This is interesting. I have some LM385-1.2 which is essentialy the same IC as LM285 but rated for a narrower temperature range. I'll test and see how it works tomorrow.
Ok, so yes, it seems like there's something wrong with your LM285 or, maybe, the way you perform measurements.
I built the same circuit with LM385Z-1.2, R1=100k, Load=300k and measured the drop across the reference with Brymen BM869s in the 500k counts mode while changing V1 from 3.0V to 15.0V:
3.0V -> 1.22525V
15.0V -> 1.22527V
Last digit is fluctuating within a range of approx. +/- 2, so I specified what seems to be the middle point. Either way, it is obvious that it is very stable across a wide range of input voltage, or, to put it correctly, the current flowing through the reference.
p.s. any chance you mixed up the anode and cathode? Make sure that you identify the pins correctly,
onsemi datasheet shows a front view pinout that's hard to misunderstand, unlike the TI's bottom view one.