Some French ciphers during the reign of Henry IV (1589-1610) are described. During this period, arbitrary symbols used in French ciphers gave way to figures. (Similarly in England. See another article)
In 1590, the French ambassador in Madrid wrote to Henry IV in the same cipher that he used in writing to Henry III in 1586, notwithstanding a compromising incident in the meantime (Garrett Mattingly, Renaissance Diplomacy p.249).
A cipher (1599) of Maximilien de Béthune, first Duke of Sully, who was a right-hand man for Henry IV (Wikipedia), is printed in Etienne Bazeries, Les Chiffres Secrets Devloiles (1901) p.28-29.
It had a much larger vocabulary than ciphers in the 1550s (see another article). Each letter of the alphabet is assigned a figure (e.g., 3 for A, 10 for B, 8 for C, ...) as well as two or three arbitrary symbols. About thirty names could be represented by a figure with two dots above and about eighty common words could be represented by single letters, single letters with two dots above, single letters with a single dot above, or figures 2-11.
A "Δ"-like sign indicates doubling the preceding letter and "y"-like and "$"-like signs indicate cancelling it.
In the first years of Henry IV's reign, François Viète broke Spanish ciphers for the King but at least this particular cipher does not appear to have incorporated features of Spanish ciphers such as the vowel indicator system for representing numerous syllables (see another article).
(This section has been rewritten in August 2015.)
Ciphers were used not only among the King and his subjects but also with an ally. A cipher between Henry IV and Maurice le Savent ("der Gelehrte"), Landgrave of Hesse-Cassel (Wikipedia) identified by Rommel is printed in Rommel, "La clef des cihffres dans la Correspondance inédite de Henri IV. avec Maurice le Savant", Allegemeine Zeitschrift für Geschichte V (1846), p.402-403 (Google).
The substitution table for a-y (plus an extra "z" column) is described in Fletcher Pratt, Secret and Urgent p.64 (with some typos) and some code words such as 33 (the Emperor) are given in James Westfall Thompson and Saul K. Padover, Secret Diplomacy, p.260 (with the overbar omitted). Pratt seems to have some other source because he gives code words 10 (le) and 39 (mon), of which Rommel leaves the former blank. Numbers 2, 4, 5, 6, 7, 8 were nulls (used always at the end of a word).
According to Pratt, the cipher was prepared in the French court to rally the alliance against the Emperor around 1609-1610. When succession of Jülich-Cleves-Berg had become a source of contention by the death of the reigning duke without issue in March 1609, the Emperor sent troops to occupy the fortress at Jülich but was compelled by the siege by Dutch, Brandenburg, and Palatine forces (June to July 1610) to withdraw (Wikipedia). It was widely believed that Henry IV was to join the alliance against the Emperor but the plan was aborted by his assassination in May 1610.
As with many ciphers at the time, j was enciphered as i and v as u. Further, q (very frequent in French) and k (very frequent in German) are both absent and were supposed to be enciphered as c.