Next, from those nouns, pick the ones that any user in any future circumstance might think of, or remember this paragraph by. It may at first seem difficult to imagine all the future circumstances or thought sequences a reader might go through. And you will find that almost any noun in your list could be eligible. The nouns that are especially eligible are nouns actually connected with the topic of that paragraph, section or chapter. However, nouns that are not directly connected with the paragraph's topic can also be useful in an index, especially if they are mentioned only once or at most a few times in a section or chapter. Readers could remember the topic by those nouns.
Conversely, you could think of this step as eliminating nouns that would not be useful. These are nouns that are either mentioned so frequently that they would not be useful to identify THAT paragraph by (example: the topic of the whole book or chapter), or nouns that are too generic in meaning and do not have any real connection with the topic being discussed.
In our example list, we would probably eliminate the word "body" since the entire book is about care of your body and so is going to be mentioned in (or the topic of) every section and every chapter of the book. Also, the words "enough" and "it" would not be useful in an index. However, the subject that they refer to are indeed of interest: water. Also, "percent" on its own is not likely to provide useful reference to this paragraph. However, it will be useful as a subentry to water, which you will see as our exercise progresses. So our list is slightly reduced, and we make a few notes about the context provided by "percent", "enough" and "it" to water:
water
percentage in body (from "percent")
body holding onto (from "it")
lack of (from "it")
not drinking enough (from "enough")
drinking enough (also from "enough" later in paragraph)
