The following table of optical elements, taken from patent GB157040A, is standardized to a focal length of 1 inch ($\approx$ 25.4 mm). The patent describes an aperture of $f$/2.
Unrelated but interesting never the less, the uncoated Cooke Speed Panchro designed by Horace W. Lee in 1920 described below, is incredibly significant in the history of cinema.
So, how do we compute the f-stop ourselves, given the constants above? Using the first equations we find on the internet, it seems straightforward:
However, quickly placing in the numbers shows that this can’t be the full story:
Note that we’re talking about the effective aperture here. This is the aperture as viewed from the sensor, which might be occluded by other lens elements. So, to take that into account we need to do some raytracing: In this case I start tracing parallel rays with increasing height until the ray is blocked by any of the lens elements.
We’re now interested in the position on the entry pupil of the last ray that was able to pass as this describes the effective aperture:
Plugging this value into the equation:
Closer.. But clearly still incorrect.
Instead, the following equation for the numerical aperture should be used, which describes the range of angles over which the system can accept or emit light:
Substituting now, this brings us much closer:
I think the main idea to take away from this, is that there’s an issue regarding the nomenclature for this popular subject. The word f-stop is passed around rather carelessly. In optics literature, usually the numerical aperture is used. In photography, the f-number is used. It is incredibly confusing they have the same f/~ notation.
The same issue goes for the focal length definitions. The effective focal length [distance between intersection(ray, optical axis) and the principal plane] is not the same as the back focal length [distance between intersection(ray, optical axis) and the entry pupil vertex].
For subjects like these, let us please use explicit terminology.