Accurate estimates of mare basalt ages are necessary to place constraints on the duration and the flux of lunar volcanism as well as on the petrogenesis of lunar mare basalts and their relationship to the thermal evolution of the Moon. We performed new crater size-frequency distribution measurements in order to investigate the stratigraphy of mare basalts in Oceanus Procellarum and related regions such as Mare Nubium, Mare Cognitum, and Mare Insularum. We used high-resolution Clementine color data to define 86 spectrally homogeneous units within these basins, which were then dated with crater counts on Lunar Orbiter IV images. Our crater size-frequency distribution measurements define mineralogical and spectral surface units and offer significant improvements in accuracy over previous analyses. Our data show that volcanism in the investigated region was active over a long period of time from ∼3.93 to 1.2 b.y., a total of ∼2.7 b.y. Volumetrically, most of the basalts erupted in the Late Imbrian Period between ∼3.3 and 3.7 b.y., and we see evidence that numerous units have been resurfaced. During the Eratosthenian Period, significantly less basalt was erupted. Depending on the absolute model ages that one can assign to the lunar chronostratigraphic systems, five units might be of Copernican age. Younger basalts are generally exposed in the center of the investigated area, that is, closer to the volcanic centers of the Aristarchus Plateau and Marius Hills. Older basalts occur preferentially along the northwestern margin of Oceanus Procellarum and in the southeastern regions of the studied area, i.e., in Mare Cognitum and Mare Nubium. Combining the new data with our previously measured ages for basalts in Mare Imbrium, Serenitatis, Tranquillitatis, Humorum, Australe, and Humboldtianum, we find that the period of active volcanism on the Moon lasted ∼2.8 b.y., from ∼4 b.y. to ∼1.2 b.y. On the basis of the basalts dated so far, which do not yet include the potentially young basalts of Mare Smythii e.g.,all investigated basins but probably also is the location of some of the youngest basalts on the lunar surface.
 Lunar mare basalts cover about 17% of the lunar surface [Head, 1976], but radiometric ages for lunar basalts are available only for spatially very limited areas, i.e., the Apollo and Luna landing sites [e.g., Basaltic Volcanism Study Project (BVSP), 1981; Stöffler and Ryder, 2001; Taylor, 1982, and references therein]. A significant portion of lunar mare basalts is exposed within Oceanus Procellarum and associated regions for which absolute radiometric age data are still lacking. Fortunately, remote sensing techniques allow one to derive ages not only for the Apollo and Luna landing sites but also for unsampled regions. Superposition of geologic units onto each other, crater degradation stages, and crater size-frequency distribution measurements have been used in order to obtain relative and absolute model ages for lunar surface units from remote sensing data [e.g., Shoemaker and Hackman, 1962; Boyce, 1976; Wilhelms, 1987; Neukum and Ivanov, 1994; Hiesinger et al., 2000].
 Here we present model ages of lunar mare basalts in Oceanus Procellarum, Mare Nubium, Mare Cognitum, and Mare Insularum (Figure 1) that are based on one of these remote sensing techniques, that is, crater counts. Compared to previous crater counts [e.g., Neukum et al., 1975; Greeley and Gault, 1970; Hartmann, 1966], we applied a new approach in that we performed crater size-frequency distribution measurements for spectrally homogeneous basalt units. A major goal of this study is to provide absolute model ages for these basalts in order to investigate their stratigraphy and to understand better the nature and evolution of lunar mare basalt volcanism.
 On the basis of our new age data we address the following questions: (1) What was the time period of active volcanism in the investigated area, i.e., when did volcanism start and when did it end? (2) Was lunar volcanism continuously active or are there distinctive periods of volcanic activity? (3) Is there a trend in the spatial distribution of basalt ages on the lunar surface? Finally, (4) What is the flux of lunar mare basalts, i.e., what volumes of basalts were erupted within a certain period of time?
2. Technique, Approach, and the Definition of Units
 Crater size-frequency distribution measurements are a powerful remote sensing technique to derive relative and absolute model ages for unsampled planetary surfaces. As this technique is described elsewhere [e.g., Neukum and Ivanov, 1994; Hiesinger et al., 2000; Stöffler and Ryder, 2001; Neukum et al., 2001; Ivanov, 2001; Hartmann and Neukum, 2001; and references therein] we will only briefly outline how crater size-frequency distribution measurements can be used to date surfaces. The technique of crater size-frequency distribution measurements on spectrally homogeneous regions, including a discussion of model assumptions, strengths and shortcomings, and an error analysis, has been described in detail by Hiesinger et al. . In short, in order to obtain the age of a photogeological unit one has to (1) measure the surface area of the unit, and (2) measure the diameters of each primary impact crater within this unit. It has been shown that lunar crater distributions measured on geologic units of different ages and in overlapping crater diameter ranges can be aligned along a complex continuous curve, the lunar production function [e.g., Neukum and Ivanov, 1994]. The lunar production function is given by
where a0 represents the amount of time during which the unit has been exposed to the meteorite bombardment [Neukum and Ivanov, 1994]. The cumulative crater density of a geologic unit taken at a fixed reference diameter (usually 1 or 10 km) is directly related to the time the unit has been exposed to the meteorite flux and therefore gives a relative age of this unit. To obtain absolute model ages from crater size-frequency distribution measurements, one has to link the radiometric ages from the returned Apollo and Luna samples with crater counts of these landing sites in order to establish the lunar cratering chronology. This is not a trivial task and has led to several more or less different chronologies [e.g., BVSP, 1981; Neukum, 1983, Neukum and Ivanov, 1994; Stöffler and Ryder, 2001; and references therein]. The empirically derived chronology of Neukum and Ivanov , which we use for this study is given by
 Once the lunar chronology is established we can derive absolute model ages for the entire lunar surface from crater size-frequency distribution measurements by solving (2) for time t for Ncum(D ≥ 1 km) measured on the geologic unit to be dated.
 The level of uncertainty of the crater retention age of a given count is given by the following formula:
in which N(1) is the crater retention age calculated for craters of 1 km diameter and A is the size of the counted area. The ±σN value gives the upper and lower limits of the error bar of the crater retention age, which are used for estimating the uncertainty of the absolute crater model age from the cratering chronology. We principally assume that the cratering chronology is free of errors. Therefore errors in our absolute model ages are only caused by errors in the determination of crater frequencies [Neukum, 1983]. Neukum et al.  estimated the systematic uncertainty of the standard distribution curve or the measurement to be <10% for 0.8 km ≤ D ≤ 3 km (this is the diameter range of most of our crater counts) and up to 25% for 0.8 km ≤ D ≤ 10 km.
 For the discussion of absolute model ages of basalt units and the application of terms like “Eratosthenian” or “Imbrian”, one must be aware that various authors defined these chronostratigraphic systems in different ways [e.g., Wilhelms, 1987; Neukum and Ivanov, 1994; Stöffler and Ryder, 2001]. A detailed discussion of this issue is given elsewhere [Hiesinger et al., 2000]. Figure 2 is a comparison of stratigraphies based on work by Wilhelms , Neukum and Ivanov , and Stöffler and Ryder . While there is general agreement on the definition of the base of the Eratosthenian system (i.e., 3.2 b.y.), the stratigraphies vary substantially for the beginning of the Copernican system (i.e., 0.8–1.5 b.y.). Wilhelms  pointed out, that there is no formal definition of the Copernican system because no extensive stratigraphic-datum horizons exist near the lower system boundary. Crater Copernicus is a good early Copernican marker, but does not mark the base of the system [Wilhelms, 1987]. The lack of a clear stratigraphic-datum horizon resulted in a vague definition of the Copernican system, a fact which is also reflected in the variations of absolute ages assigned to the base of this system.
 In this paper we adopt the system of Neukum and Ivanov , with Nectaris being 4.1 ± 0.1 b.y. (N(D = 1 km) = (1.2 ± 0.4) × 10−1), and the Imbrium basin being 3.91 ± 0.1 b.y. old (N(D = 1 km) = (3.5 ± 0.5) × 10−2). According to Neukum , the Eratosthenian System started 3.2 b.y. ago (N(D = 1 km) = 3.0 × 10−3) and the Copernican System began 1.5 b.y. ago (N(D = 1 km) = (1.3 ± 0.3) × 10−3), while radiometric dating of samples, which are thought to represent Copernicus ejecta, indicates an age of 0.85 ± 0.1 b.y. [Silver, 1971] (Figure 2). As mentioned above, the linkage of lunar sample ages to discreet basin forming events is still subject to discussion [e.g., Wilhelms, 1987; Spudis, 1993; Neukum and Ivanov, 1994; Stöffler and Ryder, 2001]. For example, on the basis of their reevaluation of lunar samples, Stöffler and Ryder  concluded that the Nectaris basin is 3.92 ± 0.03 b.y. old, hence younger than in the Neukum and Ivanov  chronology. This interpretation of Stöffler and Ryder  is consistent with the conclusions of Spudis . For the Imbrium basin Stöffler and Ryder  discuss two ages, ranging from 3.85 ± 0.02 to 3.77 ± 0.02 b.y., the later age being inconsistent with previously published ages of 3.80–3.85 b.y. for the Orientale basin [e.g., Wilhelms, 1987; Schaefer and Husain, 1974]. Application of the Neukum and Ivanov  chronostratigraphic system yielded 39 Imbrian units, 42 Eratosthenian units, and 5 Copernican units in the investigated area. Using the chronostratigraphic system of Wilhelms , the five Copernican units would be of Eratosthenian age.
 As mentioned earlier it is very important to define homogeneous units in order to obtain reliable age determinations with crater size-frequency distribution measurements. Having defined such units with Clementine images, we transferred the unit boundaries to high-resolution Lunar Orbiter IV images in order to measure the crater size-frequency distribution. This was necessary because Clementine images are not very well suited for crater counts due to their high sun angles. Figure 6 is a comparison of a Lunar Orbiter image and a Clementine image that illustrates the differences in the ability to detect small craters in these two data sets. The detailed views show that some craters that are clearly visible in the Lunar Orbiter image are barely detectable and certainly not countable in the Clementine image. Because miscounting certain craters has profound effects on the crater size-frequency distribution and hence the age of an investigated unit, we chose not to perform crater counts on Clementine data. Despite the excellent quality, we also chose not to count on Apollo metric camera or panoramic images because only Lunar Orbiter images provide a systematic coverage of all investigated mare areas. Compared to previous age determinations, our data fit spectral and lithological units and represent a major improvement in accuracy. In contrast, data from Boyce  and Boyce and Johnson  do not fit such units and the outline of their ages may be controlled or at least influenced by the applied filtering technique rather than the actual geologic diversity (Figure 5)6.
 In this paper we report on ages of 86 basalt units in Oceanus Procellarum, Mare Nubium, Mare Cognitum, and Mare Insularum. In order to facilitate the discussion of ages we did not assign type locality names to each unit. Instead, we use a simple letter/number system. The letter indicates the basin (P = Oceanus Procellarum, N = Mare Nubium, C = Mare Cognitum, IN = Mare Insularum) and the number describes the unit within a basin. The numbering is consistent with the geologic maps of the Moon with oldest units having lower numbers and younger units having higher numbers.
3.1. Oceanus Procellarum
3.1.1. Geologic Setting
 Located on the western lunar nearside, Oceanus Procellarum is the largest expanse of exposed mare basalts and is characterized by a non-circular shape (Figure 1). Despite the irregular shape it has been proposed that Oceanus Procellarum is located within a very ancient, 3200 km large impact structure, the Procellarum basin [e.g., Wilhelms and McCauley, 1971; Whitaker, 1981; Wilhelms, 1987]. The existence of this Procellarum basin is not completely accepted because the geochemical arguments for such a basin as well as the identification of ring structures of this basin are subject to alternative interpretation [e.g., Spudis, 1993; Spudis and Schultz, 1985; Wilhelms, 1987].
 However, Feldman et al.  interpreted new Lunar Prospector data to be consistent with a Procellarum basin. Investigating the spatial distribution and concentration indices of FeO, thorium, epi/thermal and fast neutrons using spherical harmonic expansion analyses, Feldman et al.  found that the dipole vectors and eigenvectors cluster closely to the center of the putative Procellarum basin. On the basis of their work, the center of the dipole axes is at 14.1° latitude and 16.4° west longitude, the center of the quadrupole axes is located at 24.6° latitude and 25.1° west longitude, both being close to the center of the Procellarum basin given by Whitaker  at 23° latitude and 15°west longitude. In addition, according to Feldman et al.  the existence of a Procellarum basin is consistent with a sharp decrease in intensities of FeO, thorium, and epi/thermal and fast neutrons at ∼50° from the center which is close to the boundary of Whitaker's  Procellarum basin. In summary, Feldman et al.  concluded that a giant impact, i.e., the Procellarum impact event, is a good candidate mechanism for producing the observed asymmetries in crustal thickness, center of mass/center of figure offset, mare basalt distribution and surface mineralogy. Haskin  argued that a giant Procellarum impact could have affected the development of the Th-rich Procellarum KREEP Terrane (PKT) by removal of overlying crust or deep injection of heat. On the basis of ejecta deposit modeling he found that the global distribution of high-Th material is consistent with the distribution expected for Imbrium ejecta deposits. Haskin et al.  argued that the Imbrium event did not form the high-Th Procellarum KREEP Terrane, but rather distributed the material globally.
 We dated 60 spectrally distinctive basalt units in Oceanus Procellarum (Figure 7). The outlines of our spectral units are generally in good agreement with the spectral units of Pieters  but our map also reveals additional detail as it subdivides some of the Pieters classes. Pieters  used the spectral information of the UV/VIS-ratio, the albedo, and the depths of the 1-μm and 2-μm absorption bands in order to establish a system of lunar basalt types (Table 1). Several units (P3, P25, P27, P29, P33, P34, P36, P37, P42, P54) were classified as “undivided” by Pieters  and 11 additional units (P11, P14, P15, P16, P20, P21, P44, P48, P50, P51, P60) were at least partly mapped as “undivided”. From these units, P11 and P50 were also characterized as hDSP basalt and units P14, P15, P16, P20, and P44 were also classified as mISP basalts. Unit P21 shows characteristics of “undivided”, mISP, and LBG- basalts, and unit P48 is mapped as “undivided” and LBG- basalts. Finally “undivided” and hDSA basalts are present in unit P51, and also in unit P60, which has an additional component of LBG- basalts. Two units (P1, P28) exhibit hDWA characteristics, five units (P32, P47, P52, P57, P58) are classified as hDSA basalts, and four units (P17, P38, P45, P46) are mapped as hDSP basalts. Units P6, P35, P43, and P49 are described as HDSA basalts in the map of Pieters , nine units (P2, P12, P13, P18, P19, P22, P23, P30, P41) are mapped as mISP basalts, four units (P4, P10, P31, P40) are shown as LBG- basalts, and units P5 and P7 are LBSP basalts. The spectral map of lunar basalt types [Pieters, 1978] indicates that unit P8 consists of mIG-, and unit P56 of LISP basalts. According to this map, several units can be characterized by two or more spectral basalt types. In unit P9 mISP and LBG- basalts are exposed, in unit P24 we found hDSA, hDSP, and mISP basalts, and in unit P26, P39, and P53 HDSA and hDSA basalts occur. Finally, units P55 and P59 are described as LBG- and hDSA basalts. A comparison of our units with the spectral units of Pieters  and several geologic maps is provided by Table 2.
 The geologic map of the nearside of the Moon [Wilhelms and McCauley, 1971] covers all basalts in Oceanus Procellarum, except units P1, P21, P22, P28, P29, P37, P55, and P57. In this map several units are mapped as Imbrian (Im) mare material (P2, P3, P4, P5, P6, P7, P10, P13, P16, P26, P30, P31, P32, P33, P36, P40, P41, P44, P46, P47, P48, P59, P60), and several units are mapped as darker Eratosthenian (Em) mare material (P8, P12, P17, P18, P23, P34, P42, P54). In this map 21 units are mapped as Im and Em basalts (P9, P11, P14, P15, P19, P20, P24, P25, P27, P35, P38, P39, P43, P45, P49, P50, P51, P52, P53, P56, P58). The geologic map of Scott et al.  covers 29 basalt units of western Oceanus Procellarum (P3, P9, P10, P12, P14, P15, P16, P18, P19, P20, P21, P22, P23, P25, P26, P29, P30, P32, P33, P34, P36, P37, P39, P44, P49, P51, P53, P58, P60) and indicates that all basalts exposed in this region are of Imbrian and/or Eratosthenian age (EIm), consistent with the map of Wilhelms and McCauley . Scott et al.  interpret albedo and color differences of these basalts to be related to differences in age and composition. Northern units in Oceanus Procellarum are covered by the geologic map of Lucchitta  and this maps shows that units P1, P10, P13, and P28 are of Imbrian age (Im) and that units P6 and P56 are of Eratosthenian age (Em). In this map, Eratosthenian basalts are characterized as dark, smooth, flat surfaces with bluish color and a lower albedo than Imbrian basalts. In the geologic map of McCauley  units P3, P15, P24, P33, P36, and P44 are classified as extensive featureless Imbrian mare plains material (Im) with albedos of 0.08–0.10. Units P18, P23, and P25 consist of Imbrian mare material (Im) and dark Eratosthenian mare material (Em) and units P34 and P54 are mapped as Eratosthenian in age (Em). According to McCauley , the albedo of Eratosthenian basalts is <0.08, and on the basis of Earth-based full-Moon images their boundaries are difficult to delineate. An Eratosthenian age (Em) was also found for unit P42 in the map of Wilshire . Wilshire  described this geologic unit Em as smooth, level plains with low albedo (0.08–0.09) and interpreted it as basaltic lava flows formed by eruption from fissures.
 Our spectrally defined units (P12, P15, P16, P18, P19, P20, P24, P44) in the Hevelius quadrangle are mostly Imbrian mare materials of different albedo (Ipm, Ipmd) [McCauley, 1967]. The albedo of the geologic unit Ipmd is <0.06 and is lower than that of Ipm. Parts of units P12, P18, and P19 also are mapped as Copernican Cavalerius Formation (Cca). The Reiner Gamma Formation which is exposed within unit P16 was mapped as Copernican in age (Cre) and has an intermediate albedo of 0.08–0.09 [McCauley, 1967]. Ulrich  mapped the J. Herschel quadrangle of the Moon and in this map unit P56 consists of Eratosthenian/Upper Imbrian mare material with very low albedo (EIm). Units P9, P10, P13, P31, P39, P40, P53, P58, and P59 are covered by the geologic map of Scott and Eggleton , which indicates an Imbrian (Im) age for these mare units. Parts of several units (P9, P39, P53, P58) also consist of Eratosthenian mare material (Em). Compared to Imbrian mare material (Im), Eratosthenian mare material (Em) is darker (<0.08–0.085) and bluer. The geologic map of Moore  which is based on telescopic observations also shows characteristics of Imbrian mare material (Ipm) for numerous basalt units (P4, P5, P7, P40, P43, P48, P51, P59, P60). Similarly, units P30, P39, P59 and P60 are shown as Imbrian in age and a range of albedos in the map of Moore . In this map most of the units (P14, P22, P26, P32, P49, P53) consist of Imbrian mare material (Ipm) and dark Imbrian mare material (Ipmd). Regions south and east of Lichtenberg (P53) are mapped as low albedo Copernican mare material (Cmd). Titley  mapped unit P17 as Imbrian (Ipm3, Ipm4) in age, with Ipm4 being the very darkest member and Ipm3 being the dark member of the Procellarum Group. In an older map by Marshall  unit P17 was mapped as Pm, hence being part of the so-called Procellarian System. Several other units (P8, P11, P27, P35, P38, P45, P46, P47, P54) are also attributed to the Procellarian System. This stratigraphic system dates back to early work of Shoemaker and Hackman  who defined the Imbrian System as equivalent to the immense sheet of material around the Imbrium basin and the Procellarian System to consist of mare material that is younger than the Imbrium ejecta sheet. The same stratigraphic system is also used in the map of Hackman , which shows that all covered basalts (i.e., units P2, P5, P6, P35, P41, P47, P50, P52) were erupted during the Procellarian System. However, for a variety of reasons this stratigraphic system was quickly abandoned [Wilhelms, 1987] and was not used in later maps. As the 1:1,000,000 USGS geologic maps of the Moon have never been updated since the end of the initial mapping program, the available maps of significant parts of the lunar nearside, i.e., the Kepler and the Letronne quadrangles of the Moon show stratigraphic systems that are long out of date.
 On the basis of our crater size-frequency distributions we conclude that basalt model ages in Oceanus Procellarum range from ∼1.2 to ∼3.93 b.y. (Figure 7; Table 3). Sixteen units, that is ∼25% of all dated units, show characteristic kinks in their crater size-frequency distributions which have been interpreted by Neukum and Horn  to indicate resurfacing, that is flooding with subsequent lavas. These kinks in the crater size-frequency distribution form when a new flow unit preferentially covers smaller craters of an older surface while larger, uncovered craters of the older surface are still detectable after the flow unit has been emplaced [Neukum and Horn, 1976]. It has been shown that the diameters at which these kinks occur can be used to estimate the thickness of the resurfacing flow unit [e.g., Neukum and Horn, 1976; Hiesinger et al., 2002]. On the basis of their study of 58 units in Oceanus Procellarum, Cognitum, Nubium, Insularum, Imbrium, Tranquillitatis, and Humorum, Hiesinger et al.  found that the average thickness of late-stage flows in these basins is ∼30–60 m.
Table 3. Comparison of Ages for Basalts in Oceanus Procellaruma
 In Oceanus Procellarum we dated 60 units. Using the chronostratigraphic system of Neukum and Ivanov , basalts of 19 units are Imbrian in age. For 3 of these 19 units, we were able to detect late-stage flooding events during the Imbrian Period. Thirty-six units in Oceanus Procellarum are Eratosthenian in age, with 13 of them also showing older Imbrian ages. Finally, if we apply the chronostratigraphic system of Neukum and Ivanov , 5 basalt units are of Copernican age. As discussed earlier, these basalts would be of Eratosthenian age if one uses the chronostratigraphic system of Wilhelms  or Stöffler and Ryder  (Figure 2).
 Earlier attempts to measure the ages of surface units relied on crater degradation processes and rates [e.g., Boyce, 1976; Boyce and Johnson, 1978]. Crater degradation ages of Boyce  and Boyce and Johnson  were performed for 1/4° squares (∼8 km), were interpolated by spatial filtering into a continuous image, and do not necessarily fit lithological or spectral units (Figure 5). These estimates were very useful when detailed spectral units were not available. The new data, however, use specifically defined units and do not require spatial filtering. Units P1, P18, P20, P21, P22, P23, P29, P30, P31, P32, P36, P38, P41, P44, P48, P50, and P59 exhibit a single degradation age; all other units show at least two, and some units (P10, P13) up to four different ages in the map of Boyce and Johnson . The implication is that ages derived from the map of Boyce and Johnson  can vary up to 1.25 b.y. for a single spectral unit. In numerous cases (P2, P3, P4, P8, P9, P11, P13, P14, P17, P24, P26, P27, P34, P35) the ages of Boyce  and Boyce and Johnson  give us an upper and lower boundary with our ages either right between or close to one or the other boundary (Table 3). In other cases (P1, P5, P7, P10, P15, P16, P19, P33), the most abundant age of Boyce  and Boyce and Johnson  is similar to our age of a particular unit. Generally we find a good agreement of our ages with ages of Boyce  and Boyce and Johnson  for units of Imbrian age. Only units P6 and P12 appear to be older than in the maps of Boyce  and Boyce and Johnson . Ages of units that are young according to our data, are systematically overestimated in age (older) in the Boyce map (P25, P28, P39, P43, P46, P47, P49, P51, P55, P56, P57, P58, P60), and we find less agreement for younger units, i.e., late Eratosthenian and Copernican units. Units P40, P45, P52, and P53 show evidence for resurfacing with both ages bracketing or being similar to the Boyce ages. Finally, units P37, P42, and P54 are not covered in the Boyce map.
 Several units for which we determined an Eratosthenian age (P23, P34, P42, P54) are also mapped as Eratosthenian in age in the map of Wilhelms and McCauley . Numerous units exhibit Eratosthenian and Imbrian ages in this map, and with our crater size-frequency distribution measurements we obtained Eratosthenian ages for all these units (P20, P24, P25, P27, P35, P38, P39, P43, P45, P49, P50, P51, P52, P53). Wilhelms and McCauley  mapped 12 units as Imbrian in age, but our data indicate an Eratosthenian age for these units (P26, P30, P31, P32, P33, P36, P40, P41, P44, P46, P47, P48). The geologic map of Scott et al.  attributes Eratosthenian and/or Imbrian ages to units that we found to be of Eratosthenian age (P20, P21, P22, P23, P25, P26, P29, P30, P32, P33, P34, P36, P37, P39, P44, P49, P51, P53). Several other geologic maps show Imbrian ages for the units that, according to our crater counts, are Eratosthenian in age, i.e., units P20–P55. For example, Moore  found Imbrian ages for units P40, P43, P48, and P51 and in the map of Moore  6 units (P22, P26, P30, P32, P39, P49) show Imbrian ages and one unit (P53) is partially Imbrian and Copernican in age. Imbrian ages are also indicated for units P20, P24, and P44 in the map of McCauley , and for unit P28 [Lucchitta, 1978], as well as for units P31, P39, P40, and P53 [Scott and Eggleton, 1973]. However, significant parts of units P39 and P53 were also mapped as Eratosthenian in age, consistent with our dating. Finally, in the map of McCauley , units P24, P33, P36, and P44 are Imbrian and units P23 and P25 are Imbrian and Eratosthenian in age. Our crater counts confirm an Eratosthenian age for unit P34 and unit P54. The map of Wilshire  indicates an Eratosthenian age for unit P42, consistent with our age for this unit. A summary of this discussion of our units in the context of the geological maps of the U. S. Geological Survey and the Boyce ages is given in Table 3.
 We defined and dated 17 spectrally distinctive units within Mare Nubium (Figure 8). We found the outlines of our spectral units to correlate well with previously spectrally defined lunar basalt types [Pieters, 1978]. However, 35% of all units within Mare Nubium (N2, N3, N4, N5, N9, N12) were mapped as “undivided” in the map of Pieters  and three more units (N11, N13, N14) were at least partially mapped as “undivided” basalts. Two of these units (N11, N13) exhibit additional characteristics of mIG- and hDG- basalts, and unit N14 partially consists of mIG- basalts. The most frequently occurring basalt type in Mare Nubium is a mIG- basalt, which is exposed in eight units (N6, N7, N10, N11, N13, N15, N16, N17). Finally, we identified LBG- basalts within unit N1 and LBSP and LBG- basalts within unit N8 (Table 2).
 This region of the Moon was geologically mapped by Wilhelms and McCauley , Holt , Howard and Masursky , Trask and Titley , and Eggleton  (Table 4). In the geologic map of Wilhelms and McCauley , units N1, N8, N11, and N12 consist of Imbrian mare materials (Im). The majority of units (N3, N4, N7, N10, N13, N15, N16, N17) is mapped as Imbrian and Eratosthenian mare materials (Im, Em), and several units (N2, N5, N6, N9, N14) are Eratosthenian mare materials (Em). Imbrian ages for unit N11 (Ipm, Ipmd), N12 (Ipm), and N16 (Ipm, Ipmd) can be derived from the geologic map of Howard and Masursky . In this map, the geologic units Ipm and Ipmd are part of the Procellarum Group with Ipmd having a lower albedo than Ipm. Eggleton  mapped units N13, N14, and N17 as Imbrian mare materials (Ipm). In the geologic map of Trask and Titley  mare material is classified according to its albedo in five classes (Ipm = undifferentiated, Ipm1 = light, Ipm2 = intermediate, Ipm3 = dark, Ipm4 = very dark). All these volcanic materials form the Imbrian age Procellarum Group which is not to be confused with the Procellarian System; the first one being a rock-stratigraphic unit, the later one being a time-stratigraphic unit. Units N1 and N17 are mapped as Ipm, Ipm1 and Ipm2, unit N2 is mapped as Ipm2 and Ipm4, and units N4 and N5 are mapped as Ipm. Unit N6 is classified as Ipm and Ipm3, unit N7 is intermediate in albedo (Ipm2), and units N8 and N15 are characterized by a wide range in albedo (Ipm1, Ipm2, Ipm3). Finally, unit N10 consists of Ipm and Ipm2 material. Holt  mapped the Purbach quadrangle of the Moon. According to this map, unit N3 consists of Eratosthenian (Em) and Imbrian (Im1, Im2) mare basalts with distinctive albedo and an Imbrian smooth plains unit (Ip). Unit N9 is a dark Imbrian basalt (Im2) and unit N11 is an Eratosthenian basalt (Em). Unit N13 is characterized by a variety of basalt types (Im1, Im2, Em), similar to unit N16 (Im2, Em).
Table 4. Comparison of Ages for Basalts in Mare Nubiuma
 On the basis of our new crater size-frequency distribution measurements we identified 14 Imbrian and 3 Eratosthenian units in Mare Nubium (Figure 8; Table 4). Compared to the crater degradation ages of Boyce  and Boyce and Johnson  we find a good agreement for several units (Table 4). We measured an Imbrian age for units N1–N14, and these ages are generally similar to the degradation ages. However, there are some discrepancies in the details. According to the map of Boyce  and Boyce and Johnson , unit N2 is 2.5 ± 0.5 b.y. old, but our data indicate that this unit is significantly older, i.e., 3.63/3.85 b.y. Similarly, very young ages (2.5 b.y.) have been shown for parts of units N6, N10, and N11. In all these cases our ages are older than the degradation ages. Very often the most abundant degradation age or at least one of the degradation ages that can be assigned to a particular unit is similar to our age for this unit (N1, N4, N5, N6, N7, N8, N10, N11, N12, N13, N16, N17). For unit N14 we derived an age of 3.76 b.y., with a resurfacing event taking place at 3.25 b.y. Crater degradation ages of Boyce  and Boyce and Johnson  do not resolve this resurfacing event and indicate ages of 3.75, 3.65, and 3.85 b.y., with 3.75 b.y. being the most abundant age of this unit. Unit N9 is significantly older in the Boyce data (3.85 b.y.) compared to our data (3.48 b.y.), as is unit N15 (3.50 b.y.) compared to our age of 3.16 b.y. Unit N3 is considerably younger (3.2 b.y.) than our age (3.63 b.y.). Eratosthenian ages of unit N16 and probably N17 in the maps of Boyce  and Boyce and Johnson  are consistent with our findings.
 The geologic map of the nearside of the Moon indicates that units N1, N8, N11, and N12 are Imbrian in age, consistent with our ages [Wilhelms and McCauley, 1971]. Several units (N3, N4, N7, N10, N13) were mapped as Imbrian and Eratosthenian in age, but we derived Imbrian ages on the basis of our crater counts. Eratosthenian ages for units N2, N5, N6, and N9 are not consistent with our data. Unit N14 is shown as Eratosthenian in the geologic map and we obtained an age of 3.25 b.y., very close to the border of the Eratosthenian and Imbrian System. Finally, units N15, N16, and N17 were mapped as Imbrian and Eratosthenian in age [Wilhelms and McCauley, 1971; Holt, 1974] and our data confirm an Eratosthenian age of these units. In addition, our data are consistent with Imbrian ages of units N11 and N12 [Howard and Masursky, 1968], N1, N2, N4, N5, N6, N7, N8, and N10 [Trask and Titley, 1966], N13 and N14 [Eggleton, 1965], and N3, N9, and N13 [Holt, 1974]. However, we cannot confirm Eratosthenian ages of units N3, N11, and N13 as shown in the map of Holt . And our data are not consistent with Imbrian ages of units N15, N16, N17 as indicated by various maps [Eggleton, 1965; Trask and Titley, 1966; Howard and Masursky, 1968; Holt, 1974]. On the basis of our crater counts we see that these units are Eratosthenian in age. A summary of our new data and a comparison between geological maps and the Boyce ages is shown in Table 4.
3.3. Mare Cognitum
3.3.1. Geologic Setting
 Mare Cognitum is located northwest of the Nubium basin (Figure 1). The rim of the pre-Nectarian small basin or large crater that contains Mare Cognitum has been mostly destroyed or obliterated since its formation by subsequent processes. It was interpreted to be superposed, hence to be younger than the postulated Procellarum basin [Wilhelms, 1987], but the existence of the Procellarum basin is not universally accepted [e.g., Spudis, 1993]. Thicknesses of basalts within Mare Cognitum are <500 m [DeHon, 1979; DeHon and Waskom, 1976], and spectral data of Pieters  indicate a rather homogeneous composition of these basalts.
3.3.2. Discussion of Units
 Compared to the spectral map of Pieters  we see that 80% of our basalt units (C1, C3, C4, C5) consist of mIG- basalts. Unit C2 was mapped as mIG- and “undivided” basalts (Table 2).
 Mare Cognitum is covered by the geologic maps of Wilhelms and McCauley  and Eggleton . We dated five spectral units in Mare Cognitum, of which Wilhelms and McCauley  mapped unit C1, C2, and C5 as Imbrian (Im) and Eratosthenian (Em) mare materials. Units C3 and C4 were mapped as Imbrian (Im). The map of Eggleton  does not differentiate these units. In this map all basalts are Imbrian mare materials (Ipm) of low reflectivity (Table 5).
Table 5. Comparison of Ages for Basalts in Mare Cognituma
 On the basis of our crater counts we see that all units in Mare Cognitum are of Imbrian age (3.32–3.49 b.y.; Figure 8; Table 5). Boyce  and Boyce and Johnson  found similar ages (3.2–3.5 b.y.) and there is an excellent agreement between the two data sets for these units. Only unit C2 appears a little younger in the Boyce  data (3.2 ± 0.2 b.y.) compared to our age (3.45 b.y.). For unit C5 we found evidence for a resurfacing event (3.32/3.65 b.y.) but this is not resolved in the Boyce data (3.2 ± 0.2 b.y.).
 Imbrian ages for the units in Mare Cognitum are consistent with the mapping of Eggleton  and Wilhelms and McCauley . However, Wilhelms and McCauley  also mapped units C1, C2, and C5 to consist, at least partially, of Eratosthenian basalts. These Eratosthenian ages are not consistent with our ages, but are within the error bars of the degradation ages (3.2 ± 0.2 b.y.).
3.4. Mare Insularum
3.4.1. Geologic Setting
 The geologic map of Wilhelms and McCauley  shows two partially preserved ring structures that outline the Insularum basin. Insularum is a pre-Nectarian impact basin with rings of 600 km and 1000 km in diameter, and is centered at 9°N and 18°W [Spudis, 1993] (Figure 1). Previous workers [e.g., Whitford-Stark, 1981] have shown the important influence of the Insularum basin on the topography and morphology of the younger Imbrium basin.
 Spectrally, large parts of Mare Insularum are influenced by superposed ejecta of crater Copernicus and Eratosthenes. In the map of spectrally defined lunar basalt types, Mare Insularum consists mainly of two basalt types [Pieters, 1978] and new lunar Prospector data indicate high concentrations of thorium and samarium in the southwestern regions of this lunar mare [Lawrence et al., 2000; Elphic et al., 2000].
3.4.2. Discussion of Units
 We identified only four basalt units within Mare Insularum. Two of these units (IN2, IN4) exhibit characteristics of mIG- basalts in the spectral map of Pieters , one unit (IN3) was mapped as “undivided”, and one unit (IN1) is covered with dark mantling material (Table 2).
 The geologic map of Wilhelms and McCauley  shows that most basalts are Imbrian mare materials (Im) with unit IN2 also showing characteristics of Eratosthenian mare materials (Em) (Table 6). Hackman  mapped units IN2 and IN3 as Pm, and Eggleton  described unit IN2 as Imbrian mare material (Ipm) with low reflectivity and small local contrast. This is consistent with the map of Schmitt et al.  in which all basalts in Mare Insularum are of Imbrian mare material (Ipm) with an albedo of 0.086–0.102.
Table 6. Comparison of Ages for Basalts in Mare Insularuma
 On the basis of the new age data, Figures 7 and 8 show the spatial distribution of basalt ages in the investigated mare regions in Oceanus Procellarum, Mare Nubium, Mare Cognitum, and Mare Insularum. Figure 9 shows the temporal distribution of basalt ages found in the investigated areas.
 In these regions the largest number of basalt units per time bin were formed in the Late Imbrian Period at ∼3.3–3.5 b.y. The general distribution can be described as asymmetrical with the peak toward the older ages. Beginning at about 4 b.y., the frequency rapidly increases to a peak at ∼3.3–3.5 b.y. and then declines generally, but perhaps episodically, to ∼1.2 b.y. Our new crater size-frequency distribution data of the remapped basalt units indicate that the ages of basalts in Oceanus Procellarum and the other investigated regions range from ∼1.20 to ∼3.93 b.y., a total duration of ∼2.7 b.y. There are 22 units which show evidence for resurfacing by late-stage flows, 16 of which occur in Oceanus Procellarum. Ages in Mare Cognitum vary from ∼3.32 to ∼3.65 b.y. We dated 5 units in Mare Cognitum but only one unit showed evidence for late-stage flooding events. In Mare Nubium we dated 17 mare basalt units. Ages in Mare Nubium are generally similar to ages obtained for basalts in Mare Cognitum but show a wider range of ages of ∼2.77–3.85 b.y., for a total duration of ∼1.08 b.y. At least two basalt units in Mare Nubium show two clearly distinguishable ages, indicating that late-stage flooding affected these units. In Mare Insularum we performed crater counts for four units. We find ages from ∼2.93 to ∼3.54 b.y. and two units are influenced by late-stage lava flooding at ∼2.95 b.y. In summary, our new crater counts indicate that active mare volcanism in Oceanus Procellarum and adjacent regions ranges over a long period of time from about 1.20 b.y. to about 3.93 b.y., a total of ∼2.7 b.y.
3.6. Link to Ages of Mare Basalts in Other Basins on the Lunar Nearside
 In a previous paper we presented model ages for ∼139 basalt units in several lunar nearside impact basins [Hiesinger et al., 2000]. The basalts for which we performed crater counts were exposed in Mare Imbrium, Serenitatis, Tranquillitatis, Humorum, Humboldtianum, and Australe. In this paper we expanded our age determinations to the Oceanus Procellarum, Mare Cognitum, Mare Nubium, and Mare Insularum regions. In these areas we dated ∼86 basalt units with crater size-frequency distribution measurements. Figure 10 shows the distribution of model ages of ∼225 basalt units of the investigated regions. The data indicate that lunar volcanism in the investigated large nearside mare started at ∼4 b.y. ago and ended at ∼1.2 b.y. Most of the investigated basalts on the lunar nearside erupted during the Late Imbrian Period, fewer basalts erupted during the Eratosthenian Period, and even fewer basalts are of Copernican age (Figure 11).
 The spatial distribution of model ages indicates that younger (Eratosthenian/Copernican) basalts occur preferentially in the Oceanus Procellarum region and in the vicinity of volcanic centers such as the Aristarchus Plateau.
3.7. Flux of Lunar Mare Basalts
 Establishing the volume of mare basalts emplaced on the surface as a function of time (the flux) is important in order to place constraints on the petrogenesis of lunar mare basalts and their relation to the thermal evolution of the Moon [Head and Wilson, 1992; Zhong et al., 1999; Wieczorek and Phillips, 2000; Parmentier et al., 2000; Wilson and Head, 2001]. Head and Wilson  reviewed and summarized different approaches to determine the extent and flux of mare basalts and found that on the basis of returned lunar samples only one would obtain a more or less Gaussian flux curve with a peak flux prior to ∼3.5 b.y. (Figure 12a). Looking at the surface exposure of units, one would derived a similar flux curve, which is somewhat wider and shifted toward younger ages (∼3–3.5 b.y.) (Figure 12b). Investigation of the stratigraphy of mare basalt volumes would yield an asymmetric flux curve with the peak at older ages at ∼3.8 b.y. (Figure 12c). Figure 12d shows a flux curve that accounts for probable volcanism in the period of early impact bombardment which is now obscured. Finally, a flux curve based on the combination of approaches is shown in Figure 12e. On the basis of their investigation, Head and Wilson  argued that the flux in the last half of lunar history is characterized by episodic rather than continuous eruptions.
 In order to calculate the flux we need accurate information on (1) the age, (2) the surface extent and (3) the thickness of a basalt flow. Early work mostly focused on estimating the total thickness of mare basalt fill in the mare basins [e.g., DeHon, 1974, 1979; DeHon and Waskom, 1976; Hörz, 1978]. These studies used a variety of techniques, which are summarized elsewhere [e.g., Head, 1982; Budney and Lucey, 1998]. Results from crater geometry techniques using pristine crater morphometric relationships and the diameter of partially to almost wholly flooded craters showed that lunar impact basins are filled with up to 2 km of basalts, with 200–400 m on average [DeHon and Waskom, 1976]. Hörz  reviewed the assumptions that underlie the thickness estimates of DeHon and Waskom  and concluded that such values were overestimates. Similarly, Budney and Lucey  concluded that basalts in Mare Humorum are generally less thick than estimated by DeHon . Using craters that excavated highland material from beneath the mare basalts in Oceanus Procellarum, Heather and Dunkin  estimates that the basalt are 160–625 m thick, with thicknesses ranging from tens to hundreds of meters near the mare/highland boundaries and several hundreds of meters closer to the center of the mare. All these estimates, which are based on flooded impact craters or impact craters that penetrated the mare basalts and excavated underlying highland material, provide the total thickness of basalts in a particular basin, but not the thickness of individual basalt flow units. Head  pointed out that two stages of filling a basin can be distinguished: Stage 1 is characterized by flooding the interior and thick deposits (∼6 km) of small to intermediate volumes covering small areas; stage 2 is characterized by flooding outward to the basin-defining scarp and thin deposits (∼2 km) of large volume covering large areas.
 To place better constraints on the flux of lunar mare basalt volcanism we need to know the thicknesses of individual mare flow units. In the past extensive work has been done to estimate the thicknesses of individual flow units [e.g., Howard et al., 1972; Schaber, 1973; Brett, 1975; Schaber et al., 1976; Gifford and El-Baz, 1978, 1981]. Gifford and El-Baz [1978, 1981] noted that despite the wide acceptance of the idea of multiple flow units filling the basins, morphometric characteristics of individual flow units have not been extensively studied. Measurements of flow unit thicknesses are complicated by (1) the limited availability of suitable data necessary for the recognition of flow fronts, i.e., high-resolution topography and near-terminator images, (2) regolith formation processes, i.e., impact cratering, which can obliterate flow fronts of up to 15 m [Head, 1976], and (3) the composition and the eruption style of lunar lavas which are thought to be responsible for the sparseness of mare flow features [Schultz et al., 1976; Head, 1976].
 A variety of techniques has been applied to estimate the thicknesses of individual flow units. Such techniques included: (1) shadow measurements in high-resolution low-sun images [e.g., Schaber, 1973; Schaber et al., 1976; Gifford and El-Baz, 1978, 1981]; (2) in situ observations of flow units by the Apollo astronauts, e.g. within the walls of Hadley Rille at the Apollo 15 landing site [Howard et al., 1972], and (3) studies of the chemical kinetic aspects of lava emplacement and cooling [Brett, 1975].
 On the basis of Apollo 14 and 15 near-terminator images of the region southeast of the crater Kunowsky, Lloyd and Head  determined a flow front height of 3–5 m for a single flow unit. Reporting on a much larger number of units, Gifford and El-Baz  found flow heights of 1–96 m with an average thickness of ∼21 m. These results are similar to thicknesses of 10–20 m observed in the wall of Hadley Rille [Howard et al., 1972]. Gifford and El-Baz  argued that a few flow units might even have actual thicknesses in excess of 100 m because only the shadowed portions of the flow fronts were measured. Schaber  found that the average thickness of the Eratosthenian flows in Mare Imbrium is 30–35 m. However, they considered these Eratosthenian flows as atypical because their thickness (10–63 m) is much larger compared to other lunar flow fronts that could be identified in imaging data (5–10 m) [Schaber et al., 1976]. Finally, chemical kinetic considerations based on Apollo samples suggest that lunar lava flow units are no thicker than ∼8–10 m at the Apollo 11, 12 and 15 sites [Brett, 1975].
 A fourth technique was applied by Neukum and Horn . They showed that endogenic lava flow processes could be identified by their characteristic effects on crater size-frequency distributions even if these individual flows were not visible in the images. This is an important result because photogeologic and morphologic recognition of individual flow fronts on the Moon is difficult and is mostly restricted to thicker flows and areas where low-sun images or samples were obtained, as discussed above. On the basis of deflections (knees) in crater size-frequency distribution curves, Neukum and Horn  estimated the thickness of Imbrian-aged flow units in Mare Imbrium. They found that these flows are about 200 m thick, and reported that Eratosthenian flows are about 60 m thick. The later value is consistent with photogeologic estimates of the same flow unit by Schaber , Schaber et al. , and Gifford and El-Baz . Hiesinger et al.  applied this technique to a much larger number of their crater counts and found that the average thickness of 58 of their investigated individual basalt flow units is ∼30–60 m with a range of flow unit thickness of ∼20–200 m. A comparison with previously published thickness estimates [e.g., Howard et al., 1972; Schaber, 1973; Brett, 1975; Schaber et al., 1976; Gifford and El-Baz, 1978, 1981] showed that crater size-frequency distribution measurements yield thicknesses that are in excellent agreement with results from these other techniques and allow one to obtain thicknesses for additional flow units that have not been detected in low-sun images. On the basis of their thickness estimates, Hiesinger et al.  also reported that the volumes of individual basalt flows range from 30–7700 km3, with an average of 590–940 km3.
 For our flux estimates we will use ages and thicknesses derived from crater size-frequency distribution measurements and surface extents of mare basalt units measured in imaging data. We have already demonstrated that crater size-frequency distribution measurements are an adequate method in order to derive the surface age of the uppermost basalt unit with sufficient accuracy [Hiesinger et al., 2000]. We also demonstrated that crater size-frequency distribution measurements can be used to estimate the thickness of mare flow units [Hiesinger et al., 2002].
 Using this approach we have to consider potential caveats. First, because older units might have been covered by younger flows, the measured surface area of older basalt flow units might be too small. Consequently, volumes estimated for these flow units must be considered as minimum estimates. Second, we also have to consider variations of flow unit thicknesses with time, that is, older flow units might be thicker than younger flow units as indicated by the work of Head . Again, this implies that volumes estimates of early flow units are minimum estimates.
 As discussed above, craters which penetrate the entire stack of basalts have been used to estimate the thicknesses of basalts within individual basin structures [e.g., DeHon, 1979]. These studies indicate that basalt thicknesses are up to 2 km with 200–400 m on average. If we assume a thickness of 30–60 m for the uppermost basalt flow unit [e.g., Hiesinger et al., 2002], this would imply that at least ∼150 m or up to ∼1900 m of basalts are covered by the youngest flow units. Similarly, thickness estimates by Heather and Dunkin , would imply that at least ∼100–550 m of basalts are buried beneath the youngest flow units. Using the Orientale basin as an example, Head  argued that the basalt thickness in the central parts of lunar basins can be up to 8 km. From the comparison of this maximum total thickness of mare basalts in a basin with the thickness of the uppermost flow units, we conclude that a large number of flow units must have filled the lunar impact basins.
Figure 13 shows a cumulative plot of the flux of basalts in all investigated areas, that is, Imbrium, Serenitatis, Tranquillitatis, Humorum, Humboldtianum, Australe, Oceanus Procellarum, Cognitum, Nubium, and Insularum. We plotted the flux curves of four scenarios, each of which assumed a different average thickness of the flow units. The thicknesses were chosen to be within reasonable limits given by the discussion above, i.e., 10 m, 25 m, 50 m, and 100 m. For comparison we also plotted a flux curve for linearly increasing thicknesses with increasing age. In this model we assumed a thickness of 10 m for flows of 1.0 b.y. age, which increases to 100 m for flows of 4.0 b.y. age. If true that the thickness of early flow units was larger [Head, 1982] and that the surface exposure of older units is underestimated, the implication is that the flux curve should be steeper at older ages, and this is what we observe for our flux curve on the basis of increasing flow unit thicknesses. While the actual amount of increase in thickness is not known to date, the model illustrates several effects on the flux curve. First, even with a 10-fold linear increase in thickness with increasing age, the overall shape of the flux curve remains similar and phases of higher and lower volcanic activity are detectable. Second, by assuming a larger thickness of older flow units, the model indirectly offsets effects introduced by the fact that older units are covered to a larger extent with subsequent units than younger units. While the models provide a qualitative basis for discussions of the lunar volcanic flux, exact quantitative measurements are left for future studies.
 Our data indicate that the flux of mare basalts was highest in the Late Imbrian ∼3.3–3.7 b.y. ago. On the basis of our data we observe a decreased flux during the Eratosthenian and Copernican Period. The decrease in flux was not continuous but was interrupted by brief phases of relatively higher volcanic activity as indicated by the steeper segments of our flux curve (Figure 13). This is consistent with predictions of Head and Wilson , who argued for episodic eruptions during the later half of lunar history (Figure 12e). Flattening of the curve at ages >3.7 b.y. might be a combination of covering older units by younger units and/or a probably lower flux.
 On the basis of our new age determinations for basalts that are exposed in Oceanus Procellarum and adjacent mare regions we conclude that (1) not considering possible cryptomaria, volcanism in this region was active over a long period of time, starting at ∼3.93 b.y. and ending at ∼1.20 b.y., (2) the largest number of basalt units were formed in the Late Imbrian Period at ∼3.3–3.5 b.y., (3) the temporal distribution of erupted mare basalts is asymmetric with the peak toward the older ages, and (4) the youngest basalts were detected in the vicinity of volcanic centers such as the Aristarchus Plateau.
 On the basis of the combination of our new data with our previous age data of several lunar nearside maria we conclude that (1) the investigated basalts on the lunar nearside erupted over a long period of time of at least ∼2.8 b.y., between ∼4 and ∼1.2 b.y., (2) if one takes into account cryptomaria, which suggest that volcanism was already active prior to the formation of the large impact basins, the period of active volcanism on the Moon might be as long as >3 b.y., (3) of all dated regions in our study, Oceanus Procellarum is the location of the youngest basalts on the lunar nearside, (4) on the basis of the regions investigated in our study, Oceanus Procellarum also exhibits the widest range in ages within a single mare region, (5) according to our crater counts there are basalt units, for example south of the Aristarchus Plateau that might be younger than the Lichtenberg and Flamsteed basalts, which were previously thought to be among the youngest lunar basalts [Schultz and Spudis, 1989], (6) there are significant differences in the flux of lunar nearside basalts, (7) the flux of lunar basalts was largest during the Late Imbrian Period, especially between ∼3.3 and ∼3.7 b.y., and (8) the flux of basalts is significantly smaller during the Eratosthenian and Copernican Period.
 We are currently expanding our efforts to date lunar mare basalt surfaces with crater size-frequency distribution measurements to additional maria, for example Mare Frigoris and Mare Nectaris, and are investigating changes in basalt mineralogy as a function of time and changes in eruption conditions, styles and locations with time. Finally, we intend to combine our results and characterize mare basalt volcanism in space and time, which will result in a new detailed stratigraphy of lunar mare basalt.
 The authors would like to thank Anne Côté and Peter Neivert for their help with the manuscript preparation. Thanks are extended to the NASA Planetary Geology and Geophysics Program, which provided support for this analysis.