- Reconsider your ant project. You might be better off studying snakes. There aren't that many. They have big, obvious features like stripes and spots.
- Get a $10,000 microscope. It is really hard with the consumer-grade scopes we are using.
- Prepare to spend an entire hour on each ant. You will go blind trying to see if there are 0.06mm hairs on the petiole. You will agonize over whether the antennal insertion is curved or bent.
- After step 3, spend another hour on the same ant.
- Nail down one species at a time. Look at multiple examples. View the face, and profile. Look for hairs and at texture.
- Meanwhile, go as far as you can in the key. Decide what features your are sure about, which are iffy. Keep notes about all of that. Pick a few possible answers.
- Look up images of the few candidate species. If you confidently match one of them, make that ant a reference specimen. If not, give it a temporary name and make it a reference specimen.
- Don't sweat the unknowns, as long as you have a reference specimen. The questions you had ("is that clypeus notched?) may become clear later, when you see a new ant ("Oh, I see, THAT's what notched looks like.)