a) top: power spectra of the noise in the TOIs for all bolometers at 143 GHz without (black) and with (red) subtraction of long glitch templates. The data corresponds to sky-subtracted TOIs, before any further processing; one can see, for instance, 4 K lines harmonics at 10 and 30 Hz which have not been removed at this stage (this is done later in the TOI processing pipeline). Two of the 143 GHz detectors have a rate of long glitches about a factor of five below that seen in those with the highest glitch rate. The power spectra for these two bolometers are those with the lowest amplitude at around 0.03 Hz; as can be seen from the figure, the template correction in these two cases has only a very small impact. As a result, these bolometers can be used to test for noise from sources other than glitches. This plot illustrates the efficiency of the template subtraction method in reducing contaminant noise in the frequency range 0.02 Hz to 2 Hz whilst simultaneously reducing the amount of flagged data. b) bottom: the auto-spectra after deglitching of 3 of the 143 GHz bolometers of the top panel, in black, are compared with their cross-spectra in blue and red. The blue curve corresponds to the cross-spectra of two bolometers in a PSB pair. This suggests that the decorrelation from common mode thermal fluctuations cannot have much of an effect on the noise at frequencies greater than the spin frequency at 0.02 Hz.