Phosphorus has been shown to have a profound influence on the evolution of silicate melt systems. It also controls the geochemical behavior of rare earths, uranium, thorium, strontium and other elements through the relationships between melt and accessory phosphates. Based on experimental data, Harrison and Watson (1984) proposed a model of saturation behavior of apatite in crustal melts as a function of temperature and silica content. Subaluminous granites fit in well with such a model, but departures are evident in the case of peraluminous granites. A systematic study of the distribution of P2O5 in granitoids from all over the world shows that there are two modes to describe phosphorus enrichment in peraluminous granitic rocks: (1) The Hoyos-type trend, which is parallel to the one in Harrison and Watson's (HW) model but with higher P2O5 concentrations at the same silica contents and (2) The Pedrobernardo-type trend, implying an increase in phosphorus in the most SiO2-rich differentiates of strongly peraluminous, low-Ca granites. Pedrobernardo-type leucogranites are characterized by the presence of rare phosphates or P-rich feldspars. Since they have excess P2O5 over CaO to form normative apatite, we here propose the term ''perphosphorous granites''. High phosphorus contents appear to be due to elevated apatite solubility in peraluminous granitic melts rather than to high amounts of restitic apatite. Both the Hoyos- and Pedrobernardo-type departures from the HW model can be ''corrected'' by using the expression: P2O5corrected = P2O5HW X e(factor), where P2O5HW is the prediction according to the HW model, and ''factor'' is (ASI-1 ) x (6429 /T), ASI being the Al2O3/(CaO + Na2O + K2O) molar ratio. We suggest that such departures could be caused by a decrease in the Ca activity in the melt as a result of increased peraluminousity. The Pedrobernardo-type trend is a limiting case in which Ca-activity in the melt is nearly zero and calcium, instead of phosphorus, becomes the essential structural constituent (ESC) of the apatite. When this happens, phosphorus starts behaving as an incompatible element that is concentrated in late fluids. High P2O5 contents are more characteristic of S- than other granite types. No significant departures from the HW model have been found for I-, A- or M-type granites. In Central Iberia, granites associated with U, Sn and W mineralizations usually have high phosphorus contents. Values Of P2O5 > 0.5 wt% at SiO2 > 70 wt.% are good indicators of economic mineralization.
