I'm writing a story set partially in historic China, and it involves a network of secret agents who seek to communicate over the mail in innocuous-looking letters. So I need a description of a pen-and-paper cipher that works for the Chinese language and (traditional) writing system. The story is not ultimately about the language or the cipher, it is just an element that I want to describe accurately.

For the plot I need to use a null cipher; a way to conceal the secret message in a large amount of regular text by interspersing the secret stuff with lots of random stuff that together looks like it is about something different entirely.

In alphabetical scripts that would work by taking every n'th word, or every n'th letter, and putting them behind each other. This example from Wikipedia:


Taking every 5th word:


I'm considering, for this story, to say that this particular cipher mixed and matched components. That is to say, if the agent wanted to encrypt "steal", 偷; he would find another character starting with 人 (e.g. 仔) and another character ending with 俞 (e.g. 瑜), and write some text between to make it flow as a sentence. Then he'd just need to do so in a regular manner; let's say picking the first component of every 3rd character, then the second component of every 5th character, and maybe some extra cases for characters with even more components. Another agent who reads the text can read it selectively and get the secret message.

That is the level of explanation I'd use in the story; so the question is basically whether that, on the surface, looks feasible. I'm not going to write any actual ciphered text. I just don't want to do disrespect to any speaker of Chinese who comes across the story when I eventually put it online.

Rather than invent my own, I would have used an actual historic cipher, but they seem hard to find. The only one I could find attested is a grille cipher. That too is a null cipher, but it operates on the level of whole characters; and I presume that hiding the entire word "steal" in a text would be more suspicious than just hiding its components. Plus using components I can use it in a scene where someone explains the Chinese script itself, for a nice little educational bonus.

  • 1
    藏頭詩 or 嵌字文 are common in Chinese.
    – Nobody
    Oct 1, 2023 at 12:43
  • @Nobody Very helpful, thank you very much!
    – KeizerHarm
    Oct 1, 2023 at 13:06
  • That's a rather obvious way to hide a message, though. Therefore it's not secure. Oct 2, 2023 at 8:36
  • 1
    Okay, why would a very rich merchant be a spy (presumably working for a foreign country)? What's in it for him? Oct 3, 2023 at 3:21
  • 1
    @KeizerHarm Its totally up to you what to write, I only mention this because you seem interested in historical accuracy: merchants were viewed very poorly in old china, like bums leeching off society. Only slaves and performers//brothels were viewed as lower, and regular working class and farmers would have been more respected in the general eye. Just keep in mind when writing a rich merchant role back then, its a tricky thing to portray casually as a spy, a person that already felt suspicious and hated, you might need a hiding in plain sight plot to be accurate. :)
    – zagrycha
    Mar 2 at 17:52

2 Answers 2


try “拆字”, or “離合”, the traditional methods to “cipher” characters

What words exist that come from deconstructed characters?

a radical-based null cipher be possible in Chinese?

no. btw, it should be “component (部件)-based”

involves a network of secret agents who seek to communicate over the mail

well, . . . a rant, or query here: the mail delivery “system” in the past is, quite slow. it may take weeks, or months to send a mail. no organisation would survive with such communication speed.

have fun :)

  • 1
    Well, there was "飛鴿傳書".
    – Nobody
    Oct 1, 2023 at 14:03
  • @Nobody, plus 魚雁 😹 Oct 1, 2023 at 14:12
  • 1
    Also, getting stopped at a border of a territory to enter another , just to deliver your mail could get quite interesting over time haha.
    – zagrycha
    Oct 1, 2023 at 22:59
  • I've always called components 偏旁, so I'm a bit surprised by 部件 here. Are they both the same thing?
    – Becky 李蓓
    Oct 3, 2023 at 6:11
  • @Becky李蓓, “Are they both the same thing?” no radical (部首) normally refers to 214 radicals used in 康熙字典. or, 540 radicals used in 說文解字. radical should be used for indexing only Oct 3, 2023 at 6:47

There are several issues that I can think of. Take your example of 偷. How would the recipient know that the character 仔 is the one to start with, and not any other ones? How would he tell that he ought to end with 瑜 and not some other character?

I think a grille cipher is more doable.

  • Counting. The algorithm could be simply to combine the first component of characters 1, 6, 11... with the second component of characters 3, 8, 13... it also can be much more complicated; special signs could change the order, maybe one needs to read the page diagonally, etc
    – KeizerHarm
    Oct 3, 2023 at 6:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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