And, just because I could... using a VNA as an SA.
2122X set to 24.576132MHz using internal timebase
-20dBm output level
The spec says -55dBc spectral purity at this frequency. I was skeptical this would truly be relative to the carrier (dBc) and thought they might mean relative to peak output (5Vpp).
Good old VNA set to 100Hz span centered on 24.576135MHz (just the peak the marker found at RBW=10, set to center)
100Hz span
1Hz RBW
40 sec sweep (!!!)
10dB/div V
10Hz/div H
This instrument is not the most agile at such a narrow span and 1Hz RBW... kind of pushing it.
It reports a peak at 24576133.5... My 53131A counter indicates it's 24576132.45Hz, so the VNA is off just a smidgeon over 1 full sweep step.
But... looks like the 2122X handily passes the spec of -55dBc for 20-40MHz.
Didn't grab a shot at RBW 10Hz, but it looks very similar, just tighter and more evenly stepped.
EDIT - here's a better one with a 100sec sweep time (4 sweeps averaged); for some reason the 3577A didn't like a 100sec sweep time until it was fully warmed up. Checking with the marker, -55dBc intersects at 5.5Hz on the high side and 6Hz on the low side.