We describe Ra-226- Th-230- U-238 (dis)equilibria and Rb-Ba-Th-U-La co
ncentration relationships in historical and prehistoric lavas from Kil
auea and Mauna Loa. (Ra-226/Th-230), (Th-230/Th-232), Th/U, Ba/Rb and
Ba/La ratios [but not (Rb, Ba, La) / (Th, U) ratios] are essentially i
dentical in both volcanoes, whereas the absolute concentrations (after
correction for olivine crystallization) differ by up to a factor of 2
, in response to varying melt fractions. This shows that bulk partitio
n coefficients of these elements are significantly smaller than melt f
ractions. Very small or absent Th-230-U-238 disequilibrium implies ver
y small or negligible magmatic fractionation between Th and U. Ra-226-
Th-230 disequilibrium are significantly larger (approximately 20% exc
ess Ra-226 on average) but are also independent of melt fraction. The
combination of significant Ra-Th fractionation together with small or
absent Th-U fractionation provides constraints on recently proposed mo
dels to explain U-series disequilibria during partial melting and melt
extraction. Instantaneous melt extraction models are rejected: (a) be
cause they are inconsistent with experimentally determined partition c
oefficients; and (b) more generally because they would require signifi
cant covariation of (Ra-226/Th-230) with melt fraction. On the other h
and, dynamic melting models involving slow fractional melting or melt
infiltration within the garnet stability region, followed by rapid mov
ement through the lithosphere, are consistent with the results and yie
ld melt porosities between 10(-3) and 10(-2) for plume upwelling veloc
ities of 1 m yr-1. In addition, we tentatively proposed alternative mo
dels for creating the Ra excesses in the magma. One such process invol
ves the mobilization of Ra within the volcanic edifice, subsequent adv
ection toward and redeposition within the roof region of the magma cha
mber, and finally incorporation into the magma itself. Another mechani
sm for incorporating excess Ra in the magma might be transport of very
small amounts of carbonate fluids or carbonatite melts (containing ve
ry large excesses of Ra-226) into partially molten regions in the mant
le. Given the currently available data and state of knowledge about ma
gma extraction processes, there is no obvious preference for either th
e purely magmatic models or those involving ''extraneous'' fluids in t
he mantle or within the volcanic edifice.