IntroductionIn [BFS] the authors introduced several notions for embeddings of smooth surfaces (or of general complete varieties) in a projective space which capture some of the higher order properties of the embeddings and whose study is possible using adjunction theoretical methods: the notions of k-spannedness, k-very ampleness and of jet ampleness. We recall (very roughly) only the first and the second one. Fix an integer k 2 0. Let X be a complete variety and i a rational map from X to a projective space n. Recall that a 0-dimensional scheme 2 c X is said to be curvilinear if for every x E Zred. 2 has embedding dimension at most 1 at x; i is called 0-spanned if it is a morphism; if k > 0, i is called k-spanned (resp. k-very ample) if it is (k -1)-spanned and for every curvilinear subscheme 2 c X with length (2) = k + 1 (resp. for every subscheme 2 c X with length (2) I k + l), the linear span (i(2)) of i(2) in I 7 has dimension k. For smooth X,i is 1-spanned if and only if it is an embedding (and if and only if it is 1-very ample). A line bundle L on X is said to be k-spanned (resp. k-very ample) if the rational map induced by H o ( X , L) is k-spanned (resp. k-very ample). Here we consider similar notions not on all X , but (essentially) on a Zariski open subset. We say that ( X , W ) (or ( X , L ) ) is generically k-spanned if for a general P E X and every curvilinear subscheme 2 of X with length (2) = k + 1 and P E Zred the restriction map from W to Ho(Z, L I 2) is surjective; this definition is due to SOMMESE. We say that ( X , W ) (or ( X , L ) ) is almost everywhere k-spanned if for a general P E X and every curvilinear subscheme 2 of X with length(2) I (k + 1) and P E Zred either the restriction map rZ;+,: W + Ho(Z, L 12) is surjective or the restriction map rZ,;+,: W + Ho(Z', LI Z ) is not surjective, where Z' is the union of the connected components of 2 not containing P. The reason to introduce the notion of almost everywhere k-spannedness is that if ( X , W ) is not (k -1)-spanned, then ( X , W ) cannot be generically k-spanned: if Z' is a length k cycle for which the restriction map W + Ho(Z', L 12') is not surjective, taking Z:=Z'U {PI with P general shows this; essentially ( X , W ) is almost everywhere k-spanned if and only if the union of minimal bad 0-dimensional cycles (for W ) of length at most k + 1 is contained in proper sub-