JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 99, NO. B9, PAGES 17,915-17,917, SEPTEMBER 10, 1994 Reply Ingi T. Bjarnason Department of Terrestrial Magnetism, Carnegie Institution of Washington, Washington, D. C. William Menke Lamont-Doherty EarthObservatory of Columbia University, Palisades, New York 01afur G. F16venz Orkustofnun, Reykjavik, Iceland The centralquestion discussed by Gudmundsson [this issue] can be succinctly stated: "Is the temperature of the shallowest upper mantle of Iceland at the peridotitesolidus" (nominally, 1200øC). The traditional view, as developed by numerous authors during the 1970sandearly 1980s (reviewed by Palmason [1986]) and to which Gudmundsson ascribes, is that it is supersolidus andpartiallymolten. For instance, Palmason [ 1981, 1986] uses a numerical model of crustal accretion and coolingto predict that the solidus is at a depth of 7 km beneath the volcanic zone (the Icelandic analogof a mid-ocean ridge), descending to 15 km for 5 Ma crust. Our tentative hypothesis, as expressed by Bjarnason et al. [1993], is that lower crustal and shallowest upper mantle temperatures, at least in southwest Iceland, are substantially cooler, -300øC below the peridotite solidus. We presented two observations that support this view: The first observation is that compressional velocities are in the range of 7.0-7.2 krn/s at depths 10-20 km, a region that we interpret as being part of the lower crust(on the basis of it being above a prominent20-24 km deepreflector that we interpretas Moho). This interpretation differs from previous work [Gebrande et al., 1980; Hermance, 1981; Hersir et al., 1984], whichassumes a considerably thinner, 10-15km thickcrust. The previous models interpreted the material with 7.0-7.2 krn/s compressional velocity as being peridotite mantleat supersolidus temperature with large amounts (>10%) of partial melt present. Our new interpretation models the 7.0-7.2 km/s materialasbeing gabbroic lower crust, without much partial melt, as the compressionalwave seismic velocity of unmelted gabbro is -7.17 km/s in the 10-20 km depth range(D. Mackensie, personal communication, 1993), close to the observed value. The compressional velocity decreases by 1-2% (0.07-0.14 km/s) per percent of melt, depending upon whetherthe melt is distributed in tubules [Mavok, 1980] or on crystal faces[Faul et al., 1994]. Hence while the observed lower crustalcompressional velocity might be compatible with very small (-1%) melt fractions, it is not compatible with the large (>10%) melt fractionhypothesized previously [Gebrande et al., 1980;Hermance, 1981;Hersir et al., 1984]. Indeed, these large melt fractions would cause a compressional velocityinversion in the lower crust andproduce a prominent seismic shadow zone. Sucha feature is not observed in our refraction lines,nor in any others of which we are aware. Copyright 1994by the American Geophysical Union. Paper number 94JB01631. 0148-0227/94/94JB-01631 $02.00 We are not the first authors to postulate a thick crustunder Iceland. Bath [1960] gives a 27.8 km crustal thickness, and Pavlenkovaand Zverev [ 1981 ] favor 30 km crustal thickness. A crustal thickness estimate, also in the 30 km range, is also available for the geologicallyrelated Iceland-Faero ridge (the Tertiary trail of the Iceland hotspot) [Bott, 1971]. However, all these estimates are based on interpretations of sparse analog data that do not show clear Moho reflections and hence must be treated cautiously. Our secondrelevant observationis a relatively fast mantle refractor velocity. This refractor, associated with seismic rays thatturnin the mantle directly beneath the western volcanic zone, has an apparent velocity of 7.74 + 0.05 (1 o) km/s. We argued that this velocitywasfaster thanwhat mightbe expected from a supersolidus mantle (i.e., >1200øC), on thebasis of laboratory measurements of the effectof temperature on the seismic velocity of peridotite [Murase and Kushiro, 1979;Sato et al., 1989a]. We would like to emphasize that the crustal and uppermost mantle velocitystructure presented by Bjarnason et al. [ 1993] is for southwest Iceland. It is clear from previouswork that all of Iceland does not have a laterally homogeneous lower crustal structure. For example, a clear7.7 km/srefractor is notobserved across central Iceland at the cross-over distance that we observed in southwest Iceland [Gebrandeet al., 1980], which may indicate that the crust is thickerthere.We would alsolike to point out that our measurements in southwest Iceland do not have the resolution to determine precisely the nature of the crust-mantle boundary directly under the western volcanic zone, as Gudmundsson assumes in his comments. The mantle reflector is observed on either side of the volcanic zone but with only few bounce pointsdirectly under it. Similarly, mostof the turning pointsof the 7.74 + 0.05 km/s refractorare to the east of the volcanic zone. Gudmundsson comments on the uncertainties in measurements of the 7.74 + 0.05 refractor velocity and arguesthat the true velocity might be lower than our unreversed measurement. We think that the measured apparent velocity might actually be a slight underestimate of the true velocity, as it is the downdip measurement of the refractorvelocity. Gudmundsson's main comment is concerned with showing that a slightly lower mantlerefractor velocitythan we observe may actually be consistent with an uppermost mantle and lower crust at solidus temperature, by developing whathe considers thebest chemical composition of the upper mantle under Iceland and applyingan equation of state that relates composition, pressure, and temperature to compressional velocity. Our opinionis that while Gudmundsson uses perfectly plausible estimates, the level 17,915