Reconstructions

PS
*tibn- - straw (Fronzaroli 1969b: 12, 27, 34; Kogan 2011: 201)
PS
*tihām(-at)- - sea (Fronzaroli 1965a: 136‒137, 144, 149; Kogan 2011: 191; Kogan 2015: 240)
PS
*tkb - to prick, to puncture (EDA II)

Cf. a likely variant root with two emphatics instead of plain stops un Gez. ṭaḳaba ‘to sew together, to stitch, to mend’ (CDG 595), Tna. ‘to mend, to patch a garment’ (TED 2447). Quite probably, the well-known PWS root *ktb ultimately belongs to the same source with metathesis. See discussion in EDA II.

PS
*tVkur(r)- - kind of bird (SED II No. 226)
PS
*tVlVˁ- - female breast (SED I No. 276; Kogan 2011: 217)
PS
*tVll- - mound (Fronzaroli 1968: 272, 287, 298; Kogan 2011: 190)
PS
*tamar-/*tamr- - date palm (Fronzaroli 1969a: 291, 279, 300)
Level? Hardly any connection with Akk. gišimmaru, nor its Sumerian parallel gišimmar (Kogan 2012a:255), including the enigmatic sa-ma-lum = GIŠ.GIŠIMMAR in VE 399 (Kogan 2015: 99)
PS
*tVrr- - kind of bird (SED II No. 229)
The PS reconstruction is not fully reliable since some of the supposed cognates can be analyzed as non-Semitic loanwords: Akk. tarru < Sum. DAR (so Salonen 1973 151, AHw. 1331), Syr. tar < Persian tar (so LSyr. 833).
PS
*tVrbal-, *bVrtal- - large-hoofed animal (SED II No. 228)
Cf. Amh. tərbe ‘baby elephant’ (AED 966), possibly related to the present root with *-al > *-ay > *-e. If this comparison is accepted, the variant proto-form *tVrbal- should be treated as the primary one. This conclusion seems to be corroborated by Bed. tírbil ‘calf just weaned’ (Blažek 2003 274) which is, admittedly, somewhat remote semantically.
PS
*tišˁ- - nine (Kogan 2009: 73)