Stavros is correct; with distinct sum indices, it works: (%i6) sum((x[j]-sum(x[i],i,1,n)/n)^2,j,1,n); (%o6) sum((x[j]-sum(x[i],i,1,n)/n)^2,j,1,n) (%i7) ev(%,n=2,'sum); (%o7) (x[2]-(x[2]+x[1])/2)^2+(x[1]-(x[2]+x[1])/2)^2 Barton -----maxima-bounces at math.utexas.edu wrote: ----- >Try using j as the inner index, for example. > > -s