--- You wrote:
There tengri was listed as
Turkic, although it was linked to some words with
quite different meanings and forms in Tungus-Manchu.
Presumably, then it is a loan-word form Turkish into
Mongolian? 
--- end of quote ---

Dear Adam,

I'm not a proponent of the Altaic theory but in this case, I think "Altaic" is an appropriate label for "tengri".

Also, in response to some of the replies, I believe "tengri" first appeared in the Orkhon Turkic sources, leading many to assume that "tengri" might have been a Turkic word.  Was "tengri" borrowed into Mongolic? It's hard to say for sure if it was a Turkic borrowing.  This word might have been in the Mongolic vocabulary and it's important to remember that the Mongols did not have a writing system until much later.

"Tengri" is attested in Sima Qian's Shiji in the chapter on the Xiongnu, together with other words spoken by the Xiongnu.  This is only a small percentage of the lexicon.  The problem with current analyses is that in defining the linguistic affinity of the Xiongnu, the Xiongnu is treated as a homogeneous tribe, if I may use "tribe" here.  Barfield's analysis shows that the Xiongnu was a confederacy and there might have been a high mixture of Turkic and Mongolic speakers.  The "Yenisean" linguistic affinity might only have been representative of one of the clans within the Xiongnu confederacy.

The etymology of Donghu etc is suspect and I can, at best, claim that it was a confederacy of sorts, like the Xiongnu.

Wayne