Make a Secret Code, Then Test It Like a Scientist

Learners invent their own cipher, swap messages, and then meet the hard question every Voynich theory faces: how would you know if you were right?

The big idea

For 100 years, people have announced they “cracked” the Voynich Manuscript — and every time, other researchers show the crack doesn’t hold up. Why is it so hard to know? This activity puts learners inside that puzzle.

Part 1 — Build a cipher (20 min)

Each pair invents a substitution cipher: pick a symbol or letter to stand for each real letter. Write your key on a hidden card.
Encode a short message (one sentence).
Swap encoded messages with another pair — but not the key. Can they read it? (Usually not, without the key — that’s the point of a code.)

Part 2 — The hard question (15 min)

Now the twist. Hand each pair a made-up “solution” to another team’s message — a reading that sounds plausible but is wrong. Ask:

Does it look like it could be right?
How could you actually prove it’s the real message and not just a good guess?

Lead them to the answer: you need the key, or an independent check that couldn’t happen by luck. Without one, “it reads like a sentence” isn’t enough — a clever guess can also read like a sentence.

Part 3 — Connect to the real manuscript (10 min)

The Voynich Manuscript has no key and no one to check against. So even a reading that looks like real words could be a coincidence. That’s why this project’s rule is: matching the manuscript’s patterns is necessary but not sufficient — it means an idea isn’t ruled out, never that it’s proven.

The takeaway

A code that “looks solved” isn’t solved. Real evidence has to be something that couldn’t easily happen by chance. This is exactly the discipline scientists use on the manuscript.

Extend it (computer)

Type your encoded message into a file and run python -m ms408 yourfile.txt (needs ~1000 words — combine the class’s messages). See which of the manuscript’s “fingerprints” your cipher matches. Discuss: matching some of them — does that mean your code is Voynichese? (No — necessary, not sufficient!)

Contributed by MS408 project. Have you run this with learners? Improvements welcome via the submission page.

Materials	Paper and pencils; A simple cipher key template (letter → symbol); Optional: a computer for the extension
Learning goals	Build and use a substitution cipher Experience why a code can look convincing but still be unverifiable Meet 'necessary but not sufficient' — the core idea of the project