In my world, I don't believe I could accurately OR precisely measure tenths with a caliper. Micrometers operate differently, etc etc, mine have a clutch mechanisms which make it easier for a ham fisted hack like me. I think the form factor of the caliper doesn't lend itself to this level of precision when handled by the likes of me.
First off calipers have a "feel" that mucks things up.
I and my guys don't have 5 minutes to measure a part.
This means sometimes you go fast.
Leighs post is interesting "parts are born to be bad".
Unfortunately too often rather than beat the parts in to submission we beat the gauge to read what we want to see.
Higher quality calipers are worth the money if you are going to use them for finish dimension and given enough care, enough time, and a new gauge there is no reason you can't hit the posted number for repeatability.
None of this reflects real world usage.
I've designed and built enough gauging systems that work fine in the lab but fail miserably on the plant floor to get the difference between what you can do and what works.
At +/-.0002 inches there is no way on God's green earth that I will trust a micrometer for measuring parts.
I have the library of 25 years worth of gauge R&R studies that prove it.
In a blind study, nobody can repeat to within a tenth let alone be accurate to this with a mic.
I know you don't want to hear it and at one time I was on the same side as you are.
Numbers here mean jumping into very expensive gauging and few shops get this fact or are willing to spend as much on something that just measures as they are willing to spend on the machine tool. Checking parts is an afterthought.
This is always a problem for us bringing in a 20 year experienced toolmaker.
Everybody wants to think they can measure closer than they can and the argument gets heated.
We have set up blind part tests to teach the errors that your brain just does not want to see.
Much of this is designed to temp you into the wrong reading.
Take a set of marked gauge pins and blocks and have them lapped and polished .0002 to .0008 undersized.
Give these to a experience machinist and have him record the numbers.
The normal trend towards "zero" becomes quite noticeable.
How much time do you have to make a check also makes a huge difference.
The home machinist may have 20 minutes to "decide" on the number or average a bunch of checks, the production machinist needs first number good so he can punch in the offsets and load the next part.
We had to implement time constraints on our R&R studies to give us shop floor numbers we could trust.
Gauge repeatability is a complicated task and after 25 years we still find things we need to change in our procedures so that we can trust the operator's and gauges' capability.
When measuring parts you make do you clean the mic anvils with alcohol and check a gauge block before every measurement?
Probably not, but when checking the accuracy of your mic you may very well do this.
We want to make good parts, and we want to trust our measurements or we cant' make good parts.
This natural human bias distorts our view of the real world.
When testing any gauge your emphasis should be on how "bad" you can make it read.
What is the worst reading that you have ever gotten?
At some point it will happily give you this error when checking parts.
Bob