The Cascades volcanic arc is the surface manifestation of volcanism on continental crust of the North American plate due to subduction of the Explorer, Juan de Fuca, and Gorda oceanic plates along the Cascadia subduction zone (Figure 1A). Volcanism is most apparent as the remarkable, high-standing edifices of calc-alkaline stratovolcanoes (e.g., Mount Rainier, Mount Hood, Mount Shasta, etc.) that make up much of the surface expression of the Cascades volcanic arc (Hildreth, 2007). In the southern part of the Cascades volcanic arc in northern California, Quaternary volcanism is concentrated at the long-lived volcanic centers of the Mount Shasta stratovolcano and around the volcanic fields of the Lassen Volcanic Center and Medicine Lake Volcano (Figure 1B). Radiocarbon and radiometric dating confirm the presence of Holocene eruptions from Mount Shasta, Medicine Lake Volcano, and the Lassen Volcanic Center (Christiansen et al., 2017; Clynne et al., 2017; Donnelly-Nolan and Grove, 2017), but the only eruption documented first-hand was at Lassen Peak in 1914–1917 (Christiansen et al., 2002; Clynne and Muffler, 2010, 2017; Muffler and Clynne, 2015).

The southernmost locus of arc-related eruptions during the Quaternary is near Lassen Peak (in the Lassen Volcanic Center). The axis of the arc extends between Mount Shasta and Medicine Lake Volcano farther north into Oregon, Washington, and southern Canada (Figure 1A). For ∼50 km north of Lassen Peak, the arc front is expressed as prominent high-standing Middle and Late Pleistocene calc-alkaline (dominantly andesite and dacite) edifices, such as Magee Volcano, Freaner Peak, Burney Mountain, and Brush Mountain (Figure 1B; Muffler and Clynne, 2015; Clynne and Muffler, 2017; Downs et al., 2020b). North of Brush Mountain for ∼70 km, Pleistocene calc-alkaline volcanic edifices are absent along the arc front. Situated east of the calc-alkaline arc front are relatively thin (typically 10–15 m thick, but thicker where they fill valleys), voluminous and laterally extensive, tholeiitic basalt flows that erupted throughout the Pleistocene (Champion et al., 2017; Muffler et al., 2017; Downs et al., 2020a). These tholeiitic basalts are eruptive products of the back-arc and overlie and surround older Quaternary and Neogene calc-alkaline volcanic edifices that are earlier expressions of the eastward-migrating Cascades volcanic arc (Guffanti et al., 1990; Borg et al., 2002).

Older tholeiitic basalts in the back-arc (predominantly those of the Early and Middle Pleistocene) tend to be deeply weathered, densely vegetated, commonly expressed as rounded boulders protruding from well-developed soils (sometimes >10 m thick). Late Pleistocene tholeiitic basalts are more conspicuous owing to the arid environment of northeastern California, with sparse vegetation, soil development, and erosion (Figure 2A). As such, most of the surface morphologies of Late Pleistocene basalt flows in the back-arc are well preserved, lending these tholeiitic basalts a young, rough appearance (Figures 2C–E). This appearance resulted in most of them being assigned to the Holocene (cf., Peterson and Martin, 1980; Miller, 1989; Grose, 1999). However, several of these back-arc tholeiitic basalts now have isotopic ages showing them to be older than Holocene (Giant Crater flow field in Donnelly-Nolan, 2010, Hat Creek Basalt in Turrin et al., 2007, and Eagle Lake tholeiitic basalts in Clynne et al., 2017).

Mapping of the Brushy Butte flow field and surrounding region was first undertaken by Peterson and Martin (1980) as a U.S. Geological Survey Wilderness Study Area, with their mapped units mostly incorporated into later mapping undertaken farther east by Grose (1999). The Brushy Butte flow field was assigned a Holocene age based on the relatively rough, sparsely vegetated flow surfaces present throughout much of its extent (Figures 2C–E; Peterson and Martin, 1980; Miller, 1989). This Holocene age assignment resulted in Miller (1989) designating the Brushy Butte flow field as having the potential to cause future ash and lava hazards equivalent to the largest basaltic flow and airfall tephra hazards (i.e., up to 10 km from vent) in the northern California back-arc. As a result, the Brushy Butte flow field was considered a ‘very low threat volcano’ in need of monitoring (Ewert et al., 2005; Ewert, 2007); however, it has recently been taken off the threat list as a result of its stratigraphic position negating the Holocene eruption age (Ewert et al., 2018) and the recognition that northern California back-arc tholeiitic basaltic eruptions are indicative of monogenetic volcanism (e.g., Wood, 1979; Németh and Kereszturi, 2015). Despite being removed from the threat list, none of the aforementioned investigations dealt with the eruptive styles, compositions, or ages of the Brushy Butte flow field or surrounding tholeiitic basalts.

During this study, tholeiitic basalt flows were mapped in the field using petrographic characteristics in an attempt to identify contacts and stratigraphic superposition. However, petrographic similarities and the unreliable nature of using visual observations based on vegetation cover and type indicated that a multidisciplinary approach for discriminating between flows would be required. Geochemical and paleomagnetic data provide two independent means of discriminating between tholeiitic basalts in the back-arc of northern California (e.g., Champion et al., 2017; Muffler et al., 2017; Downs et al., 2020a).

Our mapping was supported by 1 m-resolution lidar (available from the U.S. Geological Survey 3D Elevation Program: https://www.usgs.gov/core-science-systems/ngp/3dep) to help identify contacts between flow units and locate potential sampling sites. Processing of lidar data strips away vegetation cover to show the underlying terrain, making the surface morphology of flows and other volcanic landforms more prominent. Flows from the Early and Middle Pleistocene show smoother surface morphologies as a result of extensive soil development and erosion. Late Pleistocene flows, while moderately to sparsely vegetated along internal flow contacts, faults, and in more scoria-rich deposits, still have relatively intact primary surface morphologies (Figures 2C,D). In particular, surface features such as well-established channels and levees, channel breakouts, lava ponds, inflated pāhoehoe surfaces, scoria cones, spatter cones, and craters are evident throughout the Brushy Butte flow field (Figure 2B). This vastly aids in understanding the emplacement history of these flows when integrated with other datasets (e.g., geochemistry, paleomagnetism, and geochronology). Additionally, the lidar imagery provides a vital window into parts of the flow field that cannot be directly sampled because they lie within Pit River tribal lands, Ahjumawi Lava Springs State Park, private property, or in terrain that is difficult to access. Using the lidar imagery, we can interpret stratigraphic context, flow morphologies, and make important inferences as to eruptive styles and emplacement processes. The integration of all of these data has resulted in a detailed geologic map of the Brushy Butte flow field and the flows and other geologic units with which it is in contact (Figure 3 and see Supplementary Figure S1 in the Supplementary Material for a high-resolution version of the geologic map).

Geochemical data were helpful for distinguishing various flows and identifying the locations of contacts throughout this region. A total of 82 lava samples from the Brushy Butte region were collected for major-oxide and trace-element analyses by wavelength-dispersive X-ray fluorescence (WD-XRF) spectrometry (after Johnson et al., 1999; see the Supplementary Material for a description of the methods). Forty-three samples are from the Brushy Butte flow field, whereas the other 39 analyses are from surrounding flows and help distinguish them from the Brushy Butte lavas (see Supplementary Table S2 in the Supplementary Material for all analyses). Additionally, six samples from various lobes of the Brushy Butte flow field were analyzed by laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) as a secondary check on internal variations of different eruptive pulses of the flow field (Table 1; see the Supplementary Material for methods).

All but one of the tholeiitic basalt flows displayed on Figure 3 have very similar petrographic characteristics: diktytaxitic groundmass hosting 1–3% olivine phenocrysts of 1–2 mm diameter and rare (<<1%) plagioclase phenocrysts. The exception is the Thousand Springs flow, which has common (1–2%, to 1–2 mm) plagioclase phenocrysts that have resorbed cores with sieve textures and overgrowth rims. This flow also has a distinctive geochemistry, with a large range for many major-oxide and trace-element abundances. For example, the Thousand Springs composition ranges from 48.77 to 52.52 wt% SiO₂ and 0.24 to 1.01 wt% K₂O compared to 47.41 to 48.04 wt% SiO₂ and 0.09 to 0.15 wt% K₂O for the Brushy Butte flow field.

Both major oxides and trace elements are useful for distinguishing the other petrographically similar tholeiitic basalts as a result of magma chemistry being controlled by the degree, depth, and mineralogy of the metasomatized mantle lithosphere undergoing partial melting (e.g., Champion et al., 2017; Muffler et al., 2017; Downs et al., 2020a). Some of the major-oxide and trace-element abundances overlap, but significant differences allow distinct flows to be defined. Figure 4 shows a suite of plots used to distinguish the different tholeiitic basalt flows displayed on Figure 3. The Brushy Butte flow field tends to have relatively low, but slightly overlapping, abundances of TiO₂ (0.69–0.89 wt%) with surrounding tholeiitic basalts at their given MgO abundances (Figure 4). As a result, the Brushy Butte geochemical field (gray area in Figure 4) generally plots in a distinct position for TiO₂ versus other major-oxide and trace-elements. A distinct field is also noticeable for trace-element plots, particularly those using Zr, to discriminate the Brushy Butte geochemical field from surrounding tholeiitic basalts.

The Giant Crater flow field (Group 5 as mapped by Donnelly-Nolan et al., 1991) directly contacts the west side of the Brushy Butte flow field and closely resembles these lavas (Figures 2A,B). Our whole-rock analyses from Giant Crater overlap the limited range of values for Group 5 presented in Baker et al. (1991) and Donnelly-Nolan et al. (1991). Brushy Butte and Giant Crater major-oxide abundances plot within close proximity and commonly overlap, although there are observable differences in FeO*, MnO, CaO, and K₂O. On the other hand, trace-element values of Brushy Butte and Giant Crater rarely overlap, with characteristic differences in Sc, Cr, Ni, Sr, Rb, and in particular Zr (Figure 4; see Supplementary Table S2 in the Supplementary Material).

The high-standing (∼150 m tall) interior of the Brushy Butte flow field consists of many flows and vent systems that we have mapped as units bb1–bb29 and divided into four eruptive pulses (Figure 3). Geochemical analyses from the flow field shows that the mapped eruptive units, and eruptive pulses, are internally consistent geochemically. Brushy Butte early (?) and pulse 4 have only one analysis each, but both analyses fall within the range of Brushy Butte major-oxide and trace-element abundances from the other pulses (Figure 4). Available geochemical evidence is consistent with lavas from the Brushy Butte flow field being distinct from surrounding tholeiitic basalts with which they are in contact.

As a further evaluation of the internal consistency of the Brushy Butte flow field, six samples were analyzed by LA-ICP-MS (Table 1). These samples include four from pulse 1 (unit bb3), one from pulse 2 (unit bb14), and one from pulse 3 (unit bb18). The six samples analyzed by LA-ICP-MS all have small standard deviations and fall within analytical uncertainty of each other by this method (Conrey et al., 2019). For example, Zr abundances for these tholeiitic basalts are generally distinct and these values range from 41.86 to 48.88 ppm (standard deviation of 2.26). The combination of the whole-rock WD-XRF and LA-ICP-MS analyses supports the interpretation that lavas from the Brushy Butte flow field are magmatically identical to one another.

Paleomagnetic data provide an additional and independent test for distinguishing visually and petrographically similar flows and locating contacts. This is a result of the brief timeframe during which the Earth’s magnetic field is stationary (4–5° per century of wander; Champion and Shoemaker, 1977).

Paleomagnetic samples were collected, processed, and interpreted using the standard protocols outlined in McElhinny (1973), with a total of 28 sites from five tholeiitic basalt flows. Twenty of these sites were drilled in the Brushy Butte flow field, whereas the remaining eight sites were drilled in surrounding flows to aid in distinguishing them from the Brushy Butte flow field and each other (Table 2; see the Supplementary Material for methods). The Giant Crater flow field was not drilled for paleomagnetism as part of this study, as this flow field has been well characterized with Champion and Donnelly-Nolan (1994) presenting data from 58 drill sites.

Figure 5A shows the range of mean remanent directions of magnetization for the Brushy Butte flow field and immediately adjacent tholeiitic basalts. Average inclination and declination values for the Brushy Butte flow field are 60.4° and 357.1° (95% confidence level; α95 = 0.8°), respectively. The most visually and morphologically similar basalt in contact with the Brushy Butte flow field is Group 5 of the Giant Crater flow field, for which four drilled sites give average inclination and declination values of 59.1° and 18.6° (α95 = 1.4°), respectively (Champion and Donnelly-Nolan, 1994). The inclinations of these two tholeiitic basalts are within error of each other, but their declinations are distinct. The other surrounding tholeiitic basalts are morphologically more subdued in general and also have inclinations and declinations distinct from those of the Brushy Butte flow field (Table 2 and Figure 5A), with the Thousand Springs flow having mean inclinations and declinations of 43.6° and 3.1° (α95 = 2.1°) and the Adobe Flat flow at 55.2° and 24.2° (α95 = 5.9°), respectively. The Timbered Crater flow records an excursional remanent direction at 71.2° and 42.8° (α95 = 3.9°), whereas the Crum Reservoir flow is reversely magnetized at −37.0° and 211.2° (α95 = 2.0°). Crum Reservoir is not shown on Figure 5A because of its reverse polarity.

Previous investigations in this region used soil development and vegetation to assign age constraints for flows (Peterson and Martin, 1980; Miller, 1989). While useful for assigning relative ages in certain parts of the map area on Figure 3, this approach has limited value as much of the area has been deforested and replanted or cleared for ranch land, and differences in internal surface flow morphologies promote vegetation growth at a range of rates and plant types. Isotopic dating is preferred for understanding the time intervals between eruptions in a given region, although this also is challenging, given the difficulty in finding groundmass or organic matter suitable to analyze from young tholeiitic basalts. Tholeiitic basalts have low abundances of K (0.09–0.15 wt% K₂O for the Brushy Butte flow field; see Supplementary Table S2 in the Supplementary Material) and therefore low radiogenic ⁴⁰Ar, so that Late Pleistocene and Holocene tholeiitic basalts are particularly hard to date by the ⁴⁰Ar/³⁹Ar method. Analyses of young tholeiitic basalts from northern California by the ⁴⁰Ar/³⁹Ar method have been successful but only on samples with a nearly perfect coarsely crystalline groundmass lacking vesicles and glass (e.g., Turrin et al., 2007). The young ages of these basalts usually means that there is only minor erosion to expose dense, crystalline groundmass for ⁴⁰Ar/³⁹Ar dating unless they are cut by major, young faults. Additionally, the ¹⁴C method is commonly used for volcanic rocks that erupted <50 ka, but the same arid environment that has allowed remarkable preservation of surface morphologies resulted in the base of the flows not being exposed by erosion. This has made finding organic matter for the ¹⁴C method difficult, although not impossible, as Donnelly-Nolan (2010) collected charcoal samples dated at 12.4 ka from the Giant Crater flow field that stratigraphically overlies the Brushy Butte flow field.

A method that has proved useful for providing ages on young flows in arid environments that lack extensive vegetation cover and erosion is the ³⁶Cl cosmogenic surface-exposure dating (e.g., Vazquez and Woolford, 2015; Downs et al., 2018; Downs et al., 2019; Stelten et al., 2018; Stelten et al., 2020; Bromley et al., 2019). We utilized the ³⁶Cl cosmogenic method (see Supplementary Material for details on sample collection and analysis) for the Brushy Butte flow field to determine its emplacement age, and we couple this with ⁴⁰Ar/³⁹Ar ages of surrounding older tholeiitic basalts to better understand the eruptive history of volcanism prior to eruption of the Brushy Butte flow field.

A few dense, coarse-grained, crystalline samples were identified within the most basal exposed parts of the Brushy Butte flow field, and ⁴⁰Ar/³⁹Ar dating was attempted on five of these (Table 3; see the Supplementary Material for methods and Supplementary Table S3 for all tabulated ⁴⁰Ar/³⁹Ar data). These samples yielded ages with large analytical uncertainties (all reported at 1 σ) of 41 ± 27 ka, 48 ± 52 ka, 51 ± 17 ka, 51 ± 44 ka, and 165 ± 80 ka, which span from the Middle Pleistocene to Holocene at 95% confidence levels. Calculating a weighted mean age yields smaller analytical uncertainties of 48 ± 13 ka (using four of five analyses) and 52 ± 13 ka (using all five analyses); however, at the 95% confidence level these still span >25 kyr. The large analytical uncertainties using ⁴⁰Ar/³⁹Ar, and a lack of organic matter for ¹⁴C dating, prompted the collection of two primary surface rocks (Figure 2C) to date by the ³⁶Cl cosmogenic surface-exposure method (Table 4; see the Supplementary Material for methods and Supplementary Table S4 for input parameters and output data for calculating the ages). Both surface-exposure ages are nearly identical at 35.4 ± 2.4 ka and 36.0 ± 2.4 ka (analytical uncertainties reported at 1 σ), which yields a weighted mean age of 35.7 ± 1.7 ka. This age overlaps with the ⁴⁰Ar/³⁹Ar ages but yields a significantly smaller analytical uncertainty and clearly places the Brushy Butte flow field within the Late Pleistocene.

Four samples from surrounding tholeiitic basalts in contact with the Brushy Butte flow field were analyzed by the ⁴⁰Ar/³⁹Ar method (Table 3). These flows underlie the Brushy Butte flow field and their ⁴⁰Ar/³⁹Ar ages are consistent with their stratigraphic position. The Adobe Flat flow erupted at 309 ± 45 ka, the Timbered Crater flow at 571 ± 49 ka, the Vestal Swamp flow at 1053 ± 52 ka, and the Crum Reservoir flow at 1340 ± 59 ka. Dateable material was not identified from the Thousand Springs flow, which is stratigraphically bracketed by the Brushy Butte flow field at 35.7 ± 1.7 ka and Adobe Flat flow at 309 ± 45 ka. The Timbered Crater cone consists of palagonitized ash that formed during a phreatomagmatic eruption. This cone overlies the Timbered Crater flow, but there is no clear evidence of any eruptive relationship between the cone and flow. A single tholeiitic basalt clast analyzed for geochemistry from the cone does not match the geochemistry of the flow, but it cannot be definitively determined whether this, and similar looking, clasts are juvenile or lithic. The Giant Crater flow field has a ¹⁴C age of 12.4 ka (Donnelly-Nolan, 2010), with Ewert et al. (2018) reporting that the Giant Crater flow field overlies the Brushy Butte flow field. This relative age relation is remarkably difficult to see in the field or even using the 1 m-resolution lidar (Figures 2A,B). Our age for the Brushy Butte flow field confirms that it is older, and hence underlies, the Giant Crater flow field.

We arranged the units that make up the Brushy Butte flow field into stratigraphic order, where possible, using the 1 m-resolution lidar imagery and field studies. Based on this stratigraphic order and erupted volumes, we have defined pulses of eruptive activity from the flow field (Figure 7). Most Brushy Butte units group into four pulses. Pulse 1 consists of eight Brushy Butte units (bb3–bb10) and total 2.27 km³ in eruptive volume, including the most voluminous (2.10 km³) lava flow in unit bb3. The remaining units in pulse 1 are inferred to have erupted after bb3 based on their high-standing positions within the interior of the flow field, and each has a significantly smaller volume totalling only 0.27 km³ for the other seven units of pulse 1 (Table 5). Pulse 2 consists of seven units (bb11–bb17) totalling 0.61 km³ in eruptive volume, with the first unit (bb11) making up 0.36 km³ of that total. Pulse 3 consists of five units (bb18–bb22) totalling 0.33 km³ in eruptive volume, with the first unit (bb18) making up 0.27 km³ of that total. Pulse 4 consists of seven units (bb23–bb29) totalling 0.22 km³ in eruptive volume, with the first unit (bb23) making up 0.16 km³ of that total.

The relative ages for pulses 2 and 3 are easy to reconstruct as there are ample places where direct contacts can be used to determine stratigraphic superposition. In pulses 1 and 4, however, flows are rarely in contact with each other. Therefore, the assigned unit number for flows from these pulses is somewhat arbitrary. For pulse 1, unit bb3 is easily inferred to underlie all flows erupted from the interior of the flow field based on its areal extent, despite its stratigraphic bracketing between units bb1 and bb10. This becomes more difficult for the rest of pulse 1, in which most units have no maximum relative age bracket because they are the oldest units exposed locally, with overlying bracketing units ranging anywhere from bb10 to bb23. Additionally, flows of pulse 1 are partly buried by younger flows from the flow field, with two vents (units bb4 and bb6) in the interior remaining uncorrelated with their flows. These vents could be the source(s) for exposed but disconnected flows, but no correlation can be made with the data on hand. As a result, we have mapped them as individual eruptive units attributed to the Brushy Butte flow field. On the other hand, units assigned to pulse 4 have no overlying units to bracket their relative ages, and none of these units are in direct contact. Underlying units can only constrain their relative ages to younger than bb13 (for bb28) or to younger than bb22 (for bb23). It is notable that map units bb25–bb29 in pulse 4 are relatively well aligned along the regional grain of normal faulting.

Brushy Butte units bb1 and bb2 have not been assigned to any pulse (they are labeled Brushy Butte early (?) on Figure 3), as it is unclear how these southern vent systems relate to the units erupted from the interior of the flow field. Only one sample from bb1 was collected for geochemistry, which is situated within the Brushy Butte geochemical field (Figure 4). The other early (?) erupted Brushy Butte map unit bb2 is densely faulted, but these faults only slightly extend into the overlying unit of bb11. This hints at bb2 being older than overlying Brushy Butte lavas, but not much older.

Despite these stratigraphic limitations, we can demonstrate, based on stratigraphic position, that the first eruption in each pulse 1–3 was the most voluminous, and we rely on this evidence to infer that the most voluminous eruption in pulse 4 (unit bb23) was also the first eruption of that pulse. All other flows assigned to pulse 4 were assigned a unit name arbitrarily, since they are not in direct contact. The erupted volumes of the other units in these pulses are too small to noticeably affect the volume distribution in Figure 7.

Just as each pulse shows an internal decay of eruptive volumes, each pulse itself shows a decrease in total eruptive volume from the previous pulse. Pulse 1 totals 2.27 km³, which decreases to 0.61 km³ for pulse 2, 0.33 km³ for pulse 3, and 0.16 km³ for pulse 4 (Table 5). This represents a decrease in eruptive volume of ∼73% from pulse 1 to pulse 2, ∼46% from pulse 2 to pulse 3, and ∼32% from pulse 3 and pulse 4.

Despite no Holocene eruptions occurring outside of the major, long-lived volcanic centers of Mount Shasta, Medicine Lake Volcano, and the Lassen Volcanic Center (Figure 1B), volcanic activity between these centers is a distinct possibility. The lack of a witnessed northern California back-arc eruption makes it unclear how long such an eruption might last. Both field and laboratory evidence from the Brushy Butte flow field, when combined with analogue eruption styles, provide insights into the duration of such an eruption.

Some of the most extensively studied, first-hand accounts of tholeiitic basalt volcanism are from the East Rift Zone on the Island of Hawaiʻi, where Puʻu ʻŌʻō erupted continuously from 1983 to 2018. Over 35 years, the Puʻu ʻŌʻō eruption covered an area of ∼144 km² and erupted a volume of ∼4.4 km³, while developing many geomorphic flow features (e.g., Hon et al., 1994; Kauahikaua et al., 1998; Hoblitt et al., 2012; Patrick and Orr, 2012). Volcanic landforms are constructed over a range of rates, providing clues as to the duration of eruptive activity in older volcanic terrains. For example, inflated pāhoehoe surfaces as high as 4 m have been observed forming in Hawaiʻi over the course of ∼14 days (Hon et al., 1994; Hoblitt et al., 2012), a thickness similar to the 3–5 m inflated pāhoehoe from Brushy Butte. Unit bb3 has abundant inflated pāhoehoe surfaces, and we therefore propose that it took several weeks for the flow to be emplaced. The geomorphic features within the Brushy Butte flow field also suggest the possible duration of emplacement. These include well-established channel and levee systems with occasional breakouts and lava ponds (Figures 6A–D). Similar flow features in tholeiitic basalts of the Puʻu ʻŌʻō, as well as other K–llauea eruptions, demonstrate that these sorts of features develop and last over the course of weeks, months, or longer (Patrick and Orr, 2012; Hazlett et al., 2019).

Paleomagnetic analysis provides further clues into the eruptive duration of the Brushy Butte flow field. In particular, pulses 1 and 3 both have mean remanent directions of magnetization that yield tight precision with small α95’s. Pulse 1 encompasses 13 drill sites that yield a mean inclination of 60.9°, mean declination of 357.0°, and α95 of 1.0°. Pulse 3 yields a mean inclination of 58.8°, mean declination of 356.8°, and α95 of 1.6° from five drill sites (Figure 5B). While these two directions are very similar, an algorithm by McFadden and Jones (1981) indicates that they are distinct from each other at the 98% confidence level. Pulse 2 overlaps pulses 1 and 3 with a large α95 of 6.0° as a result of only two drill sites. This indicates that a very modest amount of time passed during these eruptions. For example, the immediately adjacent Giant Crater flow field is characterized by two distinct mean directions of remanent magnetization that are separated by 1.27° (from 58 drill sites) that has been modeled as an eruption spanning ∼15 years based on rates of geomagnetic polar wander (Champion and Donnelly-Nolan, 1994). Our results give a probable eruptive duration for the Brushy Butte flow field (at least pulses 1–3) of 10–20 years, based on the average 4–5° per century of wander (Champion and Shoemaker, 1977).