a lot of things affect this frequency response, and burn in is without a doubt not the cause of what we see. this is measurement malpractice in effect, not burn in. the low end is nonsense, even a failed seal probably wouldn't wobble like that. IDK if it's picking up noise(seems too regular for that), or if it's a messed up measurement setting that can do this? if I had to guess, I'd bet on having the device is on the desk, and he hit the desk with his hand while recording? depending on the output levels, just clicking the mouse to start the sweep could be enough(REW offers to set a delay between when we click and the actual start of the sweep, he probably could use that). but then again, trying to measure down to 2hz with an IMM-6 is brave. with the roll off, there is no way to get the necessary dynamic to avoid measuring noise in a quiet room while avoiding to distort the midrange by pushing it too loud. in any case the low end is obviously BS, the rest could be a change from placing the IEM 1 millimeters deeper. something as small and critical as trying to measure driver burn in, needs a test organized for it, not just sloppy measurements a few days apart. for starters, tips should be removed entirely when the objective is to measure variations. they're too much of a joker variable. and the IEM shouldn't move at all during the process, or it needs to have a sure way to position it exactly the same way for each take. the upper mids and trebles are reasonably close for typical FR measurements, a sign that he did try to get a similar position(maybe by measuring and moving until the result was close to the original?(that would defeat the purpose for this specific test thought). but it's nowhere close enough for when trying to demonstrate something like a variation over time. which based on my own experience with a bunch of IEMs, generally doesn't register on FR graphs aside from getting a clogged damper(filter) with dust and earwax. or if a change does show up, in my measurements it was so small I couldn't clearly isolate it from typical measurement variations(you measure twice in a row without changing anything or waiting and still you get some super tiny changes when zooming in on the FR, that's the sort of changes I measured for IEM burn in the few times I tried so far).