A couple of years ago I wrote a post on the question of whether a divisional application could effectively poison its own parent if the priority document became Article 54(3) prior art as a result of a claim to priority being invalid. At the time I thought it was a bit far-fetched, although I could not see any actual reason for the hypothesis to be wrong.
Prompted by a recent decision in the Patents Court (Nestec v Dualit - see paragraph 111, and the IPKat here for more details) in which claims of a European patent were found to lack novelty over its own priority document, I have now been pointed towards a recent decision by the EPO, T 1496/11. In this, the similar but potentially more dangerous "poisonous divisionals" (now a registered trade mark - see below) hypothesis appears to have been put to the test and found to work in practice.
The Board of Appeal found that claim 1 of patent EP0930979, as maintained by the opposition division, did not have a valid claim to priority because it had been generalised to encompass more options than were disclosed in the priority document. As a result, a divisional application became relevant for novelty under Article 54(3), and the claim was found to be anticipated. In this case the problem could be (and was) easily fixed by narrowing claim 1 to again have a valid claim to priority. I expect a fix similar to this would be possible in most cases, although it would probably result in a less satisfactory scope than the patentee was hoping to get. The decision does, however, confirm that the potential prior art effect of divisional applications should be carefully considered and not just dismissed as a theoretical possibility. More importantly, it shows that great care should be taken when filing an application from which priority will be claimed, particularly when the priority filing is narrow in its disclosure and needs to be generalised in a later filing. In some cases it might even be worth abandoning a priority filing and re-setting the clock to avoid such a problem.
Update 8 May 2013: See also the news article here from Hoffmann Eitle, K's Law here, and a (typically wordy) article from Avidity here.
Update 9 May 2013: In what seems to me to be a rather strange move, Avidity IP Limited have registered the terms 'poisonous divisional' and 'poisonous divisionals' as a series of trade marks at the UK IPO under class 45 for various intellectual property and other services. Whether they will seek to (or be able to) enforce the marks in relation to the terms being used in a blog post or otherwise is, however, unclear.
Update 16 May 2013: The generally excellent IPCopy blog from Keltie has picked up on the issue here, and added some interesting points about the relevance of Nestec v Dualit., as well as proposing an alternative to the 'poisonous divisionals' trade mark.
Update 12 June 2013: See the follow up post on my solution to the problem here.