While I was writing my grammar and lexicon of Awjila Berber, I ran into the word səndás pl. səndásən 'toilet'. While it has Berber plural morphology it is somewhat of a surprising word in Awjila Berber, as it lacks a prefix, which is unexpected.
Whenever a word looks unfamiliar in Berber, you generally start looking in Arabic. But also there, dictionaries and asking speakers of Arabic directly was not yielding any results. The word remained etymologically a mystery.
It's quite obvious that the word is no longer known in Libya, as Awjili speakers themselves cite the word on Facebook sometimes as 'typically Awjili' (and somwhat obsecene).
Only after publishing my book, I noticed that I had missed one other place where I missed this word (so those who have a copy get your pencils ready!): Paradisi also found the word in El-Foqaha Berber, with the same spelling sendâs presumably also denoting səndás. In this work however, Paradisi denotes it with a (ar.) behind the word, indicating that he believed it is Arabic. So I restarded my search once again, but still nothing.
Today, quite by chance, I finally found a cognate in Arabic. I was flipping through Corriente's Andalusi Arabic dictionary, and there it was, attested in Alcalá's 16th century lexicon: cindíç pl. cenídis. A perfect cognate having undergone a raising of ā to ī (perhaps īə like Maltese and Eastern Libyan Arabic) in non emphatic environments.
But surely, if it is found in those two places, which have no obvious connection with each other, it should be found elsewhere. My colleague Lameen Souag pointed me to an article about old Arabic words found in the Baharna Arabic dialect. The author points out that the great explorer Ibn Battutah mentions this word. The article also mentions that the word occurs in Fez, citing Idriss Dudan's al-amṯāl aš-šaʕbiyyah al-maġribiyyah. In the Baharna dialect it would also be attested as sandūs. There's not a Fezzi Arabic dictionary, but there's no reason to mistrust Dudan's account. I found a Facebook page on Maghrebi Proverbs citing (presumably the same one as Dudan's) proverb containing this word, plus explanation. But in the explanation of the word sandās, the qualification that it is from Fez is not present. Clive Holes' dictionary indeed confirms the presence in Gulf Arabic: sandūs, sindās pl. sanādis.
So there we have it, a word for toilet found as isolated Berber words in Libya, as an isolated word in the Gulf, and as an isolated word in Andalusia, and an isolated word in Fez; And the word was mentioned by a medieval explorer. That's a very widespread and ancient word which is by all accounts extremely marginal and not attested in the tradional Arabic Lexicons.
It has no obvious etymology, looking in the direction of Persia would be obvious considering its presence in the Gulf, but with my limited knowledge of Persian and a good Persian dictionary I did not manage to dig anything up.
Lameen Souag playfully suggested a connection with Latin sanitas. Which, if the Gulf wasn't so incredibly far removed from reasonable influence of early vernacular Romance (which you'd need to explain the voicing of t) would not even be that much of a stretch. For now, the word remains a very marginal but very widespread etymological mystery.
[EDIT:] Lameen Souag pointed out to me that I misrepresented the article somewhat. Fixed the text now.