I still think
@Samuel Halpern's graph on p119 of his
Centennial Reappraisal book and the related article
Why A Low Angle Break? describe the break-up most convincingly. Survivor statements like a "high angle" or "second funnel" etc are unreliable given the darkness and comparisons between such statements from different survivors would have the "vantage point" error factor built in. To a survivor still quite close to the ship just before break-up, the stern would have appeared to have reached a much higher angle than it actually did because of the optical illusion, to which we humans are quite susceptible.
As Sam showed with his calculations, the bending stress on the keel and deck plates reached its maximum when the stern reached an angle of around 11 to 12 degrees. It is very likely that was the point when the physical break-up of the
Titanic started, but it would not have been immediately obvious to onlookers. It was not as though it was a sudden "snap!" type of break; the failure of the keel and separation of deck plates would have continued over the next 2 minutes or so but towards the end of that timeframe, something else happened - the flooded bow completely lost its buoyancy and the
Titanic suddenly lost its longitudinal stability. The result of that would have been a sudden dip of the bow (the so-called "lurch") resulting in displacement of a large volume of water that washed sternwards (the "wave"). Also, though already failing, the stern section would still have been attached to the bow at that point and so the lurch at the bow would have pitched the stern quickly to a higher angle, perhaps as much as 25-degrees before the widely reported final catastrophic break occurred.