As I keep telling everyone everywhere, the real reason someone, or a group of people, downvoted your post is something only they can confirm. We can only say why a post like yours might or might not get downvoted. That said, here are my two cents.
Sub-optimal formatting
Your question's first revision isn't the worst I've seen, but you (1) forgot to enclose some of the expressions in Mathjax's dollar sign, (2) rarely made use of any formatting (except italicizing one word), and (3) due to not separating quotes from your original words, introduced paragraphing that's a bit awkward.
These are rarely problems in themselves, but they send a signal of sloppiness. And if people get the impression you did not invest time in formulating your question, they'd think you're not respecting their time they would spend answering and become more inclined to downvote.
Lack of clarity
You say
However, with these definitions, I cannot understand why $N_{\mathrm{A}}$ should be independent of substance.
but you don't say why. An answerer would have little choice other than to repeat the definition, as is illustrated in Ivan's comment under your question, and even matt_black's great answer says "The confusion, I suspect, . . . ".
Andselisk also commented under this very meta post that he did not vote at all because your question isn't all that clear for a variety of reasons, but lack of clarity is a very valid reason to downvote or close vote if needed.
Misc
There could even be other semi-valid reasons for downvoting. A lengthy debate, under the answer, on a topic that seems rudimentary to a lot of people would probably not garner lots of positive emotions. Or that you choose an outdated book with outdated concepts chemists are usually weary of doesn't make them too happy about the question either. Anyway, the current vote count is +1/-3, so that is only 1 rep lost. It's not such a drastic example.
By the way, by asking this meta post, you have incurred the meta effect and should expect more negative or positive voting activity on that post.