Here is my table of Reeva's GPRS connections with durations. Where the source is 'Phone records' the start time is accurate as is the duration in all but one instance, because Moller read these out or expanded them on the screen. The other start times are calculated from the durations that I have read from the chart. My calculated finish time is 28 seconds out by 13:07:02 (a Moller time) and 2 seconds out by 13:40:54. I realign my calculated time after each Moller time. I'm spot on by 20:04:17 but accept I may have read a 4 for a 6 or vice versa when scraping the durations, but it confirms what I thought. The GPRS durations cover the complete time Reeva's phone is on.
I do note, however, that 'events' seem to trigger a renegotiation of the cell connection. I'm guessing an event would be Reeva bringing the phone out of sleep mode (after entering her Passcode? not sure) as well as when she's on the move when the phone switches cell tower or when the phone manages to negotiate a better signal (e.g. from 2G to 3G). This would often naturally coincide with times that we know she used her phone but not all times. Periods of continuous activity (e.g. multiple calls or WhatsApps, when the phone doesn't sleep) don't trigger any renegotiation of the cell signal.
View attachment 57208
It's possible that Reeva doesn't have Auto Lock enabled on her phone and that she just switches it on and off as she chooses. This would leave the display lit if she forgets, which runs down battery faster, so is not good practise. Anyway, her non-use on the night of 13/14 Feb now becomes more interesting. If waking the phone causes the cell signal to be renegotiated then Reeva doesn't appear to attempt to use the phone (I find it hard to believe the phone was being continuously used from 20:04) and nor does OP. If the Passcode has to be entered to achieve this renegotiation then this isn't proven in his case, but I don't think it does. Any thoughts?
On to OP's GPRS data next ...