I don't mean to doubt your hearing, lejaz. And in any case, I've never heard the DT 48 (nor the DF), so I'm going purely on speculation here.
From what I understand, your ear canal naturally resonates at around 3 and 10 kHz (and 15 kHz?). Which frequencies they are exactly depends on the length of your ear canal, and whether or not you hear peaks there depends on your phones.
I'm not an expert on this, but because the ear canal naturally emphasizes the sound around those frequencies, I would assume that your brain has come to automatically account for it. I.e. the brain expects to hear a certain amount of emphasis on those frequencies, and thus it sounds normal to you instead of peaky. If the source emphasizes those frequencies either up or down, you'd hear that as a deviation, i.e. a peak or a dip.
From what I understand, if you were a headphone manufacturer and wanted your phones to conform to f.e. the free field spec, you'd get yourself a dummy head to test on and tweak the phones until the dummy head gave you a graph that matched the free field curve. If you then subtracted the free field graph from the graph of your phones, you'd get a flat line, i.e. your phones would produce sound that got bounced around a certain way, arriving at the eardrum of that dummy in a certain way and its dummy brain would interpret it as flat. (I'm not going to hazard a guess as to whether the free field curve represents anything accurately or not, because I don't know.)
Again, I'm not an expert, but I assume the problem comes in when you (being the headphone manufacturer) take your phones out of the lab and plant them on the heads of random people. Their ears vary from each other (and their brains do as well), meaning that a different type of sound would arrive at their eardrums than what you predicted in the lab, and their brain would process it in a different way. Thus your phones wouldn't sound flat anymore. The free field curve, for instance, seems to center around 3 kHz or so, but if your test subject had a shorter or longer ear canal, their 3 kHz ear canal resonance wouldn't be exactly at 3 kHz but a bit below or above that, resulting in peaks and dips at that area when they listen to your phones.
Anyway, that's how I understand it, and I don't understand it too well. (If lejaz hears a peak at 2400 Hz and EYEdROP has it at 2700 Hz, and assuming those are the 3 kHz ear canal resonances showing up, then lejaz possibly has a longer ear canal, which would be roughly 3.5 cm or so vs. 3.1 for EYEdROP. But that's just a guess.)
Edited by vid - 2/17/12 at 5:30am