I don't know how Sennheiser gets their graphs.
By the looks of it they use narrow-band noise, at least that would explain the stepped 'target' plot.
These measurement types 'remove' sharp peaks and dips in a 'smoothing' way and the filter bands are similar to the 'bands' our own hearing is 'divided' in.
They also use diffuse field compensation.
It is actually a good way to measure speakers as well as narrow resonances are equalized, better than sine-wave sweeps.
Deviations around the target are allowed, meaning they can measure differently a bit from sample to sample, so variances (not too big) are likely to exist.
Even Sennheiser is not able to make them all exactly the same as there are too many mechanical variables.
Sennheiser (like a lot of other brands) is also known to make small changes during production over the years.
It is clearly stated on the box you can download the actual measurements made by them of your individual HP (and they probably aren't measured after thousand hours of burn in and with fresh pads) and since you guys have the serial numbers perhaps download and compare...
That could be a fun exercise and since these are all measured on the same rig in the same way under the same conditions this could tell something about the spread in production.
Doesn't mean they aren't excellent headphones.
Of course Purrin has a good point about different people and perhaps different positions on the head.
I assume Sennheiser is aware of this though.
Note all of the above is speculation on my part and not based on actual facts.