Tlemcen: medieval folk etymologies and their implications

In the mid-14th century work Bughyat al-ruwwād fī dhikr il-mulūk min banī ʕAbd al-Wād, Yaḥyā Ibn Khaldūn (brother of the more famous Ibn Khaldūn) ventures two possible etymologies for the name of Tlemcen (Standard Arabic Tilimsān, dialectal Arabic Tləmsān):

تسمى بلغة البربر تلمسنين كلمة مركبة من تلم ومعناه تجمع وسين ومعناه اثنان اي الصحراء والتل فيما ذكر شيخنا العلامة ابو عبد الله الابلي رحمه الله وكان حافظا بلسان القوم ويقال ايضا تلشان وهو ايضا مركب من تل ومعناه لها وشان اي لها شان
In the Berber language it is called "T.l.msīn", a word composed of t.l.m, meaning "she/it gathers", and sīn, meaning "two" - i.e. the Sahara and the Tell - according to our shaykh the most learned Abū ʕAbd Allāh al-Ābilī, may God have mercy on him, who was well-versed in the people's tongue. It is also said "T.l.šān", which is also a compound, of t.l., meaning "she/it has", and šān, i.e. "it has status".

Both etymologies are easy enough to interpret in the light of comparative Berber data. In the nearest (barely) surviving Berber variety - Beni Snous (Aṯ Snus), some 40 km west of the town - "Tlemcen" is indeed Tləmsin, not Tləmsan (cf. Destaing's Etude, pp. 368, 370, 371, etc.) This variety, however, does not use the word sin for "two" - it uses ṯnayən, like the Rif to its west (cf. Destaing, Dictionnaire, p. 98). The closest varieties to preserve a Berber word for "two" - geographically and genetically - use sən, in common with the rest of the Zenati subgroup to which Beni Snous belongs. The nearest varieties using the form sin are Kabyle, far to the east, and Middle Atlas Tamazight and Tashlḥiyt, far to the west. For the verb, one might consider t-əlləm "she/it spun", but the gloss given better matches a widespread dialectal Arabic word that could well have been borrowed into Berber: t-ləmm "she/it gathers". The second is obviously a compound of Arabic ša'n "affair, rank, status" and the Berber verb t-la "she/it has". Today this verb survives in Beni Snous, as in Kabyle, only residually, in the construction wi-h y-il-ən "who does it belong to?" (Destaing, Grammaire, p. 88). But it may have been more productive at that time, as it still is in Middle Atlas Tamazight.

Obviously, the first of these etymologies is implausible, while the second is a self-aggrandising play on words rather than an attempt to explain the name. But the fact that the first one could seriously be suggested is strong evidence that the meaning of Tlemcen was no more transparent to 14th century Berber speakers than it is to 21st century ones - as is not unusual for placenames. A better etymology can be proposed by taking into account comparative data - and allows us to explain the cross-linguistics differences in the final vowel - but I'll leave that for another day.