One thing that occurs to me is to mine the Oyster* information for certain pattern disruption. Obviously this isn't foolproof, but might provide some pointers to go on. Obviously not everyone was using an Oyster - some may have been visitors to London. Of those, however, that are detected to make regular journeys, the information should show up that they have ceased to make regular journeys during the 7th of the 7th, and that of those, they tapped in, but not out.

Clearly this wouldn't be evident if, like me, they use a DLR station (although, unlike most, I was using a pre-pay Oyster on the day, as I'm only temping - my work isn't regular). Most DLR commuters never bother tapping in or out if they have a season ticket loaded onto the Oyster. At the typical unmanned DLR station, there's no requirement (although I've long held that there should be).

However, given those holes in the plan, mining the Oyster usage data might nevertheless throw up some information hitherto occluded. A person who has made regular journeys as far back as their records show, correlating with work commute times, and ceasing to make these journeys from 9am 7th July, should ring a bell, especially if the last recorded journey was 'incomplete'.

Of course, my journey at that time was recorded as incomplete, as was probably most other people. This is because nobody could tap out at the time in question as the system power had failed across the tube network. However, the Oyster system would have recorded that I continued to make a journey home later, then another in and out the day after.