Points are a weird and inconsistent unit of measure

· books · Source ↗

TLDR

  • LaTeX uses 1/72.27 inch points (Knuth’s TeX definition); Inkscape uses 1/72 inch points (PostScript/CSS); a 0.4% gap that breaks grid alignment across tools.

Key Takeaways

  • The typographic point has no single origin: pre-digital US standards gave values ranging from 72.27 to 72.289 points/inch across sources from 1900-1916.
  • Knuth deliberately redefined the TeX point to exactly 72.27pt/in for cleaner arithmetic, diverging from the NIST printer’s point by ~1 part in 10^8.
  • PostScript chose 1/72 inch as “default user space” for mathematical convenience, explicitly noting it was not a true printer’s point.
  • Apple shipping PostScript in the LaserWriter locked in 1/72 inch; W3C adopted it for CSS and SVG, making it the dominant digital standard.
  • Frink’s unit library itself contains a bug: its texscaledpoint definition is off by ~50 attometers (~3% of a proton width).

Hacker News Comment Review

  • Commenters mostly engaged with tangential unit humor rather than disputing the technical history; no corrections to the LaTeX/PostScript lineage were offered.
  • CSS absolute units (cm, mm, in, pc, pt) all resolve to exact SI fractions in the spec – pt = 1/72 in = 127/360 mm – confirming the PostScript lineage is now the official web standard.
  • The EMU (English Metric Unit, used in Office formats) came up as a parallel example of an awkward compromise unit chosen to satisfy multiple measurement systems simultaneously.

Notable Comments

  • @djoldman: Full CSS unit table showing pt = 1/72 in = 127/360 mm exactly – useful reference confirming PostScript point as the web canonical value.
  • @masfuerte: Early 1980s Epson dot-matrix print heads spaced dots exactly 1/72 inch apart, a likely hardware forcing function toward the PostScript standard.

Original | Discuss on HN