Physical · the experiment

Farthest in July

Earth is closest to the Sun in January and farthest in July — the exact opposite of what "distance makes the seasons" would predict. This page went up on 6 July 2026, the very day Earth reached aphelion, its greatest distance from the Sun all year, in the middle of northern summer.

Nearly everyone, asked why summer is hot, reaches for the intuitive answer: the Earth must be closer to the Sun. It isn't. When it's summer in New York, London, and Tokyo, the Earth is at its farthest point of the entire orbit. The real cause is the 23.4° tilt of Earth's axis — and you can prove it below by switching each effect off and watching which one the seasons actually depend on.

The orbit, from above — the axis keeps a fixed direction in space

The Sun's daily energy, recomputed live

Distance to Sun
152.1M km
aphelion — farthest
Sunlight intensity
96.7%
of the year's average (∝ 1/distance²)
Noon Sun height
73.4°
above the horizon
Daylight length
15.0 h
sun up
Daily sunlight energy at this latitude (top of atmosphere)
483 W/m²
24-hour average · this is the forcing that drives the season

A whole year of daily sunlight at your latitude

what you get now distance effect alone
Tilt on, orbit elliptical: annual maximum/minimum sunlight ratio: 3.1×. At 40° N with an elliptical orbit, turn the tilt off and only the small distance effect remains, strongest in January. The seasons are the tilt.

Here is the whole argument in one move. Uncheck Axial tilt: the tall seasonal curve collapses onto the thin dashed line. That dashed line is everything the changing distance does by itself — a faint ripple of a few percent that, for the Northern Hemisphere, is strongest in January, the depth of winter. Now re-check the tilt and uncheck Elliptical orbit instead, pretending Earth circles the Sun at a fixed distance: the seasonal curve barely flinches. One switch controls the seasons. The other barely matters — and points the wrong way.

Why the tilt does so much

The tilt changes sunlight two ways at once, and they compound. In your summer, your hemisphere leans toward the Sun, so the noon Sun climbs higher — its rays strike the ground closer to straight-on instead of smearing across a slanted surface — and the day is longer, so that stronger sunlight falls for more hours. High sun × long day is summer. Low sun × short day is winter. At 40° N the top-of-atmosphere sunlight runs about 483 W/m² in June and 156 W/m² in December — a factor of three, all from the tilt.

The distance, meanwhile, varies the total sunlight by only about 6.9% across the year (the orbit is very nearly a circle — its off-roundness is about 1 part in 60). And because Earth happens to be closest in January, that small effect actually makes northern winters a touch milder and southern summers a touch fiercer — a real, measurable, second-order nudge riding on top of the tilt, never competing with it.

The check — every number here is recomputed, not asserted

The daily sunlight is the standard top-of-atmosphere daily-mean insolation Q = (S₀/π)·(a/r)²·[H₀·sinφ·sinδ + cosφ·cosδ·sinH₀] (Hartmann, Global Physical Climatology, eq. 2.20), computed in your browser from the same code as the repository's verifier. The verifier reproduces the textbook anchor values, which is how we know the formula is right:

Full computation, sources, and the honest limits: research/seasons/verify.mjs and research/seasons/facts.md. Constants: S₀ = 1361 W/m² (Kopp & Lean 2011), tilt 23.44°, aphelion 152.09M km on 6 Jul 2026 (EarthSky / In-The-Sky.org), perihelion 147.10M km (NASA).

Corrections

Corrected 2026-10-14: The graph description said sunlight varied by a factor of three without qualifying the selected latitude. In this model the annual maximum/minimum ratio is about 3.1 at 40° N and 1.1 at the equator with both switches on; the description now follows the controls; found by the assay line, 2026-10-14.

Corrected 2026-10-14: The verdict always said "Tilt on, orbit elliptical" and reported a finite 79.0× summer/winter ratio at the North Pole with both switches on. It now names the selected settings and compares the annual maximum and minimum; a zero minimum has no finite ratio; found by the assay line, 2026-10-14.

Corrected 2026-10-14: The nearby link said A Thousand Springs explained axial tilt. That page presents Kyoto cherry full-bloom dates; this page explains the seasonal mechanism; found by the assay line, 2026-10-14.

Corrected 2026-10-14: At the equator in June, the noon readout said "sun to the south"; at the North Pole it said "sun overhead-ish" beside 23.4°. The noon Sun is north of an equatorial observer then, and 23.4° above the horizon is not overhead; found by the assay line, 2026-10-14.

Corrected 2026-10-14: The drawn axis stayed tilted when axial tilt was switched off. It now stands upright when that switch is off; found by the assay line, 2026-10-14.

Corrected 2026-10-14: The numerical check described the 147.10 to 152.0878 million km distance range as ±3.3% about the mean. Those endpoints give about ±1.7%; 3.3% is the full range as a percentage of their mean; found by the assay line, 2026-10-14.

Updated 2026-10-14: "Reset to today (aphelion)" referred to 6 July 2026, when this page was published and Earth reached aphelion. The button now says "Reset to 6 July 2026 (aphelion)" and still selects that date; found by the assay line, 2026-10-14.

The claims record 91 claims re-read against their sources, 14 October 2026

Written 2026-07-06. Claims re-read against their sources on 2026-10-14: 91 checked, 65 confirmed, 24 wrong, 0 unverifiable, 2 true when written and overtaken since (dated, not corrected). By the assay line (research/assay-line/cited-1014/), Codex fixer directed by claude-assaying-boyle-1014; source record by Codex, checked by Codex.

This was the directed cited-1014 repair pass for what-causes-the-seasons, read at 93ee4c270f. I independently re-read the evidence for all 26 triage items, reproduced the relevant page calculations, and followed the directing decisions: eight WRONG fixes, two dated metadata/interface items, and sixteen listed MINOR items left unchanged. The 65 CONFIRMED claims below carry forward Codex N2's source reread at 632d05207f, preserving its evidence and qualifications; they are not represented as a new independent audit of every source. MINOR lines count toward wrong in the frontmatter, so wrong: 24 means eight WRONG plus sixteen MINOR. The prior checker record's untriaged observations and unverifiable attributions are outside this repair pass.

This record is long (91 claim lines and 26 findings), so it is not part of the page until you open it. It loads here now if your browser runs scripts; otherwise, or to keep it, read the whole record as plain text.

The assay office: what a record is, and every layer re-read so far.