La Cryptographie militaire
Military cryptography — the 1883 paper that founded modern cryptanalysis: six desiderata for ciphers, the repeated-fragment key-length attack, and the maxim that a cipher must baffle even its inventor.
Mechanisms
A break doesn't expire
Against the claim that war messages only need three or four hours of secrecy: a break compounds.
Une fois qu'un cryptogramme intercepté a pu être déchiffré, toute nouvelle dépêche, écrite avec la même clef et qui subit le même sort, peut être lue instantanément. Once one intercepted cryptogram is broken, every later dispatch written with the same key can be read instantly.
Why it matters
Against the claim that war messages only need three or four hours of secrecy: a break compounds. It doesn't expire with the message — it retroactively opens everything ever sent under that key.
Use when: When you are tempted to reuse a key, password, or credential because this one use is low-stakes.
Destroy the weapon by publishing it
Publicity as an offensive move.
Ce n'était pas un des moindres services à rendre à la patrie que d'anéantir par la publicité l'arme la plus dangereuse des ennemis secrets. In 1793 Dlandol published the royalists' cipher keys — annihilating by publicity the most dangerous weapon of the Republic's secret enemies.
Why it matters
Publicity as an offensive move. A broken secret, widely known to be broken, stops being a weapon — nobody will trust a cipher the newspapers have already printed the key to.
Use when: When you need to kill a bad standard, a compromised credential, or a practice that survives on obscurity.
Every break is two jobs
For polyalphabetic systems the two unknowns are the number of alphabets and their arrangement.
Le déchiffrement d'un cryptogramme dont on n'a pas la clef comporte deux opérations bien distinctes : un calcul de probabilité et un travail de tâtonnement. Breaking a cryptogram without the key is two distinct jobs: a probability calculation and groping trial-and-error.
Why it matters
For polyalphabetic systems the two unknowns are the number of alphabets and their arrangement. Decomposing the attacker's job into countable subproblems is also how the defender prices each layer of defense.
Use when: When sizing up any adversarial problem — split it into the math and the groping.
Repetition is the leak
Take the greatest common divisor of the intervals between repeats and you have the key length.
Dans tout texte chiffré, deux polygrammes semblables sont le produit de deux groupes de lettres semblables, cryptographiés avec les mêmes alphabets ; le nombre de chiffres compris dans l'intervalle des deux polygrammes est un multiple du nombre des lettres de la clef. Two identical ciphertext fragments come from two identical plaintext groups under the same alphabets — and the distance between them is a multiple of the key length.
Why it matters
Take the greatest common divisor of the intervals between repeats and you have the key length. Redundancy betrays structure: the key is hiding in the gaps between repeats.
Use when: When hunting hidden periodicity in any data — ciphertext, logs, signals.
The traitor words
Military dispatches open predictably — le général, vous, ne — so the breaker keeps a classified list of likely openings indexed by where E falls.
Dans les cas un peu difficiles ce sont généralement les deux ou trois premiers mots qui viennent jouer le rôle de traîtres. In the hard cases it is generally the first two or three words that play the traitors.
Why it matters
Military dispatches open predictably — le général, vous, ne — so the breaker keeps a classified list of likely openings indexed by where E falls. Every message has a predictable edge; the attacker catalogs it like factory-default passwords.
Use when: When guessing the predictable parts of any structured message or protocol.
Symmetry of position
Structure leaks sideways.
Du moment que les alphabets sont disposés en nombre carré, et que la signification d'un chiffre commun a été trouvée dans deux ou plusieurs alphabets, on peut, au moyen d'une simple addition, déterminer la place que doit occuper dans ces différents alphabets tout nouveau chiffre. In square-arranged alphabets the letters keep their relative order — solve one letter in two alphabets and every other letter transfers by offset.
Why it matters
Structure leaks sideways. One solved position determines all parallel positions by simple addition — Kerckhoffs notes no cryptography book had ever pointed this out. Regularity is a side channel.
Use when: When exploiting any regular arrangement — one solved cell positions the whole grid.
One wrong digit, wrong war
Codebooks fail silently and totally: a garbled group doesn't look garbled, it looks like a different order.
Qu'un seul chiffre dans un groupe soit fautif, que le télégraphe mette un 3 où il faut un 5, et le sens sera complètement changé. One wrong digit in a code group and the meaning is completely changed — an Italian cable once turned an embassy attaché into Count Andrassy.
Why it matters
Codebooks fail silently and totally: a garbled group doesn't look garbled, it looks like a different order. In 1870 half the army's dispatches arrived partly illegible. A system whose failure mode is confident misreading will misread confidently.
Use when: When designing anything where a typo changes the order — make errors loud, not plausible.
Concepts
The enemy knows the system
Security that depends on the method staying hidden dies the day it is captured, reverse-engineered, or leaked.
Il faut qu'il n'exige pas le secret, et qu'il puisse sans inconvénient tomber entre les mains de l'ennemi. The system must not require secrecy, and must be able to fall into the enemy's hands without inconvenience.
Why it matters
Security that depends on the method staying hidden dies the day it is captured, reverse-engineered, or leaked. Design as if the enemy already has the blueprint — because eventually they will.
Use when: When you are evaluating any security system, protocol, or anyone's secret sauce.
Secrecy doesn't scale
If the machinery must stay secret, every new user is a new capture risk — so the system can never be issued beyond a handful of generals.
J'entends par secret, non la clef proprement dite, mais ce qui constitue la partie matérielle du système : tableaux, dictionnaires ou appareils mécaniques. By secrecy I mean not the key itself, but the material part of the system: tables, dictionaries, mechanical devices.
Why it matters
If the machinery must stay secret, every new user is a new capture risk — so the system can never be issued beyond a handful of generals. A secret you can't distribute is a system you can't deploy.
Use when: When deciding what in your operation must stay secret versus what must survive exposure.
The whole cipher is one letter: E
In French, E is one letter in five.
La valeur d'un chiffre se mesure aux garanties qu'il offre contre la découverte du signe correspondant à cette lettre. The value of a cipher is measured by the guarantees it offers against discovery of the sign for the letter E.
Why it matters
In French, E is one letter in five. Find the most frequent symbol and the rest of the system unravels by trial. Every substitution-like structure has a load-bearing letter — the defense lives or dies on hiding it.
Use when: When attacking any substitution-like structure — find the anchor first.
One key per message
Stack a dozen short cryptograms under one key, align them in columns, and the frequency math works across all of them at once.
Quelle que soit la disposition adoptée des alphabets, les correspondants sont toujours tenus d'écrire chacune de leurs dépêches avec une clef différente. Whatever the alphabets, correspondents must write each dispatch with a different key.
Why it matters
Stack a dozen short cryptograms under one key, align them in columns, and the frequency math works across all of them at once. Reusing a key pools many small unbreakable texts into one big breakable one.
Use when: When reusing any secret across sessions — each reuse feeds the same attack.
Teach your officers to break it
The Germans drilled every officer in cryptanalysis; the French issued ciphers nobody understood.
Ce ne sera que lorsque nos officiers auront étudié les principes de la cryptographie et appris l'art de déchiffrer, qu'ils seront en état d'éviter les nombreuses bévues qui compromettent la clef des meilleurs chiffres. Only when our officers have studied cryptography and learned the art of breaking will they avoid the blunders that compromise the key of the best ciphers.
Why it matters
The Germans drilled every officer in cryptanalysis; the French issued ciphers nobody understood. Users who can't attack the system will misuse it into insecurity — the weakest link is the operator, so train the operator as an attacker.
Use when: When rolling out any security practice to non-experts — teach the break first.
Beaufort unmasked
Most celebrated new systems are old systems in costume.
Tous ces appareils ne sont en réalité autre chose qu'un simple tableau de Vigenère. The English marveled at Admiral Beaufort's supposedly indecipherable system — it is Vigenère with a reversed alphabet, nothing more.
Why it matters
Most celebrated new systems are old systems in costume. Before awe, check for isomorphism: relabel the parts and see if the machine you already know is hiding inside.
Use when: When someone pitches you a revolutionary method — strip the costume first.
Krohn's 3,200 alphabets
What matters isn't how many alphabets you have, it's how hard it is to determine how many you used.
Ce n'est pas la possibilité de se servir d'un grand nombre d'alphabets différents qui donne de la valeur à un système, mais plutôt la difficulté plus ou moins grande de déterminer le nombre des alphabets employés. Krohn built a dictionary of 3,200 alphabets — c'est à la fois trop et trop peu: too much and too little.
Why it matters
What matters isn't how many alphabets you have, it's how hard it is to determine how many you used. Complexity theater: the metric being maximized is not the metric that matters.
Use when: When someone dazzles you with big numbers — ask which number the attacker actually needs.
Passages
Ars ipsi secreta magistro
Borrowed from Du Carlet (1644) as the closing line of the whole paper.
Un chiffre n'est bon qu'autant qu'il reste indéchiffrable pour le maître lui-même qui l'a inventé : Ars ipsi secreta magistro. A cipher is only good insofar as it remains indecipherable to the very master who invented it.
Why it matters
Borrowed from Du Carlet (1644) as the closing line of the whole paper. The ultimate test isn't whether strangers fail to break your system — it's whether you, holding every advantage, cannot read it either.
Use when: When testing your own system, argument, or lock — try to break it with the home-field advantage.