Good Question.... often pondered because like so often in photography, its a bit misleading, and it shouldn't really be "Shutter Speed" but "Shutter Duration"
As said, it depends on the type of shutter design, but the easiest to explain is the Focal-Plane, 'curtain-Shutter'...
Let me take you back to Eadweard Muybridge, in the 1870's and the Governor of California's bet whether all four horses hooves left the floor during gallop...
It was common of the era for the wet-plate cameras to have no 'timed' shutter. The photographer, composed his shot "Through-Taking-Lens" on a ground glass plate. When he was happy, he swapped the ground glass plate for a photo-plate smeared in photographic emulsion, usually kept in the dark until after fitted into the camera with tin-plate 'Dark-Slides'.. the photographer, then replaced the lens cap on the camera, removed the dark-slides, and to 'time' his exposure, took the lens cap off and started counting elephants, before putting it back on again.
BIT if a faff that, and many snappers, 'cheated' a bit; rather than put the lens cap back on, remove the dark slide, count elephants, put lens cap back on, replace dark slide, and repeat... they left the lens cap 'off'.. pulling the dark slide out from below the camera and starting to count.... pushing the dark-slide back in the top at the end of their count... oh-so-much less faff.. and for exposures counted in full seconds, accurate enough.
Back to Muybridge; his problem was that he needed a really short exposure time, to capture the horse in gallop, to be able to see if all four hooves were indeed off the floor....
His stroke of genius, then was to ponder this faff of fitting lens caps and drawing out dark-slides and counting elephants... A-N-D he concluded, that the "Exposure Duration" was the time between pulling the dark-slide from the bottom, and shoving it back in the top..... IF then he wanted a REALLY short "Exposure Duration", all he needed do, was wallop the dark slide back in the top, almost before he had pulled the one out of the bottom....
Hmmm... this suggested using two dark-slides... one to pull out the bottom, and one to start pushing in the top before it was out....
This begged a further idea... why use two dark slides... why not use just one... with a slot in it.... as you pull it out the bottom, you dont pull it out, you pull it 'through' and the bigger the slot, the more exposure you get....
And the focal-plane-curtain-shutter was invented, and much to the delight of the Governor of California, it proved that all four of a horses hooves leave the floor during gallop.
Oh-Kay.... This idea wasn't without problem; and Muybridge found most of them, and solved quite a few.
The first thing was that the slide would tend to 'jam' a bit in its slot, and the exposure interval wasn't all that reliable, depending on how much wiggling the slide did on the way down. Muybridge tackled this is various victorian ways with tin grooves and grease, and lead weights on the bottom of the slide, and leading towards the modern shutter, spring loading it to 'drive' it through its travel, faster than its own weight and gravity could.
Next, dropping a slotted slide through the camera, it fell at a constant rate due to gravity. There was little way to adjust the 'exposure duration' other than to have a lot of slides with different height slots; thinner the slot, shorter the shutter duration.
Here-in-lies the root of the anomaly between shutter speed and shutter duration... the speed is constant, the duration not.
Using a slotted slide, the top of the photo-plate was 'exposed' to light before the bottom was; the slot traveling across the slide over an interval much longer than the actual exposure time... hint make the plate smaller, slot hasn't so far to travel, shorter you can make the 'exposure furation' for any 'shutter speed'
Consider typical 35mm SLR. The film negative size is 24mm tall, by 36mm wide. The 'Shutter SPEED' may be 2m/s, (about 7Km/h or 4mph.. a brisk walking pace) then it will take just over 1/60th second for the curtain to travel from one side of the frame to the other.... So, you fire the opening curtain; it travels across the frame, and then you fire the closing curtain, and it travels across the frame. Time it takes the opening curtain, is the same as it takes the closing curtain, to cross, so the 'exposure duration' is the same both ends of the frame.. BUT whilst the 'effective' exposure is 1/60th of a second, the the 'shutter' is 'open' for twice that, it taking 1/30th of a second between the first curtain starting to move and the second curtain stopping.
Want a 'faster' shutter speed... you cant easily make the curtain go any faster... so what you do, is fire the closing curtain sooner.. now the closing curtain starts to travel before the closing curtain has finished, and you have a slot move across the frame. Still takes 1/60th of a second to get there, but the film only gets light on any one 'bit' for maybe 1/120th or whatever, depending on how big the lag twixt opening and closing shutters.
Incidentally, this is what gives rise to the 'cartoon' charecture cars that have the body stretched over the wheels and the wheels 'ovalated', the tops ahead of the bottom, because in the early days of high-speed photography, that, exaggerated, is how photo's of speeding cars looked, the image inverted on the film, the top of the frame being exposed before the bottom.
SO.... there really is your basic answer.... and the first thing is to make that distinction between "Shutter Speed" as in how fast it may actually move, which is usually 'constant', set by the distance the shutter curtain has to travel, and whatever drives it, usually a spring. Then there's the 'exposure duration' the 'effective' time the film 'sees' light, which in the FPS described is dependent on the length of the slot or lag between curtains.
Worth mentioning iris shutters at this juncture. Imagine using the aperture control as the shutter. Moment it opens, you get an equivalent of say f32, increasing through f22, f8, etc to the 'maximum' set aperture, maybe f2.... then it starts to close again, a-n-d it goes back, from f2, through f4, f8, f22, etc.. you now have some complicated maths to work out how much light the shutter has let in overall, BUT, you can do it, and get an 'effective' Exposure duration, which again, will be longer than the time any part of the shutter is letting light pass, and completely detached from the actual 'speed' that the iris leaves move.
And offers conclusion;
The "Shutter-Speed", as we call it, isn't; what we are interested in is the "Exposure Duration".
The shutter likely physically travels a heck of a lot faster or slower than anything related to what's on the dial, the "Exposure Duration" set and adjusted by the 'lag' between opening and closing curtains being triggered.
The Curtains, or leaves, are usually driven by a simple spring, and started and stopped by mechanical catches, either mechanically or electro-mechanically.
Variations and permutations of how the shutter works and bits are arranged, are pretty enormous, from the simple curtain shutter, through iris in-lens iris arrangements, with almost as many possible perms and cons on how the shutter might be driven, with springs or gravity or electrical motors, and like-wise 'triggered' both for firing and lag, using mechanical stops, that could be released by levers or catches, or magnets or electro-magnets, even pneumatic, etc.
Answer the question? Confused? Blame Muybridge's bloomin 'orse! That's what started it all!
