[Rounder]

Actually no. Take the experiment of Eratosthenes for example. He measured the difference of shadow angle in distant cities on the same day of the year. That particular experiment could easily yield an estimate for either of the following:

A. The diameter of a round earth assuming a nearly infinitely distant sun.
B. The distance of the sun from the earth assuming a flat earth.

But when you take both of those answers and "run with them", only the answer to A. ends up making predictions that repeatedly predict and fit other observations.

Option A is the only one that survives the addition of a third city to the data set.