Homophonic Cipher
Understand how one plaintext letter can map to several number symbols while every number still identifies only one plaintext letter.
What is the Homophonic Cipher?
A homophonic cipher gives common plaintext letters multiple possible ciphertext symbols. For example E might encrypt as any of several numbers, while a rare letter might have only one.
The important one-way rule is that each ciphertext number decodes to only one plaintext letter. This spreads plaintext frequency across several symbols and makes simple frequency analysis less obvious.
The cipher mechanics on this page are self-contained. Science Olympiad event formats, allowed variants, and tournament constraints can change by season; the current official Rules Manual and official clarifications take precedence.
No outside reference is assumed. Work through Foundations → Complete Reference → Encryption → Decryption in order, then use the competition and cryptanalysis sections.
What you need to know
- One plaintext letter may own multiple ciphertext tokens.
- One ciphertext token must not decode to multiple plaintext letters in the same mapping.
- Encryption may choose randomly among a letter’s allowed tokens; decryption is deterministic once the mapping is known.
Beginner glossary
| Term | Meaning |
|---|---|
| Homophone | One of several ciphertext symbols that can represent the same plaintext letter. |
| Token | A ciphertext symbol/number such as 04 or 71. |
| Reverse mapping | A token→plaintext table used for deterministic decoding. |
| Frequency flattening | Spreading common letters over multiple symbols so no single token is as frequent as the original letter. |
Keep this beside you while solving
| Plaintext | Possible ciphertext tokens (illustrative) |
|---|---|
| E | 04, 10, 20, 28, 41, … |
| T | 11, 14, 17, 44, … |
| Q | 24 |
| Z | one/few tokens |
The mapping is many-to-one only in the ciphertext→plaintext view
Several tokens may decode to E, but a token such as 20 must not also decode to another plaintext letter in the same mapping.
How encryption works
Find the plaintext letter
Example: E.
Look at its token set
E may have several assigned numbers.
Choose one allowed token
The encoder can vary the choice each time E appears.
Repeat
Different occurrences of the same plaintext letter may therefore look different in ciphertext.
How decryption works
Use the number→letter mapping
Each token belongs to exactly one plaintext letter.
Replace every token
Different numbers may collapse to the same plaintext letter.
Restore spacing if indicated
Token separators make boundaries unambiguous at the symbol level.
Read the result
Check that the full plaintext is coherent.
How to approach it in Codebusters practice
- Do not assume repeated plaintext letters produce repeated ciphertext tokens.
- For cryptanalysis, aggregate symbols that appear to behave like homophones rather than treating every number as an independent simple-substitution letter.
- Word-pattern clues may be weakened because equal plaintext letters can have different ciphertext tokens.
What the problem gives you vs. what you produce
| Part | What to expect |
|---|---|
| You may be given | Numeric/symbol ciphertext and possibly a crib or partial mapping. |
| You must find | The plaintext and/or complete mapping. |
| Fastest first move | Build a reverse token→letter table for every proven mapping and never assign one token twice. |
How to attack an unknown or partially known key
- Individual token frequency is deliberately flattened compared with ordinary substitution.
- Context, crib placement, token distributions, and the restriction that each token has one plaintext meaning become important.
- Once two tokens are proven to represent the same plaintext letter, combine their counts when reasoning about language frequency.
Follow one example from start to finish
E → {10, 20, 28}; T → {11, 17}
T E E
17 20 10
Before moving on, make sure you can answer these without another site:
- Can two different numbers both mean E? Yes. Can one number mean both E and T? No.
- Can you explain why repeated plaintext letters might not repeat in ciphertext?
- Can you combine counts of proven E-homophones when estimating E frequency?
Common mistakes
Assuming equal plaintext letters must have equal ciphertext tokens.
Allowing one number to represent two plaintext letters.
Using ordinary one-symbol-per-letter frequency analysis without combining suspected homophones.
Ignoring token boundaries.
Competition speed strategies
Build a reverse number→letter table as soon as mappings are known.
Group suspected homophones by plaintext letter rather than by numeric value.
Use cribs/context to establish high-confidence mappings before doing frequency estimates.
What to remember under time pressure
See every transformation
Use the lab to change inputs and keys, keep the relevant reference material visible, inspect each intermediate transformation, and then read the “How to reverse it” panel so encryption and decryption connect.