NATURE|Vol 436|28 July 2005 BRIEF COMMUNICATIONS ARISING E3 To put global warming into context requires knowledge about past changes in solar activity and the role of the Sun in climate change. Solanki et al. 1 propose that solar activity dur- ing recent decades was exceptionally high compared with that over the preceding 8,000 years. However, our extended analysis of the radiocarbon record reveals several periods during past centuries in which the strength of the magnetic field in the solar wind was simi- lar to, or even higher than, that of today. Solanki et al. combine radiocarbon ( 14 C) data, visually observed sunspot numbers and models to extend the historical sunspot record over the Holocene. They exclude the most recent 100 years of the 14 C record, which are influenced by 14 C-depleted fossil-fuel emis- sions and atomic-bomb tests conducted since AD 1950. We extend the analysis of the radio- carbon record to AD 1950, which allows us to link the 14 C-based solar reconstruction to instrumental measurements of solar magnetic modulation that cover the past 68 years 2,3 . The Sun influences the production rate of 14 C in the Earth’s atmosphere by modulating the galactic cosmic-ray flux through its mag- netic field. Increased magnetic field in the solar wind causes a stronger deflection of galactic cosmic rays and lower radionuclide produc- tion rates in the atmosphere and vice versa. However, the atmospheric 14 C concentration also depends nonlinearly on the geomagnetic field intensity and the global carbon cycle. These factors and their uncertainties need to be carefully included in the reconstruction of solar activity. We reconstructed 14 C production rates from radiocarbon records 4,5 (Fig. 1a) using two box-diffusion type carbon-cycle models, the box-diffusion 6 and the Bern model 7 , and a three-dimensional ocean model coupled to a four-box land biosphere. The 14 C production record was transformed 8,9 into a record of the solar modulation parameter that describes the solar influence on galactic cosmic-ray deflec- tion by normalizing it to neutron-monitor and ionization-chamber data covering the recent decades 2 . Alternatively, balloon-borne esti- mates of galactic cosmic-ray deflection 3 were used instead of the ionization-chamber data. Like Solanki et al. 1 , we assumed that natural variations in the carbon cycle were small during the past millennium, which is consis- tent with ice-core CO 2 and 13 CO 2 data and models 10 . We considered uncertainties in 14 C data, fossil CO 2 emissions, the geomag- netic field and model parameters. Results are similar across the range of models and simu- lations. Calculations using the Northern 4 or Southern 5 Hemisphere 14 C record, or both, reveal that differences between the 14 C records do not affect our main conclusions. It is standard practice to model the observed dilution of the atmospheric isotope ratios caused by the addition of isotopically depleted carbon from fossil and land-use sources. Emissions from fossil sources are prescribed, emissions from land-use sources are inferred from the atmospheric carbon budget, and the two-way exchange fluxes between reservoirs are simulated. The dilution of 14 C is governed by the same processes that affect 13 C. The good agreement between modelled 13 C and ice-core data (Fig. 1b) supports the reconstructed rate of 14 C production. The comparison between the physical quan- tities, the 14 C-production rate and the solar- modulation parameter, and the visually based sunspot record reveals similarities and striking differences (Fig. 2). The 11-year solar cycle is distinct in all records. On the other hand, solar magnetic modulation was higher or compara- ble to today during the late eighteenth (and twelfth) century and around AD 1600, whereas sunspot numbers were highest over the recent decades. Sunspot numbers fell to zero during the Maunder Minimum (AD 1650–1700), whereas 14 C production and solar modulation continued to vary. The balloon-borne mea- surements 3 imply lower values for the solar modulation parameter than the ionization- chamber data 2 . Those data seem preferable because they agree better with the solar- modulation changes inferred by Solanki et al. 1 than do the balloon-borne measurements 3 . If we include the Northern and Southern Hemi- sphere 14 C records (three-dimensional model) and normalize the data to the balloon-borne record, we obtain very similar levels of solar modulation (black curve in Fig. 2b). In any case, and irrespective of the data set applied, the recent solar activity is not excep- tionally high (Fig. 2). The 14 C results are broadly consistent with earlier reconstructions based on 10 Be data from the South Pole, which show that production rates around AD 1780 and in the twelfth century were comparable to those observed today 11 . We conclude that CLIMATE How unusual is today’s solar activity? Arising from: S. K. Solanki, I. G. Usoskin, B. Kromer, M. Schüssler & J. Beer Nature 431, 1084–1087 (2004) 0 20 –20 ∆ 14 C (‰) –6.6 –6.2 –7.0 δ 13 C (‰) Age (yr ad) 1500 1600 1700 1800 1900 2000 a b Figure 1 | Measured atmospheric 14 CO 2 : 12 CO 2 and 13 CO 2 : 12 CO 2 ratios and model results for Ȏ 13 C from AD 1500 to 1950. ǵ 14 C and Ȏ 13 C denote the per mil deviations from a standard. a, Tropospheric radiocarbon from tree-ring measurements from the Northern Hemisphere 4 (black) and Southern Hemisphere 5 (brown). Error bars show one standard deviation. b, The Law Dome (Antarctica) Ȏ 13 C record 14 (dots) compared with the output of the Bern model (black line). Error bars show one standard deviation. Figure 2 | Radiocarbon production rate and solar modulation parameter compared with the group sunspot number. a, 14 C-production rates (blue) were calculated from the Northern Hemisphere 14 C record with two carbon-cycle models 6,7 ; fossil-fuel emissions (DŽ10%) and air–sea CO 2 exchange rate (DŽ30%) were varied systematically in sensitivity calculations (various lines). The shadowed area shows the 1Ȝ errors resulting from 100 Monte Carlo simulations. We included Northern Hemisphere 4 and Southern Hemisphere 5 14 C data in a three-dimensional ocean model calculation (green line). The black curves show the 14 C-production rate corrected for the geomagnetic field intensity 8,9 . The red curve shows the group sunspot number 15 . All records are normalized and show 11-yr averages. b, The solar modulation parameter was calculated 8 from the 14 C-production rates and from ionization- chamber and neutron-monitor data 2 . The grey band shows results from annual 14 C data, including the error range based on the Monte Carlo calculations and uncertainties in the geomagnetic data 9 . The black line shows 11-yr averages (best estimate). Unfiltered (blue) and 11-yr-averaged (red) sunspot numbers 15 are shown for comparison. The green line shows the three- dimensional model results. The purple line depicts results using an alternative neutron-monitor record 3 that indicates a stronger trend from 1937 to 1950 than the ionization-chamber data 2 . 20 40 60 80 100 14 C production rate (normalized) Age (yr ad) 1500 1600 1700 1800 1900 2000 0 50 100 150 Group sunspot number Solar modulation parameter (MeV) 200 600 1,000 1,400 0.6 0.8 1.0 1.2 0.4 a b 0 1.4 © 2005 Nature Publishing Group