Not only diplomats but also generals use secret communications. Codes and ciphers used by the military are typically simpler than those used by diplomats and are often called field cipher. For concealing tactical information in the field, which might be useless if broken in say one week, convenience comes before security.
British general Lord Cornwallis and his associates used such simple ciphers during the American Revolution.
Cornwallis was given the command in the South after the British took Charleston, SC, in May 1780 by amphibious operations. Cornwallis took on an aggressive campaign and thoroughly defeated Americans under Horatio Gates in the Battle of Camden, SC, in August 1780.
The first check to his campaign was the Battle of Kings Mountain, SC, on 7 October 1780. In mid-September, when the rebel expedition to Augusta, GA (on the SC/GA border), was stopped by Cruger from Ninety-Six, SC, Major Ferguson, commanding loyalist militia, tried to intercept the retreating rebels. He boasted that the rebellion was finished in his area. When Cornwallis heard from Ferguson and Cruger, he wrote to Ferguson partly in cipher (23 Septebmer 1780). Ferguson, however, was faced with a growing rebel army and soon started to retreat. On 5 October, he reported his retreat but on 6 October he wrote of his determination that he would make a stand at a strong position of Kings Mountain. The battle on the 7th ended in a complete victory of Americans and Ferguson himself was killed in the action.
The cipher used between Cornwallis and Ferguson was a simple substitution cipher 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
One copy of instructions of Sir Henry Clinton, Commander-in-Chief of the British army in America, for Major General Leslie, who sailed from New York on 17 October 1780, was also enciphered with this cipher. Leslie was instructed to make a diversion in favor of Cornwallis' operations. The cipher copy is marked "F" to indicate that Ferguson's cipher was used. It bears a date "Oct 31 1780" by another hand.
Ferguson's cipher was also used by Lieutenant Colonel Balfour, who was put in command of Charleston when Cornwallis left the place for his aggressive campaign.
Balfour's letter to Cornwallis dated 22 October 1780 is interesting in that it uses both Ferguson's cipher and a book code. A book code encodes a word by the page, the column, and the line where the word occurs in a book (Entick's Spelling Dictionary was used in this case). For example, 47.2.19 stood for "anxiety", which appears on page 47, column 2 (i.e., right), line 19. Letters "F" and "B" were used to indicate whether Ferguson's cipher or Balfour's book code was used.
The following is a sample of their use:
The cipher used in this letter is slightly altered in that "q" is represented by "15" rather than "50" and u and v are interchanged.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
At the end of the letter, Balfour wrote in Ferguson's cipher to remark on possible compromise of Ferguson's cipher.
A letter from Cornwallis to Balfour dated 7 October 1780 also used both systems. This is a letter decoded by the American codebreaker, James Lovell (see here). The cipher found out by Lovell had "15" for "q" but u and v were not interchanged.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
Balfour refrained from using Ferguson's cipher in his letter to Cornwallis dated 13 October 1780 and used only the book code.
Soon after Balfour's expression of concern, Cornwallis-Balfour correspondence apparently switched to a new cipher that used numbers 1-29 to represent the letters (see below).
Cornwallis had proceeded to Charlottes, NC, in late September and planned to let the place be held by Major James Wemyss commanding the 63rd Regiment. On 7 October, the very day of the Battle of Kings Mountain, Cornwallis wrote partly in cipher to Wemyss about his intent. When he learned of the battle, however, he changed his plan. Cornwallis withdrew into Winnsboro, SC, about fifty kilometers west of Camden.
Cornwallis' letter to Wemyss dated 7 October 1780 used a basic substitution table including numbers 1-29: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
The cipher used in this letter invovled a more complicated scheme. First, multiples of ten (10, 20, ...) were used as nulls. Further, an insignificant tens digit was occasionally added. That is, for numbers above 29, the tens digit should be ignored and only the units digit is meaningful. For example, in "31-60-18", "31" is equivalent to "1" and "60" is a null. Thus, this is the same as "1 18" (of). In a similar manner, "31-29-35-37" is equivalent to "1-29-5-7"(only). Use of such nulls and insignificant tens digits increases the security of the cipher. American codebreaker James Lovell, however, broke this scheme anyway (see here).
This letter includes an instruction to send copies of the cipher to Lieutenant Colonel Turnbull and Balfour.
The same scheme, with nulls and insignificant tens digits, was also used in a letter of I. Money (aide de camp to Cornwallis) to Wemyss dated 10 October 1780 and a letter of Wemyss to Cornwallis dated 29 October 1780.
The same scheme, with nulls and insignificant tens digits, was also used in a letter of Turnbull to Rawdon dated 31 October 1780. Turnbull was then the commanding officer at Camden, the second most important place after Charleston, while Rawdon accompanied Cornwallis in his advance into North Carolina and the subsequent withdrawal to Winnsboro.
The same substitution table, cyclically shifted, was used by Lieutenant Colonel Banastre Tarleton.
In the early stage of the campaign in 1780, Tarleton at the head of his cavalry won the name of "Bloody Tarleton" by his atrocities in several victories. He could not relieve Ferguson in Kings Mountain because of severe illness. While yet recuperating, Tarleton was sent to suppress guerrilla activities of Francis Marion but was called back to the vicinity of Winnsboro to meet a more serious menace to Ninety-Six, some 90 km west of Winnsboro, brought up by Thomas Sumter. The battle on 20 November was a draw but Ninety-Six was relieved for the time.
The substitution table used in a letter of Tarleton to Cornwallis dated 28 November 1780 was 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
This is the same sequence of 1-29 as the Cornwallis-Wemyss cipher but the numbers are cyclically shifted and the sequence now starts with "4". It was a new element in the British cipher in the southern theater that the same substitution table can be used with different starting positions. In this letter, the starting position "4" of the cipher table was marked as /4/ (which happens to be the first figure of the letter) in the cipher text.
In a letter of Henry Haldane (aide de camp at Winnsboro) to Lieutenant Colonel Allen or the officer commanding at Ninety Six dated 22 December 1780, the mark "A.6" at the beginning of the postscript seems to show the starting position of the cipher table. The postscript suggests that instead of placing a letter as a key to the cypher, the following marks for the vowels are substituted: (A,|) (E,+), (J,V), (O,∧) (U,*).
Cornwallis's letter to Balfour dated 3 January 1781 proposed an improvement to the cipher.
Cornwallis realized that particular marks to indicate the starting position (key) tend to discover to the enemy and are less explicit to themselves. He noted that there was no figure in the cipher above 29 and proposed to use one or more figures from 30 to 49 to indicate the starting position of the cipher. That is, the figure following them is set at A. For example "34, 36, 43, 15" sets the number 15 to A. The sequence "33, 37, 11" sets 11 to A, and so on. The starting position may be changed as often as desired.
Cornwallis further told of masking words of two or three letters with the figures from 50 to 99, which are to mean nothing. Apparently, Cornwallis was aware that such short words may give clues to a codebreaker. Cornwallis' examples, "11(t), 54, 14(h), 66, 7(e)" - "17(o) 75 8 (f)" - "10(o), 88, 18(n), 76" seem to be arbitrary.
Since the high numbers 50-99 were reserved as nulls, use of insignificant tens digits as in the letter to Wemyss was not assumed in this proposal.
In January 1781, Cornwallis resumed his move to the north from Winnsboro, leaving the defense of South Carolina to Rawdon. Defeat of Tarleton at Cowpens on 17 January 1781 did not restrain him as that of Ferguson brought him back to Winnsboro before. Cornwallis conducted a vigorous pursuit until the Americans crossed the Dan to Virginia. When Cornwallis withdrew, new American commander Nathaniel Greene dared to cross the river again into North Carolina and in March faced the British army at Guilford Courthouse. Although Cornwallis gained a tactical victory, it was a strategic defeat. He retreated to Wilmington in April, where supply from sea was available.
The new scheme of indication of starting positions was used in a letter of Balfour to Cornwallis captured by the Americans on 7 March 1781 and probably written at the beginning of March. In this letter, deciphered by James Lovell, the cipher table is switched no less than 31 times (see here).
The same scheme with table switching was also used in letters of Rawdon to Balfour dated 12, 13, and 15 April 1781 and letters of Rawdon to Cornwallis dated 7 March and 2 May 1781.
Now, Cornwallis' eyes were directed to Virginia, rather than the Carolinas. He planned to make a junction with General Phillips, who had taken over the command in Virginia in March. As he moved north, Cornwallis sent Tarleton with an advance guard.
The same cipher table (substitution by 1-29 with table switching) continued to be used between Tarleton and Cornwallis, as evidenced by a letter from Tarleton to Cornwallis dated 5 May, in which Tarleton wrote in cipher, "the cypher to General Phillips I understand". This was probably in reply to a remark in Cornwallis' letter of the same date: "My letters to Phillips are in the new cypher; he has not the old one". Cornwallis had enclosed a cipher for Phillips in his letter of 24 April 1781.
Indeed, about this time, the British seems to have introduced a new substitution table.
A letter of Balfour to Cornwallis dated 26 April 1781 used a new cipher table which substitutes numbers 10-39 for the letters.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
The beginning of the cipher in this letter "43 49 51 17" indicates that "17" is aligned with "A" and the ciphertext includes higher numbers as nulls. But the cipher table is not switched in this letter.
This scheme, with table switching, was also used in letters of Craig (at command in Wilmington) to Cornwallis dated 2 May 1781 and 23 July 1781.
It was also used in a letter of Balfour to Charles Stuart dated 12 August 1781 with "A=31" (see here).
Sir Henry Clinton, Commander-in-Chief of the British army in America from 1778, is known to have used an hourglass cipher (grille) in 1777 in his letters to General Burgoyne. His subordinates used book code for contacting Benedict Arnold, who betrayed the American cause and joined the British in 1780.
The war in the north was at a standstill since Burgoyne's surrender in Saratoga in 1777. As the war was waged in the south, Virginia emerged as a strategic target. Virginia, thus far unscathed from the war, functioned as a supply base for the Americans under Greene in the south.
Thus, at the end of 1780, the turncoat Arnold landed in Virginia at the head of a British expedition. Arnold was accompanied by Lieutenant Colonels Dundas and Simcoe, whom Arnold was instructed to consult before making important decisions. At the end of March 1781, General Phillips joined Arnold and took over command. They conducted raids of towns and took Petersburg in April.
Apparently, Dundas had been given a cipher from Clinton (Clinton to Cornwallis, 29 May 1781) and handled cipher for Phillips (cf. Phillip's note on PRO 30/11/96/37-38, Clinton to Phillips, 5 April 1781). As noted above, Cornwallis had also sent Phillips a cipher when he told him his determination to make a junction with his forces (Cornwallis to Phillips, 24 April 1781).
Clinton requested Phillips to send his opinion or information in cipher (Clinton to Phillips, 24 March 1781, 24 April 1781).
The first request from Clinton to Cornwallis to write in cipher appears to have been made in a letter of Clinton to Cornwallis dated 29 May 1781, in which Clinton expressed his surprise and concern of Cornwallis' determination to join Phillips in Virginia but resigned himself to what had been done.
Lord Chewton had sailed from New York on 4 May to join Cornwallis at Wilmington, whence Cornwallis had departed on 25 April. Cornwallis joined Arnold on 20 May at Petersburg to find that his friend Phillips had died a few days before. The above letter of 29 May from Clinton was not received by Cornwallis until 12 July. Before this, on 8 July, Cornwallis received a "cyphered dispatch" from Clinton dated 28 June 1781 and apparently he had no trouble in reading it. Upon receiving this letter, Cornwallis forwarded a copy to General Leslie to make neceessary arrangements.
Failing to beat the American forces under Lafayette, Cornwallis entered Yorktown in August.
While Cornwallis was fortifying Yorktown, Washington, who had been conceiving an attack on Clinton at New York, made a determination to move his main army and allied French troops under Rochambeau to the Chesapeake to cooperate with the French fleet expected to arrive there. He successfully made his move southward without being arousing suspicion from Clinton.
Thus, it was unexpected for Cornwallis that a French fleet appeared at the mouth of the Chesapeake. He reported it in cipher to Clinton on 31 August. His message of the next day, written in cipher on a Continental note read: "An enemy's fleet within the Capes, between thirty and forty ships of war, mostly large." It was not until 2 September, when Washington's main amry was already passing through Philadelphia, that Clinton wrote to Cornwallis that Washington was moving southward to effect a junction with the French.
In September, Cornwallis and Clinton frequently sent dispatches in cipher to each other. For example, on 8 September, after reporting the movement of the allied forces, Cornwallis wrote in cipher the status of his defense. On 16 and 17 September, he wrote in cipher of his desparate situation and concluded, "if you cannot relieve me very soon, you must be prepared to hear the worst." Upon receiving this, Clinton wrote in cipher on 24 and 25 September of his relief plan and told a relief fleet would sail on 5 October. On 29 September, upon receiving this, Cornwallis wrote in cipher his satisfaction.
The cipher used in these exchanges were substitution by 1-29 with table switching, which Cornwallis had used with Balfour and Tarleton. (Their earliest use of this cipher the present author could confirm in Clinton Papers in the Clements Library is in Cornwallis to Clinton, 16 August 1781.) It is observed that Cornwallis made a point of changing the starting position of the table several times in a letter but Clinton did not.
Siege of Yorktown began. On 10 October, instead of the expected relief, Corwallis received a cipher letter dated 30 September from Clinton that told him that the relief fleet would sail by 12 October. In a cipher letter of 11 October, followed by one of 15 October, Cornwallis reported progress of the siege and his hopeless situation. On 19 October 1781, Cornwallis surrendered. This all but ended the war for independence of America.
The report Cornwallis wrote on 20 October to Clinton was in plaintext.