LetX(1), …,X(k)be the firstkordered observations of a sample of sizemfrom the distribution with p.d.f.f(x; β1, θ) = (1/θ) exp [-(x– β1)/θ] forx≥ β1≥ 0, θ > 0 and zero elsewhere (2 ≤k≤m); letY(i), …,Y(l)be the firstlordered observations of a sample of sizenfrom the distribution with p.d.f.g(y; β2, θ) = (1/θ) exp [—(y– β2)/θ] fory≥ β2≥ 0, θ > 0 and zero elsewhere (2 ≤l≤n). It is assumed that θ is unknown. A test based on (X(1), …,X(k),Y(1), …,Y(l)) is proposed for the null hypothesis β1= β2against the alternative β1≠ β2. The distribution of the test statistic under the null hypothesis is derived. The significance points for various values ofk,I,m,nare tabulated for (α = .05 and .Ol.