>>196186- Those temperatures are a mean between the respective Tlimit of 20 measurement points all over the body, Tlimit being in this case the external temperature at which a sensor temperature drops below 24°C. So it's not an overall test, but individual ones. The coldest parts in the tests are always the feet, then the head, and finally the back.
To give a better example, here is the test from an undisclosed sleeping bag from a French company:
Tlimit(torso): -27°C
Tlimit(back torso): 0°C
Tlimit(head): 3°C
Tlimit(feet): 12°C
These are the temperatures at which each point dropped below 24°C. The average of all points (not all are mentioned here) that ended up being the Tlimit was 1°C. So yeah, you can push it further than that, but actually your feet would get cold at only 12°C. Wear socks and a beanie, use a decent sleeping mat, and you will greatly increase your sleeping bag's performance.
- Most American retailers don't know how this standard should be used. Many list the extreme temperature as the one at which the bag should be used. Be careful about that.
- Temperatures given by companies on their own tend to be insane in comparison. When the first EN 15337 temps were published, you could see that some brands previously announced temperatures 10°C/20°F lower than Textreme, and surprisingly, have since reevaluated them.
Some companies, most American, give "EN 13537 compliant" temperatures, though their bags haven't actually passed the real test, but "similar" tests led by third party labs. Those labs are actually complaining that the tests aren't detailed enough, and thus can't be replicated by others, which is actually the point. It leads some companies to try to discredit this standard, so they can continue to spread bullshit. Don't buy into it. The testing procedures are confidential and patented, mainly so that companies can't analyze them and find flaws to artificially improve test performance.
