The scale type must be prominently labelled

What it is

Readers bring an unchecked default assumption to any chart: linear, ascending, starting at zero or the conventional origin. Any departure from that default — logarithmic, square-root, power, reversed, truncated, indexed (base period = 100), normalized — rewrites the conversion rules without the reader knowing; unlabelled, they proceed to misread every distance, slope, and area under linear assumptions. The labelling duty is therefore one-directional: linear needs no declaration (it is the shared default); every nonlinear scale must be declared, and the declaration's strength must match the cost of misreading.

Why it happens

The default assumption is powerful because it is silent: readers do not ask "is this axis logarithmic," any more than they ask "is this text Chinese" — presets never enter awareness. A scale declaration inserts an explicit interruption before data reading begins: text ("logarithmic scale," "indexed, base = January 2020") beats graphical hints (the uneven tick spacing itself), because subtle spacing irregularities go undetected in a scan. Placement and wording matter equally: next to the axis, no smaller than the tick labels, stating the semantics ("equal spacing = tenfold") rather than the type name alone — "log" alone says nothing to those who do not know it.

Where it stops holding

This leaf governs value axes and time axes; declaring colour-scale choices (diverging midpoint, clamping, transforms) belongs to the colour family, though the same principle holds there. The reversed axis (values increasing leftward) is a special case: strongly discouraged even when labelled, because it fights the deeper left-to-right increase reflex that labelling cannot rescue. Indexed and normalized scales, though linear, must also declare their conversion base — without knowing "what 100 means," readers cannot extract magnitude.

Applying it

• Checklist the transformations: axis transform (log/sqrt/power), reversal, truncation, indexing, normalization — any present triggers a text declaration beside the axis stating the conversion semantics.
• Declaration spec: adjacent to its axis, type size at least the tick size, same weight as the axis; repeat in the caption as a second line of defence.
• Verification: after a three-second inattentive scan, ask "what scale is this axis?"; an unanswered declaration counts as absent — escalate until answered.

Related

• Same group: U3.03.1 Log scales suit data spanning orders of magnitude · U3.03.2 Untrained readers misread logarithmic axes
• Nearby: U3.01.2 Position encodings tolerate truncation when clearly marked · U3.02.3 Break markers are often too small to notice, equal to unmarked
• Search terms: scale labelling · nonlinear scale · axis annotation · index normalization