A DH106 1A Comet has been restored at the de Havilland Aircraft Museum

I thought that too until I read this

>Many readers familiar with the Comet disasters might be wondering why, with this article drawing to its close, I have yet to utter the phrase “square windows.” But the truth is that “square windows” never had anything to do with the Comet crashes. The windows were not and never were square — in fact, you can see for yourself in the above image, which shows a Comet 1 window next to a modern Boeing 737 window. Can you tell which is which? You probably can, but not because one is any more “square” than the other.[0]

[0]https://admiralcloudberg.medium.com/neither-money-nor-manpow...

I've heard that counter-rumour before too but it never seems to make sense. The lengthy discussion by Admiral Cloudberg [1] seems to pin the blame on the corner of a window:

> De Havilland had calculated a maximum operating stress of 28,000 psi at the corners of the windows and doors, but investigators noted that this value was an average over an area of 2–3 square inches (13–19 square centimeters), meaning that in theory, highly localized stresses could be considerably greater. This “peak stress” could have been measured through the liberal application of strain gauges, but de Havilland had apparently elected not to attempt this, believing that any more precise measurements would be unreliable. Nevertheless, investigators measured it anyway, and from these data they calculated a localized peak stress at the window corners of up to 45,000 psi under normal pressurization conditions. Not only was this much greater than de Havilland’s predicted value, its relative proximity to the ultimate strength of the material (estimated to be 65,000 psi) produced an unfavorable stress ratio correlating to an expected fatigue life considerably below 10,000 cycles.

[1] https://admiralcloudberg.medium.com/neither-money-nor-manpow...

I think counter arguments come into one of two camps:

(1) when they replicated the problem in a pressure chamber at ground level it wasn't technically a window that failed but a "portal" (basically like a window but wires go through it instead of people looking out of it) - a pedantic technicality and not to say that all the failures in the air were for this same portal/window.

(2) The windows weren't don't have sharp corners anyway but rounded, not unlike some modern planes. True but you can see the failure was definely near a (rounded) corner.

Admittedly the Admiral Cloudberg article does seem to put more weight on the way the rivet holes were made than the angularity of the window corners. But it's still failing at a corner. I guess it depends how you look at it.

There were later versions of the Comet. Version 1 was underpowered, and had too much weight reduction for that reason. By version 4, the design had been debugged. With more powerful engines and structural fixes, the Comet 4 went into service and did OK. 46 Comet 4 aircraft were built. Last flight in 1997.