francophile 

Angle matters! The angle needs to be steep enough the blunt end of the cotter doesn't slip into the crank, but shallow enough you'll have flat and wider contact area with the spindle face. Cotters are fairly soft, so you don't need laser precision, but they're not so soft they won't back out if the angle is horrible and/or you apply enough force on install.

If you take a digital caliper to a spindle, you'll also find the flat isn't entirely uniform on the spindle either, there's upwards of 0.25mm variation in the spindle face surface IME. I took a caliper to a couple I have in my drawer to check before replying, for sanity's sake, and based on an inquiry in a post I believe you made and deleted previously. That lack of uniformity is one of a few reasons cotters need to be softer.

I assure you - and @ bulgie, maybe your experience is different - if you take a cotter like the # 1 I'd pictured earlier in the thread and try to hammer or press it into a crank that requires facing like the # 4 cotter, it's not going in all the way, and when it's pressed or hammered, it'll gouge like hell and work its way out prematurely.

For the record, I can't recall ever seeing gouging near the threaded side, I don't even know if it's technically possible, a cotter would likely slip too far in at the over-face required, but theoretically I believe it's what should happen. The sole purpose of my 5-minute picture which is super-poorly drawn, is to highlight the importance of angle, and express how you can "read" a cotter for newbies. It's basically what was taught to me decades ago, and it worked great for me to grasp it - YMMV

Going back to the original point, I think the majority of us who don't mind working w/cotters agree on what is critical and/or fact:
-

• You should use the correct diameter cotter
• The filed surface ideally needs to be uniformly flat
• The surface needs to be filed enough the cotter passes through to expose adequate threads
• The cotter is softer metal, so ultimate precision is not a big deal, but a fast, extreme pressure is critical
• You should never use the nut on the threaded end to "pull in" the cotter, the threads will strip before enough pressure is exerted

The following is my opinion, which some agree are truths, others may argue:
-

• Getting at least one old cotter intact and use it as a facing template is ideal
• Under-filing, steep angles are usually a recipe for failure, but over-filing is equally bad
• If you're starting from scratch with no old reference cotter, or previous cotters weren't faced right, press in and remove the new cotter \at least once to look for gouges and re-face as necessary
• The job is possible with a hammer, newbies or blind & lazy people like me can get more success w/less damage out of a cantilever-style press (ala VAR 07) or a crank-down press (ala Bikesmith

As with anything else requiring some level of skill and practice to master (maybe some special tools too), it's going to feel like black magic sorcery to someone who's never done it or has tried and failed miserably. Furthermore, those of us comfortable with the process are liable not to agree on each others' perspective fully.

However, that said, I'm totally confident I can complete my entire process of cotter install in 5 minutes or less without having the older cotter as template, and it will stay in until the next BB service, regardless how many thousands of miles get ridden, I've done it more times than I can count. The exact process and logic I use in my process is in this thread - take it or leave it - the general concepts and cues I adhere to are above. They may not be true, but it illustrates my personal experience.

As with anything I say, it's mostly true, I'm human so it may contain errors, try it, apply it, and take away what you will. If I'm wrong, you can prove it, and I can replicate it, I'll totally admit my mistake.

PS - My drawings usually suck, don't read into them too far. I'm not an artist, and it's worse when I'm limited to what Microsoft Paint is capable of doing.