It already needs some time to get a stable temperature
RTFM
Note 2: All testing is done at 25°C. Pulse testing is used for LTZ1000A to
minimize temperature rise during testing.
And against other data sheets like LT1021 or LT1236 there is no 100% noise testing specified.
The newer data sheet of the newer "3rd generation buried zener" LT1027 already specifies a method with reduced testing time
Note 5: RMS noise is measured with an 8-pole bandpass filter with a
center frequency of 30Hz and a Q of 1.5. The filter output is then rectified
and integrated for a fixed time period, resulting in an average, as opposed
to RMS voltage. A correction factor is used to convert average to RMS.
This value is then used to obtain RMS noise voltage in the 10Hz to 1000Hz
frequency band. This test also screens for low frequency "popcorn" noise
within the bandwidth of the filter. Consult factory for 100% 0.1Hz to 10Hz
noise testing.
So the LT1027 is tested from around 22 -41 Hz (with minimum 50 ms testing time).
(no 100% test specified).
In the LTZ1000 datasheet there is no noise testing specified.
(it is always more essential to know what is missing in a data sheet as that what is specced).
with best regards
Andreas