^{page 773 note 1} The rimes of the Shï king have been determined by Ku Yen-wu and Tuan Yü-ts'ai, and Legge conveniently has recorded the rimes according to the latter authority. It seems to me that both Ku and Tuan are inclined to exaggerate and see rimes even in places where it is very doubtful if the poet ever intended the lines to rime, but on the whole their researches are clever and convincing.

^{page 780 note 1} If Maspero had taken into account not only the words of the “second division” of the rime tables, i.e. words of the type Anc. ka, etc., but also words of the “third division”, i.e. words of the type Anc. įa, tsįa, he would have obtained, with the same interpretation of the Archaic vocalism, (Mand. ye) Arch, iò, (Mand. chê) Arch. tśiò. According to his own reconstruction system, however, (Mand. yü) was Arch, iò and (Mand. chu) was Arch. tśįò, and it would be unreasonable to suppose one iò to have become ye, another iò to have become yü. Hence it is evident that Maspero has had in view onl words of the type Anc. ka (second division).

^{page 783 note 1} kuophon. in kâ; and dz'â phon. in ts'įwo.

^{page 784 note 1} This -įa' < -k' then was distinguished from the įak of the rime table (Ts'ie yün rime ): ngįak, etc., by the length of the vowel.

^{page 786 note 1} Le dialecte de Tch'ang-ngan sous les T'ang, p. 79.

^{page 788 note 1} “Zur Rekonstruktion der altchinesischen Endkonsonanten,” Mitteilungen des Seminars f. Orientalische Sprachen, Bd. xxx, Abt. I, 21 pp.

^{page 790 note 1} See Hirth, F., China and the Roman Orient, 1885, p. 139Google Scholar.

^{page 795 note 1} It is not a proof of the same binding order as the conclusion that lįäi has had a final dental because it has liät for phonetic.

^{page 797 note 1} E. g. zįêu: tâu; xâu: xįäu.

^{page 798 note 1} One might imagine that his theory would help understanding why e.g. Anc. yuo and Anc. ka rime in the Shï king (cf. p. 779 above); it would be an Arch. yuoy riming with an Arch. kay. This, however, is of no value, for just as well as âng and -uong do not rime in the Shi king, a presumed -uoy and a -ay would not rime. So the rime : does not find a satisfactory explanation by Simon's theory; its reason is given, p. 786 above.

^{page 800 note 1} I limit the investigation here to the Shī king, as being the oldest collection of poetry. Supporting materials can be found in Ch'u ts'ï, Chuang-tsï, etc.

^{page 802 note 1} Observe that nang in another sense has the reading nâi. Also that has two readings, tang and tâi, and that it may have the same stem as d'âi, as Simon cleverly remarks (p. 13).

^{page 804 note 1} In the category , i.e. words with medial w, there is only one rime for β and λ in Anc. Chin. We can now easily see the reason for this: Arch, kji became > Anc. kji and kjig became > kjii; but after w which made the syllable longer, this difference in length was not kept up: Arch, kjwig > kjwii: was therefore contracted into kjwi and thus coincided with Anc. kjwi < Arch. kjwi (rime β). So rime λ in ho k'ou did no longer exist, but all the words were ranged under rime β (Arch, kjwi) and (Arch, kjwig), etc.

^{page 805 note 1} The cases proposed already by Simon, p. 13.

^{page 809 note 1} Simon has added an interesting example: kįwo “wild boar, to fight”. I had supposed, Dict., p. 163, that ξuo was phonetic and that g'įαk “sport” was a logical compound: “fight with sword”. Simon proposes that had an Arch. guttural, and was simply phonetic in . I believe he is right, for there is a parallel g'įak (riming with zįäk in Shi king, p. 476). This entails kįwo < -k' also in the derivates .

^{page 810 note 1} I do not count then (Legge, p. 329), where , contrary to Legge's indication, obviously does not belong to the rime; nor , which I take for a rime pattern a: b: b: a.

^{page 810 note 2} Which would be highly interesting, for then ńźįwo < -g and hźįak could be suspected to be two phases of the same stem.

^{page 810 note 3} This is a bad rime, for the -p and -k of these words are attested in all kinds of sources and in still living dialects. But sometimes Shī king throws an unexpected light on phonetic problems otherwise obscure. There is Anc. tsįek, the final -k of which seems strange because it has tsiet as phonetic and is itself phonetic in tsiet. Shï king shows a -t instead of the traditional -k: rimes lįet: śįět: (Legge, 142); miět: (Legge, 489).