For each 250-μm source, we select the component with the highest assigned flux fraction. We plot these brightest components as a function of the total 250-μm flux in small background points. Horizontal error bars on the larger, binned points indicate the width of the bin, and the numbers below each point indicate the number of brightest components within each bin. Pale vertical error bars show the standard deviation on the distribution of all possible medians, including both the flux uncertainty and the dispersion of brightest component contained within the bin. Black vertical error bars (typically of comparable size to the data point) are calculated by dividing the pale error bars by |$\sqrt{n}$|⁠, applicable as each data point is an independent measurement and measurement error should not introduce a systematic bias. We refer to the black |$\sqrt{n}$| error bars for the remainder of this work. The brightest components show a trend towards higher fractions of 250-μm flux as total 250-μm flux decreases. A Spearman Rank correlation test on the unbinned data gives a p-value of 1.86 × 10⁻¹², which excludes the null hypothesis of no correlation at >6σ.