I confess I'd been scrapping data for my RDF election site from files on the Press Association site without taking any notice of their terms and conditions. Too obsessed with exploring my model-driven approach you see. When the question of the availability of open election data was raised on the mySociety developers mailing list, much discussion ensued about what laws applied to this data - whether copyright (seems doubtful since these are facts in the public domain) and/or database rights (which do seem to apply here in the UK because of the effort required to gather the data).
The Electoral Commission will publish ratified results sometime in the future as open data, but what is needed is openly available election results as they are reported. As far as I know results are only available commercially from the PA (or one of its client organisations such as the Guardian). There is a nice trade-off here - data becomes freely available when its value has diminished, but at a higher level of quality because it will have been checked and ratified.
So penitently I emailed Chris Mead, the elections editor at the Press Association confessing my sins. I was very pleasantly surprised when he and his team replied that, as long as it remains an academic project with no commercial element and the PA continues to be credited, they were willing to allow this use of their data. This is very generous and many thanks to the PA.
In the future, other possibilities arise for collecting this data from Council websites along the lines of the work that Chris Taggart is working on in the Open Election Data project or from crowd-sourcing results. Perhaps another solution is for the government to outsource the provision of open data and buy the data from PA on behalf of its citizens.