I don't have a time for a detailed reply, but:
(1) You _never_ take the F-statistic quantities (spmF, ess) to the second level.
(2) The simplest thing is to take single contrasts (also known as "t-contrasts") to the second level. (These are the con files.) The one problem though is that you can't do ANOVA on within-subject effects that way. If you want to do an ANOVA over the within-subject factors at the second level, I'd suggest searching the mail archives using keywords related to the multiple basis functions, and look in particular for posts by Rik Henson. You can also look at the monograph "ANOVAs and SPM," at http://www.fil.ion.ucl.ac.uk/~wpenny/publications/rik_anova.pdf, though that's highly technical. I seem to recall there might be a short monograph (or perhaps example discussion in an SPM manual) on the particular case of ANOVAs with multiple basis functions, probably by Rik, but don't remember the name of it.