Bergne, in his two page-long accounts on Sarts, shows how uncertain Russian ethnographers were regarding the origins of the Sarts. They were divided according to their assumptions. Noteworthy features of Sarts are: they are settled, urban dwellers and bilingual. Some called themselves Tajik, some Uzbek, but spoke both languages. But the term Sart remains unclear. Here is how Bergne describes Sart identity:
“In this Islamic Central Asian environment, where ethnicity was of little consequence, the process of assimilation between the later arrivals, the Uzbeks, and the Iranian/Persian/Turkic/Arab cocktail of peoples whom they found, produced in due course a composite identity of mixed ethnic make-up. In terms of language, its representatives were usually bi-lingual, but eventually preferred a Turkic language strongly influenced by Persian in both vocabulary but also in vocalisation. This composite identity became known as ‘Sart'” (Bergne 2007: 7).
Bergne in his endnotes (p. 136) writes about Bartold’s view on the origins of the word Sart. According to him, Sart is a Hindi word used by Mongols to refer to all Central Asians involved in trade. This, indeed, is a very vague description, given that Mongolia very often is seen as part of larger Central Asia. And who are ‘Central Asians’ for the Mongols?
From Bergne’s chapter ‘Central Asian Identities before 1917′ one can draw two conclusions. First, the term’s meaning changed throughout the time. As Bergne suggests, during the Timurid dynasty (14th century) Sart used to denote ‘Tajik’, or any non-Turkic peoples, i.e. Iranians. As the only non-Turks in the region were mainly Iranians: “[A]t the end of 14th century, the language and literature of the Sarts was described as being what wasn’t Turkish i.e. Iranian”. But what meant ‘Tajik’ during the Timurid period? Did it really mean the same as now? The term Tajik used to mean different notions in different times. It was generally meant to refer to the non-Turkic sedentary population of Central Asia, earlier it was used to mean Arabs. In Tibet (China), all Persians are known to be called Tajiks, in the courts of India, Tajiks were the ones who spoke both Persian and Arabic. In this context, Bergne, probably, refers to ‘Tajik’ as to the Iranian sedentary population of Central Asia. He notes that a German traveller, J. Klaproth in the 19th century, mentions the Bukharan people referring to themselves as Tajiks, whereas Turks would call them Sart. In the nineteenth century the term Sart acquires somewhat “derogatory connotations”. According to Bergne, the word Sart derives from Turkish ‘‘sari it” i.e. “yellow dog” (cf. Bergne 2007: 7). My observation in this regard is that the word Sart is mainly used [in the 20th and 21st century] to refer to Uzbeks primarily by the Kyrgyz and Kazakh. As I have asked my Kyrgyz and Kazakh fellows on a number of occasions of what the term “Sart” means, they first were embarrassed, saying that was a very offensive word to call Uzbeks. Not all of them could really explain what literally the word means. But according to some of the responses Sart is referred to nomads who have dismounted their horses (the horse being an important symbol of nomadism) and settled down, by doing so they betrayed their life style.
The second conclusion is that several records of Russian orientalists show them having difficulties identifying ethnic affiliations of the peoples of Central Asia. The shared view in this respect was the fact that Sarts were a mixture of Turkic and Iranian people. But as who they were, more Uzbeks or Tajiks, this was largely disputed. Grobenkin, a Russian ethnographer classified them as Uzbeks due to their preferred use of Turkic languages. Andreev “identified them as a mix of various Turkic elements”. He also noted that “Uzbek-Sarts” live in all the settlements together with a small admixture of an assimilated Tajik element, who have lost the conception of any sort of division into tribes and no longer remember their ethnic origins, but who do not consider themselves to be ‘Turks'”. Skvartskij, however “called them Tajiks, albeit turkicised” and confessed that they deny being Tajiks and consider themselves to be Sarts. Bartol’d, one of the most cited scholars, called a Sart an “Uzbekised urban Tajik” (cf. ibid. 8). Zarubin, however makes it clear that “they do not really know what they are. They call themselves Turks. But their Turkmen and Kyrgyz neighbours call them “Sart” which word they also use for Tajiks” (ibid.).
Sarts, according to Sengupta, in Khorezm, Ferghana and Tashkent were primarily Turkic speaking, but sometimes bilingual. Whereas in Transoxiana they were called “Tajiks or Chagatais, and were Tajik speaking or bilingual” (ibid.20). She believes Sarts remained distinct from both Tajiks and Uzbeks (ibid.).
This uncertainty about ethnic/national identities of Central Asian people shows how difficult it was to separate these people along the ethno-linguistic line. Alisher Ilkhamov, according to Bergne, argues that “the traditional image of the “Sarts” did not accord with the social engineering embarked on by the new communist regime” (Bergne 2007: 8). By the 1920s though, the designation ‘Sart’ was replaced by ‘Uzbek’, and so they were all counted as Uzbeks in all censuses, regardless of, as Zarubin puts it, the “lack of clarity as to the term’s meaning” (cf. ibid. 9).
 On the word Tat, see for more information Sengupta 2002.