KIFIS, I can identify certain features in the messages that might help to identify the code used.
In the first telegraph from Changsha, it says there are 32 words in the message. That number corresponds exactly with the number of 4-digit blocks, as long as you also include the words "Peking" and "seal" in the word count. The second telegraph says it's 282 words long. But you correctly pointed out the word "seal" is missing from the end of the second message, which suggests that the second message is incomplete, as there's nowhere near 282 word blocks in that message.
So each 4-digit block seems to correspond with a single word. You correctly pointed out that certain blocks are repeated in the second message. This could mean that a "page" cypher was used, where a select list of words were written (or typed) on a page (or pages) and each one was assigned a 4-digit block code to represent them. There may have been many such lists in the "code book" with different "pages" being used for different days, as none of the blocks in the first message seem to repeat in the second message (though I didn't check them all).
Now whether English words or Chinese words were used I can't tell, but it does seems that whole "words" were encoded in the 4-digit blocks. The fact that the rest of the message is in English, and the fact that the telegraph system being used was meant to exchange messages from all around the world suggests that the encoded words were English words.
The other possibility is that this was some form of early system to transmit Chinese words in morse code, as there might have been a simple "number book" with each number corresponding to one Chinese word character. 4-digit blocks would allow up to 10,000 Chinese characters (words) to be encoded. The word "Peking" at the beginning of both messages suggests that maybe a number code called "Peking" was used to transmit the Chinese words. In fact, isn't there a character code (memory storage code) for Chinese characters in the PC Windows environment called "Peking" or something like that?
Later...
I checked a Windows 95 programming book, and there's a code called Chinese (PROC, Singapore) that uses code page 936. This code uses 4 digit blocks (hexidecimal digits instead of base 10 digits) to represent all of the Chinese characters. I wonder if the "Peking" code was similar?
------------------
Safe flying to you...
[This message has been edited by Flight Safety (edited 31 January 2001).]