Do palatalized initials exist in third-division rimes and do medials exist in fourth-division rimes in Ancient Chinese?  What actually are the differences between the fan-ch'ieh doublets in third-division rimes?  All controversies regarding these questions will be resolved if we analyze the fan-ch'ieh's in Chi-yun.

In the series of rime-books related to the Ch'ieh-yun──the Kuang-yun and the Complete Manuscripts of Wang-yun, in particular──the initial letter which appears in all four division rimes has two different groups of fan-ch'ieh first-words: the first-, second- and fourth-divisions form one group (Group A); and the third division forms the other (Group B).  These 2 groups remain distinct most of the time with only a few exceptions.  Take the Complete Manuscripts of Wang-yun for example, the initials 見溪疑影曉 appear a total of 1002 times.  Of these, 449 occur as third-division rimes where only one of their fan-ch'ieh first-word is from Group A; 553 are first-, second- or fourth-division rimes where only 3 fan-ch'ieh first-words are from Group B.  In other words, the ratio of Group A characters acting as fan-ch'ieh first-words for third-division rimes and vice versa is 1002 :4.  It is no wonder that most scholars agree that the initials of third-division rimes are palatalized. For the initials 幫滂並明, the ratio of Group A characters acting as the fan-ch'ieh first-word for third-division rimes and Group B characters acting as fan-ch'ieh first-word for first-, second- and fourth-division rimes is 512:55.  This ratio can be broken down as follows:  the number of instances where Group B characters are used as fan-ch'ieh first-words of first-, second- and fourth-division rimes is 46 out of 262, and the number of instances for Group A characters being used for third-division rimes is 9 out of 250. Although the ratios are slightly higher than the previous example, the latter can still be passed off as exceptions.  The circumstances in Kuang-yun are similar.  One point that is perhaps worth mentioning is that for the initials 幫滂並明, the ratio of Group A characters acting as fan-ch'ieh first-words for third-division rimes is 260:6. This is lower than what is found in the Complete Manuscripts of Wang-yun.

The situation is decidedly different with the fan-ch'ieh's in Chi-yun. The ratio for the initials 見 etc is 1218:212. The ratio for the initials 幫 etc is 497:75.  Such ratios can no longer be called exceptions.  To further illustrate this point, let us take as examples the characters 古 and 居.  In the Complete Manuscripts of Wang-yun, all 132 occurrences of 古 are found spelling the initial for first-, second- and fourth- division rimes; while all 71 occurrences of 居 are in third- division rimes.  In the Chi-yun, however, 古 is used 54 times in first-, second- and fourth-division rimes and 5 times in third-division rimes; while 居 is used 39 times in third-division rimes and 36 times in first-, second- and fourth-division rimes.  Thus, Chi-yun does not support the theory that the initials of third-division rimes are palatalized.

Moreover, there is a close relationship between the characters listed in the fourth division of the rime table of fan-ch'ieh doublets (Fan-ch'ieh Doublet B) in the Chi-yun and the characters of fourth-division rimes. On the other hand, the characters listed in the third division (Fan-ch'ieh Doublet A) have absolutely no connection with the characters of fourth-division rimes. Let us illustrate this with the initials 見溪疑影曉. Characters of fourth-division rimes are not used as fan-ch'ieh first-words for Fan-ch'ieh Doublet A characters although they are used 16 times for Fan-ch'ieh Doublet B characters.  Fan-ch'ieh Doublet B characters are not used as fan-ch'ieh first words for Fan-ch'ieh Doublet A characters although they are used 35 times for fourth-division rimes and 48 times for Fan-ch'ieh Doublet B characters.  Fan-ch'ieh Doublet A characters are not used as fan-ch'ieh first-words for Fan-ch'ieh Doublet B characters or fourth-division rimes. Characters of third-division rimes are used 34 times for Fan-ch'ieh Doublet B characters but only 3 times for fourth-division rimes (of the 3, two are from Ching-tien-shih-wen). These show that the situation of the medial of the Fan-ch'ieh Doublet B characters must be the same as that of the characters of fourth-division rimes. That is to say if a medial exists in the former, an identical one will be present in the latter and vice versa; if the former does not have a medial then the latter will not have one either and vice versa.  Since we cannot say that the Fan-ch'ieh Doublet B characters do not have a medial, we, therefore, arrive at the following conclusion: the medial of both Fan-ch'ieh Doublet B characters and those of fourth-division rime is -i-, while that of the Fan-ch'ieh Doublet A characters and those of third-division rimes is -j-.  This conclusion together with the first one show that -j- is only a part of the rime; it has not palatalized the initial.