Like the title implies, two out of three ain't bad. I bought three of these and tested them all. The punch line is there were two "good" ones and one not-so-good one. Details below:My test setup uses a MT5 to MT3 reducing bushing that I scribed an index mark on to insert in the same orientation in the lathe spindle every time. I also reground its MT3 taper socket to ensure concentricity with the lathe axis when inserted in the indexed position. Measuring the runout with a 1 µm resolution Mahr dial test indicator shows ~0.5 micron runout at the taper surface. Also, it prints very nicely on my MT3 plug gage, and also prints very nicely on the spindle itself. I've inserted and removed it many times: it's super repeatable (kind of the point of a good taper...).I stoned the taper of the three holders I bought with my precision ground flat stones to remove any burrs or surface irregularities and help the tapers to seat well in the spindle. I then measure the runout at multiple points in the inside ER collet seat to check for concentricity and skew. The two good ones were between 1 and 4 micron TIR at the collet seat: pretty good I'd say, and also coaxial. The third one was more like 15 microns TIR and was not coaxial. I tested the performance as well at ~3x diameter with a 0.346" gage pin and a Techniks 3/8 ER16 collet (what I had lying around ready to hand). The two "good" ones were between 6 and 8 microns TIR at 3x diameter, and note that compressing a 3/8 collet to 0.346 is not an ideal setup, and the collets are supposed to be 0.0002" runout on their best day, so there's probably some error there. Still, 6-8 microns at 3 diameters is more than good enough to be getting on with, I'd say. The "bad" one was around 30 micron TIR at 3 diameters. Although all three stuck in the MT3 socket just fine during testing, I also blued them up and checked the print with my MT3 socket gage. I attached pics of the best print, though its a bit hard to photograph them well. I would say the two good ones had a pretty reasonable print, maybe ~40-50% coverage and this with a reasonable span, so I would expect them to perform well. The "bad" one was not too much worse, suggesting that it was the collet seat grind that experienced the error. I also did check these runout numbers a couple times by inserting and removing from the spindle and re-measuring, and the measurements repeat, so I've basically now got two tools I trust for tailstock tool holding and one that is basically a paperweight for my purposes. If you want to use it to hold drills in a drill press, though, all three would be more than sufficient for your needs. In my experience, the good ones of these are ace for holding taps and power tapping with the tailstock unclamped, and they also make excellent drill holders, much more grip than a drill chuck and more concentric as well. I bought one on amazon years ago and use it all the time. Liked it so much I thought I'd try my hand again, in fact, but the original vendor has disappeared. I chose these because at least they listed a specification for runout (0.0005", which is 12.7 microns for you metrique folk)So, all-in-all, I'm happy with them, but the usual caveats apply. If you need these to be in spec for sure, you must test them yourself to verify. I have the gear to do the test, so I can buy el cheapo chinesium tools on amazon and make it work. If that isn't you, well, honestly you don't have any choice because I couldn't find any reputable manufacturers that supplied ER collet holders with MT shanks. They only seem to be available on eBay or amazon, and they are all suspiciously similar-looking. It's either buy them and test or make them yourself. I figure the testing is faster than the making, and at ~40 bucks a toolholder, I think it was a good bet. It ended up being more like $60 a toolholder with the one I won't end up using, but given how long it would take to set everything up to make a handful of these, I'd rather just buy.