Tironian Notes

IPhenomenon

A Roman scribe taking down a speech in the Senate did not write SENATUS CONSULTO. He wrote two marks. The system assigned a sign to a whole concept, and the sign had no phonetic relationship to the spoken syllables it stood for.

This bench writes notae as bracketed stems: [ET], [CON], [EST]. The brackets are a modern convention — the original marks were single strokes — but the structure is faithful: a small lexicon of high-frequency stems, plus ordinary letters for the remainder.

The consequence is a frequency effect running the other way. In ordinary text the letter E is most common; in notae the sign for ET is most common, and its frequency is higher than any letter's. A cryptanalyst who assumes letters will misread the whole document.

IIKey space

There is no key and no algorithm. Security is entirely a property of the lexicon: if you know the notae you can read the document, and if you do not, you are looking at scratches.

That makes the effective key the dictionary, and dictionaries leak. A scribe's training manual, a reused document, a glossary in a monastery library — any of these converts the whole system from unreadable to readable in an afternoon.

Note the trade-off clearly, because it recurs constantly. Notae buy speed and pay in obscurity. Obscurity is not a key and cannot be rotated.

IIIEnciphering

• Agree the lexicon. In the Roman system this was a taught body of signs; in the bench it is a fixed list of high-frequency stems.
• Write the message word by word.
• If a word's first three letters appear in the lexicon, replace that stem with its nota and write the remaining letters in ordinary script.
• If the word is not in the lexicon, write it in ordinary letters. The bench marks these with [·] so the substitution is visible.
• Keep word spacing. Shorthand is meant to be read back quickly, and word boundaries are part of the speed.
• Expand nothing. A scribe writing at speed leaves the expansion to the reader.

IVDeciphering

• Read the document and mark every bracketed stem or, in the original, every non-alphabetic stroke.
• Expand each nota to its full word using the lexicon.
• Read the remaining ordinary letters in sequence and join them to the expansions.
• Where a stem is ambiguous — CON can begin consul, contra or concedo — resolve from the sense of the passage, not from the marks.
• Expect to be slow the first time. Shorthand is designed for a reader who already knows the text's subject matter.

VWeakness

The lexicon is the whole secret. Every copy of it that circulates is a total compromise, and no rekeying is possible without changing the writing system itself.

The symbol set is large and public. Several thousand signs is far too many to memorise and far too many to hide. The system was taught openly and used by clerks who had no interest in secrecy.

Frequency analysis adapted, and then worked. Medieval scholars analysing notae found that even when words are the symbols, the most frequent symbols still correspond to the most frequent words, and word-frequency lists are as effective as letter-frequency lists.

It was replaced by something strictly better. Ordinary letters with a few abbreviations won, because a small alphabet can be learned by anyone and still written quickly with practice.

VIPractice

·History note

The system is attributed to Marcus Tullius Tiro, Cicero's secretary and freedman, who is said to have devised it around 63 BC to record speeches in the Senate. It grew over the following centuries into a lexicon of several thousand signs and remained in use in chancery and monastic documents into the Middle Ages; the surviving examples are concentrated in ninth- and tenth-century manuscripts, where copyists used notae to fit more text into expensive parchment. Charlemagne's court promoted it, and it was finally displaced by ordinary abbreviation around the twelfth century.

·Further reading

The modern descendants of the idea are everywhere: & (a surviving Tironian ET), the ampersand's sibling ⁊ still printed in older Irish texts, medical abbreviations, and every standards body's controlled vocabulary. Each is a small learned lexicon that speeds up the people who share it and locks out everyone else.

The cryptographic point stands unchanged: a private vocabulary is not a key. If your scheme's security rests on an attacker not having read a document, your scheme has no security parameter at all.

Next on the loom: nomenclator-cipher — the same idea, but with the lexicon kept deliberately secret and rotated per correspondent.

+Weave into