These graphs are cramped on a phone. The short version is on the argument page. Rotate the screen or use a desk for the full series.
What was the climate at Stonehenge, 5000–2000 BC?
There is no thermometer on Salisbury Plain in the Neolithic. The honest product is a range: Mauri et al. 2015 pollen-grid anomalies for the 1° cell over Wessex added to Boscombe Down 1971–2000. Bands are published reconstruction standard errors. Four millennial slices only — 5050, 4050, 3050 and 2050 BC. That is the same maritime temperate world as now, not a planning climate for metre-thick river ice.
Compared with 1991–2020 Boscombe Down (annual mean 10.4 °C, 783 mm), mid-Holocene Wessex sits in the same maritime temperate world. Winters around 5000 BC may have been ~1 °C colder than the late-20th-century mean; summers then look similar or slightly warmer. Rainfall is more uncertain than temperature. Gold’s formula wants about a metre of good ice to carry a 40-tonne sarsen; these winters do not grow that ice on the Avon. A hard January with lying snow on the downs is a different, smaller claim.
Stonehenge — estimated temperature
Solid lines are the pollen-grid reconstruction at Wessex. Shaded band is ±1 published SE on the seasonal anomalies, transferred onto January and July. Gold dotted lines are Boscombe Down 1971–2000. The 1991–2020 means sit a little above those dots.
Stonehenge — estimated rainfall
Seasonal Mauri anomalies (mm day⁻¹) converted to DJF and JJA totals and to a simple annual estimate (mean of the two seasons × 365). Annual precipitation in the GeoTIFF is not used: its units are internally inconsistent with the seasonal fields. The annual total assumes spring and autumn anomalies equal the mean of winter and summer — a possible bias if MAM/SON diverge. Shaded bands are ±1 SE. Peat wet-shifts elsewhere in Britain record water balance, not these millimetre totals.
Recipe, not a new climate model
1. Take Mauri EPOCH-2 anomalies at 51°N, 2°W (Stonehenge is 15 km from that cell centre).
2. Add them to Boscombe Down 1971–2000: year 9.85 °C, January 4.0 °C, July 16.75 °C, year 736 mm, DJF 215 mm, JJA 153 mm.
3. Draw the published SE as the band (~±1.0 °C; ~±25–45 mm per season).
4. Do not interpolate fake decadal weather between the four slices. The lines between dots are only a guide.
What the band is hiding
Pollen transfer functions are coarse. After ~4000 BC, Neolithic clearance mixes a human signal into southern English pollen. Mauri slices are 1000-year means, so a wet decade or a cold century is invisible. Chironomids at Talkin Tarn (Cumbria) give July temperatures of about 13–14 °C in the later window — cooler than Wessex, as today — with ±1.2 °C errors that swallow most of the wiggle.
How to read these graphs
Time runs older → younger, left to right, in years BC. The lines are measurements. The vertical washes are not. They are published time-windows other people have named — climate events, wet shifts, flood-phase compilations, dating spans. A coloured block does not mean the lines inside it jumped. Use the toggles under this card to switch each family on or off.
Hover a line for values. Click a legend item to hide a series. Do not convert a wet bog into a frozen river.
Vertical colour blocks are named time-windows (see the key above), not spikes in the data. Turn each family off with the boxes.
1. British and Irish hydroclimate
Peat water tables record summer effective precipitation. Higher Charman 2006 σ ≈ wetter. Charman 2000 is plotted inverted so upward is also wetter. Thorne Moors oak and Fenland yew ring widths are site hydrology / vitality, not rainfall mm. Fenland yew δ18O and Lancaster Hole cave δ18O are the two public British isotope series through this window. Macklin/Lewin UK flood-phase bands (toggle) are national 14C CPDFs, not an Avon hydrograph. The 2006 national peat stack only begins near 2493 BC. The first swing is poorly constrained (few cores) and sits on a known ~2500 BC wet shift also recorded as an Irish/continental bog-oak generation change — it is not a separate Irish peat curve.
1b. July temperature — Cumbria and western Ireland chironomids
Not Stonehenge. Talkin Tarn and Bigland Tarn are Cumbria; Meenachrinna is Donegal. Dotted lines: lake modern July, and Boscombe Down July as the Wessex present-day comparison (warmer, as expected).
2. Irish cave δ¹⁸O (Crag Cave CC3)
Rainfall source + cave temperature. Not a °C scale. The 2001 laser “8.2 ka crash” was an artefact; this is the robust 1999 / SISAL series.
3. Solar forcing (context)
Steinhilber et al. 2009 ΔTSI from ice-core 10Be, with published 1σ envelope. Forcing, not British weather.
4. Greenland temperature (context)
GISP2 Summit air temperature. Polar, amplified, and not Britain. Useful only as a North Atlantic backbone.
5. North Atlantic drift ice (Bond stack)
Higher % hematite-stained grains ≈ more drift ice / cooler subpolar Atlantic. Holocene “Bond cycles” are themselves contested.
6. Overlay — each series as a z-score inside the visible window
Each line is that proxy as a z-score in this window only. That is pattern-spotting, not a shared physical unit. The Charman peat stack is omitted before 2200 BC on this panel: those first centuries are few cores plus one outlier, and they coincide with Beaker-period archaeology. Do not read a climate crash off the start of a stack.
Series catalogue
Every plotted series, what it measures, and why the uncertainty matters.
Claims vs evidence
A frozen Avon as the sarsen road
Gold’s formula (Canadian ice roads: P = A h²) wants ~1 m of good floating ice for a 40-tonne load, more on a river (A ≈ 3.5). Mid-Holocene Wessex January in this reconstruction is a few degrees above freezing as a millennial mean — the same world as now. That does not accumulate the thousands of freezing-degree-days a metre of bearing ice requires. Lying snow on the downs (Hill 1961) is a separate, smaller claim. Pleistocene glaciers carrying the stones is a different debate and not what these graphs measure.
Centennial wet/dry swings
Raised-bog stacks are the strongest British evidence. They record water balance, mostly in the warm season. Events are real at centennial scale; exact calendar years are not.
Climate crash, then monuments, then Beakers
Sarsen-building is ~2500 BC; Beaker ancestry replacement starts ~2450–2400 BC; the 4.2 ka label is ~2200 BC. The big stones are earlier than that interval, not a response to it. Roland et al. (2014) found no coherent British–Irish peat event at 4.2 ka. Fenland yew ends ~2200 BC in wet/inundation terms, not a freeze. Replacement is real; a British climate catastrophe that explains it is not in these series.
Greenland cold snaps and Bond drift-ice peaks
These are North Atlantic series. They can share forcing with Britain (solar, ocean, storm-track) but they are not a thermometer for Wessex or Orkney.
Authoritative sources not yet plotted
How to cite
Cite the original papers and NOAA archives for any series you reuse. This site is a viewer, not a new reconstruction.
Suggested site citation: Daw / Grok (2026), Climate of Britain, 5000–2000 BC: referenced proxy explorer, v0.10. GitHub Pages.
Raw files used here were downloaded from NOAA WDS Paleoclimatology on 12 September 2026. See CITATIONS.md in the repository.