Yes, chronograph readings can be inaccurate, and there's no way to absolutely know if you're getting good readings or not with a single chronograph. Unless, perhaps you have the Oehler model with the proof option which actually provides two separate readings for every shot fired.
This became very clear to me when I finally purchased a 2nd chronograph and constructed a skyscreen rail which allowed me to mount two sets of skyscreens, one set for each separate chronograph.
I run an Oehler 33 with a 4' skyscreen spacing and a PACT Pro with a 4' 6" skyscreen spacing. The results from this two-chrono setup have proven very enlightening, educational, and most importantly - useful. With my dual chronograph setup, I can tell with high confidence when I'm getting credible data, simply by comparing the difference in the readings from shot to shot. When I'm getting high quality data, the difference in the readings between the two over a string of shots will be very uniform. It's not so important to me that I get the same exact number. What's important is that the difference in velocities reported between the two chronies remains consistent and within a fairly narrow window.
I've learned chronographs can provide very accurate data when they're funtioning well. I've also learned that they can and will throw a bad number every now and then, and that without having two running chronos, a person would have no way to identify the bad data from the good.
If I get 2910, 2900, 2917, 2923, and 2908 on the Oehler, and I get 2917, 2907, 2924, 2930, and 2915 on the PACT, then I conclude I'm getting very good, valid data because the velocity difference was 7 fps for each of the five shots. That would be the ultimate in chrono consistency.
If the PACT Pro reads 2915, 2909, 2927, 2931 and 2914, then I'm again getting really good velocity data. The difference between the two chrono's is a minimum of 5 fps for the first shot (2915 -2910) and a maximum of 10 fps on the third shot (2927-2917). I say this is really good data because the difference in readings due to all errors in the equipment and setup is only 5 fps (10-5) for this five shot string.
How do I know when I've got a glitch in the data? When the differences in velocity produced by the chronies for any shot is much different than is typically present. For example, if I shoot 7 shots and the difference between the velocities recorded for these seven shots is 12, 16, 11, 14, 17, 39, and 65 fps, I know I received at least one bogus reading from at least one chronograph with each of the last two shots fired. I can't know if one chrono or both spit out bad data. But I can know that if I remove the last two shots from the data pool, that the velocities from the first 5 shots fired are pretty darn accurate.
This is how I get around the uncertainty associated with chronograph data. Based on my experiences and observations, I would never invest a lot of time and effort, and place a lot of confidence, in the velocity data produced by a single chronograph setup. After you run two chronos in series, you'll find it's not uncommon for bogus velocity readings to crop up every now and then. Now that I can identify bad readings from good readings, I do rely pretty heavily on my chrono's for load development. I can do that only because, with the dual setup, I am able to identify and toss out the bad data and then reach valid conclusions based on the knowledge that the remaining velocity data is relatively accurate and precise.
Lastly, my Oehler 33 is a better performing chronograph than my PACT - for those that might want to know. The PACT pukes out in dimmer lighting conditions quite a bit sooner than the Oehler, and tends to act up a little more than the Oehler. But they both work well enough most of the time to produce high quality velocity data.
