Holocene CO₂ & Temperature — the meme’s own sources

What the meme actually plots. The chart loads in the configuration the meme circulates — the bold red “Vinós 2022 — meme version” temperature curve and the pale-blue Monnin Dome C CO₂ line. Everything else is hidden until you click it on in the legend.

Where the Vinós curve comes from. Javier Vinós is a retired biochemist; the curve labelled “Vinós 2022” appears in his self-published Climate of the Past, Present and Future (2022), not in any peer-reviewed paper. Its construction has two steps:

1. Differencing (defensible) — take the 73 individual proxy time series from Marcott et al. 2013, convert each one to an anomaly relative to its own Holocene mean, then average across all proxies present in each 20-yr bin. The result is a wigglier stack than Marcott’s published curve because none of his area weighting or Monte-Carlo dating perturbations are applied. The wiggles are real proxy variability that Marcott smooths down by design.
2. ×2 rescaling (the manipulation) — the natural conversion from Vinós’s Z-score back to °C is to multiply by the stack’s standard deviation. Vinós instead multiplies by roughly twice that, doubling the apparent amplitude so the Holocene Climate Optimum reads ≈1.2 °C warmer than the Little Ice Age, his stated target. The proxy data itself supports only ≈0.6 °C. He admits the rescaling on Judith Curry’s blog as matching “biological, glaciological, and marine sedimentary evidence” — evidence the 73 proxies already are.

Turn on “Vinós method — honest scale (×1)” in the legend to see step 1 alone: same wiggles, half the amplitude. The gap between the two red curves is the manipulation.

The honest references. Marcott’s own published Standard5×5 and RegEM global stacks (orange and amber) are the area-weighted, dating-perturbed versions of the same proxy data. RegEM in particular explicitly corrects for proxy drop-off at the modern end — the exact problem Vinós claims to be fixing. Both show a HCO–LIA delta consistent with the honest ×1 curve, not the meme’s ×2.

The missing modern record. Monnin 2004 ends in 1777 CE and the Marcott proxy stack effectively ends around 1800 CE, so a chart built only from those two records crops out the entire industrial CO₂ rise and the corresponding warming. Toggle “HadCRUT5 — 100-yr smoothed” and “Mauna Loa” on, then zoom to 200 yr: even after smoothing HadCRUT down to Marcott’s effective resolution, the post-1850 warming alone exceeds the entire Holocene swing in any of the proxy reconstructions.

Two y-axes. Temperature (°C anomaly vs 1800–1900) on the left; CO₂ (ppm) on the right. Marcott’s stacks are rebaselined from their native 1961–1990 reference by subtracting their mean over the 50–150 yr BP window (≈ 1800–1900 CE) so every temperature curve shares HadCRUT5’s baseline. Full methodology in VINOS_DECONSTRUCTION.md.