Given 0 < p, q, r < ∞ and q < r ≤ ∞ we consider the natural embedding ℓp,q → ℓp,r between Lorentz sequence spaces. We introduce a new method of proving that this non-compact embedding is always strictly singular but not finitely strictly singular.1. Introduction. Let T : X → Y be a linear operator between Banach spaces. We recall that T is strictly singular if there is no infinite-dimensional closed subspace Z of X such that T : Z → T (Z), the restriction of T to Z, is an isomorphism. Moreover, we say that T is finitely strictly singular if, for every ε > 0, there exists n ε ≥ 1 such that for every subspace E of X with dimension greater than n ε , there exists x in the unit sphere of E such that ∥T (x)∥ Y ≤ ε.Strictly singular operators and finitely strictly singular operators, which encompass compact operators, possess similar properties which are habitually connected with compact operators. For example, it is well known that Fredholm operators are invariant when perturbed by strictly singular operators (i.e. if T is Fredholm and S is strictly singular then T + S is Fredholm; see [1, Theorem 4.63]).Many of the non-compact operators in Analysis are strictly singular or finitely strictly singular. For example, the Fourier transform, which is obviously non-compact when considered as a map from L p into L p ′ , is finitely strictly singular for 1 < p < 2 and strictly singular when p = 1 (see [6]).The natural embedding of sequence spaces I : ℓ p → ℓ q for p < q, which is non-compact, is finitely strictly singular (see [8]).