# Part 5 — Three Claims to Evaluate

Each of the three claims below is a statement *some* geophysics student
might write in a homework answer or report. Your job is to figure out —
without your AI assistant's help at first — whether each one is
correct, and then ask your AI assistant to evaluate them and compare
notes.

**One of the three claims is wrong.** The other two are essentially
correct. The wrong one is wrong in a *subtle* way — it has the right
shape of an equation but the wrong content.

Do *not* read the answer key (your instructor has it) until after you
have submitted your worksheet.

---

## Claim A

> Critically refracted head waves only exist when the lower layer has a
> higher seismic velocity than the upper layer (V₂ > V₁). If V₂ < V₁,
> the transmitted ray bends *toward* the normal upon entering the
> lower layer, and no critically refracted head wave can be generated
> along the interface.

---

## Claim B

> In a flat two-layer refraction problem, the intercept time on the
> travel-time vs. offset plot — the time at which the refracted-wave
> linear branch would intersect the time axis if extended to x = 0 —
> equals 2h / V₁, where h is the depth to the refractor and V₁ is the
> velocity of the upper layer. This is the same as the vertical
> two-way travel time of a P-wave through the upper layer.

---

## Claim C

> When a regional seismic network records a small crustal earthquake,
> the *focal depth* of the earthquake is typically more poorly
> constrained than its *epicentral coordinates*, because all stations
> are located near the Earth's surface. The ray geometry from a
> shallow source to surface stations has poor vertical resolution
> compared to its horizontal resolution, particularly when no station
> is within roughly one focal depth of the epicenter.

---

## Your task

For each claim:

1. **Without using AI**, write down whether you think it's right or wrong,
   and why. (One or two sentences each, in your own words.)
2. **Then** ask your AI assistant to evaluate each claim, one at a time,
   using the "Always push back" pattern from the AI Literacy Guide.
3. **Then** cross-check the claim you suspect is wrong against
   Lowrie & Fichtner (Ch. 3 or 4), the relevant ESS 314 lecture, or
   your Lab 3 notebook. Quote the specific page, equation number, or
   notebook cell that confirms or rejects the claim.
4. Document the verification in your `ai_logs/error_log.md`.

The point is not that you find the wrong claim — it's that you
articulate the *reason* it's wrong with a specific primary-source
citation.