An aperture is a circle, and a circle's area goes as the square of its diameter. Since the f-number is focal length over diameter, doubling the light means multiplying the diameter by √2 and the f-number by 1/√2 — so the scale that steps through halving and doubling is not 1, 2, 3, 4 but the powers of √2: 1.4142135623731, 2, 2.8284271247462, 4, 5.6568542494924, 8, printed on the lens ring as 1.4, 2, 2.8, 4, 5.6, 8. Half-stops are the fourth root of 2, which the second link prints as 1.1892071150027, and third-stops the sixth root, 1.1224620483094 on the third.
The rounding is why some numbers on a lens ring look wrong. The half-stop between f/2 and f/2.8 is printed 2.4 and is really 2^(5/4) = 2.3784142300054, the fourth link; the one between f/4 and f/5.6 is printed 4.8 and is 2^(9/4) = 4.7568284600109. Nothing is mislabelled — the ring is a rounded name for an exact ratio — and the reason the exact ratio matters is that exposure is counted in stops, which are multiplications, not in millimetres of opening, which are not. The same √2 fixes the size of the front element: an f/2.8 lens of 400 mm focal length has a 142.8 mm aperture in front of it, and that is why a fast long lens is a barrel.
Answer
1.4142135623731
How to read it
√2 on the workbench, one full stop of light. The second link is a half-stop, the third a third-stop, the fourth the f-number printed as 2.4.
Source
the f-number is focal length divided by aperture diameter, so the image brightness goes as 1/N²; the full-stop scale 1.4, 2, 2.8, 4, 5.6, 8, 11, 16, 22 is the sequence of powers of √2 rounded, and half and third stops are powers of 2^(1/4) and 2^(1/6)
Work it in the calculator
These links open the right calculator with this problem's numbers already typed in. The line above is what the screen should show — check it yourself.