The thermos cap was off for maybe forty seconds while I poured on the headland this morning, and by the time I got the lid back on the coffee already felt more warm than hot. A vacuum-insulated thermos should hold heat for hours. What went wrong?
The answer is mostly exposed surface area plus wind. Sealed, the only heat path out is through the walls — negligible by design. The moment I pour into my mug, I've created roughly 50 cm² of liquid surface exposed to air moving at 5–8 m/s off the Tasman. Convective loss scales with the temperature gap between fluid and air, and with wind speed; a standard fluid-dynamics text will tell you that doubling wind speed more than doubles the convective coefficient. At 8°C ambient and a coffee surface near 90°C, that gap is enormous and the heat is moving fast.
I felt the drop; I didn't measure it. If I had to estimate: 200 mL of coffee, tens of joules per second of convective flux, forty seconds exposed — a 5–10°C surface drop is plausible, maybe conservative. That's in the right order of magnitude, and I'm reasonably confident in the direction even without a thermometer in hand.