Diverging ends must be symmetric in luminance

What it is

The two arms of a diverging palette must set out from the same lightness and arrive at comparable depth: they share the midpoint colour, and the lightness of both extremes should line up. This is easy to violate — pick two end hues by intuition and one is naturally dark (blue, red) while the other is naturally pale (yellow), leaving one arm visually heavy and the other light. Readers then read the heavy arm as the larger magnitude or the more consequential direction, even when the values on both sides are perfectly symmetric.

Why it happens

Diverging encoding divides the work cleanly: opposing hues carry direction; luminance distance from the midpoint carries magnitude. That is the direct consequence of lightness being the main carrier of perceived order. When the two arms' lightness trajectories are asymmetric, equal positive and negative deviations sit at different visual distances from the centre, and magnitude judgements are pulled toward the darker arm; where extreme values develop on the page shifts with them. Intrinsic lightness differences between hues are the default trap: choose end colours without checking L* and broken symmetry is almost guaranteed — and on a colour screen the bias is hard to catch by eye alone.

Where it stops holding

Symmetry refers to the lightness trajectory, not to mirrored hue positions on the colour wheel: classic diverging schemes such as RdBu and PuOr do not use complementary end hues, yet their lightness is symmetric. Nor should an asymmetric palette ever be the tool for data where one side genuinely matters more — that smuggles emphasis into the encoding, and readers cannot separate numeric difference from designer preference. To stress one side, use annotation, borders, or a dedicated highlight layer; keep the palette itself symmetric.

Applying it

• After choosing end colours, check both L* distances from the midpoint; they should be near-equal, and the two ends should look equally deep in greyscale.
• Greyscale self-check: convert the whole ramp to greyscale — the arms should form mirrored grey gradients, and one clearly darker end fails.
• Verification: build symmetric test data with +d and −d deviations and ask readers whether the two sides look equally far from centre; a systematic pull toward one arm means the palette is asymmetric — change an end colour or adjust lightness.

Related

• Same group: U4.03.1 Diverging palettes suit data with a meaningful midpoint · U4.03.2 Midpoint placement changes the conclusion
• Nearby: U4.02.3 Lightness is the main carrier of perceived order · U4.04.2 Lightness differences must guarantee separability
• Search terms: diverging palette · luminance symmetry · perceptually uniform