At a glanceCiphertext-only break · complete
- The telegrams
- Four telegrams from Tokyo to Shanghai, 3 to 13 June 1916 (N05, N13, N16, N26), 239 four-digit groups: Huang Xing on the North–South settlement after Yuan Shikai’s death, and Zhang Xiaozhun on a small payment.
- Where
- Diplomatic Archives of the Ministry of Foreign Affairs of Japan, via JACAR: B03050089200, B03050089300 and B03050089400. Telegraph-office forms copied by the police.
- The cipher
- Chinese telegraph code numbers in groups of four digits. The first two groups of a form are in clear and name the code book: 6591 1378 = 辰密. N05 has no prefix.
- How it was deciphered
- The key was found by annealing, with no crib, on three telegrams; the fourth reads with it unchanged and the day code at the end of N16 matches the date on its form.
- Credit
- Brice Dudout, with Claude (mainly Opus 5.5): digits re-read on the scans, the key (published 19 October 2026), translations and context. Satoshi Tomokiyo, Cryptiana: the identification of the key as a renumbered 明密 code book; his section credits the formula to Dudout. This project checked the key with its own decoder.
- What it says
- Huang Xing declines to word a joint telegram to Li Yuanhong and the provinces and asks the Shanghai friends to take a firm line (N13), and Zhang Xiaozhun suggests a reliable division for the new president (N16).
- Still open
- Seven characters are probable readings resting on one digit or on the sense. The key is a reconstruction: no copy of the 辰密 book has been found.
01 The forms
When Yuan Shikai died on 6 June 1916, Huang Xing was in Tokyo, living at Aoyama Minami-machi 3-53, and his allies were in Shanghai. The Japanese police copied the telegrams that passed between them onto the telegraph office’s own forms, and the copies survive in a Foreign Ministry series of mail and telegrams of watched persons. Each form holds a grid of digits written as kanji numerals. Brice Dudout reads the grid in columns, right to left, each column from the top, and cuts the stream into four-digit groups.
The first two groups are in clear and name the code. Hsueh Chun-tu’s edition of Huang Xing’s 1916 drafts notes a code name at the head of each telegram, one per correspondent. On that evidence 辰密 is a code book issued to particular correspondents, and it needed its own key. The same files hold telegrams in two other numeric keys with their own pages: 容密 and 共密.
| telegram | N05 (no prefix) | N13 | N16 | N26 | all |
|---|---|---|---|---|---|
| date, 1916 | 3 June | 9 June | 10 June | 13 June ? | |
| groups | 38 | 88 | 104 | 9 | 239 |
Group counts measured on the transcription. Of the 239, 232 read without doubt and 7 are probable readings (Daniel’s rule, counted on reading.md).
02 The 辰密 key
This key adds nothing to the number: it is the telegraph code itself, reordered. The code is laid out in pages of a hundred characters. In 辰密 the first two digits give the page, counted backwards over the numbers that contain no zero: 99 is page 00, 98 page 01, down to 91 = page 08, then 89 = page 09 (90 is skipped), and so on to 11 = page 80. The tens and units are permuted:
| digit | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
|---|---|---|---|---|---|---|---|---|---|---|
| tens | 3 | 9 | 8 | 7 | 1 | 0 | 2 | 6 | 4 | 5 |
| units | 1 | 9 | 5 | 4 | 3 | 2 | 6 | 7 | 8 | 10 |
Example. 9950 4220 6219. In 9950, page 99 is 00, tens 5 is 0, units 0 is 1: 0001, 一. In 4220, page 42 is 52, tens 2 is 8, units 0 is 1: 5281, 興 (Huang Xing’s signature). In 6219, page 62 is 34, tens 1 is 9, and units 9 counts as 10, carrying into the tens: 3400 + 90 + 10 = 3500, 灰, the day code for the 10th.
How it was found. No 辰密 telegram had a known plaintext, and the 共密 and 容密 keys did not fit. Two features pointed to a reordered book: no 0 in the first two places, and a quarter of the text on three “pages” (99, 96, 93). Annealing a model with permuted pages, tens and units on three telegrams (N13, N16, N26: 201 groups) gave readable text, scored with a language model of telegraph-code characters. The page table it found followed the zero-free countdown, which was then imposed. Two wrong characters, 劈 for 力 and 滠 for 灰, both from groups ending in 9, gave the rule that 9 stands for 10.
As a renumbered book. Satoshi Tomokiyo, in the Cryptiana section on the mingmi (明密) format codes, reads this key, and the 共密 and 容密 keys, not as arithmetic but as one standard 明密 code book whose pages, columns and rows were rewritten by hand, as in a June 1916 edition he shows. In his picture the tens digit is a column number and units 9 stands for the tenth row, which is why xxx0 sits at the foot of the previous column. The two descriptions give the same table; the repository file key.md states the key both ways.
Dating. Dudout published the 辰密 page, key and four telegrams on 19 October 2026 (GitHub commit of the Wenmi repository, 15:15 UTC). Tomokiyo’s section is undated, the page was last modified on 24 October 2026, and Wenmi cites it as of 23 October. The section credits the formulas to Dudout (“He discovered the pages are numbered backwards, skipping any number containing a zero”) and adds the renumbered-book reading.
03 The checks
His checks. N05 was left out of the search and has no prefix, yet it decodes straight through. N16 ends with 灰, the 10th, the date on its header. The names that come out belong to people holding the right offices in June 1916 (Zhao Hengti, Tang Hualong). Every digit was re-read blind on the scans by Brice Dudout, without seeing the decipherment.
The check made here (20 October 2026). His verify.py reproduces the published raw texts from his digits and scores the 辰密 text at 3.77 against 200 random keys of the same form (best −1.59) and 200 shufflings of the digits (best −1.18). A decoder written here from the key statement alone, using the Unihan telegraph-code table already in this repository, gives the same character as his published text for all 239 groups. The 辰密 search itself was not rerun here. This confirms that the key as stated turns his digits into his text; it does not re-read the scans. Only the first dozen digits of the N16 form shown above were compared by eye, and they match.
04 What they say
Hsueh Chun-tu’s edition of Huang Xing’s 1916 drafts has nothing between the pieces of 8 and 20 June, and these telegrams are mostly dated in between. Translations below are Dudout’s, shortened.
On 3 June (N05, no prefix) Huang Xing tells Liewu that if the Shandong side wants him to take charge of unifying matters he may accept directly, and that Puquan (Zhang Ji) should go with him. On 9 June (N13) he writes to three friends in Shanghai that the telegram to Li Yuanhong, Cen Chunxuan and the provinces is very hard to word: if the positions are not unanimous the party risks looking divided, and if it writes as their telegram says people will see it as opportunism. “Japan’s attitude is very much wait-and-see” (日態度頗觀望); the friends in Shanghai should take a firm stand.
On 10 June Zhang Xiaozhun asks the same group to give the new president troops he can rely on (N16, shown above): Zhao Hengti, “always trusted by Li”, then training troops under Governor Tang, should form an elite division at his disposal, and arms “should be possible”. The last telegram (N26, 13 June?) is a single line: three hundred yuan to Baili, signed 準.
05 What is open
- Seven characters are probable readings, four resting on a single digit and the others on the sense: 一不 in N13 is probably a slip for 不一, and 盻 stands for 盼. Nothing is left unresolved.
- The key is a reconstruction: no historical copy of the 辰密 book has been found. The 1916 edition of the telegraph code is not identified.
- The texts have not been found in the 黃興集 or in Li Yuanhong’s correspondence, which are still to be searched. The translations are proposals.
Full texts
The transcription and the decryption of every document on this page, so that the reading can be used and checked. Short files are shown here; all are in the repository.
辰密 (Chenmi), 4 telegrams, 3 June - 13 June 1916
Four telegrams from Tokyo to the Hunanese group's address in Shanghai (ヒツキン, 277 Avenue Joffre): Shandong and Bo Wenwei; the joint telegram to Li Yuanhong, Cen Chunxuan and the provinces; an elite division under Zhao Hengti for Li Yuanhong; 300 yuan for Jiang Fangzhen.
Transcription
# 辰密 (Chenmi), June-July 1916: cipher groups as re-read by Brice Dudout on the JACAR police copies. # The forms carry the digits in kanji numerals in a grid, read in columns right to left, top to bottom, sheet after sheet, # cut into groups of four. One telegram per line: label, then its groups. The clear prefix 6591 1378 (= 辰密) is left out. # "?" = illegible digit. Sheets and dates: decryption.txt. N05 2777 6162 6736 9608 2254 8435 7101 7392 1909 8735 6788 9608 5620 4667 9950 9997 9847 9378 5499 7659 3231 3294 6355 6564 9794 9527 6733 2202 5620 5487 9519 7576 9453 8735 9881 5430 3359 4220 N13 8449 3255 1522 9952 9608 2254 4278 1345 8438 9383 9337 5473 2114 1863 2120 3840 4920 4123 9900 8168 9950 9958 7998 7221 6738 2912 9907 8634 4127 8735 9830 2114 9970 9970 9926 7950 3348 5903 9388 8373 3294 9388 3104 8988 2740 9320 7885 5692 8442 3659 7475 5134 7684 5511 5903 5475 7339 7876 8294 1863 3303 7218 6307 9320 9926 8592 3454 5292 8169 5321 9900 8168 2740 9464 2158 8461 1845 7313 5450 2655 9697 7339 3101 5915 9987 2114 5292 4220 N16 9947 2777 4651 1522 8449 3232 9670 2254 1345 9670 1906 6413 9730 9950 9991 8317 9378 1907 9672 9519 7418 9851 7998 9958 7316 8512 3527 2915 8708 9425 4729 5903 1345 9670 7736 9757 9700 9914 9385 8970 6425 5356 9957 4626 9672 9378 9313 4239 6283 6736 4924 9213 2495 3339 6567 9749 9676 2718 4626 4707 9672 9950 8317 9859 9730 1345 9670 1647 2639 6736 9115 9950 8470 4989 9696 9519 8996 9907 5620 9378 2726 9590 7256 9683 2114 3294 9388 3104 8988 2740 9320 7885 9858 7475 5134 5511 2114 9950 4403 7218 2684 3527 6413 6219 N26 9986 9859 9952 5455 9607 9933 5455 2481 6413
Decryption
# 辰密 (Chenmi): decryption, group by group # cipher group -> CTC number under key C92 (key.md) -> character; status as on the Wenmi site: # ok; douteux = doubtful (one-digit uncertainty on the copy, or an anomalous code read as the sense requires); illisible = illegible digit ## N05 | 3 June 1916 | Tokyo → Shanghai | JACAR B03050089200, PDF pp. 25R, 25L, 26R, 26L | no prefix 1 2777 6567 轉 ok 2 6162 3525 烈 ok 3 6736 2976 武 ok 4 9608 0338 兄 ok 5 2254 7003 鑒 ok 6 8435 1472 山 ok 7 7101 2639 東 ok 8 7392 2455 方 ok 9 1909 7240 面 ok 10 8735 1172 如 ok 11 6788 2948 欲 ok 12 9608 0338 兄 ok 13 5620 3981 當 ok 14 4667 4827 統 ok 15 9950 0001 一 ok 16 9997 0057 事 ok 17 9847 0117 任 ok 18 9378 0668 可 ok 19 5499 4160 直 ok 20 7659 2110 承 ok 21 3231 6179 諾 ok 22 3294 6153 請 ok 23 6355 3302 溥 ok 24 6564 3123 泉 ok 25 9794 0253 偕 ok 26 9527 0487 劃 douteux 27 6733 2974 此 ok 28 2202 7035 間 ok 29 5620 3981 當 ok 30 5487 4147 盡 ok 31 9519 0500 力 ok 32 7576 2266 援 ok 33 9453 0504 助 ok 34 8735 1172 如 ok 35 9881 0149 何 ok 36 5430 4171 盻 douteux 37 3359 6010 覆 ok 38 4220 5281 興 ok ## N13 | 9 June 1916 | Tokyo → Shanghai | JACAR B03050089300, PDF pp. 3L, 4R, 4L, 5R, 5L, 6R, 6L, 7R 1 8449 1420 小 ok 2 3255 6102 詠 ok 3 1522 7685 鳳 ok 4 9952 0005 三 ok 5 9608 0338 兄 ok 6 2254 7003 鑒 ok 7 4278 5268 致 ok 8 1345 7812 黎 ok 9 8438 1478 岑 ok 10 9383 0644 及 ok 11 9337 0677 各 ok 12 5473 4164 省 ok 13 2114 7193 電 ok 14 1863 7324 頗 ok 15 2120 7181 難 ok 16 3840 5511 著 ok 17 4920 4581 筆 ok 18 4123 5384 苟 ok 19 9900 0031 主 ok 20 8168 1728 張 ok 21 9950 0001 一 douteux 22 9958 0008 不 douteux 23 7998 1858 恐 ok 24 7221 2589 有 ok 25 6738 2978 歧 ok 26 2912 6395 趨 ok 27 9907 0037 之 ok 28 8634 1273 嫌 ok 29 4127 5387 若 ok 30 8735 1172 如 ok 31 9830 0171 來 ok 32 2114 7193 電 ok 33 9970 0061 云 ok 34 9970 0061 云 ok 35 9926 0086 人 ok 36 7950 1801 必 ok 37 3348 6018 視 ok 38 5903 3634 爲 ok 39 9388 0648 取 ok 40 8373 1564 巧 ok 41 3294 6153 請 ok 42 9388 0648 取 ok 43 3104 6233 護 ok 44 8988 0948 國 ok 45 2740 6511 軍 ok 46 9320 0681 同 ok 47 7885 1942 意 ok 48 5692 3955 畎 douteux 49 8442 1415 尊 ok 50 3659 5710 處 ok 51 7475 2362 擬 ok 52 5134 4473 稿 ok 53 7684 2143 拍 ok 54 5511 4099 發 ok 55 5903 3634 爲 ok 56 5475 4162 盼 ok 57 7339 2480 日 ok 58 7876 1966 態 ok 59 8294 1653 度 ok 60 1863 7324 頗 ok 61 3303 6034 觀 ok 62 7218 2598 望 ok 63 6307 3337 滬 ok 64 9320 0681 同 ok 65 9926 0086 人 ok 66 8592 1355 宜 ok 67 3454 5903 表 ok 68 5292 4355 示 ok 69 8169 1730 強 ok 70 5321 4289 硬 ok 71 9900 0031 主 ok 72 8168 1728 張 ok 73 2740 6511 軍 ok 74 9464 0523 務 ok 75 2158 7108 院 ok 76 8461 1429 尤 ok 77 1845 7312 須 ok 78 7313 2494 明 ok 79 5450 4101 白 ok 80 2655 6602 近 ok 81 9697 0357 兩 ok 82 7339 2480 日 ok 83 3101 6239 變 ok 84 5915 3692 狀 ok 85 9987 0047 乞 ok 86 2114 7193 電 ok 87 5292 4355 示 ok 88 4220 5281 興 ok ## N16 | 10 June 1916 | Tokyo → Shanghai | JACAR B03050089300, PDF pp. 11L, 12R, 12L, 13R, 13L, 14R, 14L, 15R, 15L 1 9947 0017 並 ok 2 2777 6567 轉 ok 3 4651 4809 組 ok 4 1522 7685 鳳 ok 5 8449 1420 小 ok 6 3232 6175 諸 ok 7 9670 0361 公 ok 8 2254 7003 鑒 ok 9 1345 7812 黎 ok 10 9670 0361 公 ok 11 1906 7236 非 ok 12 6413 3294 準 ok 13 9730 0271 備 ok 14 9950 0001 一 ok 15 9991 0059 二 ok 16 8317 1597 師 ok 17 9378 0668 可 ok 18 1907 7237 靠 ok 19 9672 0365 兵 ok 20 9519 0500 力 ok 21 7418 2398 政 ok 22 9851 0109 令 ok 23 7998 1858 恐 ok 24 9958 0008 不 ok 25 7316 2496 易 ok 26 8512 1395 實 ok 27 3527 5887 行 ok 28 2915 6392 趙 ok 29 8708 1138 夷 ok 30 9425 0582 午 ok 31 4729 4790 素 ok 32 5903 3634 爲 ok 33 1345 7812 黎 ok 34 9670 0361 公 ok 35 7736 2076 所 ok 36 9757 0207 信 ok 37 9700 0231 倚 ok 38 9914 0093 今 ok 39 9385 0642 又 ok 40 8970 0961 在 ok 41 6425 3282 湯 ok 42 5356 4206 督 ok 43 9957 0007 下 ok 44 4626 4886 練 ok 45 9672 0365 兵 ok 46 9378 0668 可 ok 47 9313 0694 否 ok 48 4239 5280 與 ok 49 6283 3444 濟 ok 50 6736 2976 武 ok 51 4924 4583 等 ok 52 9213 0794 商 ok 53 2495 6852 量 ok 54 3339 6080 設 ok 55 6567 3127 法 ok 56 9749 0220 俾 ok 57 9676 0366 其 ok 58 2718 6598 迅 ok 59 4626 4886 練 ok 60 4707 4737 精 ok 61 9672 0365 兵 ok 62 9950 0001 一 ok 63 8317 1597 師 ok 64 9859 0110 以 ok 65 9730 0271 備 ok 66 1345 7812 黎 ok 67 9670 0361 公 ok 68 1647 7517 驅 ok 69 2639 6680 遣 ok 70 6736 2976 武 ok 71 9115 0892 器 ok 72 9950 0001 一 ok 73 8470 1461 層 ok 74 4989 4550 竭 ok 75 9696 0356 全 ok 76 9519 0500 力 ok 77 8996 0956 圖 ok 78 9907 0037 之 ok 79 5620 3981 當 ok 80 9378 0668 可 ok 81 2726 6586 辦 ok 82 9590 0451 到 ok 83 7256 2506 昨 ok 84 9683 0344 克 ok 85 2114 7193 電 ok 86 3294 6153 請 ok 87 9388 0648 取 ok 88 3104 6233 護 ok 89 8988 0948 國 ok 90 2740 6511 軍 ok 91 9320 0681 同 ok 92 7885 1942 意 ok 93 9858 0108 代 ok 94 7475 2362 擬 ok 95 5134 4473 稿 ok 96 5511 4099 發 ok 97 2114 7193 電 ok 98 9950 0001 一 ok 99 4403 5034 羶 douteux 100 7218 2598 望 ok 101 2684 6643 速 ok 102 3527 5887 行 ok 103 6413 3294 準 ok 104 6219 3500 灰 ok ## N26 | 13 June 1916 ? | Tokyo → Shanghai | JACAR B03050089400, PDF pp. 31R 1 9986 0046 九 douteux 2 9859 0110 以 ok 3 9952 0005 三 ok 4 5455 4102 百 ok 5 9607 0337 元 ok 6 9933 0074 交 ok 7 5455 4102 百 ok 8 2481 6849 里 ok 9 6413 3294 準 ok
Reading
# 辰密 (Chenmi): reading
Chinese text with punctuation, as published in the Wenmi edition (Brice Dudout with Claude). {x}: a probable reading replacing an
anomalous decoded character (a one-digit variant of the group or the context; the decryption keeps the figure's
own value). 〔x〕: the decoded character kept, not understood. □ / 〔…〕: illegible group. Translations in translation.md.
## N05 — 3 June 1916, Tokyo → Shanghai
轉烈武兄鑒：山東方面如欲兄當統一事任，可直承諾；請溥泉偕{行}。此間當盡力援助。如何，{盼}覆。興。
## N13 — 9 June 1916, Tokyo → Shanghai
小、詠、鳳三兄鑒：致黎、岑及各省電頗難著筆。苟主張{不}{一}，恐有歧趨之嫌；若如來電云云，人必視爲取巧。請取護國軍同意，{由}尊處擬稿拍發爲盼。日態度頗觀望，滬同人宜表示強硬主張，軍務院尤須明白。近兩日變狀，乞電示。興。
## N16 — 10 June 1916, Tokyo → Shanghai
並轉組、鳳、小諸公鑒：黎公非準備一二師可靠兵力，政令恐不易實行。趙夷午素爲黎公所信倚，今又在湯督下練兵，可否與濟武等商量設法，俾其迅練精兵一師，以備黎公驅遣。武器一層，竭全力圖之，當可辦到。昨克電請取護國軍同意，代擬稿發電，一{節}望速行。準。灰。
## N26 — 13 June 1916 ?, Tokyo → Shanghai
{乞}以三百元交百里。準。Translation
# 辰密 (Chenmi): English translation Working translations from the Wenmi page (https://bdudout.github.io/wenmi/en/chenmi.html, Brice Dudout with Claude, CC BY 4.0). Identifications of people are candidates; uncertain passages are marked as on the site. ## N05 — 3 June 1916 Please pass on to brother Liewu: if, on the Shandong side, they want you to take charge of unifying matters, you may accept directly; ask Puquan (Zhang Ji) to go with you. From here we will help with all our strength. What do you think? Awaiting your reply. Xing. Note: Read as a blind check: not used to find the key. Digits re-read by Brice Dudout. ## N13 — 9 June 1916 To the three brothers Xiao, Yong and Feng: the telegram to Li [Yuanhong], Cen [Chunxuan] and the provinces is very hard to word. If our positions are not unanimous, we risk looking divided; if we write as your telegram says, people will see it as opportunism. Please obtain the agreement of the National Protection Army and have the draft written and sent from your side. Japan’s attitude is very much wait-and-see; the comrades in Shanghai should take a firm stand, and the Military Council above all must be clear. Please cable me how things have changed over the last two days. Xing. Note: Digits re-read by Brice Dudout. ## N16 — 10 June 1916 Please also pass on to Zu, Feng, Xiao and the others: unless Li [Yuanhong] has one or two reliable divisions, his orders will hardly be carried out. Zhao Yiwu (Zhao Hengti) has always had Li’s trust and is now training troops under Governor Tang; could you discuss with Jiwu (Tang Hualong) and the others how to have him quickly train an elite division at Li’s disposal? As for weapons, we will spare no effort and it should be possible. Yesterday Ke[qiang] cabled asking for the Protection Army’s agreement and for the draft to be written and sent on his behalf: please settle that point quickly. Zhun. The 10th. Note: 灰 = 10 June, as on the header. Digits re-read by Brice Dudout. ## N26 — 13 June 1916 ? Please hand three hundred yuan to Baili (Jiang Fangzhen). Zhun. Note: Digits re-read by Brice Dudout.
Key: key.md
06 Sources
- Japan Center for Asian Historical Records (JACAR), Diplomatic Archives of the Ministry of Foreign Affairs of Japan, series 特定人ノ発受郵便物取調: B03050089200 (N05), B03050089300 (N13, N16), B03050089400 (N26).
- Brice Dudout, Wenmi (October 2026, CC BY 4.0): 辰密 and “June 1916: what Huang Xing’s telegrams say”; github.com/bdudout/wenmi, solver/numeric1916; pull request #266 to this repository.
- Satoshi Tomokiyo, Cryptiana, section “Mingmi (明密) Format Code: Gongmi, Rongmi, Chenmi”.
- Hsueh Chun-tu (薛君度), ed., 黃興等電稿, as cited by Brice Dudout.
- Related pages, each a different key or system: 容密 (Rongmi), 共密 (Gongmi), Sun Yat-sen’s intercepted telegrams, 1916–17 (code condensers) and Huang Xing to Lin Hu and Li Genyuan, 25 May 1916 (romanised kana).
The digit transcription, decryption, reading, translation and key file are in the repository under targets/chenmi1916/; the independent check made on 20 October 2026 is in targets/numeric1916/. The JACAR scans stay in Brice Dudout’s repository.
Cite this page
Brice Dudout, Satoshi Tomokiyo and Daniel Bourdeau, “Huang Xing’s 辰密 telegrams, June 1916”, Cyphersolver, posted 5 November 2026, https://cyphersolver.com/chenmi1916.html.