38
$\begingroup$

Heinrich is a scientist and a scholar, one of the greatest minds of the Austrian-Hungarian Empire around 1880. He studied mathematics, physics and engineering. He patented a few of his ideas and became very rich.

He writes a diary with all his observations, experiments and plans. Some of his ideas and inventions are potentially dangerous for humanity and he knows that. This is why he keeps his diary encrypted in order to protect his thoughts in case the diary is stolen.

He arrives at his lab in the morning and wants to continue his work. A few pages of the diary need to be decrypted - Heinrich needs to read last pages in order to refresh his yesterday's ideas and plans. Decryption of 10 pages shouldn't take more than one hour and it shouldn't require anything more than just paper and ink. He has to encrypt new pages and destroy all the plaintext paper trace in the evening.

What encryption schemes were known in the 19th century?

What encryption would Heinrich use?


I tag this question [military], because I think encryption was mainly used in army those days and military procedures may be a valuable source of knowledge.

$\endgroup$
  • 2
    $\begingroup$ I don't have the time to answer this thoroughly, but the immediate thought I have is a Da Vinci's cryptographic discoveries. $\endgroup$ – roberrrt-s Dec 28 '16 at 8:51
  • 6
    $\begingroup$ Here is a list of classical encryption techniques. You'd have to choose the one Heinrich's competitors wouldn't know, but anyone familiar with encryption, on seeing an encrypted diary, might research known classical techniques and solve them. Heinrich could develop his own language & alphabet if he's super concerned about security, I've read about people doing that, or instead of encrypting his native language, Heinrich could encrypt a known language that he knows but his competitors wouldn't. $\endgroup$ – Thom Blair III Dec 28 '16 at 10:05
  • 1
    $\begingroup$ Then why do you say "encrypt the whole diary again"? Why should you even to need to encrypt the 10 pages? There is not reason to destroy the encryption. $\endgroup$ – paparazzo Dec 28 '16 at 11:40
  • 14
    $\begingroup$ When evaluating the answers, they should be compared against a simple heavy locked safe for utility and security. $\endgroup$ – James Reinstate Monica Polk Dec 28 '16 at 16:18
  • 1
    $\begingroup$ A few more examples, and links, at users.telenet.be/d.rijmenants/en/handciphers.htm . Codebooks were also used at that time, see en.wikipedia.org/wiki/Codebook . $\endgroup$ – jose_castro_arnaud Dec 28 '16 at 21:45
29
$\begingroup$

Something like a Vigenère cipher with a very long keyword. If the keyword is long enough, it approaches a "reused one-time pad". While reusing an one-time pad is a bad thing, proper exploits require a lot of computer power that might not be available to the attacker.

$\endgroup$
  • 7
    $\begingroup$ And for added spice, since he is in Autria-Hungary the plain text would be in one of the minority languages, for example Ruthene, written with an ordinary monoalphabetic substitution... $\endgroup$ – AlexP Dec 28 '16 at 11:40
  • 6
    $\begingroup$ Practical applications of the Vigenere cipher in the 19th century through WWI were relatively easily broken. $\endgroup$ – kingledion Dec 28 '16 at 13:44
  • 2
    $\begingroup$ @kingledion, it was not practical to use an overly long keyword (or key phrase, or key paragraph) because that would have to be transmitted and stored safely. The mad scientist writing for himself can use a long key. $\endgroup$ – o.m. Dec 28 '16 at 16:07
  • 5
    $\begingroup$ "A reused one-time pad"? You mean... you know.... not a one-time pad? $\endgroup$ – Jasper Dec 28 '16 at 16:41
  • 2
    $\begingroup$ People, I hope you noticed my use of quotation marks. $\endgroup$ – o.m. Dec 29 '16 at 9:12
20
$\begingroup$

Invent his own language - although known before, late 19th century is when the constructed languages took off for real. Add an unusual alphabet of his own design, and everything is set. Though it could be "decrypted" with a lot of effort, especially if there are pictures or an occasional word from a natural language, it's still rather improbable.

Though this method is susceptible to the Rubber hose cryptanalysis, so if he suspects the adversary is capable of this kind of attack, he might not bother and use a simple substitution cipher to distract casual readers.

$\endgroup$
  • 1
    $\begingroup$ This is probably the best, as once you learn the language you can read and write in it with ease. $\endgroup$ – cybernard Dec 28 '16 at 20:26
  • 3
    $\begingroup$ This one is not very difficult to break if the whole diary is analysed. $\endgroup$ – Pere Dec 28 '16 at 23:33
  • 3
    $\begingroup$ Doesn't that depend on the complexity of the language? Ancient Egyptian wasn't "decrypted" until the Rosetta Stone was found. $\endgroup$ – Charles Burge Dec 29 '16 at 1:09
  • 1
    $\begingroup$ This depends very much on how the conlang is built. A language with similar grammar and construction to the scientist's first language will be pretty simple to puzzle out. A language with radically different grammar and syntax could be very hard to work out - for example, there are various combinations of declension, conjugation, syntax etc. that don't exist in any natural language, but are perfectly valid. Similarly, a writing system that makes it hard to spot patterns would complicate it further - pictograms or ideograms are much more complex than an alphabet-type system. $\endgroup$ – anaximander Dec 29 '16 at 11:21
  • 2
    $\begingroup$ @Ovi Darn it, you beat me to it. Navajo was used to devastating effect in WWII because it was relatively obscure and didn't have a written component and the symbology that goes with it unless you count sand paintings. Combine Navajo put down with english phonetics and then subject that to a substitution cypher. that should be easy enough to decrypt and be useful, but be secure enough to thwart most everyone. $\endgroup$ – Paul TIKI Dec 29 '16 at 15:05
13
$\begingroup$

It seems to me that the Playfair cipher would be the obvious choice.

It was invented in 1854, and at the time quite a few people believed it to be unbreakable (and, in fact, it wasn't broken until around 1920, if memory serves).

If you want to get slightly more elaborate, there's a slightly modification of Playfair called two-square, which adds a little difficulty in carrying out the encryption/decryption, but adds a fair amount of extra security.

Both Playfair and two-square operate on pairs of characters (digraphs) instead single characters. This makes frequency analysis substantially more difficult. You have to look at frequencies of 676 possible letter pairs, rather than the 26 individual letters for most typical substitution ciphers (e.g., Vigenere).

In fact, Playfair was strong enough that it was used fairly heavily through WW I, and even continued to be used as a tactical cipher in WW II. With WW II technology, it still took a few hours to break it, so if you were encrypting something like "charge in 24 minutes" it was still secure. In 1880, it was still pretty widely viewed as simply unbreakable (and, as already noted, did remain unbroken for another 40 years or so).

$\endgroup$
  • $\begingroup$ The problem with Playfair is that if you've got an entire diary to work from, you've got more than enough text for frequency analysis of digraphs. $\endgroup$ – Mark Dec 28 '16 at 23:39
  • 4
    $\begingroup$ @Mark: You do--but as of 1880, it's not at all clear anybody even had a clue of how to do that (in fact, available evidence tends to indicate that nobody knew how to start doing it). $\endgroup$ – Jerry Coffin Dec 29 '16 at 4:23
11
$\begingroup$

Build a Machine Cipher

A machine cypher, similar to what would become the Enigma machine, is not out of the realm of possibility in 1880, especially to a billionaire inventor.

Enigma relies on rotors moving in such a way as to form a varying electrical circuit. The varying electrical circuits provide the cipher text after the operator types the plain-text into a keyboard. The machine illuminates the cipher key for each character as it is typed.

The light bulb was invented in 1879 by Edison, the first commerical keyboard produced in 1870. Babbage's mechanical difference engine was first built in 1859. So the various components were present at the cutting edge of technology. It is not unreasonable that Edison, or your Herr Heinrich, or any of Jules Verne's polymaths could have created a personal enigma-style cipher.

While Enigma was not unbreakable, the computing resources to break it were generally not available in the 1880s. Heinrich' secrets would be safest with an encryption machine of his own design.

$\endgroup$
  • 2
    $\begingroup$ Indeed, you can make an enigma out of paper :) wiki.franklinheath.co.uk/index.php/Enigma/Paper_Enigma $\endgroup$ – Sobrique Dec 28 '16 at 14:08
  • 6
    $\begingroup$ @Sobrique Enigma's encryption by itself is just a polyalphabetic system, not much different from the Vigenere cipher mentioned in another answer. The point is that by mechanizing the encryption, you can encrypt more, faster, and better. That is the power of the this method compared to the others mentioned. $\endgroup$ – kingledion Dec 28 '16 at 14:18
  • $\begingroup$ How is the secret settings to be kept secure? $\endgroup$ – James Reinstate Monica Polk Dec 28 '16 at 16:34
  • 2
    $\begingroup$ @JamesKPolk Memorized by our genius-inventor? That seems like a very Jules Verne-ian answer. $\endgroup$ – kingledion Dec 28 '16 at 18:35
  • 1
    $\begingroup$ Yes, he could memorize them. $\endgroup$ – James Reinstate Monica Polk Dec 28 '16 at 20:55
5
$\begingroup$

The state of the art in detective fiction of the 19th century was the simple substitution cipher, as seen in Poe's The Gold-Bug (1843) and Doyle's The Adventure of the Dancing Men (1903). According to Wikipedia, the state of the art in real life at the time was the Vigenère cipher — le chiffre indéchiffrable — until it was in fact déchiffred circa mid-century.

Another cipher that would be mechanically plausible for the 19th century, but anachronistic in its level of sophistication, would be the Solitaire (a.k.a. Pontifex) cipher devised by Bruce Schneier and Neal Stephenson for Stephenson's novel Cryptonomicon. It uses a deck of playing cards to produce a stream of pseudorandom values, which can then be used as a non-repeating "keystream" in the same way as the Vigenère cipher uses the letters of its key word over and over.

The most effective way for your eccentric inventor to protect his enciphered message would be to make it hard for the attacker to tell what is the message and what isn't. For example, instead of encrypting "SECRETFORMULA" to "OOAUAIOIAUOUEAIUOIIIUAIEAA", he could encrypt it to "motor acquaint voilá hum you're radium soil lithium nailed array". This would be very time-consuming to encrypt, although I guess not too bad if you're good at free-associating words quickly. Your genius inventor might even be able to disguise the cipher as a plausible-sounding diary entry! In general, this is known as steganography, and state-of-the-art methods for the 19th century would have included acrostics and the Cardan grille. With either of those methods, especially the Cardan grille, the inventor could read and write essentially at full speed.

$\endgroup$
4
$\begingroup$

I don't think our great scientist would really spend one hour every day to decrypt yesterday's annotations, and probably even more time to encrypt today's.

This is a diary where he will be writing lots of things, and he will need some agility both writing and reading so it doesn't delay his investigations.

He would probably use just a Caesar cipher¹, additionally replacing some keywords with different words².

He will end up being quite proficient with caesar substitutions, so he needs very little effort to read and write his diary. He may prepare in the morning the key translating today's alphabet (or use a mechanical wheel that can be reused) and write directly using that as a hint, and wouldn't need to use an intermediate plaintext sheet at all for reading.

This doesn't preclude that he may use more complex encryption methods for really dangerous inventions, after he completes his experiments and they are no longer unrealistic fantasies.

¹ It may be enhanced by using a different offset (key) each day, eg. calculated performing some operations on the month, day of week and moon phase.

² Thus, a formula becomes nonsense without knowing what the actual elements are.

$\endgroup$
  • $\begingroup$ Also note that almost everyone overestimates the strength of his own ciphering strategies... $\endgroup$ – Ángel Dec 29 '16 at 23:41
2
$\begingroup$

One approach to making encryption much harder to crack, at the risk of having mistakes in encryption garble a message unrecoverably, is to use movable tiles to keep track of the encryption mapping and rearrange them during the encryption process. For example, when using the Playfair cipher, one could say that after processing each pair of letters, one would swap within the matrix the plaintext version of the first character and the ciphertext version of the second. Digraph frequency analysis would generally go out the window, since the scrambling method would split up which letters shared rows and columns with other letters. The effort required to keep track of the continuously-changing key would not be unreasonable, but a character would need to be very careful to avoid making mistakes.

$\endgroup$
2
$\begingroup$

As far as mathematical options go, the best ones I know have already by suggested. I will suggest a different option, used by the Spartans. The Scytale. Basically you wind your medium around a rod and write along the rod. When the medium is unwound you get nothing but shuffled letters. The description in Wikipedia makes this most basic implementation appear easily breakable. While I am no expert, I suppose a few twists could be added:

Don't align letters with stripes.

If you always place your letters on the stripe perfectly, you have nothing but reshuffled letters. This is easily breakable, by statistical analysis of the letter frequency. Instead use very, very thin stripes, so that mere dots are found on each, offering no clue to which letter they belong. This might be difficult to reproduce by hand, but Heinrich will let a machine do the winding for him. Very neatly and reproducibly.

Don't use a simple rod

In the event that the stripes are found, the cryptanalyst will have the patience to test all kinds of rod diameters. Therefore, don't use a rod of fixed diameter, but rather a complicated shape, similar to what you get by spinning an elaborate key around the long axis, or the ones found here.

Disguise the data as everyday objects

Unfortunately, I do not know the shape and size of 19th century sewing spools, so I will take today's situation as a guide. If I had enough trusted technicians, I would use sewing yarn, or something resembling it as a medium. I would manufacture a machine typing on the yarn, while it is of course wound around the intricately shaped rod. It would type in tiny letters, wind and unwind the yarn for me and reading would be easy with a magnifying glass. When storing the data, I would wind it on an ordinary spool. I would take care to leave some yarn blank, so the outside of the yarn (when on an ordinary spool) would be blank and inauspicious.

If my typing/printing technique is good enough, I can even hide my yarn inside Gobelins, carpets, clothes, for longer travel.

Problems

Yarn has an axis of rotation. To avoid that becoming a problem, you must have the ink seep through the yarn. That might be difficult to achieve uniformly. And you must leave blank layers, to avoid damaging layers below.

$\endgroup$
  • 1
    $\begingroup$ Yarn can stretch. When you go to read it the lines won't line up. $\endgroup$ – Rick Ryker Dec 29 '16 at 23:21
  • 1
    $\begingroup$ Great point. I tried it with different yarns here. They snapped but didn't stretch visibly. Maximum 1cm/m. Do you mean sewing yarn or knitting yarn? $\endgroup$ – Ludi Dec 30 '16 at 11:29
1
$\begingroup$

Solitaire cipher is strong (modern) cipher that is hard to break even using computers. However @o.m. answer would be more possible as Vinegre cipher was already known in 19th century.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.