Latin builds qualitative abstracts in -tās, and similar forms are found in Gk. -της, Ved. -tāt-/-tāti- , and Av. -tāt-, suggesting that *-tāt- belongs to PIE. Descriptively, *-tāt- built abstracts primarily to thematic adjectives from an early date, with subsequent reanalysis of the thematic vowel as a part of the suffix, as reflected in abstracts built to consonant stems, e.g. Lat. auctōritās. Moreover, some abstracts reflect collective semantics when built to a nominal base. While some synchronic studies of the tāt -suffix have been made, e.g. Paucker (1977) for Latin, Renou and Benveniste (1960) for Vedic, and Petersen (1922) for Greek, no serious work on the history of *-tāt- has been done.
The study of PIE *-tāt- goes as far back as Aufrecht (1852), who believed that the suffix *-tāti- (cf. Ved. -tāti-) was older than “syncopated” *-tāt- . He speculated that the suffix *-tāti- reflects a combination of an underlying abstract in *-tā- with a ti-stem, since several tāt-abstracts correspond to abstracts in *-tā-, e.g. Skt. gurutā- ‘weight’ beside Gk. [special characters omitted] ‘id.’ and Lat. gravitās ‘id.’, etc. Though his survey was brief, subsequent work has largely accepted his view on the origin of *-tāt- and *-tāti- and their relationship.
Recently, however, Nussbaum (1996, 2004b) has uncovered a number of derivational patterns involving secondary t-stems, including formations built directly to case forms, as in kwsp-én-t- ‘nighttime’ (Hitt. išpant- ‘night’), continuing a t-stem addition to the loc. sg. of the root noun *kwsep- ‘the dark’ (Ved. ks&dotbelow;áp- and YAv. xšap- ‘night’). Lat. salūs ‘health’ < *sl&dotbelow;h 2-ú-h1-t- likely reflects a similar formation, but with a secondary t-suffix added to an instr. sg. Such a derivational pattern may also underlie PIE “*-tāt-", based on its individualizing function, which could continue (as *teh 2-h1-t-) the instr. sg. of an abstract in *-teh2- subsequently “individualized” with a t-stem suffix.
If so, the history PIE *-tāt- becomes clear: it would indeed reflect a secondary addition to an original teh 2-abstract, but in a very different way than Aufrecht thought. A secondary t-stem origin of this sort underlying *-tāt- might shed light on other complex abstract suffixes found in Latin (and, in part, elsewhere), including -tūs (with comparanda in Celtic and Germanic) and -tūdō.