Let K be a number field and a i (t) ∈ K[t] polynomials of degree 2. We consider the family of elliptic curves E t := y 2 = a 3 (t)x 3 + a 2 (t)x 2 + a 1 (t)x + a 0 (t) ⊂ A 2 xy ( * ) parametrized by t ∈ K. Our aim is to show that there are many values t ∈ K for which the corresponding elliptic curve E t has rank ≥ 1. Conjecturally this should hold for a positive proportion of them; see the survey [RS02]. We prove rank ≥ 1 for about the square root of all t ∈ K, listed by height. We are only interested in nontrivial families, when at least two of the curves E t are smooth, elliptic and not isomorphic to each other over K. Thus a 3 (t) is not identically 0 and not all the a i (t) are constant multiples of the same square (t − c) 2 . We view the whole family as a single algebraic surface in A 3 xyt and look at the distribution of K-points. The resulting surface has degree 5 but its closure in P 3 is very singular. We prove the following. Theorem 1. Let k be any field of characteristic = 2 and a 0 (t), . . . , a 3 (t) ∈ k[t] polynomials of degree 2 giving a nontrivial family of elliptic curves. Then the surfaceis unirational over k.The proof has two parts. First assume that we know a k-point p ∈ S that is not a 6-torsion point on the corresponding elliptic curve. Then we have geometrically clear and quite explicit formulas to prove unirationality. The second, harder part is to show that there are such k-points. We have not been able to turn this part of the proof into explicit formulas; see Remark 40.From any sufficiently general k-point of S we obtain families of elliptic curves of rank ≥ 1.Corollary 2. Let k be an infinite field of characteristic = 2. Then there are infinitely many different rational functions q(u) ∈ k(u) that are quotients of degree 2 polynomials such that the rank of the elliptic curve E t as in ( * ) is ≥ 1 for all but finitely many values t = q(u) where u ∈ k.Unirationality can also be used to exhibit points of small height. For a Zariski open subset U ⊂ S, let N (U, B) be the number of K-points of height ≤ B in U . Manin's conjecture [FMT89] suggests that N (U, B) should grow at least like B 3/2 , using the naive height function ht(x, y, t) := ht(x) + ht(t). Theorem 1 implies that N (U, B) grows at least like a power of B. The proof gives an explicit value for ǫ but it seems to be small. 4 (Connection with conic bundles). We can rewritewhere A, B, C are cubics. Projection to the x-axis exhibits S as the family of conicsyt . The conic F x is singular iff x is a root of the discriminant B(x) 2 − 4A(x)C(x); in general this happens for 6 different values of x. We get a possible 7th singular fiber at infinity. After suitable birational transformations the fiber at infinity is isomorphic tois the homogenization of a 3 (t). Thus S is birational to a conic bundle with ≤ 7 singular fibers; see Definition 6.A very degenerate case is when B(x) 2 − 4A(x)C(x) ≡ 0. These are easy to enumerate by hand. The only non-rational surface occurs when a i (t) = c i (t − α) 2 for every i. Then setting z := y/(t − α)...