The scatter on L-T seems dodgy

I have a suspicion it may be related to how I'm using all observations to get those measurements, particularly for these large clusters any spectra which pass the checks XGA does for quality but are for the outskirts would really affect the simultaneous fit. 