New data show that island arc rocks have (210Pb/226Ra)o ratios which range from as low as 0.24 up to 2.88. In contrast, (228Ra/232Th) appears always within error of 1 suggesting that the large 226Ra-excesses observed in arc rocks were generated more than 30 years ago. This places a maximum estimate on melt ascent velocities of around 4000 m/year and provides further confidence that the 226Ra excesses reflect deep (source) processes rather than shallow level alteration or seawater contamination. Conversely, partial melting must have occurred more than 30 years prior to eruption. The 210Pb deficits are most readily explained by protracted magma degassing. Using published numerical models, the data suggest that degassing occurred continuously for periods up to several decades just prior to eruption but no link with eruption periodicity was found. Longer periods are required if degassing is discontinuous, less than 100% efficient or if magma is recharged or stored after degassing. The long durations suggest much of this degassing occurs at depth with implications for the formation of hydrothermal and copper-porphyry systems. A suite of lavas erupted in 1985-1986 from Sangeang Api volcano in the Sunda arc are characterised by deficits of 210Pb relative to 226Ra from which 6-8 years of continuous 222Rn degassing would be inferred from recent numerical models. These data also form a linear (210Pb)/Pb- (226Ra)/Pb array which might be interpreted as a 71-year isochron. However, the array passes through the origin suggesting displacement downwards from the equiline in response to degassing and so the slope of the array is inferred not to have any age significance. Simple modelling shows that the range of (226Ra)/Pb ratios requires thousands of years to develop consistent with differentiation occurring in response to cooling at the base of the crust. Thus, degassing post-dated, and was not responsible for magma differentiation. The formation, migration and extraction of gas bubbles must be extremely efficient in mafic magma whereas the higher viscosity of more siliceous magmas retards the process and can lead to 210Pb excesses. A possible negative correlation between (210Pb/226Ra)o and SO2 emission rate requires further testing but may have implications for future eruptions.