The N-S–trending Andes surmount South American lithosphere above the east-descending Nazca plate. The highest mountains, underlain by ∼70-km-thick crust, cap the Cordillera at 25°S, an extremely arid region. In contrast, the precipitation-drenched fjordland at 45°S is supported by ∼35-km-thick crust. The Cascade and Sierra Nevada ranges in the western United States display comparable N-S trends and latitudinal rainfall patterns. Westerly winds supply abundant moisture to the northwest, but precipitation diminishes southward, producing increased aridity where the Sierra achieves its maximum regional elevation around Mount Whitney. Overthickened, now delaminating, ∼42–55-km-thick crust of the southern Sierra exceeds that of the ∼35-km-thick northern Sierra and the active Cascade arc. Contrasts in orogenic crustal thickness in California are not as marked as in the Andes because Sierran arc construction ceased near the end of Cretaceous time. Geologic, geochemical, and stable isotopic data suggest that a Nevadaplano occupied central + southern Nevada + western Utah in the rain shadow of the Late Cretaceous–Paleogene arc. The ∼40–45-km-thick Colorado Plateau crust lies well downwind from the southern Sierra, and, depending on when it became elevated, it might represent part of a broader highland that collapsed during Neogene Basin and Range extension.


A topic of vigorous geomorphic-neotectonic research involves quantifying the competing influences of contractional-accretionary tectonic processes and uplift versus surficial, precipitation-induced erosion in sculpting the topography of active mountain belts (Koons, 1989, 2009; Molnar, 2003; Burbank et al., 2003; Reiners et al., 2003; Wobus et al., 2003; Lamb and Davis, 2003; Hodges et al., 2004; Willett et al., 2006). As a consequence of regional gravitational equilibration, crustal thickness in the vicinity of a young, high mountain belt is also partly a function of rain and snowfall, as well as contraction. Reflecting the orographic effect, climate and surface elevation appear to be intimately coupled (e.g., Molnar, 2009; Strecker et al., 2009). Using the Andes, the Himalayas, the Sierra Nevada, and the Japanese island arc as examples, the thickness of continental crust in active orogens has been linked to modern precipitation patterns (Ernst, 2004). An oversimplification, igneous activity, sedimentary + tectonic accretion, compressional shortening, crustal removal through lithospheric delamination + subduction erosion, exhumation, and climate nevertheless all play crucial roles in determining the elevation and regional sialic crustal thickness of active mountain belts, as schematically illustrated in Figure 1.

In general, for young convergent orogens, a dry climate results in feeble surface erosion, allowing the accumulation of thick, isostatically compensated continental crust and high relief, whereas a wet climate produces more vigorous surface degradation and thus lower elevation supported by a moderate sialic crustal thickness. This review briefly compares three young, N-S–trending volcanic-plutonic orogens, the Andes, the Sierra Nevada, and the Cascade ranges. It emphasizes that climate-driven erosion to some degree influences crustal thickness and the existence or absence of a downwind plateau. The climate specific to each orogen displays an important latitudinal temperature gradient, and each longitudinally disposed terrane lies athwart the prevailing wind patterns. Mountain ranges are exceedingly complex, dynamic products of a broad range of processes not comprehensively treated here, but assuming the regional operation of isostasy, erosion due to rain and snowfall must be related somehow to the overall crustal thickness. However, as recently shown by Pelletier et al. (2010), climate exerts a subsidiary role in controlling the elevation of the Andes.



The thickness of sialic crust surmounting convergent plate junctions is a dynamic function of competing constructional and destructional processes (Molnar and England, 1990; Koons, 1995; Beaumont et al., 1996, 1999; Lamb and Davis, 2003; Anders et al., 2006; Roe et al., 2006; Stolar et al., 2006). Constructional mechanisms involve primary calc-alkaline magmatic additions derived from the subducting lithosphere and/or the upper mantle, as well as from plate underflow (contraction), transform motion (strike-slip), terrane suturing, and accretionary prism offloading. Destructional processes include plate divergence (except where metasomatism and juvenile igneous additions to the backarc overwhelm the effects of extension), mass wastage, surface weathering + erosion, subduction-induced subcrustal removal, and delamination + foundering of the lower, mafic continental crust. Pacific-type paired mountain chains consist of an outboard subduction complex and a subparallel inboard Andean-type arc of massive proportions, whereas Alpine-type orogens chiefly display the effects of continental collision and lack a substantial coeval volcanic-plutonic arc. Ancient cratons have attained thermal and structural stasis throughout the lithosphere, due to tectonic quiescence maintained by long-term decoupling from asthenospheric flow, reflecting the establishment of a steady-state mantle-crust geothermal gradient. In contrast, geologically young mountain belts forming along convergent plate junctions may possess thinner, or more commonly thicker, crust than old cratons because, although regional isostatic compensation is approximated, the complex interplay between constructional and destructional processes is influenced to some extent by local erosion rates (e.g., Simoes et al., 2007; Malavieille, 2010), and the latter necessarily lag constructional topographic expression.

The regional elevation of an active orogen is a dynamic function of the petrotectonically added volume per unit area minus the erosional loss (Horton, 1999; Willett, 1999; Willett and Brandon, 2002; Whipple and Meade, 2004; Hilley et al., 2004; Hilley and Strecker, 2004). Erosion rates in modern mountain ranges more or less track with precipitation (Strecker et al., 2003). However, the mass and mean elevation of a young mountain belt are the net result of a much more complex set of variables (e.g., Burbank et al., 2003; Pelletier et al., 2010), involving chemical and physical weathering, nature and erosional resistance of the geologic substrate, proximity to erosional base levels, etc. Steeper topographic gradients promote more rapid erosion for a given precipitation rate, but as rain and snowfall decrease, contraction and uplift may block external drainage systems and produce elevated, internally draining plateaus (Burbank et al., 1996; Whipple and Tucker, 1999; Sobel et al., 2003). Local and regional climates are complex functions of Earth's evolving atmospheric and oceanic flow regimes: coupled wind and surface ocean circulation patterns (functions of latitude, regional albedo, obstructing and/or deflecting land masses, Coriolis effect, etc.). Annual global precipitation and major oceanic surface gyres are depicted schematically in Figure 2.

How do climate and erosion impact constructional features of an active mountain belt? At roughly 30°N and 30°S, descending Hadley cell air masses heat and desiccate adiabatically, establishing belts of global aridity. Due to convergent plate motions, mountain chains and scattered desert plateaus at such latitudes mark the eastern Pacific Rim (Mojave-Colorado, Atacama-Altiplano), whereas the circum-Atlantic and circum-Indian continental margins are divergent, lacking young mountain belts but exhibiting the widespread development of deserts (e.g., Sahara–Persian Gulf–central Asia, Kalahari, western Australia).

The Andes

The volcanogenic Chilean Andes chain parallels the outboard Chile-Peru Trench, and overlies the eastward-inclined Nazca oceanic-crust–capped plate above the convergent plate junction (Allmendinger et al., 1997; Montgomery et al., 2001; Strecker et al., 2007). Generalized topographic and climatic relationships are illustrated in Figure 3. The locus of towering peaks and intensely contracted, thickest continental crust (55–75 km) occurs in the north-central Andes at ∼25°S (James, 1971; Zeil, 1979; Sheffels, 1990; Wigger et al., 1994; Baby et al., 1997; Horton, 1999; Lamb and Davis, 2003; Mamani et al., 2010). Directly offshore, cold water upwells from the north-flowing Humboldt Current. Along this low-latitude sector of the Chilean coast, relatively cool onshore winds are weak, being overpowered by the southeasterly trade winds. Landward, atmospheric circulation consists of descending high-pressure, hot, dry Hadley cells. Thus, the low humidity in the Atacama Desert is legendary, and precipitation is miniscule to nonexistent. The SE trade winds lose moisture as they pass over the Eastern Cordillera, and yearly precipitation is only ∼0.2–0.3 m in the High Andes (Montgomery et al., 2001; Hoke et al., 2004; Ehlers and Poulsen, 2009). Due to the orographic effect (Smith, 2006), air parcels passing northwestward sink over the Altiplano (3.6 km average elevation), and humidity falls, yielding annual precipitation rates of ∼0.3 m. Low precipitation translates to low erosion rates, partially accounting for the elevated calc-alkaline volcanic-plutonic arc and internally draining high plateau between the Eastern and Western Cordillera (Vandervoort et al., 1995; Sobel et al., 2003). Minor amounts of clastic debris are transported toward the Amazon Basin and the Chile-Peru Trench. Reflecting intense aridity, erosion is weakly developed along the north-central part of the Cordillera, so little volcaniclastic material is carried downslope to the Nazca plate subduction channel, a realm where increased friction on the slab results in intense contraction (Lamb and Davis, 2003). Sediment-starved subduction zones should favor active subcrustal erosion, as is typical of some segments of the Chile-Peru continental margin (von Huene and Scholl, 1991). However, in spite of the operation of this process, the great thickness of continental crust at ∼25°S evidently is mainly a product of calc-alkaline igneous activity combined with tectonic shortening, lower-crustal flow, and weak erosion. The present crustal thickness evidently cannot be due to accretionary offloading or partial fusion of subducted sialic material, because very little continental debris is supplied to the trench, nor can it be a result of delamination, which would thin—not thicken—the sialic crust (Mamani et al., 2010). Multiple stages of mountain building during Phanerozoic underflow of paleo-Pacific lithospheric plates apparently have been reflected in episodic plateau uplift events behind the arc at ca. 400–350, 80–50, and 30–5 Ma (Carrapa et al., 2009).

The climate is quite different at ∼45°S: The temperature contrast between the Humboldt Current and the land is reversed, and the land is cooler than the ocean. Moreover, low-pressure Ferrell cells rise through the atmosphere at this latitude, conducted by abundant H2O-laden westerly winds at ground level to the relatively cool Andes. Thus, the rugged, but less severely shortened, lower-elevation fjordland terrane receives ∼2–4 m of annual precipitation, producing ice caps and glaciated valleys (Montgomery et al., 2001); this alpine landscape surmounts a continental crust only ∼35 km thick (Davidson et al., 1997; Ramos et al., 2002). Due to moderate rainfall downwind beyond the Andes, eastward stream drainage is external, and a high plateau is lacking as sediments are transported toward the Atlantic margin. Western slopes of the Cordillera receive oodles of precipitation, promoting vigorous stream flow and active erosion, in the process transporting voluminous amounts of debris westward to the Chile-Peru Trench, where it is subducted (Lamb and Davis, 2003). In spite of this loading of the subduction zone, the southern Andean crust is only about half as thick as that at ∼25°S; apparently intense surface erosion, rather than sediment supply ± anatexis, is partly responsible for the normal crustal thickness. Although divided into rollback segments of differing dip angles, the physicochemical nature of the Nazca oceanic lithosphere is roughly similar in both north-central and southern Andes, so the primary volcanic-plutonic arc materials added to the crust prior to ca. 15 Ma should be roughly comparable in both northern and southern tracts of the mountain belt. However, reduced convergence and calc-alkaline magma production in the southern Andes apparently reflect Chile Ridge impingement since the middle Miocene (Gorring and Kay, 2001). A substantial sediment cushion in the subduction zone at 45°S might be expected to inhibit subcrustal erosion and instead cause offloading and accretionary growth, but if such mechanisms are operating (von Huene and Scholl, 1991), they have failed to generate a thick crust.

Western United States

The N-S–trending, volcanically active Cascade Range in western Washington + Oregon and the Cretaceous Sierran calc-alkaline arc in eastern California lie inboard of the southeastward-flowing California Current. Supplied by the marine waters, westerly winds bring abundant moisture to the relatively cool Pacific Northwest and to northern California. Southward, the land surface is substantially warmer, and Sierra Nevada crust reaches its maximum thickness in the high regional elevation of the Kern Plateau (i.e., the Mount Whitney area). Crustal thickness of the southern Sierra, ∼42–55 km, apparently exceeds that of the northern Sierra, ∼35 km, based on limited geologic and geophysical data (Fuis et al., 1987; Mooney and Weaver, 1989; Fliedner et al., 1996, 2000; Zhu and Kanamori, 2000; Heimgartner et al., 2005). Longitudinal contrasts in Sierran crustal thickness are not as marked as those that characterize the Andes, because mountain building in eastern California ceased around the end of the Cretaceous. Reflecting the dynamic overthickening of the southern Sierran arc attending Cretaceous ± earliest Tertiary subduction, delamination of a portion of the lower crust—the Sierran microplate—is now in the process of sinking into the deep upper mantle (Ducea and Saleeby, 1996; Saleeby et al., 2003, 2007; Zandt et al., 2004).

Like the Andean arc at ∼45°S, the active Cascade Range is currently supplying abundant detritus to the sediment-overwhelmed Cascadia Trench. Modest crustal thickness (e.g., Langston, 1979; Schultz and Crosson, 1996; Miller et al., 1997) probably reflects vigorous erosional decapitation in the Cascade Range. Similar to the southern Andes, abundant precipitation evidently moderates the average elevation of the constructional volcanic-plutonic arc and, reflecting the operation of isostasy, prevents the accumulation of a great crustal thickness as well as a downwind high-standing orogenic plateau behind the arc. Of course, reflecting a relatively anhydrous mantle wedge (Hacker et al., 2003), and the present slow oceanic plate subduction (Kreemer and Holt, 2001), the rate of Cascades calc-alkaline arc production is considerably less than that of the northern Andes.

Present-day average annual precipitation for the western United States is illustrated in Figure 4. Episodes of Miocene and younger transtensional faulting have severely extended the Basin and Range continental crust, so it seems possible that, in addition to igneous constructional processes, erosion linked to climatic patterns has influenced the regional crustal thickness in both the Eocene Sierra Nevada (Hren et al., 2010) and a behind-the-arc highland (C.P. Chamberlain, 2009, personal commun.). In addition, changing styles of lithospheric underflow (e.g., subduction angle, plate segmentation; Saleeby, 2003) apparently influenced the Laramide tectonics of this realm, so attribution of the Nevadaplano mainly to an arid paleoclimate and weak erosion must be viewed with caution. Does the existence of the Colorado Plateau (2.4 km average elevation) attest to causal relationships involving climate, topography, and crustal thickness? Sited downwind from the southern Sierra Nevada, it possesses relatively thick crust, ∼40–45+ km (Das and Nolet, 1998; McQuarrie and Chase, 2000; Frassetto et al., 2006) that currently supports the high plateau. It lies well to the leeward of the Sierra Nevada, but the Miocene and younger crustal stretching of what may have been an initially larger Late Cretaceous–Paleogene plateau has produced the intervening extensional Basin and Range Province. If the Colorado Plateau achieved its elevated position during early Cenozoic time, it might well be a surviving remnant of the Nevadaplano. However, the timing of uplift of the plateau is still unresolved (McQuarrie and Chase, 2000; Sahagian et al., 2002, 2003; Libarkin and Chase, 2003; Karlstrom et al., 2007; Flowers et al., 2008).


Chase et al. (1998), Dilek and Moores (1999), and DeCelles (2004) presented evidence supporting the late Mesozoic–Paleogene existence in Nevada of a broad, high plain underlain by thick continental crust consisting of high-grade metamorphic rocks. Terming this Late Cretaceous–Eocene hinterland metamorphic belt the Nevadaplano, DeCelles (2004, p. 147) estimated its average Paleogene elevation as probably exceeding 3 km. Sited in the rain shadow of the then-active Sierran volcanic-plutonic arc (House et al., 2001), the orographic effect on westerly wind-driven precipitation coupled with limited erosion and active contractional tectonism similar to that described for the Andes-Altiplano, and other internally draining high plains, could have stabilized a thick sialic crust-supported Nevadaplano (∼40–50 km, Gans, 1987; ∼45–65 km, Wernicke et al., 1996) well into Paleogene time. Later E-W extension and NW-directed shear during the Neogene to present-day (Wernicke and Snow, 1998; Bennett et al., 2003; Hammond and Thatcher, 2004) may reflect divergent asthenospheric flow, resulting in high-temperature thinning of the preexisting crust and gradual, piecemeal collapse of the plateau. The onset of Basin and Range transtension somewhat lagged the ca. 28 Ma initial over-running of the East Pacific Rise heat source by the North American lithosphere (Atwater, 1970; Atwater and Stock, 1998), but it may well be related to this thermal event.

Geologic and structural field relationships and metamorphic thermobarometry of basement rocks, combined with radiometric data, suggest the presence of an Eocene to mid-Miocene elevated upland of relatively low relief in the western Basin and Range and Sierra Nevada (Dalrymple, 1964; Christensen, 1966; Huber, 1981, 1990; Unruh, 1991; Chase et al., 1998; Lewis et al., 1999; Wakabayashi and Sawyer, 2001; Jayko, 2009; Lee et al., 2009; Saleeby et al., 2009). This widespread peneplain stood high, judging from stable isotope paleoelevation studies (Horton et al., 2004; Horton and Chamberlain, 2006; Mulch et al., 2006; Crowley et al., 2008; C.P. Chamberlain, 2009, personal commun.). Its surface was covered by a series of lahars, trachyandesites, basaltic lava flows, and associated volcanogenic sediments that extended E-W across the region for a distance of ∼200 km (Henry, 2008, 2009; Pluhar et al., 2009; Gorny et al., 2009). This elevated plateau was subsequently normal-faulted and dissected by late Miocene and more recent erosion.

Petrochemical data (major element, trace element, and 87Sr/86Sinitial values) support the existence of an elevated, laterally extensive Nevadaplano during Paleogene time, as documented by Best et al. (2009). These authors compared bulk-rock compositions of 376 Great Basin flows of the late Eocene–early Miocene ignimbrite flare-up in a compilation of more than 6000 mid-Cenozoic continental arc lavas worldwide, the geochemistry of which tracks with crustal thickness and elevation. Based on this comparison, the mid-Cenozoic crust of central + southern Nevada and western Utah was ∼60–70 km thick (Coney and Harms, 1984), and evidently was isostatically compensated as a high-standing plateau. In addition, a N-S–trending topographic high acted as a divide between the western slope and an eastern, volcanically smoothed surface.

In a much different type of study, Cassel et al. (2009a, 2009b) used trace and rare earth element analyses of volcanic glasses to determine the origin, dispersal, and paleoaltimetry of Oligocene rhyolitic ignimbrites in the northern and central Sierra Nevada. These deposits transect the modern topographic range well into the western Sierran Foothills and are correlated with 30 Ma ignimbrites in west-central Nevada. Based on palinspastic reconstruction, the volcanics were transported Pacificward ∼200 km from source calderas across what is now the Sierran crest; hence during Oligocene time, a drainage divide was not present between the calderas in west-central Nevada, and the ignimbrite blanket dispersed far to the west. Stable isotope bulk-rock analyses of Sierran ignimbrite glasses reflect the effect of this described western slope on meteoric water hydrogen and deuterium (H-D) values and indicate that the range had elevations similar to those of today. Evidently, source calderas were located in a region of high elevation east of the Oligocene Sierra Nevada Range, and the voluminous ignimbrites flowed down the western regional slope. For such long Pacificward flow of ignimbrites, ash-flow tuffs, and lahars, the source area must have been a highland of at least comparable elevation to the Paleogene Sierran arc.

Rather than east-central Nevada being a plateau of subdued topography, Druschke et al. (2009a, 2009b) showed that episodic faulting during early Tertiary time deformed basinal uppermost Cretaceous-Eocene alluvial fans and lakebeds in the region. Current data record westward paleoflow, as surmised in the aforementioned studies, and suggest that the central Nevada and western Utah area consisted of a series of long-lived, fault-bounded highlands. The plateau was more rugged than generally recognized, and it underwent episodic extension throughout latest Cretaceous–Paleogene time. Local deviations from a Pacificward regional slope of the western Nevadaplano are also present in the northeastern corner of California, where Egger et al. (2009) mapped a thick sequence of uppermost Eocene to Lower Oligocene volcaniclastic and sedimentary rocks. Paleocurrent indicators, stratigraphy, and clast petrology suggest that NNE-flowing streams deposited the section in an alluvial plain adjacent to a volcanic arc. Comparison with other Eocene sedimentary units and integration with paleofloral and geophysical data allow the definition of drainage divides, and suggest that the mapped sequence accumulated in a depressed intra-arc basin (Myers, 2003). This localized deposition differs markedly from coeval drainages to the south, which transported material westward from west-central Nevada to the North American paleoshoreline, and it indicates that ongoing volcanism ± faulting had a strong influence on paleogeography during the Paleogene; it also suggests that NW Nevada probably was typified by more normal crustal thicknesses during this time.

The evidence noted here supports the presence of a high plateau in central + southern Nevada + SW Utah during Late Cretaceous and Paleogene time (Coney and Harms, 1984; Wernicke et al., 1996; Best et al., 2009). Colgan and Henry (2009) documented substantial Paleozoic-Mesozoic E-W shortening and thickening of the pre-Cenozoic rocks in central Nevada. Major regional extension of this thickened crust began at ca. 16–17 Ma, with middle Miocene stretching partitioned into high-strain (50%–100%) domains separated by almost unextended crustal blocks. Later faulting, typified by widely spaced, high-angle normal faults, cut both extended and unextended domains. Important widening of the Basin and Range at this latitude evidently took place during a brief interval in the middle Miocene. Because of a near lack of coeval shortening west of the Sierra Nevada, Colgan and Henry (2009) suggested that the change in tectonics from microplate subduction to transtensional slip probably played an important role in this extension.

However, as also seems clear, paleobotanical, geological, structural, and geophysical data (Myers, 2006; Colgan et al., 2006; Lerch et al., 2007; Van Buer et al., 2009) indicate that the early Tertiary continental crust in northern and eastern Nevada was not more than ∼38 km thick. The conjectured Paleogene crustal thickness is sketched in Figure 5. Now relatively thin throughout Nevada, Neogene extension must have been greatest in the south, where the Eocene crust apparently was much thicker, as supported by prior geologic studies and by new deformational data (McQuarrie and Oskin, 2010). Van Buer et al. (2009) estimated Miocene-Recent extension as ranging from ∼7.5% in NE California to ∼20% in northern Nevada, while that of southern Nevada amounted to ∼50%; the Walker Lane area of westernmost Nevada was even more stretched. In support of this estimate, during late Cenozoic time, the central + southern Basin and Range appears to have been extended and downdropped much more than in the north, judging from oxygen isotope data from clay minerals (Poage and Chamberlain, 2002; Horton et al., 2004).


Continental crust is generated by juvenile calc-alkaline magmatism in island arcs and continental margins above convergent plate junctions; the recycling of preexisting sialic materials occurs due to underplating and anatexis of eroded volcanogenic sediments and old crustal fragments, but some igneous rocks apparently represent net additions to the crust. Growth of an active orogen is largely a function of petrotectonic processes, such as volcanic-plutonic magmatism, exotic terrane accretion, and contraction. Ongoing mountain building results in regional crustal thicknesses that are similar to, or exceed, those of old, stable cratons (e.g., Mooney et al., 1998). However, reflecting erosional removal, the thickness of the continental crust in a geologically youthful contractional orogen is also partly a function of climatic patterns. For a given convergent plate-tectonic regime, the mean elevation of a young mountain chain and the regional thickness of isostatically compensated crust are to some extent inversely correlated with precipitation and consequent running water + glacial flow–induced erosion. In addition, tectonic shortening and attendant rapid uplift in rain-shadow realms can defeat rapid degradation and promote the formation of internally draining, high-elevation plateaus typified by modest sedimentary deposition and/or erosion.

An Andean-type N-S–trending Sierran calc-alkaline arc and behind-the-arc Nevadaplano apparently typified the arid American Southwest in central + southern Nevada + SW Utah during Late Cretaceous and Paleogene time. The volcanic-plutonic arc remains active today in the Pacific Northwest where the Farallon plate is descending beneath the stable North American lithosphere. However, the N-S–trending Cascades are not characterized by a high mean elevation because the region receives copious amounts of rain and snowfall, driving vigorous erosion and wearing down the arc; the moderate crustal thickness is inadequate to form an Altiplano-like highland in the rain shadow. Farther south, precipitation declines markedly, and the Sierran Nevadan and hinterland crust thickened considerably attending nearly orthogonal Late Cretaceous–Eocene plate convergence and contraction. Arrival of the East Pacific Rise at the continental margin near the end of Oligocene time evidently caused a change in Pacific–North American plate configurations, the regional cessation of sialic crustal growth, and the onset of transtensional faulting. Overthickened southern Sierran arc crust is now delaminating and is sinking back into the deep upper mantle. Hinterland effects included the Neogene extensional destruction of the Nevadaplano—complete except for a possible remnant, the Colorado Plateau.

This synthesis is based on Great Basin research reports by numerous earth scientists as well as on a special issue of International Geology Review entitled, “Rise and Fall of the Nevadaplano” (Ernst, 2009). I thank the many contributors to that topical issue as well as earlier workers in the field. University of California–Los Angeles (UCLA) supported my own petrotectonic studies for 30 yr, and Stanford for 20 yr more. Page Chamberlain, Paul Davis, George Hilley, Elizabeth Miller, and Norm Sleep provided constructive feedback on this review. I thank these researchers and institutions for help.