It has been suggested (Coblin 1986:40) that PTB *d-wam ‘bear’ can be compared with Chinese xióng 熊 jung. Coblin reconstructed PST **dwjəm > OC *gwjəm. It seems more prudent, however, to presume that the root was simply PST *wəm

PST *tswan ‘perforate’
Chinese ‘to bore, perforate, penetrate’ 鑽 *ùtson > tsuân > zuān [G *tsuan, B *tson]
PTB *tswan > WT mtshon ‘any pointed or cutting instrument’

PST *lwi ‘river’
Chinese ‘water’ 水 *ìhluj÷ > świ > shuǐ [G *hljədx, B *h[l]juj÷]
PTB *lwi & lwəy ‘flow, stream’

PST *sum ‘three’
Chinese ‘three’ 三 *səm > sâm > sān [G *səm, B *sum]
PTB *g-sum ‘three’

---

PST *Kᵂa- > OC *Kᵂa-, WT Ko-
PST *Kᵂə- > OC *Kᵂə-, WT Ko-
PST *-aKᵂ > OC *-aKᵂ, WT -o(g)
PST *-ua- > OC *-ua-, WT -o-

PST PTB OC (B) OC (G)
*Kwan > *Kwan *Kᵂan *Kᵂan
*Twan > *Twan *Ton *Tuan
*Kwaŋ > *Kwaŋ *Kᵂang *Kᵂang
*Twaŋ > *Twaŋ *Tong *Tung
*Kon > *Kon *Kon *Kᵂan
*Ton > *Ton *Ton *Tuan
*Koŋ > *Kuŋ *Kong *Kung
*Toŋ > *Tuŋ *Tong *Tung

---

A final note on the dental-initial roots: It has been suggested (Coblin 1986:40) that
PTB *d-wam ‘bear’ can be compared with Chinese xióng 熊 jung. Coblin reconstructed
PST **dwjəm > OC *gwjəm. It seems more prudent, however, to presume that the
root was simply PST *wəm, as not all TB languages show the effects of prefixal *d-.
With Baxter’s reconstruction of xióng 熊 as *w(r)j/um, the comparison looks quite good.
On the strength of the comparison I would reconstruct OC *ìwəm. (This is in contrast
to Baxter’s reconstruction of *w- in many of the roots in §5.2, where the comparative
evidence favors the reconstruction of an OC velar stop initial.) As far as Chinese and
PST are concerned, then, this root should not be considered as having a dental initial.

The last set of PTB roots has initial *h-. We might expect initial *h- to pattern
with velar initials in terms of correspondence with Chinese, so that PTB *hwa- would
correspond to OC *ha- or *xa- and go back to PST *hwa-. However, I have not found
any Chinese cognates of these PTB roots.

In Baxter’s reconstruction, the OC vowel system is more similar to that of PST.
The vowels *o and *u remain distinct from *wa and *w. This similarity cannot, however,
be taken by itself as evidence that Baxter’s system is more accurate. In deciding which
approach to OC vowel reconstruction is correct, the crux of the matter remains the
interpretation of Shījīng rhyming evidence. In terms of my PST system and Gong’s OC
system, PST *kwan and *kon would have merged first to *kwan, and then developed
to OC *kan. In Baxter’s system the merger never occurs, leading to two distinct OC
forms *Kan and *Kon which rhyme only imperfectly in the Shījīng. Both proposals
involve plausible phonological developments.
Since the MC reflexes of these syllables are identical, Baxter’s proposal can only
be confirmed if one accepts the statistical arguments made in Baxter (1992) which
demonstrate that there are distinct rhymes in *-an and *-on. If one accepts Baxter’s
arguments, it would be possible to retain Gong’s system by distinguishing OC syllables
*kan and *kuan (Baxter’s *o elsewhere corresponds to Gong’s *ua), and arguing that
the two finals *-an and *-uan tend not to rhyme. Such a tendency could be explained
by arguing that, for example, Gong’s *ua was realized [uø]. But this revision to Gong’s
system still would not resolve all the problems if we accept his PST reconstructions. The
contrasting OC syllables *kan and *kuan would both be reflexes of PST *kwan, since
Gong does not reconstruct *o as part of his PST vowel system. There is no conditioning
factor for this split. This problem can ultimately be traced back to Gong’s reliance on
Tibetan comparisons.
In summary, I believe that on balance the evidence pertaining to the reconstruction
of medial *-w- in PST favors Baxter’s OC reconstruction; for this reason, I have for the
most part adopted Baxter’s vowel system. Listed below are the developments proposed
in this chapter, as ordered rules. Baxter’s and Gong’s reconstructions are also provided,
but it should be noted that the ordered rules as formulated here cannot be reconciled
with Gong’s system.