As some of you know, my wife and I are translating Grothendieck's Tohoku paper. Originally, it was suggested that it be published as a TAC reprint, but Grothendieck refuses permission because he "does not believe in copyright". So I thought to retype it and post it on my own site (so sue). I then realized that translation would be easier than retyping. Which brings me to a translation problem. I am not expert in spectral sequences and what I know is from Cartan-Eilenberg. I cannot related G's definition to theirs. G defines a spectral sequence as a pair E=(E^{p,q}_r,E^n), all indexed objects of an abelian category subject to five conditions. The first three refer only to the E^{p,q}_r, including that for each pair p,q the E^{p,q}_r stabilize vis-vis r to a term he calls E^{p,q}_\infty (and not E^{p,q}). The fourth assumes "isomorphisms $\beta^{p,q}:E^{pq}\to G^p(E^{p+q})$. The family $(E^n)$ without filtrations is called the \emph{l'aboutissement} of the spectral sequence $E$." The E^n are assumed filtered and G^p is the associated grading: G^p(A)=F^p(A)/F^{p-1}(A). Now this makes no sense. The only thing called E^{pq} would be the term E^n for n=pq and this is really unlikely. I strongly suspect the domain of \beta^{p,q} is intended to be E^{p,q}_\infty. Finally does anyone have any idea how "aboutissement" is to be translated. It means something like limit, but the usual term for that is of course "limite". The Cartan-Eilenberg development is different enough that there seems not to be any corresponding word. Michael