How accurate is the M's mechanical shutter?
The mechanical shutters in M series cameras are nowhere near as accurate as those in modern, electronically controlled cameras. No surprises there.
Typically, you can expect the Ms shutter to be 1/6th to 1/3rd of a stop adrift of the correct speed. Which may sound bad, but as "Jay" remarked on a mailing list:
I own the Calumet Digital Shutter Tester and 6x M bodies and up to 1/3-stop accuracy is about normal for a Leica M shutter "out of the box". A good repairman can get it to withing about 1/6 stop (closer to that 10% mark) but this becomes more difficult as you progress upward to the faster the speeds. The speeds will vary from shot to shot also, and at the fastest speeds (1/500 and 1/1000) the variance is sometimes up to 1/3 stop no matter what you do. The M's have a separate combined adjustment for those 2 speeds and I've spent lots of time carefully "tweaking" my M6's and they'll be dead-on and then after a few firings I'll get a few that are still about 1/3-stop off. I've quit obsessing over it. Light meters and film emulsions aren't going to be "dead-on" either so even with electronically-controlled shutters (the one in my F5 supposedly checks and re-calibrates itself after every shot!) if I'm shooting slide film and I've got time to bracket, I do. In any event, 1/3-stop is well within the range of even narrow-latitude slide film.
[… S]ometimes if a mechanical shutter is left sitting the speeds (particularly the slow speeds under 1/15 in the case of the M Leicas) can slow down, and sometimes working it can improve things (in cases where the Leica slow speeds are really hanging, though, it usually doesn't work)…
Wondering why the M's max shutter speed is stuck at a puny 1/1000th when other modern cameras run up to 1/4000th at least? In Oct 2001 Dante Stella noted the following:
It is not inherent to rangefinders but inherent to the M-series design, which assumes a large, heavy, springloaded cloth curtain that moves across the film plane from left to right. The maximum (let's call it) crossing speed - the speed at which the shutter can be completely open - is 1/55 of a second. All faster speeds have slits that travel across the film plane. There are only two ways to increase the speed from 1/1000 sec. One is to make the springs tighter - but spring tension won't reliably pull the M-shutter faster than 1/800 or 1/900 sec. The other is to change the slit width, but when the slit sizes get small, it is hard to control error. The fastest horizontal shutters (the Nikon titanium foil version) hit about 1/2000 sec and 1/90 synch.
More modern multibladed metal and composite shutters (vertical travel) have much higher crossing speeds (1/125 and 1/250 sec). This translates into higher flash synch speeds. It also means that errors in slit width are minimized, because the necessary slit is now 2-4x the size, meaning an error of a fraction of a mm is not nearly as significant. They are also timed by oscillators, which eliminates the timing reliability of springs. Vertical travel (across 24mm and not 36mm) also helps significantly.
It is surprising that Leitz never developed a new shutter that could allow their lenses to be shot wide open or at the optimum aperture in bright sunlight. With 100-speed film, the Sunny-16 rule (and modern film speed measurement), you need faster than 1/1000 sec to shoot a 35/2 Summicron at its optimum of between ƒ4.0 and ƒ5.6. (Same goes for Elmarits, Sonnars, Jupiters, Nikkors and any other lens with this optimum setting). To shoot a Summilux wide-open you need 1/16000 sec... I wonder how much the 1/1000 sec limitation has informed Leica rangefinder lens design.
Of course the disclaimer is that film speeds were much lower (Tri-X, at 200, was "fast") back in the 1950s, when the modern M shutter was designed. But it has less of a place now, where the distinctive fingerprints of some famous of Leica lenses (Summilux 75, Noctilux, etc) are lost by ƒ2.8.
None of this is an issue on the R, which now has a fast metal shutter.
Store shutter cocked or uncocked?
Okay - now are you wondering if it is better to leave the shutter cocked or uncocked when storing the camera for long periods. After all, if you leave the thing wound, won't the springs detension over time? Well no. All mechanical Leicas are designed such that their shutter springs are always under tension. When wound, they are at 75% tension, when uncocked, they relax to 50% tension. Since the shutter springs never get to zero tension, it makes no difference either way.
Here is Sal DiMarco Jnr (Jan 2003) take on this:
When I attended the Leica School in 1971, this question was asked, and I remember the answer very clearly. John Brooks, the instructor said it didn't really matter. The spring tension on a cocked M camera is 75%, and on an uncocked camera, it's 50%. It was a personal choice. Now, today with battey drain being a factor, I would think, uncocked and if you remember it, on the M6TTL & M7 switch the camera to "off" […]
Intermediate shutter speeds possible?
On the majority of mechanical Leicas, yes, except for between 1/8th and 1/15th.
Some caveats: With the M7 it won't work (electro-mechanical shutter control); the M5 supports all intermediate speeds because it has a stepless shutter (and matching lightmeter); whereas the lightmeter in the M6TTL won't work when set on intermediate speeds.
See this Feb 2005 discussion at: <Photo.net: #00BJmC>
Shutter speed dial "clicks"
Finally, some users will hear an unusual "click" when changing shutter speeds from 1/2 to 1 sec when the shutter is uncocked - is this a problem? No. It is a standard part of the M shutter's design and will go away as you shoot more and more film (you bought the camera to take pictures, remember?)