Cab made some great comments on one of the Hemi threads about tach accuracy.

I think this is a very important topic.

Which tachs are the most accurate / reliable and which to avoid ? Electric or mechanical.

a very good question.
i have seen very accurate cheepo tachs, and high quality tachs screwed up by bad installations.
i have a super old SW direct read tach [that was last calibrated around 1969] that i could press against the spinning damper to check an electric tach, but it only goes to about 3500 or so.
besides that, i would probably get tangled up in anything else spinning around in the vicinity, thereby causing an erratic reading.

I run Autometer Ultimate playback tachs and last I checked, they were within about 150 rpm at the upper end. I have a Mallory ignition tester that I checked with. Their tachs have lasted 10+ years.

MSD 8998 is a great tool to have, borrow, rent, go in partners on, etc., to check tachs, chips, ignitions, etc.

I haven't been impressed with the linearity of several tachs (including Autometer and Bosch), checked on my bench with an accurate frequency counter and function generator. They are usually close at some high RPM like 6K or 8K, probably because the factory calibrates them at that point... but they have only one trimpot so you can get it perfect at a single RPM but can't do anything about the errors at other points. For example, my 5" Autometer I set to be most accurate at 6000, but at 3000 it reads almost 150 low, at 2000 it reads 50 high, at 1000 it reads low again.

I wonder if a mechanical tach would be inherently more linear or not... it's been a looong time since I had physics

back in the day it was "supposed" to be something like + or - 2% accuracy. we just ignored it because we had no way of checking it. later years we started to recognize that the electrics had a sweet spot generally thought to be around the middle of the tach's rpm range.