After cleaning and lubricating my 9 speed KMC chain, I checked the wear and to my surprise the checker shown 0.5%. I moved it further and the checker shown almost no wear. Trying to figure out what I am doing and how that can be, I noticed that when measuring over the missing/quick link, the indicated wear is larger. I thought it must be the missing/quick prematurely worn out. So I bought a few new, two KMC and two SRAM 9 speed links. Again, to my surprise, when measuring over the link the wear is 0.5%, anywhere else over the chain is minimal. It happens no matter what missing/quick link I use. Even measuring from the one side of the missing/quick link and away, there is no wear. It is as if that link itself is a bit longer than the others of the chain. No skipping, no strange noises, just that weird chain wear measurements.
What am I doing wrong, what is going on, how can that be?

Hard to know if you're doing anything wrong. I've noticed it's not unusual for the QL to wear more than the rest of the chain. I don't worry about it much. Only thing I can think of is that the QL might not get lubed as thoroughly when installing and creates more wear on that pair of links so even a new QL won't eliminate the difference but, that is just my completely unscientific guess. Also, most chain checkers will show more wear on your chain than you actually have. Use a ruler to measure as a backup to your checker.

Strange the same happens even with brand new quick links, and of two different brands (KMC, SRAM).
I suspect it could be the not so tight fitting of the pins of the quick links to facilitate fitting and removal.

Yes, this could also be true. I have never measured a new chain and link to find out but try it next time you have a new chain to know for sure. Like I say it's not something to worry about and not likely to create any noticeable wear in your drivetrain.

When gaging chain wear I avoid the connecting link on purpose. Andy

Stop confusing yourself and measure with a ruler. You can see if the ruler measurement is different when the quick link is in the measured segment. Chain should show no more than 1/16" elongation in 12" of original length. Chain checkers have a questionable reputation for accuracy, which should surprise no one since they are measuring over such a short section of chain.

I mostly use KMC chains and they always show more wear when measuring over the QL. Now, I never measure over the link.

Using a ruler is out of question. I don't care for the exact stretch. I prefer other means that quickly show you it is time for a chain to leave the floor. Chain checkers are pretty accurate. There is also a nice test for the accuracy of some of them.
Just for the history, I eventually found out that you do not measure over the quick link. The pins of the quick link are grooved so that the interlock with the sockets in their plates. The total difference in length can be of 1mm or more and hence the discrepancy when using a chain checker over them.
As all others said, I shouldn't do this over the quick link.

I concur. When these topics regarding chain wear, and measuring chains come up, it puzzles me as to why there are always some who insist that chain checking tools are inferior, and the ONLY proper way to measure a chain is by using a steel ruler. As if companies like Park Tool…otherwise known for quality tools…is suddenly going to produce a tool that is substandard. — Dan

Park tools absolutely does make a bad one, though. I have one of theirs that has a little triangle that drops into the chain that's really bad. Sometimes reports brand new chains as worn.

I also have one of theirs with three points that properly positions the rollers so that play in them doesn't hurt the result - the biggest reason for the chain wear tools to mess up.

I got the SHIMANO TL-CN42. At least it is a well reputed one delivering spot on estimates on the wear of the chain. The PT CC-4.2 was my choice fist place, but it was not available at the time of purchase in my local bike shop.

I have both those Park Tools chain checkers. The first time I tried it on a new chain, it wouldn't even go into the links. I queried Park Tools and they told me the tool is calibrated for greatest accuracy at the "1%" mark. I've since found that it over-estimates chain wear slightly, as do many such devices that don't compensate for roller wear. The thing I like about it is that I can easily track the progression of wear, and when it gets close to 1%, I can use the other checker. I find that the PT CC-4.2 also slightly over-estimates wear.
Because of the possible error (noted above by Kerry Irons) due to the short length of chain sampled, I always check in at least 3 places, and never across the quick link. It's still faster, easier and cleaner than using a ruler, and if I change the chain a bit early it's no big deal because "chains are cheaper than gears".

Why is roller wear not basically part of the phenomenon of chain stretch and therefore a valid part of the measurement? I mean, those are the contact points on the rings and cassette?

The two point chain checkers don't pull the rollers in the same direction, I think, so are actually measuring roller slop, not wear. Different brands ship with different amounts of that out of the box.

Roller wear *contributes* to chain elongation measurement, which is why the "2-point" checkers have reduced accuracy. But elongation is the result of wear on the pins and their complementary bearing surfaces on the inner links. It seems to me that the rollers could all be worn or unworn and the chain would not contribute to excessive sprocket wear as long as the pin-to-pin spacing was 1/2 inch.

the rollers are sandwiched between the pin and tooth though, right? So wouldn’t wear on those create additional misalignment?

They can "flop around" on the pins, but the pin-to-pin distance is unaffected. I'm assuming all the rollers are worn (or not) to the same extent. Anyway, this explains (to my satisfaction, at least) why the "ruler" method is the Gold Standard to which all others are compared.

Yes, the slop on the rollers is irrelevant except that it makes the bad chain wear checking tools report new chains as bad. It's quite a large distance. There's diagrams here:

https://pardo.net/bike/pic/fail-004/0...l#park-rohloff

People don't use these tools to measure a number, write down the number on a sticky note, and post it next to the chain, then compare later. So it's kind of irrelevant that if you knew the initial number you could subtract and get the actual chain wear. Worse, these checkers usually just have a couple numbers on them like .25, .5, and .75 and that's it. So they are really go/no go gauges, not something that gives you a precise number you can add and subtract from and they don't include higher numbers just so you can measure that high and subtract the slop later.

I haven't had any of my chain checkers report new chains as bad. There may be some manufacturing tolerance variations that make this happen though.
My Park Tools CC-2 won't even go into a new chain. As the chain wears, progressing from 0.25% to 0.5% to 1.0% on the checker, I start using another checker that compensates for roller wear; I have the Park CC-4.2 and a similar Shimano checker. These are both "Go-NoGo". When those indicate 1% wear (by dropping into the chain), I change the chain knowing there's probably a little wear left.
EDIT: That ProLink gauge doesn't compensate for roller wear either; It's basically a "Go-NoGo" gauge with varying shades of "Go", like the Park CC-2.

Yep... That's normal and of minimal concern on the overall chain. Remember its just a measurement.

Do note this though, I used to reuse my quick links but have had some failures. One came apart after I dropped in the driveway. So I no longer reuse my quick links after a little wear...

This has been my experience too. I usually change at 0.75% though

And in other news, that 36" belt around your waist doesn't include the connecting buckle in the total length measurement. Buckles can be slightly different.

Off the OP and going into a chain gauge: I have been using a 3D printed Chain Gauge and have found it quite handy. My only problem is printing it in PLA. The PLA does not go well with my Central Texas heat in the garage. I am considering printing it in ABS...