“No high point in the piping installation without possible release of air and steam (on the ground, in ascent or in horizontal flight).
The radiator (or feeder) must be loaded on top of the cylinders, both in climbing and horizontal flight.
Water outlet pipes with a minimum internal diameter of 34 mm must run upwards to the radiator (or feeder) in both climb and horizontal flight. (...)”
This passage tells us that it was not unusual to proceed with a water tank on top of the cylinders ("nourrice"), as the cold and hot water stream pipes had to be placed and routed in a way that favored fluid dynamics, circulation and the evacuation of steam in the event of overheating.
Agreed.
I was also able to check that the capacity of the water jackets was 19 liters for a total of 26 liters including the piping (not counting the radiators? I don't know for sure...).
A comment in the manual also states that in the event of a water shortage, the operating time without abnormal overheating is only a few minutes...
The question is, in the case of l'Oiseau Blanc, how long is a few minutes? That logically depends on the power setting (how much heat are the cylinders generating?); outside air temperature (the colder the better) and, primarily, how fast it's losing coolant. Witnesses in the town of St. Mary's saw the plane "on fire" (trailing steam mistaken for smoke) headed west across St. Mary's Bay and last seen in the vicinity of Red Head on the eastern coast of the Cape Shore peninsula. That's a distance of 14 miles. From Red Head to Gull Pond is 8 miles, for a total of 22 miles. At l'Oiseau Blanc's cruising speed of 100 mph, that would take 13 minutes – call it 15 minutes with a headwind.
In your opinion, what is the capacity of the 'feeder' on the cylinder tops ? Judging by the proportions, I'd say between 120 and 140 liters.
It's an important point. The trail of steam suggests the pipe delivering hot water to one of the radiators burst. With one of the radiators still working, the cylinders are still being cooled to some degree until the coolant is exhausted. How fast coolant is lost depends on the water pressure and the diameter of the failed pipe. Once the water in the reservoir is depleted the engine will overheat until it eventually seizes. We don't have enough data to quantify the time between the failure of the pipe and engine seizure, so we're left with a gut guess. Is 15 minutes a reasonable ?