affairs and an expert in cryptography, was active in deciphering British secret communications around the time of the Yorktown Campaign in 1781.
On 25 August 1781, Nathaniel Greene, commanding the southern army, had forwarded two intercepted letters in cipher. It took Lovell only a few days to decipher them after they reached Congress on 17 September. On 21 September, Lovell reported his results to Greene and Washington.
The two letters Lovell deciphered are as follows.
PCC Roll 65, Page 733 (translation on Page 709)
In this, the letters of the alphabet totalling 25 (there is no "j") are substituted by code numbers 10-39 (10, 11, 20, 21, 30 are not used).
a b c d e f g h i k l m n o p q r s t u v w x y z
31 12 22 32 13 23 33 14 24 34 15 25 35 16 26 36 17 27 37 18 28 38 19 29 39
Code numbers not included in the substitution table also occur in this letter. They are nulls (baulks as Lovell called it) for distracting the interceptors. As Lovell observes, such nulls are often placed between words but sometimes within a word. Especially, double letters are often hidden by splitting such letters by a null (see "will" and "officers" in the examples given below). This has an effect of defying cryptographers who may use double letters as a clue to guess, for example, 27 18 22 22 13 27 27 stands for "success".
It is noted the cipher message is preceded by a sequence "91 45 31 60". This sequence, or rather the code "31" therein, is an indicator of the start position of the substitution table. It will be further described in the next section.
PCC Roll 65, Page 703 (translation on Page 759)
The cipher used in this letter is based on the same substitution table as above with shifted mapping between the letters and code numbers. Actually, three variations are used in this single letter.
a b c d e f g h i k l m n o p q r s t u v w x y z
26 36 17 27 37 18 28 38 19 29 39 31 12 22 32 13 23 33 14 24 34 15 25 35 16
34 15 25 35 16 26 36 17 27 37 18 28 38 19 29 39 31 12 22 32 13 23 33 14 24
15 25 35 16 26 36 17 27 37 18 28 38 19 29 39 31 12 22 32 13 23 33 14 24 34
To convey the start position of the code arrangement, the letter begins with an indicator 44 57 26 61. Of these, 44, 57, and 61, which are not included in the substitution table, are nulls but 26 indicates the substitution table should start with 26.
In his letter of 21 September, Lovell attached the above substitution tables and explained the system as follows.
Lovell's deciphering proved useful for Washington at this critical juncture at least on one occasion.
In the above letter of 21 September to Greene, Lovell added a postscript next day.
The substitution table is as follows.
a b c d e f g h i k l m n o p q r s t u v w x y z
19 09 17 13 16 07 12 08 14 15 26 04 18 21 03 02 11 05 24 29 01 25 23 22 06
Here, the 25 letters of the alphabet (there is no "j") are substituted by code numbers 1-29 (10, 20, 27, 28 are not used). (Note that, hereinafter, preceding 0s are included by the author merely for convenience of tabulation.) While this is based on the same simple substitution with table switching as above, the deciphering involved "greater pain" because the substitution table was switched more than twenty times, almost every sentence. Thus, the letter contains the following indicators.
The first indicator shows "a" should be enciphered as "19", as shown in the above table. The second and third indicators show "a" should be enciphered as "23" and "05", respectively. Thus, the substitution table for these indicators will be as follows.
a b c d e f g h i k l m n o p q r s t u v w x y z
23 22 06 19 09 17 13 16 07 12 08 14 15 26 04 18 21 03 02 11 05 24 29 01 25
05 24 29 01 25 23 22 06 19 09 17 13 16 07 12 08 14 15 26 04 18 21 03 02 11
Tables for the other indicators can be constructed similarly.
PCC Roll 65 Page 529 (translation on Page 760)
On 28 September, Lovell reported to Greene of his success on some letters Greene forwarded a year before (see below for quotation). Of these, a letter to Wemyss was based on the same table for "substitution by 1-29" as above.
a b c d e f g h i k l m n o p q r s t u v w x y z
08 14 15 26 04 18 21 03 02 11 05 24 29 01 25 23 22 06 19 09 17 13 16 07 12
At first sight of the letter, the system may look very different from the simple substitution as above. The letter "a" is enciphered as either 8, 38, 48, or 68, "e" as 4, 44, 64, or 74, "h" as 3 or 73, and so on. Thus, some polyalphabetic principles more intricate than switched simple substitution seem to be used.
However, Lovell found out that for two-digit numbers not listed in the substitution table, the left-side digit is insignificant. Thus, 38, 48, and 68 have all the same value as 8.
For the reader's convenience, single-digit codes are:
1(o) 2(i) 3(h) 4(e) 5(l) 6(s) 7(y) 8(a) 9(u).
Lovell's explanation is as follows.
(The cipher of this Cornwallis-Wemyss letter of 7 October 1780 was independently broken by Peter P. Fagone in 1977, who reported the fact and printed the cipher with hints (not the decipherment) for the readers in 'A Message in Cipher Written by General Cornwallis during the Revolutionary War', Cryptologia, Vol. 1, No. 4, pp. 392-395.)
Lovell's letter of 28 September reported two more ciphers.
The "second" cipher is simple substitution similar to above. Lovell's reconstruction of the substitution table is as follows.
a b c d e f g h i k l m n o p q r s t u v w x y z
06 05 04 03 02 01 12 11 10 09 08 07 18 17 16 15 14 13 24 23 22 21 20 19 25*
*Lovell actually wrote "I presume 25" for "z".
The letter was marked "F" and Lovell regarded it as an indicator of the start position of the substitution table.
As to the "first" of the two ciphers used for Balfour, Lovell considered it was a book code based on Entick's New Spelling Dictionary.
A book code encodes a word by the page and the line where the word occurs in a book (a dictionary is typically used because it facilitates finding a word to encode). When a page consists of two or more columns, the column is indicated by a letter ("a", "b" or "1", "2" etc.), an underline, an overdot, an overline, or the word position may be reckoned in the whole page without regard to columns.
Interestingly, the same dictionary was used by the Americans during the Revolution. For specific examples, see another article of mine. Compared with complex versions used by John Jay, Cornwallis' use of Entick's Dictionary was straightforward.
Lovell confirmed his conjecture two weeks afterwards.
The deciphered letters reported at the end of the above quotation were no less than ones sent by Sir Henry Clinton, Commander-in-Chief of the British forces in New York, to Lord Cornwallis in Yorktown. By late September, Cornwallis' situation was desperate. The British planned to relieve Cornwallis but the close watch kept by the Americans made it hard for the dispatch boats to get through.
The capture of Clinton's letters was reported by Thomas McKean, the President of Congress, to Washington on 12 October.
Two days later, McKean could report Lovell's deciphering of the captured letters. In such a feat, Lovell was helped by British commanders' tendency to use the same cipher. It has been verified by David Kahn that Cornwallis used the same cipher table (presumably, the "substitution by 1-29" as indicated herein) with the start position a=7.
A circumstantial description of the event is given in a later memoir of Elias Boudinot.
The President's letter of 14 October further reported about the British preparation of relieving Cornwallis. Actually, it was only on 17 October that the fleet set sail. When the fleet arrived near the Chesapeake, Cornwallis had already surrendered on 19 October 1781.
Apparently, Washington received the President's letter after this. On 20 October, he lost "not an instant" to send his ally, Admiral de Grasse, the two intercepted letters (without knowing that the President had already forwarded copies of these letters to de Grasse).
At the time, the British forces under Clinton were expected to make a counterattack. After all, however, the victory of Yorktown effectively concluded the Revolutionary War.
After five years' hard working, without even visiting his wife and children in Boston, Lovell resigned in 1782. On his way home, he took a letter in cipher with him, which he "copied by desire of the gentleman at my House's" and deciphered it on the way. Though it contained little information, he reported the feat to Robert Livingston for future use (19 April 1782).
This was a passage cipher similar to Dumas' cipher (see another article). In a passage cipher, each letter in a passage (phrase or sentence) is assigned a number sequentially.
A letter is enciphered with any of the assigned numbers. Letter r may be enciphered as 7, 20, 26, 29, 33, or 45 in this example. Thus, peace may be enciphered as 34 6 30 31 10 but there could be many other ways to represent the same word.
The passage cipher used in this particular letter used several such passages. Thus, a letter is represented by two numbers: one indicating the passage and the other indicating the letter. The passage given above is passage 16 and peace should actually be enciphered as 16-34 6 30 31 10. (Apparently, the passage number was written only at the beginning.)
Lovell included a table of assigned numbers and explained the cipher thus: "you'll observe that there is regularity of sense in the alphabets when placed by the columns of numbers from 1 upwards".
It is possible that these passages 1-18 are not independent but actually lines from one contiguous passage.