Timed entries, or timed events, relied upon in this case, need to be looked at, and treated accurate to within a maximum of 60 seconds either way, scaled down second by second to a minimum adjustment of 1 second either way. For example, let's take the 3.30am call which Jeremy made to Julie. We say the call took place at this particular time, but was it made 1 second, 2 seconds, 3 seconds, on a scale all the way up to the 60th second of 3.30am? This is very interesting because if for example Jeremy's call to Julie, had started at 03:30:01, and the call had lasted barely a minute, it could be argued that his call to Julie had been made at 3.31am. If the call lasted 2 minutes, it could be argued that Jeremy had called Julie at 3.32am. If the duration had been three minutes, it could be argued that Jeremy called Julie at 3.33am. Alternatively, if Jeremy's call to Julie had started at 03:30:60, and it barely had lasted 1 minute, it could be argued that Jeremy had called Julie at 3.32am. If the call had lasted 2 minutes, it could have been argued, Jeremy had called Julie at 3.33am. Similarly, if Jeremy's call to Julie had lasted 3 minutes, it could have been argued that Jeremy had called Julie at 3.34am...
The purpose of trying to demonstrate how it might be possible to show that any timed reference of an event might only be accurate to within a minute either way, depending upon the duration of event. Where duration exceeds 1 minute in its entirety, the accuracy of a timed event could be out by half of the duration period either way. To simplify matters, adopting the principle, of rather than saying Jeremy's call to Julie took place at 3.30am, another way of saying the same thing is that the call lasted between 3.29 and 3.31am, or 3.28 and 3.32, or 3.27 and 3.33am. But adopting this approach is still problematic. This problem may only be rectified if the precise time, in hours, minutes, and seconds are known at the start of event!!!