Each four-digit group printed under a Chinese name on a Hong Kong identity card is a telegraph code, 中文電碼 (Chinese telegraph code), one number per character, and by our count the two codebooks Unicode records almost never disagree about a character until its code reaches 8000. Of the 4,572 characters listed in both of Unicode's telegraph codebooks, 4,151 have every code below 8000, and in Unicode's record of the two books only 2 of those 4,151 carry different numbers in each. Of the 421 with a code of 8000 or above, 416 do. The Immigration Department calls the groups the "Chinese Commercial Code" and lists them as a field to be filled "if any".
Hong Kong's card-face sheet preserves a field for the commercial code
The Immigration Department (ImmD) explains the printed side of the card, the card face, in form ROP133, revision 10/2025. One of its fields is labelled "Name in Chinese Commercial Code (if any)", in Chinese 中文姓名電碼（如有者） (Chinese name telegraph code, if any). The words "if any" mean the field is not filled on every card.
The sheet does not name the codebook the numbers come from. Unicode records two, a mainland book and a Taiwan book.
The Royal Danish Library archives a Shanghai codebook dated 1872
The Royal Danish Library's catalogue entry for 電報新書 (catalogued as Dianbao xinshu), shelfmark OA 102-299, describes a "new numbercode for the telegraph in Chinese made by S. A. Viguier in 1871", based on a code by "the Dane H.C.F.C Schjellerup". The catalogue dates its Shanghai publication to the Tongzhi shiyi year, which it gives as 1872.
A second entry, OA 102-301, records 電信新法 (catalogued as Dianxin xinfa), revised by 德明在 on the basis of Viguier's code and dated by the catalogue to 1871. So the revision is catalogued a year before the book it is said to revise. We report both dates as the library gives them; the records we opened are catalogue entries, and we did not open the scanned books.
Unicode stores both books from a single Tokyo conversion table
Unicode's Han database, Unihan, keeps one field for each book. UAX #38, revision 41 of 1 September 2026, defines kMainlandTelegraph as "The PRC telegraph code for this ideograph" and kTaiwanTelegraph as "The Taiwanese telegraph code for this ideograph". It says both are derived from a 1984 conversion table, 漢字電報コード変換表 (Chinese character telegraph code conversion table), by Lin Jinyi, published in Tokyo by KDD Engineering and Consulting. Unicode marks both fields Provisional. Below, "the mainland book" and "the Taiwan book" mean these two fields.
The same document lists a 1983 mainland codebook, 标准电码本（修订本） (Standard Telegraph Codebook, revised), among the sources Unicode drew characters from. That makes it a source for which characters were encoded. The code values themselves come from the Tokyo table.
In Unihan 18.0.0, 7,078 characters (Unicode code points) have a mainland code and 9,024 have a Taiwan code. 4,572 appear in both. Two Taiwan entries carry two codes each, among them 塵 (dust) at 1057 and 7775.
Codes of 8000 or above shelter almost every mismatched character
Of the 4,572 shared characters, 4,154 (90.9%) have the same number in both books and 418 do not. By our count, 416 of those 418 involve a code of 8000 or above. In 233 both codes sit that high; in the other 183 one book uses a low number and the other a high one. The only two exceptions below 8000 are 夔 (read kuí), 5688 on the mainland and 1117 in Taiwan, and 廿 (twenty), 0579 against 1700.
The high range is also where the books grow apart in size. The mainland book uses 321 codes in the 8000s and 178 in the 9000s, topping out at 9694. The Taiwan book uses 733 and 542, topping out at 9798. These are counts from Unicode's record of the 1984 table and say nothing directly about any printed codebook.
A single number retains 乒 on the mainland and 丘 in Taiwan
A card holder, though, starts from the numbers, and counted by number the two books part ways far more often. Of the 6,770 code numbers used in both books, 4,154 name the same character in each and 1,878 name a simplified and traditional pair. The remaining 738 name characters that the variant links in the data we used do not connect; that is an upper bound, since some are variants the links miss. 391 of those numbers are below 8000. Take 8003: 乒 (the "ping" of ping-pong) in the mainland book, 丘 (hill) in the Taiwan book. Or 1001: 坨 (a lump) on the mainland, while the Taiwan book places 坨 at 9737 and gives 1001 to 垚 (yao, lofty).
The University of Hong Kong's name registration sheet treats decoding the same way, asking students to "Input the CC Codes as on your HKID, and choose the correct Chinese characters".
Our Chinese telegraph code lookup shows both books side by side. Until 14 October 2026 its copy line took the first code wherever a character had more than one, which affected 37 Taiwan-book and 2 mainland-book entries. For 发 it copied 4099 (發, to issue) and dropped 7569 (髮, hair). Twenty characters in Unicode's Extensions H and J could not be looked up at all. The revision made that day writes 4099/7569 in the line and names the character with a warning, and reaches every character in its data.
Simplified forms save their traditional partner's number in most pairs
We counted 2,256 Taiwan-book characters linked in Unicode's data to a simplified form that the mainland book lists. 378 of them carry a different number from that form; the rest share it. 陳, the surname romanised in Hong Kong as Chan, is 7115 in the Taiwan book, and its simplified form 陈 is 7115 in the mainland book.
The Taiwan book sometimes lists a simplified form under a number of its own. The mainland book's 吴 and the Taiwan book's 吳, the surname Ng or Wu, share 0702, but the Taiwan book also lists 吴 as a character of its own, at 8160. Someone schooled in simplified characters, in Singapore or Malaysia for instance, who looks up 吴 will find 8160 against it in the Taiwan book. 后 (queen; behind) is 0683 in both books, while the Taiwan book gives its traditional partner 後 (behind) its own number, 1775; the mainland book has no entry for 後. Where such variant forms come from is traced in our guide to how mainland China, Taiwan and Unicode choose a character.
Read against a card one character at a time, a four-digit group either appears in one book's column, in both, or in neither, and the card-face sheet does not say which book, or which edition, was used to print it.
