Here we show the ratio between our 2D power spectrum and the error bars estimated by the empirical covariance method of D15. On the top left, we show our data calibrated using our initial calibration (see Section 2.2) with no attempt made to correct for standing wave structure in the MWA bandpass. Bright, band-like structures are clearly visible at the delays associated with reflections. On the top right, we show a first attempt to correct for cable reflections by fitting a sinusoidal model to rather noisy calibration solutions that had been integrated over a night of observing (1.5 h each night). While the bands appear weaker, they are still quite visible above the noise. In the bottom right panel, we show the same plot with calibration solutions using scaled autocorrelations described in Section 3.4. In the lower left panel we show a power spectrum with calibration solutions using autocorrelations for the amplitudes but without any attempt to correct reflections in the phase solutions. Pronounced reflection features are visible in this power spectrum, indicating that any mismodelled reflection structure in the phases will contaminate our measurement.