Tufte — The Complete Canon

Consolidated from 27 single-topic skills. Sources: The Visual Display of Quantitative Information (VDQI, 1983/2001), Envisioning Information (EI, 1990), Visual Explanations (VE, 1997), Beautiful Evidence (BE, 2006), Seeing With Fresh Eyes (SWFE, 2020). One ## section per source concept.

Graphical Excellence (VDQI ch. 1)

Quality = true ideas delivered per unit of viewer time, ink, and space. Five criteria (all constraints — failing one disqualifies):

1. Substance + statistics + design together; none compensates for another.
2. Clarity, precision, efficiency.
3. Maximum ideas, minimum time/ink/space.
4. Multivariate by default — "Graphical excellence is nearly always multivariate."
5. Truth — distortion voids everything else.

Four exemplary forms: data maps (densest — U.S. cancer atlas: 3,056 counties on one map; Snow's cholera map), time-series (>75% of published graphics 1974–80; reserve for genuinely variable data), narrative graphics of space-and-time (Minard, Marey), relational/scatterplot ("the greatest of all graphical designs"; ~40% of scientific graphics).

Minard's Napoleon map — "may well be the best statistical graphic ever drawn" — six variables in one band: army size (width), longitude, latitude, direction (color), temperature (linked lower graph), time (dates).

Graphics vs. tables: small N, labeled, non-comparative → table. Large N, multivariate, comparative, pattern-driven → graphic (NYC weather chart: 1,888 numbers in one display). Playfair invented the modern forms (line/bar 1786, pie 1801); the oldest known time-series is a 10th–11th-c. planetary-orbit manuscript graph.