Where can I find a clear exposé of the so called "standard reduction to the local artinian (with algebraically closed residue field", a sentence I read everywhere but that is never completely unfold? --- Thank you for your answear dear BCnrd. Here are some precisions since I'm still not able to do it on my own. EDIT: properness assumption for $f$ has been removed. Questions are still uncorrect. I say that $(f:X\longrightarrow S,Z,E)$ is a 'context' if: (1) $f:X\longrightarrow S$ is a morphism of preschemes which is surjective, separated, smooth of pure relative dimension $1$ and finitely presented. (2) $Z$ is a closed sub-prescheme that is supposed finite etale and surjective over $S$ (finitely presented over $S$ If needed) (3) $E$ is a $O_X$-Module which is locally free of finite rank. There is some extra data $D$ defined over any 'context' $(X/S,Z,E)$ as before. I can base-change $D$ along any $S'\longrightarrow S$ and localize it $D$ at any point $x\in X$. (Yes, the base-change and the localization of a context are still a context) The claim I want to prove is something like: If some hypothesis $H(X/S,Z,E,D)$ is true then $D$ is zero. The hypothesis is stable under any base change $S'\longrightarrow S$ and is stable by localizing over $X$ at any point. I call a context $(X/S,Z,E)$ 'limited' if $S=spec(A)$, $X=spec(B=A[t])$ (or $X=spec(B=A[[t]])$), $Z$ is defined by $spec(tB)$, $E=spec(B^n)$ for ANY ring $A$. First question: Is it possible to reduce the proof of the claim on any context to the proof of the limited case? If it is, how? Second question: If I had some pull-back of the data along any etale map $X'\longrightarrow X$ and I forget the assumption that $f$ is fintely presented would I be able to make the reduction to the limited case also? (maybe with a different argument). If it is, how?